Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * xml.c
4 : * XML data type support.
5 : *
6 : *
7 : * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group
8 : * Portions Copyright (c) 1994, Regents of the University of California
9 : *
10 : * src/backend/utils/adt/xml.c
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 :
15 : /*
16 : * Generally, XML type support is only available when libxml use was
17 : * configured during the build. But even if that is not done, the
18 : * type and all the functions are available, but most of them will
19 : * fail. For one thing, this avoids having to manage variant catalog
20 : * installations. But it also has nice effects such as that you can
21 : * dump a database containing XML type data even if the server is not
22 : * linked with libxml. Thus, make sure xml_out() works even if nothing
23 : * else does.
24 : */
25 :
26 : /*
27 : * Notes on memory management:
28 : *
29 : * Sometimes libxml allocates global structures in the hope that it can reuse
30 : * them later on. This makes it impractical to change the xmlMemSetup
31 : * functions on-the-fly; that is likely to lead to trying to pfree() chunks
32 : * allocated with malloc() or vice versa. Since libxml might be used by
33 : * loadable modules, eg libperl, our only safe choices are to change the
34 : * functions at postmaster/backend launch or not at all. Since we'd rather
35 : * not activate libxml in sessions that might never use it, the latter choice
36 : * is the preferred one. However, for debugging purposes it can be awfully
37 : * handy to constrain libxml's allocations to be done in a specific palloc
38 : * context, where they're easy to track. Therefore there is code here that
39 : * can be enabled in debug builds to redirect libxml's allocations into a
40 : * special context LibxmlContext. It's not recommended to turn this on in
41 : * a production build because of the possibility of bad interactions with
42 : * external modules.
43 : */
44 : /* #define USE_LIBXMLCONTEXT */
45 :
46 : #include "postgres.h"
47 :
48 : #ifdef USE_LIBXML
49 : #include <libxml/chvalid.h>
50 : #include <libxml/entities.h>
51 : #include <libxml/parser.h>
52 : #include <libxml/parserInternals.h>
53 : #include <libxml/tree.h>
54 : #include <libxml/uri.h>
55 : #include <libxml/xmlerror.h>
56 : #include <libxml/xmlsave.h>
57 : #include <libxml/xmlversion.h>
58 : #include <libxml/xmlwriter.h>
59 : #include <libxml/xpath.h>
60 : #include <libxml/xpathInternals.h>
61 :
62 : /*
63 : * We used to check for xmlStructuredErrorContext via a configure test; but
64 : * that doesn't work on Windows, so instead use this grottier method of
65 : * testing the library version number.
66 : */
67 : #if LIBXML_VERSION >= 20704
68 : #define HAVE_XMLSTRUCTUREDERRORCONTEXT 1
69 : #endif
70 :
71 : /*
72 : * libxml2 2.12 decided to insert "const" into the error handler API.
73 : */
74 : #if LIBXML_VERSION >= 21200
75 : #define PgXmlErrorPtr const xmlError *
76 : #else
77 : #define PgXmlErrorPtr xmlErrorPtr
78 : #endif
79 :
80 : #endif /* USE_LIBXML */
81 :
82 : #include "access/htup_details.h"
83 : #include "access/table.h"
84 : #include "catalog/namespace.h"
85 : #include "catalog/pg_class.h"
86 : #include "catalog/pg_type.h"
87 : #include "commands/dbcommands.h"
88 : #include "executor/spi.h"
89 : #include "executor/tablefunc.h"
90 : #include "fmgr.h"
91 : #include "lib/stringinfo.h"
92 : #include "libpq/pqformat.h"
93 : #include "mb/pg_wchar.h"
94 : #include "miscadmin.h"
95 : #include "nodes/execnodes.h"
96 : #include "nodes/miscnodes.h"
97 : #include "nodes/nodeFuncs.h"
98 : #include "utils/array.h"
99 : #include "utils/builtins.h"
100 : #include "utils/date.h"
101 : #include "utils/datetime.h"
102 : #include "utils/lsyscache.h"
103 : #include "utils/rel.h"
104 : #include "utils/syscache.h"
105 : #include "utils/xml.h"
106 :
107 :
108 : /* GUC variables */
109 : int xmlbinary = XMLBINARY_BASE64;
110 : int xmloption = XMLOPTION_CONTENT;
111 :
112 : #ifdef USE_LIBXML
113 :
114 : /* random number to identify PgXmlErrorContext */
115 : #define ERRCXT_MAGIC 68275028
116 :
117 : struct PgXmlErrorContext
118 : {
119 : int magic;
120 : /* strictness argument passed to pg_xml_init */
121 : PgXmlStrictness strictness;
122 : /* current error status and accumulated message, if any */
123 : bool err_occurred;
124 : StringInfoData err_buf;
125 : /* previous libxml error handling state (saved by pg_xml_init) */
126 : xmlStructuredErrorFunc saved_errfunc;
127 : void *saved_errcxt;
128 : /* previous libxml entity handler (saved by pg_xml_init) */
129 : xmlExternalEntityLoader saved_entityfunc;
130 : };
131 :
132 : static xmlParserInputPtr xmlPgEntityLoader(const char *URL, const char *ID,
133 : xmlParserCtxtPtr ctxt);
134 : static void xml_errsave(Node *escontext, PgXmlErrorContext *errcxt,
135 : int sqlcode, const char *msg);
136 : static void xml_errorHandler(void *data, PgXmlErrorPtr error);
137 : static int errdetail_for_xml_code(int code);
138 : static void chopStringInfoNewlines(StringInfo str);
139 : static void appendStringInfoLineSeparator(StringInfo str);
140 :
141 : #ifdef USE_LIBXMLCONTEXT
142 :
143 : static MemoryContext LibxmlContext = NULL;
144 :
145 : static void xml_memory_init(void);
146 : static void *xml_palloc(size_t size);
147 : static void *xml_repalloc(void *ptr, size_t size);
148 : static void xml_pfree(void *ptr);
149 : static char *xml_pstrdup(const char *string);
150 : #endif /* USE_LIBXMLCONTEXT */
151 :
152 : static xmlChar *xml_text2xmlChar(text *in);
153 : static int parse_xml_decl(const xmlChar *str, size_t *lenp,
154 : xmlChar **version, xmlChar **encoding, int *standalone);
155 : static bool print_xml_decl(StringInfo buf, const xmlChar *version,
156 : pg_enc encoding, int standalone);
157 : static bool xml_doctype_in_content(const xmlChar *str);
158 : static xmlDocPtr xml_parse(text *data, XmlOptionType xmloption_arg,
159 : bool preserve_whitespace, int encoding,
160 : XmlOptionType *parsed_xmloptiontype,
161 : xmlNodePtr *parsed_nodes,
162 : Node *escontext);
163 : static text *xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt);
164 : static int xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
165 : ArrayBuildState *astate,
166 : PgXmlErrorContext *xmlerrcxt);
167 : static xmlChar *pg_xmlCharStrndup(const char *str, size_t len);
168 : #endif /* USE_LIBXML */
169 :
170 : static void xmldata_root_element_start(StringInfo result, const char *eltname,
171 : const char *xmlschema, const char *targetns,
172 : bool top_level);
173 : static void xmldata_root_element_end(StringInfo result, const char *eltname);
174 : static StringInfo query_to_xml_internal(const char *query, char *tablename,
175 : const char *xmlschema, bool nulls, bool tableforest,
176 : const char *targetns, bool top_level);
177 : static const char *map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid,
178 : bool nulls, bool tableforest, const char *targetns);
179 : static const char *map_sql_schema_to_xmlschema_types(Oid nspid,
180 : List *relid_list, bool nulls,
181 : bool tableforest, const char *targetns);
182 : static const char *map_sql_catalog_to_xmlschema_types(List *nspid_list,
183 : bool nulls, bool tableforest,
184 : const char *targetns);
185 : static const char *map_sql_type_to_xml_name(Oid typeoid, int typmod);
186 : static const char *map_sql_typecoll_to_xmlschema_types(List *tupdesc_list);
187 : static const char *map_sql_type_to_xmlschema_type(Oid typeoid, int typmod);
188 : static void SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result,
189 : char *tablename, bool nulls, bool tableforest,
190 : const char *targetns, bool top_level);
191 :
192 : /* XMLTABLE support */
193 : #ifdef USE_LIBXML
194 : /* random number to identify XmlTableContext */
195 : #define XMLTABLE_CONTEXT_MAGIC 46922182
196 : typedef struct XmlTableBuilderData
197 : {
198 : int magic;
199 : int natts;
200 : long int row_count;
201 : PgXmlErrorContext *xmlerrcxt;
202 : xmlParserCtxtPtr ctxt;
203 : xmlDocPtr doc;
204 : xmlXPathContextPtr xpathcxt;
205 : xmlXPathCompExprPtr xpathcomp;
206 : xmlXPathObjectPtr xpathobj;
207 : xmlXPathCompExprPtr *xpathscomp;
208 : } XmlTableBuilderData;
209 : #endif
210 :
211 : static void XmlTableInitOpaque(struct TableFuncScanState *state, int natts);
212 : static void XmlTableSetDocument(struct TableFuncScanState *state, Datum value);
213 : static void XmlTableSetNamespace(struct TableFuncScanState *state, const char *name,
214 : const char *uri);
215 : static void XmlTableSetRowFilter(struct TableFuncScanState *state, const char *path);
216 : static void XmlTableSetColumnFilter(struct TableFuncScanState *state,
217 : const char *path, int colnum);
218 : static bool XmlTableFetchRow(struct TableFuncScanState *state);
219 : static Datum XmlTableGetValue(struct TableFuncScanState *state, int colnum,
220 : Oid typid, int32 typmod, bool *isnull);
221 : static void XmlTableDestroyOpaque(struct TableFuncScanState *state);
222 :
223 : const TableFuncRoutine XmlTableRoutine =
224 : {
225 : .InitOpaque = XmlTableInitOpaque,
226 : .SetDocument = XmlTableSetDocument,
227 : .SetNamespace = XmlTableSetNamespace,
228 : .SetRowFilter = XmlTableSetRowFilter,
229 : .SetColumnFilter = XmlTableSetColumnFilter,
230 : .FetchRow = XmlTableFetchRow,
231 : .GetValue = XmlTableGetValue,
232 : .DestroyOpaque = XmlTableDestroyOpaque
233 : };
234 :
235 : #define NO_XML_SUPPORT() \
236 : ereport(ERROR, \
237 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), \
238 : errmsg("unsupported XML feature"), \
239 : errdetail("This functionality requires the server to be built with libxml support.")))
240 :
241 :
242 : /* from SQL/XML:2008 section 4.9 */
243 : #define NAMESPACE_XSD "http://www.w3.org/2001/XMLSchema"
244 : #define NAMESPACE_XSI "http://www.w3.org/2001/XMLSchema-instance"
245 : #define NAMESPACE_SQLXML "http://standards.iso.org/iso/9075/2003/sqlxml"
246 :
247 :
248 : #ifdef USE_LIBXML
249 :
250 : static int
251 0 : xmlChar_to_encoding(const xmlChar *encoding_name)
252 : {
253 0 : int encoding = pg_char_to_encoding((const char *) encoding_name);
254 :
255 0 : if (encoding < 0)
256 0 : ereport(ERROR,
257 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
258 : errmsg("invalid encoding name \"%s\"",
259 : (const char *) encoding_name)));
260 0 : return encoding;
261 : }
262 : #endif
263 :
264 :
265 : /*
266 : * xml_in uses a plain C string to VARDATA conversion, so for the time being
267 : * we use the conversion function for the text datatype.
268 : *
269 : * This is only acceptable so long as xmltype and text use the same
270 : * representation.
271 : */
272 : Datum
273 900 : xml_in(PG_FUNCTION_ARGS)
274 : {
275 : #ifdef USE_LIBXML
276 900 : char *s = PG_GETARG_CSTRING(0);
277 : xmltype *vardata;
278 : xmlDocPtr doc;
279 :
280 : /* Build the result object. */
281 900 : vardata = (xmltype *) cstring_to_text(s);
282 :
283 : /*
284 : * Parse the data to check if it is well-formed XML data.
285 : *
286 : * Note: we don't need to worry about whether a soft error is detected.
287 : */
288 900 : doc = xml_parse(vardata, xmloption, true, GetDatabaseEncoding(),
289 900 : NULL, NULL, fcinfo->context);
290 852 : if (doc != NULL)
291 840 : xmlFreeDoc(doc);
292 :
293 852 : PG_RETURN_XML_P(vardata);
294 : #else
295 : NO_XML_SUPPORT();
296 : return 0;
297 : #endif
298 : }
299 :
300 :
301 : #define PG_XML_DEFAULT_VERSION "1.0"
302 :
303 :
304 : /*
305 : * xml_out_internal uses a plain VARDATA to C string conversion, so for the
306 : * time being we use the conversion function for the text datatype.
307 : *
308 : * This is only acceptable so long as xmltype and text use the same
309 : * representation.
310 : */
311 : static char *
312 23750 : xml_out_internal(xmltype *x, pg_enc target_encoding)
313 : {
314 23750 : char *str = text_to_cstring((text *) x);
315 :
316 : #ifdef USE_LIBXML
317 23750 : size_t len = strlen(str);
318 : xmlChar *version;
319 : int standalone;
320 : int res_code;
321 :
322 23750 : if ((res_code = parse_xml_decl((xmlChar *) str,
323 : &len, &version, NULL, &standalone)) == 0)
324 : {
325 : StringInfoData buf;
326 :
327 23750 : initStringInfo(&buf);
328 :
329 23750 : if (!print_xml_decl(&buf, version, target_encoding, standalone))
330 : {
331 : /*
332 : * If we are not going to produce an XML declaration, eat a single
333 : * newline in the original string to prevent empty first lines in
334 : * the output.
335 : */
336 23702 : if (*(str + len) == '\n')
337 6 : len += 1;
338 : }
339 23750 : appendStringInfoString(&buf, str + len);
340 :
341 23750 : pfree(str);
342 :
343 23750 : return buf.data;
344 : }
345 :
346 0 : ereport(WARNING,
347 : errcode(ERRCODE_DATA_CORRUPTED),
348 : errmsg_internal("could not parse XML declaration in stored value"),
349 : errdetail_for_xml_code(res_code));
350 : #endif
351 0 : return str;
352 : }
353 :
354 :
355 : Datum
356 23486 : xml_out(PG_FUNCTION_ARGS)
357 : {
358 23486 : xmltype *x = PG_GETARG_XML_P(0);
359 :
360 : /*
361 : * xml_out removes the encoding property in all cases. This is because we
362 : * cannot control from here whether the datum will be converted to a
363 : * different client encoding, so we'd do more harm than good by including
364 : * it.
365 : */
366 23486 : PG_RETURN_CSTRING(xml_out_internal(x, 0));
367 : }
368 :
369 :
370 : Datum
371 0 : xml_recv(PG_FUNCTION_ARGS)
372 : {
373 : #ifdef USE_LIBXML
374 0 : StringInfo buf = (StringInfo) PG_GETARG_POINTER(0);
375 : xmltype *result;
376 : char *str;
377 : char *newstr;
378 : int nbytes;
379 : xmlDocPtr doc;
380 0 : xmlChar *encodingStr = NULL;
381 : int encoding;
382 :
383 : /*
384 : * Read the data in raw format. We don't know yet what the encoding is, as
385 : * that information is embedded in the xml declaration; so we have to
386 : * parse that before converting to server encoding.
387 : */
388 0 : nbytes = buf->len - buf->cursor;
389 0 : str = (char *) pq_getmsgbytes(buf, nbytes);
390 :
391 : /*
392 : * We need a null-terminated string to pass to parse_xml_decl(). Rather
393 : * than make a separate copy, make the temporary result one byte bigger
394 : * than it needs to be.
395 : */
396 0 : result = palloc(nbytes + 1 + VARHDRSZ);
397 0 : SET_VARSIZE(result, nbytes + VARHDRSZ);
398 0 : memcpy(VARDATA(result), str, nbytes);
399 0 : str = VARDATA(result);
400 0 : str[nbytes] = '\0';
401 :
402 0 : parse_xml_decl((const xmlChar *) str, NULL, NULL, &encodingStr, NULL);
403 :
404 : /*
405 : * If encoding wasn't explicitly specified in the XML header, treat it as
406 : * UTF-8, as that's the default in XML. This is different from xml_in(),
407 : * where the input has to go through the normal client to server encoding
408 : * conversion.
409 : */
410 0 : encoding = encodingStr ? xmlChar_to_encoding(encodingStr) : PG_UTF8;
411 :
412 : /*
413 : * Parse the data to check if it is well-formed XML data. Assume that
414 : * xml_parse will throw ERROR if not.
415 : */
416 0 : doc = xml_parse(result, xmloption, true, encoding, NULL, NULL, NULL);
417 0 : xmlFreeDoc(doc);
418 :
419 : /* Now that we know what we're dealing with, convert to server encoding */
420 0 : newstr = pg_any_to_server(str, nbytes, encoding);
421 :
422 0 : if (newstr != str)
423 : {
424 0 : pfree(result);
425 0 : result = (xmltype *) cstring_to_text(newstr);
426 0 : pfree(newstr);
427 : }
428 :
429 0 : PG_RETURN_XML_P(result);
430 : #else
431 : NO_XML_SUPPORT();
432 : return 0;
433 : #endif
434 : }
435 :
436 :
437 : Datum
438 0 : xml_send(PG_FUNCTION_ARGS)
439 : {
440 0 : xmltype *x = PG_GETARG_XML_P(0);
441 : char *outval;
442 : StringInfoData buf;
443 :
444 : /*
445 : * xml_out_internal doesn't convert the encoding, it just prints the right
446 : * declaration. pq_sendtext will do the conversion.
447 : */
448 0 : outval = xml_out_internal(x, pg_get_client_encoding());
449 :
450 0 : pq_begintypsend(&buf);
451 0 : pq_sendtext(&buf, outval, strlen(outval));
452 0 : pfree(outval);
453 0 : PG_RETURN_BYTEA_P(pq_endtypsend(&buf));
454 : }
455 :
456 :
457 : #ifdef USE_LIBXML
458 : static void
459 132 : appendStringInfoText(StringInfo str, const text *t)
460 : {
461 132 : appendBinaryStringInfo(str, VARDATA_ANY(t), VARSIZE_ANY_EXHDR(t));
462 132 : }
463 : #endif
464 :
465 :
466 : static xmltype *
467 22566 : stringinfo_to_xmltype(StringInfo buf)
468 : {
469 22566 : return (xmltype *) cstring_to_text_with_len(buf->data, buf->len);
470 : }
471 :
472 :
473 : static xmltype *
474 78 : cstring_to_xmltype(const char *string)
475 : {
476 78 : return (xmltype *) cstring_to_text(string);
477 : }
478 :
479 :
480 : #ifdef USE_LIBXML
481 : static xmltype *
482 22664 : xmlBuffer_to_xmltype(xmlBufferPtr buf)
483 : {
484 22664 : return (xmltype *) cstring_to_text_with_len((const char *) xmlBufferContent(buf),
485 : xmlBufferLength(buf));
486 : }
487 : #endif
488 :
489 :
490 : Datum
491 42 : xmlcomment(PG_FUNCTION_ARGS)
492 : {
493 : #ifdef USE_LIBXML
494 42 : text *arg = PG_GETARG_TEXT_PP(0);
495 42 : char *argdata = VARDATA_ANY(arg);
496 42 : int len = VARSIZE_ANY_EXHDR(arg);
497 : StringInfoData buf;
498 : int i;
499 :
500 : /* check for "--" in string or "-" at the end */
501 180 : for (i = 1; i < len; i++)
502 : {
503 144 : if (argdata[i] == '-' && argdata[i - 1] == '-')
504 6 : ereport(ERROR,
505 : (errcode(ERRCODE_INVALID_XML_COMMENT),
506 : errmsg("invalid XML comment")));
507 : }
508 36 : if (len > 0 && argdata[len - 1] == '-')
509 6 : ereport(ERROR,
510 : (errcode(ERRCODE_INVALID_XML_COMMENT),
511 : errmsg("invalid XML comment")));
512 :
513 30 : initStringInfo(&buf);
514 30 : appendStringInfoString(&buf, "<!--");
515 30 : appendStringInfoText(&buf, arg);
516 30 : appendStringInfoString(&buf, "-->");
517 :
518 30 : PG_RETURN_XML_P(stringinfo_to_xmltype(&buf));
519 : #else
520 : NO_XML_SUPPORT();
521 : return 0;
522 : #endif
523 : }
524 :
525 :
526 : Datum
527 30 : xmltext(PG_FUNCTION_ARGS)
528 : {
529 : #ifdef USE_LIBXML
530 30 : text *arg = PG_GETARG_TEXT_PP(0);
531 : text *result;
532 30 : volatile xmlChar *xmlbuf = NULL;
533 : PgXmlErrorContext *xmlerrcxt;
534 :
535 : /* First we gotta spin up some error handling. */
536 30 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
537 :
538 30 : PG_TRY();
539 : {
540 30 : xmlbuf = xmlEncodeSpecialChars(NULL, xml_text2xmlChar(arg));
541 :
542 30 : if (xmlbuf == NULL || xmlerrcxt->err_occurred)
543 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
544 : "could not allocate xmlChar");
545 :
546 30 : result = cstring_to_text_with_len((const char *) xmlbuf,
547 : xmlStrlen((const xmlChar *) xmlbuf));
548 : }
549 0 : PG_CATCH();
550 : {
551 0 : if (xmlbuf)
552 0 : xmlFree((xmlChar *) xmlbuf);
553 :
554 0 : pg_xml_done(xmlerrcxt, true);
555 0 : PG_RE_THROW();
556 : }
557 30 : PG_END_TRY();
558 :
559 30 : xmlFree((xmlChar *) xmlbuf);
560 30 : pg_xml_done(xmlerrcxt, false);
561 :
562 30 : PG_RETURN_XML_P(result);
563 : #else
564 : NO_XML_SUPPORT();
565 : return 0;
566 : #endif /* not USE_LIBXML */
567 : }
568 :
569 :
570 : /*
571 : * TODO: xmlconcat needs to merge the notations and unparsed entities
572 : * of the argument values. Not very important in practice, though.
573 : */
574 : xmltype *
575 22316 : xmlconcat(List *args)
576 : {
577 : #ifdef USE_LIBXML
578 22316 : int global_standalone = 1;
579 22316 : xmlChar *global_version = NULL;
580 22316 : bool global_version_no_value = false;
581 : StringInfoData buf;
582 : ListCell *v;
583 :
584 22316 : initStringInfo(&buf);
585 66954 : foreach(v, args)
586 : {
587 44638 : xmltype *x = DatumGetXmlP(PointerGetDatum(lfirst(v)));
588 : size_t len;
589 : xmlChar *version;
590 : int standalone;
591 : char *str;
592 :
593 44638 : len = VARSIZE(x) - VARHDRSZ;
594 44638 : str = text_to_cstring((text *) x);
595 :
596 44638 : parse_xml_decl((xmlChar *) str, &len, &version, NULL, &standalone);
597 :
598 44638 : if (standalone == 0 && global_standalone == 1)
599 0 : global_standalone = 0;
600 44638 : if (standalone < 0)
601 44626 : global_standalone = -1;
602 :
603 44638 : if (!version)
604 44620 : global_version_no_value = true;
605 18 : else if (!global_version)
606 12 : global_version = version;
607 6 : else if (xmlStrcmp(version, global_version) != 0)
608 0 : global_version_no_value = true;
609 :
610 44638 : appendStringInfoString(&buf, str + len);
611 44638 : pfree(str);
612 : }
613 :
614 22316 : if (!global_version_no_value || global_standalone >= 0)
615 : {
616 : StringInfoData buf2;
617 :
618 6 : initStringInfo(&buf2);
619 :
620 6 : print_xml_decl(&buf2,
621 6 : (!global_version_no_value) ? global_version : NULL,
622 : 0,
623 : global_standalone);
624 :
625 6 : appendBinaryStringInfo(&buf2, buf.data, buf.len);
626 6 : buf = buf2;
627 : }
628 :
629 22316 : return stringinfo_to_xmltype(&buf);
630 : #else
631 : NO_XML_SUPPORT();
632 : return NULL;
633 : #endif
634 : }
635 :
636 :
637 : /*
638 : * XMLAGG support
639 : */
640 : Datum
641 22292 : xmlconcat2(PG_FUNCTION_ARGS)
642 : {
643 22292 : if (PG_ARGISNULL(0))
644 : {
645 18 : if (PG_ARGISNULL(1))
646 0 : PG_RETURN_NULL();
647 : else
648 18 : PG_RETURN_XML_P(PG_GETARG_XML_P(1));
649 : }
650 22274 : else if (PG_ARGISNULL(1))
651 0 : PG_RETURN_XML_P(PG_GETARG_XML_P(0));
652 : else
653 22274 : PG_RETURN_XML_P(xmlconcat(list_make2(PG_GETARG_XML_P(0),
654 : PG_GETARG_XML_P(1))));
655 : }
656 :
657 :
658 : Datum
659 6 : texttoxml(PG_FUNCTION_ARGS)
660 : {
661 6 : text *data = PG_GETARG_TEXT_PP(0);
662 :
663 6 : PG_RETURN_XML_P(xmlparse(data, xmloption, true));
664 : }
665 :
666 :
667 : Datum
668 0 : xmltotext(PG_FUNCTION_ARGS)
669 : {
670 0 : xmltype *data = PG_GETARG_XML_P(0);
671 :
672 : /* It's actually binary compatible. */
673 0 : PG_RETURN_TEXT_P((text *) data);
674 : }
675 :
676 :
677 : text *
678 180 : xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent)
679 : {
680 : #ifdef USE_LIBXML
681 : text *volatile result;
682 : xmlDocPtr doc;
683 : XmlOptionType parsed_xmloptiontype;
684 : xmlNodePtr content_nodes;
685 180 : volatile xmlBufferPtr buf = NULL;
686 180 : volatile xmlSaveCtxtPtr ctxt = NULL;
687 180 : ErrorSaveContext escontext = {T_ErrorSaveContext};
688 180 : PgXmlErrorContext *volatile xmlerrcxt = NULL;
689 : #endif
690 :
691 180 : if (xmloption_arg != XMLOPTION_DOCUMENT && !indent)
692 : {
693 : /*
694 : * We don't actually need to do anything, so just return the
695 : * binary-compatible input. For backwards-compatibility reasons,
696 : * allow such cases to succeed even without USE_LIBXML.
697 : */
698 36 : return (text *) data;
699 : }
700 :
701 : #ifdef USE_LIBXML
702 :
703 : /*
704 : * Parse the input according to the xmloption.
705 : *
706 : * preserve_whitespace is set to false in case we are indenting, otherwise
707 : * libxml2 will fail to indent elements that have whitespace between them.
708 : */
709 144 : doc = xml_parse(data, xmloption_arg, !indent, GetDatabaseEncoding(),
710 : &parsed_xmloptiontype, &content_nodes,
711 144 : (Node *) &escontext);
712 144 : if (doc == NULL || escontext.error_occurred)
713 : {
714 30 : if (doc)
715 0 : xmlFreeDoc(doc);
716 : /* A soft error must be failure to conform to XMLOPTION_DOCUMENT */
717 30 : ereport(ERROR,
718 : (errcode(ERRCODE_NOT_AN_XML_DOCUMENT),
719 : errmsg("not an XML document")));
720 : }
721 :
722 : /* If we weren't asked to indent, we're done. */
723 114 : if (!indent)
724 : {
725 18 : xmlFreeDoc(doc);
726 18 : return (text *) data;
727 : }
728 :
729 : /*
730 : * Otherwise, we gotta spin up some error handling. Unlike most other
731 : * routines in this module, we already have a libxml "doc" structure to
732 : * free, so we need to call pg_xml_init() inside the PG_TRY and be
733 : * prepared for it to fail (typically due to palloc OOM).
734 : */
735 96 : PG_TRY();
736 : {
737 96 : size_t decl_len = 0;
738 :
739 96 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
740 :
741 : /* The serialized data will go into this buffer. */
742 96 : buf = xmlBufferCreate();
743 :
744 96 : if (buf == NULL || xmlerrcxt->err_occurred)
745 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
746 : "could not allocate xmlBuffer");
747 :
748 : /* Detect whether there's an XML declaration */
749 96 : parse_xml_decl(xml_text2xmlChar(data), &decl_len, NULL, NULL, NULL);
750 :
751 : /*
752 : * Emit declaration only if the input had one. Note: some versions of
753 : * xmlSaveToBuffer leak memory if a non-null encoding argument is
754 : * passed, so don't do that. We don't want any encoding conversion
755 : * anyway.
756 : */
757 96 : if (decl_len == 0)
758 84 : ctxt = xmlSaveToBuffer(buf, NULL,
759 : XML_SAVE_NO_DECL | XML_SAVE_FORMAT);
760 : else
761 12 : ctxt = xmlSaveToBuffer(buf, NULL,
762 : XML_SAVE_FORMAT);
763 :
764 96 : if (ctxt == NULL || xmlerrcxt->err_occurred)
765 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
766 : "could not allocate xmlSaveCtxt");
767 :
768 96 : if (parsed_xmloptiontype == XMLOPTION_DOCUMENT)
769 : {
770 : /* If it's a document, saving is easy. */
771 42 : if (xmlSaveDoc(ctxt, doc) == -1 || xmlerrcxt->err_occurred)
772 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
773 : "could not save document to xmlBuffer");
774 : }
775 54 : else if (content_nodes != NULL)
776 : {
777 : /*
778 : * Deal with the case where we have non-singly-rooted XML.
779 : * libxml's dump functions don't work well for that without help.
780 : * We build a fake root node that serves as a container for the
781 : * content nodes, and then iterate over the nodes.
782 : */
783 : xmlNodePtr root;
784 : xmlNodePtr oldroot;
785 : xmlNodePtr newline;
786 :
787 48 : root = xmlNewNode(NULL, (const xmlChar *) "content-root");
788 48 : if (root == NULL || xmlerrcxt->err_occurred)
789 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
790 : "could not allocate xml node");
791 :
792 : /*
793 : * This attaches root to doc, so we need not free it separately...
794 : * but instead, we have to free the old root if there was one.
795 : */
796 48 : oldroot = xmlDocSetRootElement(doc, root);
797 48 : if (oldroot != NULL)
798 0 : xmlFreeNode(oldroot);
799 :
800 48 : if (xmlAddChildList(root, content_nodes) == NULL ||
801 48 : xmlerrcxt->err_occurred)
802 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
803 : "could not append xml node list");
804 :
805 : /*
806 : * We use this node to insert newlines in the dump. Note: in at
807 : * least some libxml versions, xmlNewDocText would not attach the
808 : * node to the document even if we passed it. Therefore, manage
809 : * freeing of this node manually, and pass NULL here to make sure
810 : * there's not a dangling link.
811 : */
812 48 : newline = xmlNewDocText(NULL, (const xmlChar *) "\n");
813 48 : if (newline == NULL || xmlerrcxt->err_occurred)
814 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
815 : "could not allocate xml node");
816 :
817 126 : for (xmlNodePtr node = root->children; node; node = node->next)
818 : {
819 : /* insert newlines between nodes */
820 78 : if (node->type != XML_TEXT_NODE && node->prev != NULL)
821 : {
822 24 : if (xmlSaveTree(ctxt, newline) == -1 || xmlerrcxt->err_occurred)
823 : {
824 0 : xmlFreeNode(newline);
825 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
826 : "could not save newline to xmlBuffer");
827 : }
828 : }
829 :
830 78 : if (xmlSaveTree(ctxt, node) == -1 || xmlerrcxt->err_occurred)
831 : {
832 0 : xmlFreeNode(newline);
833 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
834 : "could not save content to xmlBuffer");
835 : }
836 : }
837 :
838 48 : xmlFreeNode(newline);
839 : }
840 :
841 96 : if (xmlSaveClose(ctxt) == -1 || xmlerrcxt->err_occurred)
842 : {
843 0 : ctxt = NULL; /* don't try to close it again */
844 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
845 : "could not close xmlSaveCtxtPtr");
846 : }
847 :
848 : /*
849 : * xmlDocContentDumpOutput may add a trailing newline, so remove that.
850 : */
851 96 : if (xmloption_arg == XMLOPTION_DOCUMENT)
852 : {
853 36 : const char *str = (const char *) xmlBufferContent(buf);
854 36 : int len = xmlBufferLength(buf);
855 :
856 72 : while (len > 0 && (str[len - 1] == '\n' ||
857 36 : str[len - 1] == '\r'))
858 36 : len--;
859 :
860 36 : result = cstring_to_text_with_len(str, len);
861 : }
862 : else
863 60 : result = (text *) xmlBuffer_to_xmltype(buf);
864 : }
865 0 : PG_CATCH();
866 : {
867 0 : if (ctxt)
868 0 : xmlSaveClose(ctxt);
869 0 : if (buf)
870 0 : xmlBufferFree(buf);
871 0 : xmlFreeDoc(doc);
872 :
873 0 : if (xmlerrcxt)
874 0 : pg_xml_done(xmlerrcxt, true);
875 :
876 0 : PG_RE_THROW();
877 : }
878 96 : PG_END_TRY();
879 :
880 96 : xmlBufferFree(buf);
881 96 : xmlFreeDoc(doc);
882 :
883 96 : pg_xml_done(xmlerrcxt, false);
884 :
885 96 : return result;
886 : #else
887 : NO_XML_SUPPORT();
888 : return NULL;
889 : #endif
890 : }
891 :
892 :
893 : xmltype *
894 22448 : xmlelement(XmlExpr *xexpr,
895 : Datum *named_argvalue, bool *named_argnull,
896 : Datum *argvalue, bool *argnull)
897 : {
898 : #ifdef USE_LIBXML
899 : xmltype *result;
900 : List *named_arg_strings;
901 : List *arg_strings;
902 : int i;
903 : ListCell *arg;
904 : ListCell *narg;
905 : PgXmlErrorContext *xmlerrcxt;
906 22448 : volatile xmlBufferPtr buf = NULL;
907 22448 : volatile xmlTextWriterPtr writer = NULL;
908 :
909 : /*
910 : * All arguments are already evaluated, and their values are passed in the
911 : * named_argvalue/named_argnull or argvalue/argnull arrays. This avoids
912 : * issues if one of the arguments involves a call to some other function
913 : * or subsystem that wants to use libxml on its own terms. We examine the
914 : * original XmlExpr to identify the numbers and types of the arguments.
915 : */
916 22448 : named_arg_strings = NIL;
917 22448 : i = 0;
918 22496 : foreach(arg, xexpr->named_args)
919 : {
920 54 : Expr *e = (Expr *) lfirst(arg);
921 : char *str;
922 :
923 54 : if (named_argnull[i])
924 0 : str = NULL;
925 : else
926 54 : str = map_sql_value_to_xml_value(named_argvalue[i],
927 : exprType((Node *) e),
928 : false);
929 48 : named_arg_strings = lappend(named_arg_strings, str);
930 48 : i++;
931 : }
932 :
933 22442 : arg_strings = NIL;
934 22442 : i = 0;
935 44860 : foreach(arg, xexpr->args)
936 : {
937 22418 : Expr *e = (Expr *) lfirst(arg);
938 : char *str;
939 :
940 : /* here we can just forget NULL elements immediately */
941 22418 : if (!argnull[i])
942 : {
943 22418 : str = map_sql_value_to_xml_value(argvalue[i],
944 : exprType((Node *) e),
945 : true);
946 22418 : arg_strings = lappend(arg_strings, str);
947 : }
948 22418 : i++;
949 : }
950 :
951 22442 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
952 :
953 22442 : PG_TRY();
954 : {
955 22442 : buf = xmlBufferCreate();
956 22442 : if (buf == NULL || xmlerrcxt->err_occurred)
957 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
958 : "could not allocate xmlBuffer");
959 22442 : writer = xmlNewTextWriterMemory(buf, 0);
960 22442 : if (writer == NULL || xmlerrcxt->err_occurred)
961 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
962 : "could not allocate xmlTextWriter");
963 :
964 22442 : if (xmlTextWriterStartElement(writer, (xmlChar *) xexpr->name) < 0 ||
965 22442 : xmlerrcxt->err_occurred)
966 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
967 : "could not start xml element");
968 :
969 22490 : forboth(arg, named_arg_strings, narg, xexpr->arg_names)
970 : {
971 48 : char *str = (char *) lfirst(arg);
972 48 : char *argname = strVal(lfirst(narg));
973 :
974 48 : if (str)
975 : {
976 48 : if (xmlTextWriterWriteAttribute(writer,
977 : (xmlChar *) argname,
978 48 : (xmlChar *) str) < 0 ||
979 48 : xmlerrcxt->err_occurred)
980 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
981 : "could not write xml attribute");
982 : }
983 : }
984 :
985 44860 : foreach(arg, arg_strings)
986 : {
987 22418 : char *str = (char *) lfirst(arg);
988 :
989 22418 : if (xmlTextWriterWriteRaw(writer, (xmlChar *) str) < 0 ||
990 22418 : xmlerrcxt->err_occurred)
991 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
992 : "could not write raw xml text");
993 : }
994 :
995 22442 : if (xmlTextWriterEndElement(writer) < 0 ||
996 22442 : xmlerrcxt->err_occurred)
997 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
998 : "could not end xml element");
999 :
1000 : /* we MUST do this now to flush data out to the buffer ... */
1001 22442 : xmlFreeTextWriter(writer);
1002 22442 : writer = NULL;
1003 :
1004 22442 : result = xmlBuffer_to_xmltype(buf);
1005 : }
1006 0 : PG_CATCH();
1007 : {
1008 0 : if (writer)
1009 0 : xmlFreeTextWriter(writer);
1010 0 : if (buf)
1011 0 : xmlBufferFree(buf);
1012 :
1013 0 : pg_xml_done(xmlerrcxt, true);
1014 :
1015 0 : PG_RE_THROW();
1016 : }
1017 22442 : PG_END_TRY();
1018 :
1019 22442 : xmlBufferFree(buf);
1020 :
1021 22442 : pg_xml_done(xmlerrcxt, false);
1022 :
1023 22442 : return result;
1024 : #else
1025 : NO_XML_SUPPORT();
1026 : return NULL;
1027 : #endif
1028 : }
1029 :
1030 :
1031 : xmltype *
1032 138 : xmlparse(text *data, XmlOptionType xmloption_arg, bool preserve_whitespace)
1033 : {
1034 : #ifdef USE_LIBXML
1035 : xmlDocPtr doc;
1036 :
1037 138 : doc = xml_parse(data, xmloption_arg, preserve_whitespace,
1038 : GetDatabaseEncoding(), NULL, NULL, NULL);
1039 90 : xmlFreeDoc(doc);
1040 :
1041 90 : return (xmltype *) data;
1042 : #else
1043 : NO_XML_SUPPORT();
1044 : return NULL;
1045 : #endif
1046 : }
1047 :
1048 :
1049 : xmltype *
1050 72 : xmlpi(const char *target, text *arg, bool arg_is_null, bool *result_is_null)
1051 : {
1052 : #ifdef USE_LIBXML
1053 : xmltype *result;
1054 : StringInfoData buf;
1055 :
1056 72 : if (pg_strcasecmp(target, "xml") == 0)
1057 12 : ereport(ERROR,
1058 : (errcode(ERRCODE_INVALID_XML_PROCESSING_INSTRUCTION),
1059 : errmsg("invalid XML processing instruction"),
1060 : errdetail("XML processing instruction target name cannot be \"%s\".", target)));
1061 :
1062 : /*
1063 : * Following the SQL standard, the null check comes after the syntax check
1064 : * above.
1065 : */
1066 60 : *result_is_null = arg_is_null;
1067 60 : if (*result_is_null)
1068 12 : return NULL;
1069 :
1070 48 : initStringInfo(&buf);
1071 :
1072 48 : appendStringInfo(&buf, "<?%s", target);
1073 :
1074 48 : if (arg != NULL)
1075 : {
1076 : char *string;
1077 :
1078 24 : string = text_to_cstring(arg);
1079 24 : if (strstr(string, "?>") != NULL)
1080 6 : ereport(ERROR,
1081 : (errcode(ERRCODE_INVALID_XML_PROCESSING_INSTRUCTION),
1082 : errmsg("invalid XML processing instruction"),
1083 : errdetail("XML processing instruction cannot contain \"?>\".")));
1084 :
1085 18 : appendStringInfoChar(&buf, ' ');
1086 18 : appendStringInfoString(&buf, string + strspn(string, " "));
1087 18 : pfree(string);
1088 : }
1089 42 : appendStringInfoString(&buf, "?>");
1090 :
1091 42 : result = stringinfo_to_xmltype(&buf);
1092 42 : pfree(buf.data);
1093 42 : return result;
1094 : #else
1095 : NO_XML_SUPPORT();
1096 : return NULL;
1097 : #endif
1098 : }
1099 :
1100 :
1101 : xmltype *
1102 60 : xmlroot(xmltype *data, text *version, int standalone)
1103 : {
1104 : #ifdef USE_LIBXML
1105 : char *str;
1106 : size_t len;
1107 : xmlChar *orig_version;
1108 : int orig_standalone;
1109 : StringInfoData buf;
1110 :
1111 60 : len = VARSIZE(data) - VARHDRSZ;
1112 60 : str = text_to_cstring((text *) data);
1113 :
1114 60 : parse_xml_decl((xmlChar *) str, &len, &orig_version, NULL, &orig_standalone);
1115 :
1116 60 : if (version)
1117 24 : orig_version = xml_text2xmlChar(version);
1118 : else
1119 36 : orig_version = NULL;
1120 :
1121 60 : switch (standalone)
1122 : {
1123 18 : case XML_STANDALONE_YES:
1124 18 : orig_standalone = 1;
1125 18 : break;
1126 12 : case XML_STANDALONE_NO:
1127 12 : orig_standalone = 0;
1128 12 : break;
1129 12 : case XML_STANDALONE_NO_VALUE:
1130 12 : orig_standalone = -1;
1131 12 : break;
1132 18 : case XML_STANDALONE_OMITTED:
1133 : /* leave original value */
1134 18 : break;
1135 : }
1136 :
1137 60 : initStringInfo(&buf);
1138 60 : print_xml_decl(&buf, orig_version, 0, orig_standalone);
1139 60 : appendStringInfoString(&buf, str + len);
1140 :
1141 60 : return stringinfo_to_xmltype(&buf);
1142 : #else
1143 : NO_XML_SUPPORT();
1144 : return NULL;
1145 : #endif
1146 : }
1147 :
1148 :
1149 : /*
1150 : * Validate document (given as string) against DTD (given as external link)
1151 : *
1152 : * This has been removed because it is a security hole: unprivileged users
1153 : * should not be able to use Postgres to fetch arbitrary external files,
1154 : * which unfortunately is exactly what libxml is willing to do with the DTD
1155 : * parameter.
1156 : */
1157 : Datum
1158 0 : xmlvalidate(PG_FUNCTION_ARGS)
1159 : {
1160 0 : ereport(ERROR,
1161 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
1162 : errmsg("xmlvalidate is not implemented")));
1163 : return 0;
1164 : }
1165 :
1166 :
1167 : bool
1168 24 : xml_is_document(xmltype *arg)
1169 : {
1170 : #ifdef USE_LIBXML
1171 : xmlDocPtr doc;
1172 24 : ErrorSaveContext escontext = {T_ErrorSaveContext};
1173 :
1174 : /*
1175 : * We'll report "true" if no soft error is reported by xml_parse().
1176 : */
1177 24 : doc = xml_parse((text *) arg, XMLOPTION_DOCUMENT, true,
1178 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
1179 24 : if (doc)
1180 12 : xmlFreeDoc(doc);
1181 :
1182 24 : return !escontext.error_occurred;
1183 : #else /* not USE_LIBXML */
1184 : NO_XML_SUPPORT();
1185 : return false;
1186 : #endif /* not USE_LIBXML */
1187 : }
1188 :
1189 :
1190 : #ifdef USE_LIBXML
1191 :
1192 : /*
1193 : * pg_xml_init_library --- set up for use of libxml
1194 : *
1195 : * This should be called by each function that is about to use libxml
1196 : * facilities but doesn't require error handling. It initializes libxml
1197 : * and verifies compatibility with the loaded libxml version. These are
1198 : * once-per-session activities.
1199 : *
1200 : * TODO: xmlChar is utf8-char, make proper tuning (initdb with enc!=utf8 and
1201 : * check)
1202 : */
1203 : void
1204 94904 : pg_xml_init_library(void)
1205 : {
1206 : static bool first_time = true;
1207 :
1208 94904 : if (first_time)
1209 : {
1210 : /* Stuff we need do only once per session */
1211 :
1212 : /*
1213 : * Currently, we have no pure UTF-8 support for internals -- check if
1214 : * we can work.
1215 : */
1216 : if (sizeof(char) != sizeof(xmlChar))
1217 : ereport(ERROR,
1218 : (errmsg("could not initialize XML library"),
1219 : errdetail("libxml2 has incompatible char type: sizeof(char)=%zu, sizeof(xmlChar)=%zu.",
1220 : sizeof(char), sizeof(xmlChar))));
1221 :
1222 : #ifdef USE_LIBXMLCONTEXT
1223 : /* Set up libxml's memory allocation our way */
1224 : xml_memory_init();
1225 : #endif
1226 :
1227 : /* Check library compatibility */
1228 40 : LIBXML_TEST_VERSION;
1229 :
1230 40 : first_time = false;
1231 : }
1232 94904 : }
1233 :
1234 : /*
1235 : * pg_xml_init --- set up for use of libxml and register an error handler
1236 : *
1237 : * This should be called by each function that is about to use libxml
1238 : * facilities and requires error handling. It initializes libxml with
1239 : * pg_xml_init_library() and establishes our libxml error handler.
1240 : *
1241 : * strictness determines which errors are reported and which are ignored.
1242 : *
1243 : * Calls to this function MUST be followed by a PG_TRY block that guarantees
1244 : * that pg_xml_done() is called during either normal or error exit.
1245 : *
1246 : * This is exported for use by contrib/xml2, as well as other code that might
1247 : * wish to share use of this module's libxml error handler.
1248 : */
1249 : PgXmlErrorContext *
1250 24782 : pg_xml_init(PgXmlStrictness strictness)
1251 : {
1252 : PgXmlErrorContext *errcxt;
1253 : void *new_errcxt;
1254 :
1255 : /* Do one-time setup if needed */
1256 24782 : pg_xml_init_library();
1257 :
1258 : /* Create error handling context structure */
1259 24782 : errcxt = (PgXmlErrorContext *) palloc(sizeof(PgXmlErrorContext));
1260 24782 : errcxt->magic = ERRCXT_MAGIC;
1261 24782 : errcxt->strictness = strictness;
1262 24782 : errcxt->err_occurred = false;
1263 24782 : initStringInfo(&errcxt->err_buf);
1264 :
1265 : /*
1266 : * Save original error handler and install ours. libxml originally didn't
1267 : * distinguish between the contexts for generic and for structured error
1268 : * handlers. If we're using an old libxml version, we must thus save the
1269 : * generic error context, even though we're using a structured error
1270 : * handler.
1271 : */
1272 24782 : errcxt->saved_errfunc = xmlStructuredError;
1273 :
1274 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1275 24782 : errcxt->saved_errcxt = xmlStructuredErrorContext;
1276 : #else
1277 : errcxt->saved_errcxt = xmlGenericErrorContext;
1278 : #endif
1279 :
1280 24782 : xmlSetStructuredErrorFunc(errcxt, xml_errorHandler);
1281 :
1282 : /*
1283 : * Verify that xmlSetStructuredErrorFunc set the context variable we
1284 : * expected it to. If not, the error context pointer we just saved is not
1285 : * the correct thing to restore, and since that leaves us without a way to
1286 : * restore the context in pg_xml_done, we must fail.
1287 : *
1288 : * The only known situation in which this test fails is if we compile with
1289 : * headers from a libxml2 that doesn't track the structured error context
1290 : * separately (< 2.7.4), but at runtime use a version that does, or vice
1291 : * versa. The libxml2 authors did not treat that change as constituting
1292 : * an ABI break, so the LIBXML_TEST_VERSION test in pg_xml_init_library
1293 : * fails to protect us from this.
1294 : */
1295 :
1296 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1297 24782 : new_errcxt = xmlStructuredErrorContext;
1298 : #else
1299 : new_errcxt = xmlGenericErrorContext;
1300 : #endif
1301 :
1302 24782 : if (new_errcxt != errcxt)
1303 0 : ereport(ERROR,
1304 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
1305 : errmsg("could not set up XML error handler"),
1306 : errhint("This probably indicates that the version of libxml2"
1307 : " being used is not compatible with the libxml2"
1308 : " header files that PostgreSQL was built with.")));
1309 :
1310 : /*
1311 : * Also, install an entity loader to prevent unwanted fetches of external
1312 : * files and URLs.
1313 : */
1314 24782 : errcxt->saved_entityfunc = xmlGetExternalEntityLoader();
1315 24782 : xmlSetExternalEntityLoader(xmlPgEntityLoader);
1316 :
1317 24782 : return errcxt;
1318 : }
1319 :
1320 :
1321 : /*
1322 : * pg_xml_done --- restore previous libxml error handling
1323 : *
1324 : * Resets libxml's global error-handling state to what it was before
1325 : * pg_xml_init() was called.
1326 : *
1327 : * This routine verifies that all pending errors have been dealt with
1328 : * (in assert-enabled builds, anyway).
1329 : */
1330 : void
1331 24782 : pg_xml_done(PgXmlErrorContext *errcxt, bool isError)
1332 : {
1333 : void *cur_errcxt;
1334 :
1335 : /* An assert seems like enough protection here */
1336 : Assert(errcxt->magic == ERRCXT_MAGIC);
1337 :
1338 : /*
1339 : * In a normal exit, there should be no un-handled libxml errors. But we
1340 : * shouldn't try to enforce this during error recovery, since the longjmp
1341 : * could have been thrown before xml_ereport had a chance to run.
1342 : */
1343 : Assert(!errcxt->err_occurred || isError);
1344 :
1345 : /*
1346 : * Check that libxml's global state is correct, warn if not. This is a
1347 : * real test and not an Assert because it has a higher probability of
1348 : * happening.
1349 : */
1350 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1351 24782 : cur_errcxt = xmlStructuredErrorContext;
1352 : #else
1353 : cur_errcxt = xmlGenericErrorContext;
1354 : #endif
1355 :
1356 24782 : if (cur_errcxt != errcxt)
1357 0 : elog(WARNING, "libxml error handling state is out of sync with xml.c");
1358 :
1359 : /* Restore the saved handlers */
1360 24782 : xmlSetStructuredErrorFunc(errcxt->saved_errcxt, errcxt->saved_errfunc);
1361 24782 : xmlSetExternalEntityLoader(errcxt->saved_entityfunc);
1362 :
1363 : /*
1364 : * Mark the struct as invalid, just in case somebody somehow manages to
1365 : * call xml_errorHandler or xml_ereport with it.
1366 : */
1367 24782 : errcxt->magic = 0;
1368 :
1369 : /* Release memory */
1370 24782 : pfree(errcxt->err_buf.data);
1371 24782 : pfree(errcxt);
1372 24782 : }
1373 :
1374 :
1375 : /*
1376 : * pg_xml_error_occurred() --- test the error flag
1377 : */
1378 : bool
1379 78 : pg_xml_error_occurred(PgXmlErrorContext *errcxt)
1380 : {
1381 78 : return errcxt->err_occurred;
1382 : }
1383 :
1384 :
1385 : /*
1386 : * SQL/XML allows storing "XML documents" or "XML content". "XML
1387 : * documents" are specified by the XML specification and are parsed
1388 : * easily by libxml. "XML content" is specified by SQL/XML as the
1389 : * production "XMLDecl? content". But libxml can only parse the
1390 : * "content" part, so we have to parse the XML declaration ourselves
1391 : * to complete this.
1392 : */
1393 :
1394 : #define CHECK_XML_SPACE(p) \
1395 : do { \
1396 : if (!xmlIsBlank_ch(*(p))) \
1397 : return XML_ERR_SPACE_REQUIRED; \
1398 : } while (0)
1399 :
1400 : #define SKIP_XML_SPACE(p) \
1401 : while (xmlIsBlank_ch(*(p))) (p)++
1402 :
1403 : /* Letter | Digit | '.' | '-' | '_' | ':' | CombiningChar | Extender */
1404 : /* Beware of multiple evaluations of argument! */
1405 : #define PG_XMLISNAMECHAR(c) \
1406 : (xmlIsBaseChar_ch(c) || xmlIsIdeographicQ(c) \
1407 : || xmlIsDigit_ch(c) \
1408 : || c == '.' || c == '-' || c == '_' || c == ':' \
1409 : || xmlIsCombiningQ(c) \
1410 : || xmlIsExtender_ch(c))
1411 :
1412 : /* pnstrdup, but deal with xmlChar not char; len is measured in xmlChars */
1413 : static xmlChar *
1414 192 : xml_pnstrdup(const xmlChar *str, size_t len)
1415 : {
1416 : xmlChar *result;
1417 :
1418 192 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1419 192 : memcpy(result, str, len * sizeof(xmlChar));
1420 192 : result[len] = 0;
1421 192 : return result;
1422 : }
1423 :
1424 : /* Ditto, except input is char* */
1425 : static xmlChar *
1426 2424 : pg_xmlCharStrndup(const char *str, size_t len)
1427 : {
1428 : xmlChar *result;
1429 :
1430 2424 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1431 2424 : memcpy(result, str, len);
1432 2424 : result[len] = '\0';
1433 :
1434 2424 : return result;
1435 : }
1436 :
1437 : /*
1438 : * Copy xmlChar string to PostgreSQL-owned memory, freeing the input.
1439 : *
1440 : * The input xmlChar is freed regardless of success of the copy.
1441 : */
1442 : static char *
1443 112828 : xml_pstrdup_and_free(xmlChar *str)
1444 : {
1445 : char *result;
1446 :
1447 112828 : if (str)
1448 : {
1449 112828 : PG_TRY();
1450 : {
1451 112828 : result = pstrdup((char *) str);
1452 : }
1453 0 : PG_FINALLY();
1454 : {
1455 112828 : xmlFree(str);
1456 : }
1457 112828 : PG_END_TRY();
1458 : }
1459 : else
1460 0 : result = NULL;
1461 :
1462 112828 : return result;
1463 : }
1464 :
1465 : /*
1466 : * str is the null-terminated input string. Remaining arguments are
1467 : * output arguments; each can be NULL if value is not wanted.
1468 : * version and encoding are returned as locally-palloc'd strings.
1469 : * Result is 0 if OK, an error code if not.
1470 : */
1471 : static int
1472 70122 : parse_xml_decl(const xmlChar *str, size_t *lenp,
1473 : xmlChar **version, xmlChar **encoding, int *standalone)
1474 : {
1475 : const xmlChar *p;
1476 : const xmlChar *save_p;
1477 : size_t len;
1478 : int utf8char;
1479 : int utf8len;
1480 :
1481 : /*
1482 : * Only initialize libxml. We don't need error handling here, but we do
1483 : * need to make sure libxml is initialized before calling any of its
1484 : * functions. Note that this is safe (and a no-op) if caller has already
1485 : * done pg_xml_init().
1486 : */
1487 70122 : pg_xml_init_library();
1488 :
1489 : /* Initialize output arguments to "not present" */
1490 70122 : if (version)
1491 69474 : *version = NULL;
1492 70122 : if (encoding)
1493 0 : *encoding = NULL;
1494 70122 : if (standalone)
1495 69474 : *standalone = -1;
1496 :
1497 70122 : p = str;
1498 :
1499 70122 : if (xmlStrncmp(p, (xmlChar *) "<?xml", 5) != 0)
1500 69900 : goto finished;
1501 :
1502 : /*
1503 : * If next char is a name char, it's a PI like <?xml-stylesheet ...?>
1504 : * rather than an XMLDecl, so we have done what we came to do and found no
1505 : * XMLDecl.
1506 : *
1507 : * We need an input length value for xmlGetUTF8Char, but there's no need
1508 : * to count the whole document size, so use strnlen not strlen.
1509 : */
1510 222 : utf8len = strnlen((const char *) (p + 5), MAX_MULTIBYTE_CHAR_LEN);
1511 222 : utf8char = xmlGetUTF8Char(p + 5, &utf8len);
1512 222 : if (PG_XMLISNAMECHAR(utf8char))
1513 12 : goto finished;
1514 :
1515 210 : p += 5;
1516 :
1517 : /* version */
1518 210 : CHECK_XML_SPACE(p);
1519 420 : SKIP_XML_SPACE(p);
1520 210 : if (xmlStrncmp(p, (xmlChar *) "version", 7) != 0)
1521 0 : return XML_ERR_VERSION_MISSING;
1522 210 : p += 7;
1523 210 : SKIP_XML_SPACE(p);
1524 210 : if (*p != '=')
1525 0 : return XML_ERR_VERSION_MISSING;
1526 210 : p += 1;
1527 210 : SKIP_XML_SPACE(p);
1528 :
1529 210 : if (*p == '\'' || *p == '"')
1530 210 : {
1531 : const xmlChar *q;
1532 :
1533 210 : q = xmlStrchr(p + 1, *p);
1534 210 : if (!q)
1535 0 : return XML_ERR_VERSION_MISSING;
1536 :
1537 210 : if (version)
1538 192 : *version = xml_pnstrdup(p + 1, q - p - 1);
1539 210 : p = q + 1;
1540 : }
1541 : else
1542 0 : return XML_ERR_VERSION_MISSING;
1543 :
1544 : /* encoding */
1545 210 : save_p = p;
1546 372 : SKIP_XML_SPACE(p);
1547 210 : if (xmlStrncmp(p, (xmlChar *) "encoding", 8) == 0)
1548 : {
1549 54 : CHECK_XML_SPACE(save_p);
1550 54 : p += 8;
1551 54 : SKIP_XML_SPACE(p);
1552 54 : if (*p != '=')
1553 0 : return XML_ERR_MISSING_ENCODING;
1554 54 : p += 1;
1555 54 : SKIP_XML_SPACE(p);
1556 :
1557 54 : if (*p == '\'' || *p == '"')
1558 54 : {
1559 : const xmlChar *q;
1560 :
1561 54 : q = xmlStrchr(p + 1, *p);
1562 54 : if (!q)
1563 0 : return XML_ERR_MISSING_ENCODING;
1564 :
1565 54 : if (encoding)
1566 0 : *encoding = xml_pnstrdup(p + 1, q - p - 1);
1567 54 : p = q + 1;
1568 : }
1569 : else
1570 0 : return XML_ERR_MISSING_ENCODING;
1571 : }
1572 : else
1573 : {
1574 156 : p = save_p;
1575 : }
1576 :
1577 : /* standalone */
1578 210 : save_p = p;
1579 318 : SKIP_XML_SPACE(p);
1580 210 : if (xmlStrncmp(p, (xmlChar *) "standalone", 10) == 0)
1581 : {
1582 108 : CHECK_XML_SPACE(save_p);
1583 108 : p += 10;
1584 108 : SKIP_XML_SPACE(p);
1585 108 : if (*p != '=')
1586 0 : return XML_ERR_STANDALONE_VALUE;
1587 108 : p += 1;
1588 108 : SKIP_XML_SPACE(p);
1589 216 : if (xmlStrncmp(p, (xmlChar *) "'yes'", 5) == 0 ||
1590 108 : xmlStrncmp(p, (xmlChar *) "\"yes\"", 5) == 0)
1591 : {
1592 60 : if (standalone)
1593 60 : *standalone = 1;
1594 60 : p += 5;
1595 : }
1596 96 : else if (xmlStrncmp(p, (xmlChar *) "'no'", 4) == 0 ||
1597 48 : xmlStrncmp(p, (xmlChar *) "\"no\"", 4) == 0)
1598 : {
1599 36 : if (standalone)
1600 36 : *standalone = 0;
1601 36 : p += 4;
1602 : }
1603 : else
1604 12 : return XML_ERR_STANDALONE_VALUE;
1605 : }
1606 : else
1607 : {
1608 102 : p = save_p;
1609 : }
1610 :
1611 198 : SKIP_XML_SPACE(p);
1612 198 : if (xmlStrncmp(p, (xmlChar *) "?>", 2) != 0)
1613 0 : return XML_ERR_XMLDECL_NOT_FINISHED;
1614 198 : p += 2;
1615 :
1616 70110 : finished:
1617 70110 : len = p - str;
1618 :
1619 76842 : for (p = str; p < str + len; p++)
1620 6732 : if (*p > 127)
1621 0 : return XML_ERR_INVALID_CHAR;
1622 :
1623 70110 : if (lenp)
1624 70110 : *lenp = len;
1625 :
1626 70110 : return XML_ERR_OK;
1627 : }
1628 :
1629 :
1630 : /*
1631 : * Write an XML declaration. On output, we adjust the XML declaration
1632 : * as follows. (These rules are the moral equivalent of the clause
1633 : * "Serialization of an XML value" in the SQL standard.)
1634 : *
1635 : * We try to avoid generating an XML declaration if possible. This is
1636 : * so that you don't get trivial things like xml '<foo/>' resulting in
1637 : * '<?xml version="1.0"?><foo/>', which would surely be annoying. We
1638 : * must provide a declaration if the standalone property is specified
1639 : * or if we include an encoding declaration. If we have a
1640 : * declaration, we must specify a version (XML requires this).
1641 : * Otherwise we only make a declaration if the version is not "1.0",
1642 : * which is the default version specified in SQL:2003.
1643 : */
1644 : static bool
1645 23816 : print_xml_decl(StringInfo buf, const xmlChar *version,
1646 : pg_enc encoding, int standalone)
1647 : {
1648 23816 : if ((version && strcmp((const char *) version, PG_XML_DEFAULT_VERSION) != 0)
1649 23780 : || (encoding && encoding != PG_UTF8)
1650 23780 : || standalone != -1)
1651 : {
1652 96 : appendStringInfoString(buf, "<?xml");
1653 :
1654 96 : if (version)
1655 72 : appendStringInfo(buf, " version=\"%s\"", version);
1656 : else
1657 24 : appendStringInfo(buf, " version=\"%s\"", PG_XML_DEFAULT_VERSION);
1658 :
1659 96 : if (encoding && encoding != PG_UTF8)
1660 : {
1661 : /*
1662 : * XXX might be useful to convert this to IANA names (ISO-8859-1
1663 : * instead of LATIN1 etc.); needs field experience
1664 : */
1665 0 : appendStringInfo(buf, " encoding=\"%s\"",
1666 : pg_encoding_to_char(encoding));
1667 : }
1668 :
1669 96 : if (standalone == 1)
1670 48 : appendStringInfoString(buf, " standalone=\"yes\"");
1671 48 : else if (standalone == 0)
1672 24 : appendStringInfoString(buf, " standalone=\"no\"");
1673 96 : appendStringInfoString(buf, "?>");
1674 :
1675 96 : return true;
1676 : }
1677 : else
1678 23720 : return false;
1679 : }
1680 :
1681 : /*
1682 : * Test whether an input that is to be parsed as CONTENT contains a DTD.
1683 : *
1684 : * The SQL/XML:2003 definition of CONTENT ("XMLDecl? content") is not
1685 : * satisfied by a document with a DTD, which is a bit of a wart, as it means
1686 : * the CONTENT type is not a proper superset of DOCUMENT. SQL/XML:2006 and
1687 : * later fix that, by redefining content with reference to the "more
1688 : * permissive" Document Node of the XQuery/XPath Data Model, such that any
1689 : * DOCUMENT value is indeed also a CONTENT value. That definition is more
1690 : * useful, as CONTENT becomes usable for parsing input of unknown form (think
1691 : * pg_restore).
1692 : *
1693 : * As used below in parse_xml when parsing for CONTENT, libxml does not give
1694 : * us the 2006+ behavior, but only the 2003; it will choke if the input has
1695 : * a DTD. But we can provide the 2006+ definition of CONTENT easily enough,
1696 : * by detecting this case first and simply doing the parse as DOCUMENT.
1697 : *
1698 : * A DTD can be found arbitrarily far in, but that would be a contrived case;
1699 : * it will ordinarily start within a few dozen characters. The only things
1700 : * that can precede it are an XMLDecl (here, the caller will have called
1701 : * parse_xml_decl already), whitespace, comments, and processing instructions.
1702 : * This function need only return true if it sees a valid sequence of such
1703 : * things leading to <!DOCTYPE. It can simply return false in any other
1704 : * cases, including malformed input; that will mean the input gets parsed as
1705 : * CONTENT as originally planned, with libxml reporting any errors.
1706 : *
1707 : * This is only to be called from xml_parse, when pg_xml_init has already
1708 : * been called. The input is already in UTF8 encoding.
1709 : */
1710 : static bool
1711 1014 : xml_doctype_in_content(const xmlChar *str)
1712 : {
1713 1014 : const xmlChar *p = str;
1714 :
1715 : for (;;)
1716 36 : {
1717 : const xmlChar *e;
1718 :
1719 1140 : SKIP_XML_SPACE(p);
1720 1050 : if (*p != '<')
1721 214 : return false;
1722 836 : p++;
1723 :
1724 836 : if (*p == '!')
1725 : {
1726 72 : p++;
1727 :
1728 : /* if we see <!DOCTYPE, we can return true */
1729 72 : if (xmlStrncmp(p, (xmlChar *) "DOCTYPE", 7) == 0)
1730 42 : return true;
1731 :
1732 : /* otherwise, if it's not a comment, fail */
1733 30 : if (xmlStrncmp(p, (xmlChar *) "--", 2) != 0)
1734 0 : return false;
1735 : /* find end of comment: find -- and a > must follow */
1736 30 : p = xmlStrstr(p + 2, (xmlChar *) "--");
1737 30 : if (!p || p[2] != '>')
1738 0 : return false;
1739 : /* advance over comment, and keep scanning */
1740 30 : p += 3;
1741 30 : continue;
1742 : }
1743 :
1744 : /* otherwise, if it's not a PI <?target something?>, fail */
1745 764 : if (*p != '?')
1746 758 : return false;
1747 6 : p++;
1748 :
1749 : /* find end of PI (the string ?> is forbidden within a PI) */
1750 6 : e = xmlStrstr(p, (xmlChar *) "?>");
1751 6 : if (!e)
1752 0 : return false;
1753 :
1754 : /* advance over PI, keep scanning */
1755 6 : p = e + 2;
1756 : }
1757 : }
1758 :
1759 :
1760 : /*
1761 : * Convert a text object to XML internal representation
1762 : *
1763 : * data is the source data (must not be toasted!), encoding is its encoding,
1764 : * and xmloption_arg and preserve_whitespace are options for the
1765 : * transformation.
1766 : *
1767 : * If parsed_xmloptiontype isn't NULL, *parsed_xmloptiontype is set to the
1768 : * XmlOptionType actually used to parse the input (typically the same as
1769 : * xmloption_arg, but a DOCTYPE node in the input can force DOCUMENT mode).
1770 : *
1771 : * If parsed_nodes isn't NULL and we parse in CONTENT mode, the list
1772 : * of parsed nodes from the xmlParseBalancedChunkMemory call will be returned
1773 : * to *parsed_nodes. (It is caller's responsibility to free that.)
1774 : *
1775 : * Errors normally result in ereport(ERROR), but if escontext is an
1776 : * ErrorSaveContext, then "safe" errors are reported there instead, and the
1777 : * caller must check SOFT_ERROR_OCCURRED() to see whether that happened.
1778 : *
1779 : * Note: it is caller's responsibility to xmlFreeDoc() the result,
1780 : * else a permanent memory leak will ensue! But note the result could
1781 : * be NULL after a soft error.
1782 : *
1783 : * TODO maybe libxml2's xmlreader is better? (do not construct DOM,
1784 : * yet do not use SAX - see xmlreader.c)
1785 : */
1786 : static xmlDocPtr
1787 1320 : xml_parse(text *data, XmlOptionType xmloption_arg,
1788 : bool preserve_whitespace, int encoding,
1789 : XmlOptionType *parsed_xmloptiontype, xmlNodePtr *parsed_nodes,
1790 : Node *escontext)
1791 : {
1792 : int32 len;
1793 : xmlChar *string;
1794 : xmlChar *utf8string;
1795 : PgXmlErrorContext *xmlerrcxt;
1796 1320 : volatile xmlParserCtxtPtr ctxt = NULL;
1797 1320 : volatile xmlDocPtr doc = NULL;
1798 1320 : volatile int save_keep_blanks = -1;
1799 :
1800 : /*
1801 : * This step looks annoyingly redundant, but we must do it to have a
1802 : * null-terminated string in case encoding conversion isn't required.
1803 : */
1804 1320 : len = VARSIZE_ANY_EXHDR(data); /* will be useful later */
1805 1320 : string = xml_text2xmlChar(data);
1806 :
1807 : /*
1808 : * If the data isn't UTF8, we must translate before giving it to libxml.
1809 : *
1810 : * XXX ideally, we'd catch any encoding conversion failure and return a
1811 : * soft error. However, failure to convert to UTF8 should be pretty darn
1812 : * rare, so for now this is left undone.
1813 : */
1814 1320 : utf8string = pg_do_encoding_conversion(string,
1815 : len,
1816 : encoding,
1817 : PG_UTF8);
1818 :
1819 : /* Start up libxml and its parser */
1820 1320 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_WELLFORMED);
1821 :
1822 : /* Use a TRY block to ensure we clean up correctly */
1823 1320 : PG_TRY();
1824 : {
1825 1320 : bool parse_as_document = false;
1826 : int res_code;
1827 1320 : size_t count = 0;
1828 1320 : xmlChar *version = NULL;
1829 1320 : int standalone = 0;
1830 :
1831 : /* Any errors here are reported as hard ereport's */
1832 1320 : xmlInitParser();
1833 :
1834 : /* Decide whether to parse as document or content */
1835 1320 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1836 294 : parse_as_document = true;
1837 : else
1838 : {
1839 : /* Parse and skip over the XML declaration, if any */
1840 1026 : res_code = parse_xml_decl(utf8string,
1841 : &count, &version, NULL, &standalone);
1842 1026 : if (res_code != 0)
1843 : {
1844 12 : errsave(escontext,
1845 : errcode(ERRCODE_INVALID_XML_CONTENT),
1846 : errmsg_internal("invalid XML content: invalid XML declaration"),
1847 : errdetail_for_xml_code(res_code));
1848 12 : goto fail;
1849 : }
1850 :
1851 : /* Is there a DOCTYPE element? */
1852 1014 : if (xml_doctype_in_content(utf8string + count))
1853 42 : parse_as_document = true;
1854 : }
1855 :
1856 : /* initialize output parameters */
1857 1308 : if (parsed_xmloptiontype != NULL)
1858 144 : *parsed_xmloptiontype = parse_as_document ? XMLOPTION_DOCUMENT :
1859 : XMLOPTION_CONTENT;
1860 1308 : if (parsed_nodes != NULL)
1861 144 : *parsed_nodes = NULL;
1862 :
1863 1308 : if (parse_as_document)
1864 : {
1865 : int options;
1866 :
1867 : /* set up parser context used by xmlCtxtReadDoc */
1868 336 : ctxt = xmlNewParserCtxt();
1869 336 : if (ctxt == NULL || xmlerrcxt->err_occurred)
1870 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1871 : "could not allocate parser context");
1872 :
1873 : /*
1874 : * Select parse options.
1875 : *
1876 : * Note that here we try to apply DTD defaults (XML_PARSE_DTDATTR)
1877 : * according to SQL/XML:2008 GR 10.16.7.d: 'Default values defined
1878 : * by internal DTD are applied'. As for external DTDs, we try to
1879 : * support them too (see SQL/XML:2008 GR 10.16.7.e), but that
1880 : * doesn't really happen because xmlPgEntityLoader prevents it.
1881 : */
1882 336 : options = XML_PARSE_NOENT | XML_PARSE_DTDATTR
1883 336 : | (preserve_whitespace ? 0 : XML_PARSE_NOBLANKS);
1884 :
1885 336 : doc = xmlCtxtReadDoc(ctxt, utf8string,
1886 : NULL, /* no URL */
1887 : "UTF-8",
1888 : options);
1889 :
1890 336 : if (doc == NULL || xmlerrcxt->err_occurred)
1891 : {
1892 : /* Use original option to decide which error code to report */
1893 144 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1894 138 : xml_errsave(escontext, xmlerrcxt,
1895 : ERRCODE_INVALID_XML_DOCUMENT,
1896 : "invalid XML document");
1897 : else
1898 6 : xml_errsave(escontext, xmlerrcxt,
1899 : ERRCODE_INVALID_XML_CONTENT,
1900 : "invalid XML content");
1901 96 : goto fail;
1902 : }
1903 : }
1904 : else
1905 : {
1906 : /* set up document that xmlParseBalancedChunkMemory will add to */
1907 972 : doc = xmlNewDoc(version);
1908 972 : if (doc == NULL || xmlerrcxt->err_occurred)
1909 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1910 : "could not allocate XML document");
1911 :
1912 : Assert(doc->encoding == NULL);
1913 972 : doc->encoding = xmlStrdup((const xmlChar *) "UTF-8");
1914 972 : if (doc->encoding == NULL || xmlerrcxt->err_occurred)
1915 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1916 : "could not allocate XML document");
1917 972 : doc->standalone = standalone;
1918 :
1919 : /* set parse options --- have to do this the ugly way */
1920 972 : save_keep_blanks = xmlKeepBlanksDefault(preserve_whitespace ? 1 : 0);
1921 :
1922 : /* allow empty content */
1923 972 : if (*(utf8string + count))
1924 : {
1925 1896 : res_code = xmlParseBalancedChunkMemory(doc, NULL, NULL, 0,
1926 948 : utf8string + count,
1927 : parsed_nodes);
1928 948 : if (res_code != 0 || xmlerrcxt->err_occurred)
1929 : {
1930 60 : xml_errsave(escontext, xmlerrcxt,
1931 : ERRCODE_INVALID_XML_CONTENT,
1932 : "invalid XML content");
1933 12 : goto fail;
1934 : }
1935 : }
1936 : }
1937 :
1938 1224 : fail:
1939 : ;
1940 : }
1941 96 : PG_CATCH();
1942 : {
1943 96 : if (save_keep_blanks != -1)
1944 48 : xmlKeepBlanksDefault(save_keep_blanks);
1945 96 : if (doc != NULL)
1946 48 : xmlFreeDoc(doc);
1947 96 : if (ctxt != NULL)
1948 48 : xmlFreeParserCtxt(ctxt);
1949 :
1950 96 : pg_xml_done(xmlerrcxt, true);
1951 :
1952 96 : PG_RE_THROW();
1953 : }
1954 1224 : PG_END_TRY();
1955 :
1956 1224 : if (save_keep_blanks != -1)
1957 924 : xmlKeepBlanksDefault(save_keep_blanks);
1958 :
1959 1224 : if (ctxt != NULL)
1960 288 : xmlFreeParserCtxt(ctxt);
1961 :
1962 1224 : pg_xml_done(xmlerrcxt, false);
1963 :
1964 1224 : return doc;
1965 : }
1966 :
1967 :
1968 : /*
1969 : * xmlChar<->text conversions
1970 : */
1971 : static xmlChar *
1972 1470 : xml_text2xmlChar(text *in)
1973 : {
1974 1470 : return (xmlChar *) text_to_cstring(in);
1975 : }
1976 :
1977 :
1978 : #ifdef USE_LIBXMLCONTEXT
1979 :
1980 : /*
1981 : * Manage the special context used for all libxml allocations (but only
1982 : * in special debug builds; see notes at top of file)
1983 : */
1984 : static void
1985 : xml_memory_init(void)
1986 : {
1987 : /* Create memory context if not there already */
1988 : if (LibxmlContext == NULL)
1989 : LibxmlContext = AllocSetContextCreate(TopMemoryContext,
1990 : "Libxml context",
1991 : ALLOCSET_DEFAULT_SIZES);
1992 :
1993 : /* Re-establish the callbacks even if already set */
1994 : xmlMemSetup(xml_pfree, xml_palloc, xml_repalloc, xml_pstrdup);
1995 : }
1996 :
1997 : /*
1998 : * Wrappers for memory management functions
1999 : */
2000 : static void *
2001 : xml_palloc(size_t size)
2002 : {
2003 : return MemoryContextAlloc(LibxmlContext, size);
2004 : }
2005 :
2006 :
2007 : static void *
2008 : xml_repalloc(void *ptr, size_t size)
2009 : {
2010 : return repalloc(ptr, size);
2011 : }
2012 :
2013 :
2014 : static void
2015 : xml_pfree(void *ptr)
2016 : {
2017 : /* At least some parts of libxml assume xmlFree(NULL) is allowed */
2018 : if (ptr)
2019 : pfree(ptr);
2020 : }
2021 :
2022 :
2023 : static char *
2024 : xml_pstrdup(const char *string)
2025 : {
2026 : return MemoryContextStrdup(LibxmlContext, string);
2027 : }
2028 : #endif /* USE_LIBXMLCONTEXT */
2029 :
2030 :
2031 : /*
2032 : * xmlPgEntityLoader --- entity loader callback function
2033 : *
2034 : * Silently prevent any external entity URL from being loaded. We don't want
2035 : * to throw an error, so instead make the entity appear to expand to an empty
2036 : * string.
2037 : *
2038 : * We would prefer to allow loading entities that exist in the system's
2039 : * global XML catalog; but the available libxml2 APIs make that a complex
2040 : * and fragile task. For now, just shut down all external access.
2041 : */
2042 : static xmlParserInputPtr
2043 18 : xmlPgEntityLoader(const char *URL, const char *ID,
2044 : xmlParserCtxtPtr ctxt)
2045 : {
2046 18 : return xmlNewStringInputStream(ctxt, (const xmlChar *) "");
2047 : }
2048 :
2049 :
2050 : /*
2051 : * xml_ereport --- report an XML-related error
2052 : *
2053 : * The "msg" is the SQL-level message; some can be adopted from the SQL/XML
2054 : * standard. This function adds libxml's native error message, if any, as
2055 : * detail.
2056 : *
2057 : * This is exported for modules that want to share the core libxml error
2058 : * handler. Note that pg_xml_init() *must* have been called previously.
2059 : */
2060 : void
2061 12 : xml_ereport(PgXmlErrorContext *errcxt, int level, int sqlcode, const char *msg)
2062 : {
2063 : char *detail;
2064 :
2065 : /* Defend against someone passing us a bogus context struct */
2066 12 : if (errcxt->magic != ERRCXT_MAGIC)
2067 0 : elog(ERROR, "xml_ereport called with invalid PgXmlErrorContext");
2068 :
2069 : /* Flag that the current libxml error has been reported */
2070 12 : errcxt->err_occurred = false;
2071 :
2072 : /* Include detail only if we have some text from libxml */
2073 12 : if (errcxt->err_buf.len > 0)
2074 12 : detail = errcxt->err_buf.data;
2075 : else
2076 0 : detail = NULL;
2077 :
2078 12 : ereport(level,
2079 : (errcode(sqlcode),
2080 : errmsg_internal("%s", msg),
2081 : detail ? errdetail_internal("%s", detail) : 0));
2082 0 : }
2083 :
2084 :
2085 : /*
2086 : * xml_errsave --- save an XML-related error
2087 : *
2088 : * If escontext is an ErrorSaveContext, error details are saved into it,
2089 : * and control returns normally.
2090 : *
2091 : * Otherwise, the error is thrown, so that this is equivalent to
2092 : * xml_ereport() with level == ERROR.
2093 : *
2094 : * This should be used only for errors that we're sure we do not need
2095 : * a transaction abort to clean up after.
2096 : */
2097 : static void
2098 204 : xml_errsave(Node *escontext, PgXmlErrorContext *errcxt,
2099 : int sqlcode, const char *msg)
2100 : {
2101 : char *detail;
2102 :
2103 : /* Defend against someone passing us a bogus context struct */
2104 204 : if (errcxt->magic != ERRCXT_MAGIC)
2105 0 : elog(ERROR, "xml_errsave called with invalid PgXmlErrorContext");
2106 :
2107 : /* Flag that the current libxml error has been reported */
2108 204 : errcxt->err_occurred = false;
2109 :
2110 : /* Include detail only if we have some text from libxml */
2111 204 : if (errcxt->err_buf.len > 0)
2112 204 : detail = errcxt->err_buf.data;
2113 : else
2114 0 : detail = NULL;
2115 :
2116 204 : errsave(escontext,
2117 : (errcode(sqlcode),
2118 : errmsg_internal("%s", msg),
2119 : detail ? errdetail_internal("%s", detail) : 0));
2120 108 : }
2121 :
2122 :
2123 : /*
2124 : * Error handler for libxml errors and warnings
2125 : */
2126 : static void
2127 398 : xml_errorHandler(void *data, PgXmlErrorPtr error)
2128 : {
2129 398 : PgXmlErrorContext *xmlerrcxt = (PgXmlErrorContext *) data;
2130 398 : xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) error->ctxt;
2131 398 : xmlParserInputPtr input = (ctxt != NULL) ? ctxt->input : NULL;
2132 398 : xmlNodePtr node = error->node;
2133 398 : const xmlChar *name = (node != NULL &&
2134 398 : node->type == XML_ELEMENT_NODE) ? node->name : NULL;
2135 398 : int domain = error->domain;
2136 398 : int level = error->level;
2137 : StringInfo errorBuf;
2138 :
2139 : /*
2140 : * Defend against someone passing us a bogus context struct.
2141 : *
2142 : * We force a backend exit if this check fails because longjmp'ing out of
2143 : * libxml would likely render it unsafe to use further.
2144 : */
2145 398 : if (xmlerrcxt->magic != ERRCXT_MAGIC)
2146 0 : elog(FATAL, "xml_errorHandler called with invalid PgXmlErrorContext");
2147 :
2148 : /*----------
2149 : * Older libxml versions report some errors differently.
2150 : * First, some errors were previously reported as coming from the parser
2151 : * domain but are now reported as coming from the namespace domain.
2152 : * Second, some warnings were upgraded to errors.
2153 : * We attempt to compensate for that here.
2154 : *----------
2155 : */
2156 398 : switch (error->code)
2157 : {
2158 30 : case XML_WAR_NS_URI:
2159 30 : level = XML_ERR_ERROR;
2160 30 : domain = XML_FROM_NAMESPACE;
2161 30 : break;
2162 :
2163 54 : case XML_ERR_NS_DECL_ERROR:
2164 : case XML_WAR_NS_URI_RELATIVE:
2165 : case XML_WAR_NS_COLUMN:
2166 : case XML_NS_ERR_XML_NAMESPACE:
2167 : case XML_NS_ERR_UNDEFINED_NAMESPACE:
2168 : case XML_NS_ERR_QNAME:
2169 : case XML_NS_ERR_ATTRIBUTE_REDEFINED:
2170 : case XML_NS_ERR_EMPTY:
2171 54 : domain = XML_FROM_NAMESPACE;
2172 54 : break;
2173 : }
2174 :
2175 : /* Decide whether to act on the error or not */
2176 398 : switch (domain)
2177 : {
2178 314 : case XML_FROM_PARSER:
2179 :
2180 : /*
2181 : * XML_ERR_NOT_WELL_BALANCED is typically reported after some
2182 : * other, more on-point error. Furthermore, libxml2 2.13 reports
2183 : * it under a completely different set of rules than prior
2184 : * versions. To avoid cross-version behavioral differences,
2185 : * suppress it so long as we already logged some error.
2186 : */
2187 314 : if (error->code == XML_ERR_NOT_WELL_BALANCED &&
2188 30 : xmlerrcxt->err_occurred)
2189 30 : return;
2190 : /* fall through */
2191 :
2192 : case XML_FROM_NONE:
2193 : case XML_FROM_MEMORY:
2194 : case XML_FROM_IO:
2195 :
2196 : /*
2197 : * Suppress warnings about undeclared entities. We need to do
2198 : * this to avoid problems due to not loading DTD definitions.
2199 : */
2200 284 : if (error->code == XML_WAR_UNDECLARED_ENTITY)
2201 6 : return;
2202 :
2203 : /* Otherwise, accept error regardless of the parsing purpose */
2204 278 : break;
2205 :
2206 84 : default:
2207 : /* Ignore error if only doing well-formedness check */
2208 84 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_WELLFORMED)
2209 66 : return;
2210 18 : break;
2211 : }
2212 :
2213 : /* Prepare error message in errorBuf */
2214 296 : errorBuf = makeStringInfo();
2215 :
2216 296 : if (error->line > 0)
2217 296 : appendStringInfo(errorBuf, "line %d: ", error->line);
2218 296 : if (name != NULL)
2219 0 : appendStringInfo(errorBuf, "element %s: ", name);
2220 296 : if (error->message != NULL)
2221 296 : appendStringInfoString(errorBuf, error->message);
2222 : else
2223 0 : appendStringInfoString(errorBuf, "(no message provided)");
2224 :
2225 : /*
2226 : * Append context information to errorBuf.
2227 : *
2228 : * xmlParserPrintFileContext() uses libxml's "generic" error handler to
2229 : * write the context. Since we don't want to duplicate libxml
2230 : * functionality here, we set up a generic error handler temporarily.
2231 : *
2232 : * We use appendStringInfo() directly as libxml's generic error handler.
2233 : * This should work because it has essentially the same signature as
2234 : * libxml expects, namely (void *ptr, const char *msg, ...).
2235 : */
2236 296 : if (input != NULL)
2237 : {
2238 296 : xmlGenericErrorFunc errFuncSaved = xmlGenericError;
2239 296 : void *errCtxSaved = xmlGenericErrorContext;
2240 :
2241 296 : xmlSetGenericErrorFunc(errorBuf,
2242 : (xmlGenericErrorFunc) appendStringInfo);
2243 :
2244 : /* Add context information to errorBuf */
2245 296 : appendStringInfoLineSeparator(errorBuf);
2246 :
2247 296 : xmlParserPrintFileContext(input);
2248 :
2249 : /* Restore generic error func */
2250 296 : xmlSetGenericErrorFunc(errCtxSaved, errFuncSaved);
2251 : }
2252 :
2253 : /* Get rid of any trailing newlines in errorBuf */
2254 296 : chopStringInfoNewlines(errorBuf);
2255 :
2256 : /*
2257 : * Legacy error handling mode. err_occurred is never set, we just add the
2258 : * message to err_buf. This mode exists because the xml2 contrib module
2259 : * uses our error-handling infrastructure, but we don't want to change its
2260 : * behaviour since it's deprecated anyway. This is also why we don't
2261 : * distinguish between notices, warnings and errors here --- the old-style
2262 : * generic error handler wouldn't have done that either.
2263 : */
2264 296 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_LEGACY)
2265 : {
2266 2 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2267 2 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf->data,
2268 : errorBuf->len);
2269 :
2270 2 : destroyStringInfo(errorBuf);
2271 2 : return;
2272 : }
2273 :
2274 : /*
2275 : * We don't want to ereport() here because that'd probably leave libxml in
2276 : * an inconsistent state. Instead, we remember the error and ereport()
2277 : * from xml_ereport().
2278 : *
2279 : * Warnings and notices can be reported immediately since they won't cause
2280 : * a longjmp() out of libxml.
2281 : */
2282 294 : if (level >= XML_ERR_ERROR)
2283 : {
2284 288 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2285 288 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf->data,
2286 : errorBuf->len);
2287 :
2288 288 : xmlerrcxt->err_occurred = true;
2289 : }
2290 6 : else if (level >= XML_ERR_WARNING)
2291 : {
2292 6 : ereport(WARNING,
2293 : (errmsg_internal("%s", errorBuf->data)));
2294 : }
2295 : else
2296 : {
2297 0 : ereport(NOTICE,
2298 : (errmsg_internal("%s", errorBuf->data)));
2299 : }
2300 :
2301 294 : destroyStringInfo(errorBuf);
2302 : }
2303 :
2304 :
2305 : /*
2306 : * Convert libxml error codes into textual errdetail messages.
2307 : *
2308 : * This should be called within an ereport or errsave invocation,
2309 : * just as errdetail would be.
2310 : *
2311 : * At the moment, we only need to cover those codes that we
2312 : * may raise in this file.
2313 : */
2314 : static int
2315 6 : errdetail_for_xml_code(int code)
2316 : {
2317 : const char *det;
2318 :
2319 6 : switch (code)
2320 : {
2321 0 : case XML_ERR_INVALID_CHAR:
2322 0 : det = gettext_noop("Invalid character value.");
2323 0 : break;
2324 0 : case XML_ERR_SPACE_REQUIRED:
2325 0 : det = gettext_noop("Space required.");
2326 0 : break;
2327 6 : case XML_ERR_STANDALONE_VALUE:
2328 6 : det = gettext_noop("standalone accepts only 'yes' or 'no'.");
2329 6 : break;
2330 0 : case XML_ERR_VERSION_MISSING:
2331 0 : det = gettext_noop("Malformed declaration: missing version.");
2332 0 : break;
2333 0 : case XML_ERR_MISSING_ENCODING:
2334 0 : det = gettext_noop("Missing encoding in text declaration.");
2335 0 : break;
2336 0 : case XML_ERR_XMLDECL_NOT_FINISHED:
2337 0 : det = gettext_noop("Parsing XML declaration: '?>' expected.");
2338 0 : break;
2339 0 : default:
2340 0 : det = gettext_noop("Unrecognized libxml error code: %d.");
2341 0 : break;
2342 : }
2343 :
2344 6 : return errdetail(det, code);
2345 : }
2346 :
2347 :
2348 : /*
2349 : * Remove all trailing newlines from a StringInfo string
2350 : */
2351 : static void
2352 882 : chopStringInfoNewlines(StringInfo str)
2353 : {
2354 1474 : while (str->len > 0 && str->data[str->len - 1] == '\n')
2355 592 : str->data[--str->len] = '\0';
2356 882 : }
2357 :
2358 :
2359 : /*
2360 : * Append a newline after removing any existing trailing newlines
2361 : */
2362 : static void
2363 586 : appendStringInfoLineSeparator(StringInfo str)
2364 : {
2365 586 : chopStringInfoNewlines(str);
2366 586 : if (str->len > 0)
2367 368 : appendStringInfoChar(str, '\n');
2368 586 : }
2369 :
2370 :
2371 : /*
2372 : * Convert one char in the current server encoding to a Unicode codepoint.
2373 : */
2374 : static pg_wchar
2375 18350 : sqlchar_to_unicode(const char *s)
2376 : {
2377 : char *utf8string;
2378 : pg_wchar ret[2]; /* need space for trailing zero */
2379 :
2380 : /* note we're not assuming s is null-terminated */
2381 18350 : utf8string = pg_server_to_any(s, pg_mblen(s), PG_UTF8);
2382 :
2383 18350 : pg_encoding_mb2wchar_with_len(PG_UTF8, utf8string, ret,
2384 : pg_encoding_mblen(PG_UTF8, utf8string));
2385 :
2386 18350 : if (utf8string != s)
2387 0 : pfree(utf8string);
2388 :
2389 18350 : return ret[0];
2390 : }
2391 :
2392 :
2393 : static bool
2394 3652 : is_valid_xml_namefirst(pg_wchar c)
2395 : {
2396 : /* (Letter | '_' | ':') */
2397 3658 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2398 7310 : || c == '_' || c == ':');
2399 : }
2400 :
2401 :
2402 : static bool
2403 14698 : is_valid_xml_namechar(pg_wchar c)
2404 : {
2405 : /* Letter | Digit | '.' | '-' | '_' | ':' | CombiningChar | Extender */
2406 15590 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2407 892 : || xmlIsDigitQ(c)
2408 256 : || c == '.' || c == '-' || c == '_' || c == ':'
2409 12 : || xmlIsCombiningQ(c)
2410 31180 : || xmlIsExtenderQ(c));
2411 : }
2412 : #endif /* USE_LIBXML */
2413 :
2414 :
2415 : /*
2416 : * Map SQL identifier to XML name; see SQL/XML:2008 section 9.1.
2417 : */
2418 : char *
2419 3668 : map_sql_identifier_to_xml_name(const char *ident, bool fully_escaped,
2420 : bool escape_period)
2421 : {
2422 : #ifdef USE_LIBXML
2423 : StringInfoData buf;
2424 : const char *p;
2425 :
2426 : /*
2427 : * SQL/XML doesn't make use of this case anywhere, so it's probably a
2428 : * mistake.
2429 : */
2430 : Assert(fully_escaped || !escape_period);
2431 :
2432 3668 : initStringInfo(&buf);
2433 :
2434 22040 : for (p = ident; *p; p += pg_mblen(p))
2435 : {
2436 18372 : if (*p == ':' && (p == ident || fully_escaped))
2437 16 : appendStringInfoString(&buf, "_x003A_");
2438 18356 : else if (*p == '_' && *(p + 1) == 'x')
2439 6 : appendStringInfoString(&buf, "_x005F_");
2440 21638 : else if (fully_escaped && p == ident &&
2441 3288 : pg_strncasecmp(p, "xml", 3) == 0)
2442 : {
2443 0 : if (*p == 'x')
2444 0 : appendStringInfoString(&buf, "_x0078_");
2445 : else
2446 0 : appendStringInfoString(&buf, "_x0058_");
2447 : }
2448 18350 : else if (escape_period && *p == '.')
2449 0 : appendStringInfoString(&buf, "_x002E_");
2450 : else
2451 : {
2452 18350 : pg_wchar u = sqlchar_to_unicode(p);
2453 :
2454 36700 : if ((p == ident)
2455 3652 : ? !is_valid_xml_namefirst(u)
2456 14698 : : !is_valid_xml_namechar(u))
2457 18 : appendStringInfo(&buf, "_x%04X_", (unsigned int) u);
2458 : else
2459 18332 : appendBinaryStringInfo(&buf, p, pg_mblen(p));
2460 : }
2461 : }
2462 :
2463 3668 : return buf.data;
2464 : #else /* not USE_LIBXML */
2465 : NO_XML_SUPPORT();
2466 : return NULL;
2467 : #endif /* not USE_LIBXML */
2468 : }
2469 :
2470 :
2471 : /*
2472 : * Map XML name to SQL identifier; see SQL/XML:2008 section 9.3.
2473 : */
2474 : char *
2475 176 : map_xml_name_to_sql_identifier(const char *name)
2476 : {
2477 : StringInfoData buf;
2478 : const char *p;
2479 :
2480 176 : initStringInfo(&buf);
2481 :
2482 968 : for (p = name; *p; p += pg_mblen(p))
2483 : {
2484 792 : if (*p == '_' && *(p + 1) == 'x'
2485 22 : && isxdigit((unsigned char) *(p + 2))
2486 22 : && isxdigit((unsigned char) *(p + 3))
2487 22 : && isxdigit((unsigned char) *(p + 4))
2488 22 : && isxdigit((unsigned char) *(p + 5))
2489 22 : && *(p + 6) == '_')
2490 22 : {
2491 : char cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
2492 : unsigned int u;
2493 :
2494 22 : sscanf(p + 2, "%X", &u);
2495 22 : pg_unicode_to_server(u, (unsigned char *) cbuf);
2496 22 : appendStringInfoString(&buf, cbuf);
2497 22 : p += 6;
2498 : }
2499 : else
2500 770 : appendBinaryStringInfo(&buf, p, pg_mblen(p));
2501 : }
2502 :
2503 176 : return buf.data;
2504 : }
2505 :
2506 : /*
2507 : * Map SQL value to XML value; see SQL/XML:2008 section 9.8.
2508 : *
2509 : * When xml_escape_strings is true, then certain characters in string
2510 : * values are replaced by entity references (< etc.), as specified
2511 : * in SQL/XML:2008 section 9.8 GR 9) a) iii). This is normally what is
2512 : * wanted. The false case is mainly useful when the resulting value
2513 : * is used with xmlTextWriterWriteAttribute() to write out an
2514 : * attribute, because that function does the escaping itself.
2515 : */
2516 : char *
2517 134916 : map_sql_value_to_xml_value(Datum value, Oid type, bool xml_escape_strings)
2518 : {
2519 134916 : if (type_is_array_domain(type))
2520 : {
2521 : ArrayType *array;
2522 : Oid elmtype;
2523 : int16 elmlen;
2524 : bool elmbyval;
2525 : char elmalign;
2526 : int num_elems;
2527 : Datum *elem_values;
2528 : bool *elem_nulls;
2529 : StringInfoData buf;
2530 : int i;
2531 :
2532 6 : array = DatumGetArrayTypeP(value);
2533 6 : elmtype = ARR_ELEMTYPE(array);
2534 6 : get_typlenbyvalalign(elmtype, &elmlen, &elmbyval, &elmalign);
2535 :
2536 6 : deconstruct_array(array, elmtype,
2537 : elmlen, elmbyval, elmalign,
2538 : &elem_values, &elem_nulls,
2539 : &num_elems);
2540 :
2541 6 : initStringInfo(&buf);
2542 :
2543 24 : for (i = 0; i < num_elems; i++)
2544 : {
2545 18 : if (elem_nulls[i])
2546 0 : continue;
2547 18 : appendStringInfoString(&buf, "<element>");
2548 18 : appendStringInfoString(&buf,
2549 18 : map_sql_value_to_xml_value(elem_values[i],
2550 : elmtype, true));
2551 18 : appendStringInfoString(&buf, "</element>");
2552 : }
2553 :
2554 6 : pfree(elem_values);
2555 6 : pfree(elem_nulls);
2556 :
2557 6 : return buf.data;
2558 : }
2559 : else
2560 : {
2561 : Oid typeOut;
2562 : bool isvarlena;
2563 : char *str;
2564 :
2565 : /*
2566 : * Flatten domains; the special-case treatments below should apply to,
2567 : * eg, domains over boolean not just boolean.
2568 : */
2569 134910 : type = getBaseType(type);
2570 :
2571 : /*
2572 : * Special XSD formatting for some data types
2573 : */
2574 134910 : switch (type)
2575 : {
2576 66 : case BOOLOID:
2577 66 : if (DatumGetBool(value))
2578 60 : return "true";
2579 : else
2580 6 : return "false";
2581 :
2582 48 : case DATEOID:
2583 : {
2584 : DateADT date;
2585 : struct pg_tm tm;
2586 : char buf[MAXDATELEN + 1];
2587 :
2588 48 : date = DatumGetDateADT(value);
2589 : /* XSD doesn't support infinite values */
2590 48 : if (DATE_NOT_FINITE(date))
2591 0 : ereport(ERROR,
2592 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2593 : errmsg("date out of range"),
2594 : errdetail("XML does not support infinite date values.")));
2595 48 : j2date(date + POSTGRES_EPOCH_JDATE,
2596 : &(tm.tm_year), &(tm.tm_mon), &(tm.tm_mday));
2597 48 : EncodeDateOnly(&tm, USE_XSD_DATES, buf);
2598 :
2599 48 : return pstrdup(buf);
2600 : }
2601 :
2602 36 : case TIMESTAMPOID:
2603 : {
2604 : Timestamp timestamp;
2605 : struct pg_tm tm;
2606 : fsec_t fsec;
2607 : char buf[MAXDATELEN + 1];
2608 :
2609 36 : timestamp = DatumGetTimestamp(value);
2610 :
2611 : /* XSD doesn't support infinite values */
2612 36 : if (TIMESTAMP_NOT_FINITE(timestamp))
2613 6 : ereport(ERROR,
2614 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2615 : errmsg("timestamp out of range"),
2616 : errdetail("XML does not support infinite timestamp values.")));
2617 30 : else if (timestamp2tm(timestamp, NULL, &tm, &fsec, NULL, NULL) == 0)
2618 30 : EncodeDateTime(&tm, fsec, false, 0, NULL, USE_XSD_DATES, buf);
2619 : else
2620 0 : ereport(ERROR,
2621 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2622 : errmsg("timestamp out of range")));
2623 :
2624 30 : return pstrdup(buf);
2625 : }
2626 :
2627 24 : case TIMESTAMPTZOID:
2628 : {
2629 : TimestampTz timestamp;
2630 : struct pg_tm tm;
2631 : int tz;
2632 : fsec_t fsec;
2633 24 : const char *tzn = NULL;
2634 : char buf[MAXDATELEN + 1];
2635 :
2636 24 : timestamp = DatumGetTimestamp(value);
2637 :
2638 : /* XSD doesn't support infinite values */
2639 24 : if (TIMESTAMP_NOT_FINITE(timestamp))
2640 0 : ereport(ERROR,
2641 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2642 : errmsg("timestamp out of range"),
2643 : errdetail("XML does not support infinite timestamp values.")));
2644 24 : else if (timestamp2tm(timestamp, &tz, &tm, &fsec, &tzn, NULL) == 0)
2645 24 : EncodeDateTime(&tm, fsec, true, tz, tzn, USE_XSD_DATES, buf);
2646 : else
2647 0 : ereport(ERROR,
2648 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2649 : errmsg("timestamp out of range")));
2650 :
2651 24 : return pstrdup(buf);
2652 : }
2653 :
2654 : #ifdef USE_LIBXML
2655 36 : case BYTEAOID:
2656 : {
2657 36 : bytea *bstr = DatumGetByteaPP(value);
2658 : PgXmlErrorContext *xmlerrcxt;
2659 36 : volatile xmlBufferPtr buf = NULL;
2660 36 : volatile xmlTextWriterPtr writer = NULL;
2661 : char *result;
2662 :
2663 36 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
2664 :
2665 36 : PG_TRY();
2666 : {
2667 36 : buf = xmlBufferCreate();
2668 36 : if (buf == NULL || xmlerrcxt->err_occurred)
2669 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2670 : "could not allocate xmlBuffer");
2671 36 : writer = xmlNewTextWriterMemory(buf, 0);
2672 36 : if (writer == NULL || xmlerrcxt->err_occurred)
2673 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2674 : "could not allocate xmlTextWriter");
2675 :
2676 36 : if (xmlbinary == XMLBINARY_BASE64)
2677 30 : xmlTextWriterWriteBase64(writer, VARDATA_ANY(bstr),
2678 30 : 0, VARSIZE_ANY_EXHDR(bstr));
2679 : else
2680 6 : xmlTextWriterWriteBinHex(writer, VARDATA_ANY(bstr),
2681 6 : 0, VARSIZE_ANY_EXHDR(bstr));
2682 :
2683 : /* we MUST do this now to flush data out to the buffer */
2684 36 : xmlFreeTextWriter(writer);
2685 36 : writer = NULL;
2686 :
2687 36 : result = pstrdup((const char *) xmlBufferContent(buf));
2688 : }
2689 0 : PG_CATCH();
2690 : {
2691 0 : if (writer)
2692 0 : xmlFreeTextWriter(writer);
2693 0 : if (buf)
2694 0 : xmlBufferFree(buf);
2695 :
2696 0 : pg_xml_done(xmlerrcxt, true);
2697 :
2698 0 : PG_RE_THROW();
2699 : }
2700 36 : PG_END_TRY();
2701 :
2702 36 : xmlBufferFree(buf);
2703 :
2704 36 : pg_xml_done(xmlerrcxt, false);
2705 :
2706 36 : return result;
2707 : }
2708 : #endif /* USE_LIBXML */
2709 :
2710 : }
2711 :
2712 : /*
2713 : * otherwise, just use the type's native text representation
2714 : */
2715 134700 : getTypeOutputInfo(type, &typeOut, &isvarlena);
2716 134700 : str = OidOutputFunctionCall(typeOut, value);
2717 :
2718 : /* ... exactly as-is for XML, and when escaping is not wanted */
2719 134700 : if (type == XMLOID || !xml_escape_strings)
2720 22358 : return str;
2721 :
2722 : /* otherwise, translate special characters as needed */
2723 112342 : return escape_xml(str);
2724 : }
2725 : }
2726 :
2727 :
2728 : /*
2729 : * Escape characters in text that have special meanings in XML.
2730 : *
2731 : * Returns a palloc'd string.
2732 : *
2733 : * NB: this is intentionally not dependent on libxml.
2734 : */
2735 : char *
2736 112814 : escape_xml(const char *str)
2737 : {
2738 : StringInfoData buf;
2739 : const char *p;
2740 :
2741 112814 : initStringInfo(&buf);
2742 708908 : for (p = str; *p; p++)
2743 : {
2744 596094 : switch (*p)
2745 : {
2746 0 : case '&':
2747 0 : appendStringInfoString(&buf, "&");
2748 0 : break;
2749 36 : case '<':
2750 36 : appendStringInfoString(&buf, "<");
2751 36 : break;
2752 24 : case '>':
2753 24 : appendStringInfoString(&buf, ">");
2754 24 : break;
2755 0 : case '\r':
2756 0 : appendStringInfoString(&buf, "
");
2757 0 : break;
2758 596034 : default:
2759 596034 : appendStringInfoCharMacro(&buf, *p);
2760 596034 : break;
2761 : }
2762 : }
2763 112814 : return buf.data;
2764 : }
2765 :
2766 :
2767 : static char *
2768 24 : _SPI_strdup(const char *s)
2769 : {
2770 24 : size_t len = strlen(s) + 1;
2771 24 : char *ret = SPI_palloc(len);
2772 :
2773 24 : memcpy(ret, s, len);
2774 24 : return ret;
2775 : }
2776 :
2777 :
2778 : /*
2779 : * SQL to XML mapping functions
2780 : *
2781 : * What follows below was at one point intentionally organized so that
2782 : * you can read along in the SQL/XML standard. The functions are
2783 : * mostly split up the way the clauses lay out in the standards
2784 : * document, and the identifiers are also aligned with the standard
2785 : * text. Unfortunately, SQL/XML:2006 reordered the clauses
2786 : * differently than SQL/XML:2003, so the order below doesn't make much
2787 : * sense anymore.
2788 : *
2789 : * There are many things going on there:
2790 : *
2791 : * There are two kinds of mappings: Mapping SQL data (table contents)
2792 : * to XML documents, and mapping SQL structure (the "schema") to XML
2793 : * Schema. And there are functions that do both at the same time.
2794 : *
2795 : * Then you can map a database, a schema, or a table, each in both
2796 : * ways. This breaks down recursively: Mapping a database invokes
2797 : * mapping schemas, which invokes mapping tables, which invokes
2798 : * mapping rows, which invokes mapping columns, although you can't
2799 : * call the last two from the outside. Because of this, there are a
2800 : * number of xyz_internal() functions which are to be called both from
2801 : * the function manager wrapper and from some upper layer in a
2802 : * recursive call.
2803 : *
2804 : * See the documentation about what the common function arguments
2805 : * nulls, tableforest, and targetns mean.
2806 : *
2807 : * Some style guidelines for XML output: Use double quotes for quoting
2808 : * XML attributes. Indent XML elements by two spaces, but remember
2809 : * that a lot of code is called recursively at different levels, so
2810 : * it's better not to indent rather than create output that indents
2811 : * and outdents weirdly. Add newlines to make the output look nice.
2812 : */
2813 :
2814 :
2815 : /*
2816 : * Visibility of objects for XML mappings; see SQL/XML:2008 section
2817 : * 4.10.8.
2818 : */
2819 :
2820 : /*
2821 : * Given a query, which must return type oid as first column, produce
2822 : * a list of Oids with the query results.
2823 : */
2824 : static List *
2825 36 : query_to_oid_list(const char *query)
2826 : {
2827 : uint64 i;
2828 36 : List *list = NIL;
2829 : int spi_result;
2830 :
2831 36 : spi_result = SPI_execute(query, true, 0);
2832 36 : if (spi_result != SPI_OK_SELECT)
2833 0 : elog(ERROR, "SPI_execute returned %s for %s",
2834 : SPI_result_code_string(spi_result), query);
2835 :
2836 108 : for (i = 0; i < SPI_processed; i++)
2837 : {
2838 : Datum oid;
2839 : bool isnull;
2840 :
2841 72 : oid = SPI_getbinval(SPI_tuptable->vals[i],
2842 72 : SPI_tuptable->tupdesc,
2843 : 1,
2844 : &isnull);
2845 72 : if (!isnull)
2846 72 : list = lappend_oid(list, DatumGetObjectId(oid));
2847 : }
2848 :
2849 36 : return list;
2850 : }
2851 :
2852 :
2853 : static List *
2854 36 : schema_get_xml_visible_tables(Oid nspid)
2855 : {
2856 : StringInfoData query;
2857 :
2858 36 : initStringInfo(&query);
2859 36 : appendStringInfo(&query, "SELECT oid FROM pg_catalog.pg_class"
2860 : " WHERE relnamespace = %u AND relkind IN ("
2861 : CppAsString2(RELKIND_RELATION) ","
2862 : CppAsString2(RELKIND_MATVIEW) ","
2863 : CppAsString2(RELKIND_VIEW) ")"
2864 : " AND pg_catalog.has_table_privilege (oid, 'SELECT')"
2865 : " ORDER BY relname;", nspid);
2866 :
2867 36 : return query_to_oid_list(query.data);
2868 : }
2869 :
2870 :
2871 : /*
2872 : * Including the system schemas is probably not useful for a database
2873 : * mapping.
2874 : */
2875 : #define XML_VISIBLE_SCHEMAS_EXCLUDE "(nspname ~ '^pg_' OR nspname = 'information_schema')"
2876 :
2877 : #define XML_VISIBLE_SCHEMAS "SELECT oid FROM pg_catalog.pg_namespace WHERE pg_catalog.has_schema_privilege (oid, 'USAGE') AND NOT " XML_VISIBLE_SCHEMAS_EXCLUDE
2878 :
2879 :
2880 : static List *
2881 0 : database_get_xml_visible_schemas(void)
2882 : {
2883 0 : return query_to_oid_list(XML_VISIBLE_SCHEMAS " ORDER BY nspname;");
2884 : }
2885 :
2886 :
2887 : static List *
2888 0 : database_get_xml_visible_tables(void)
2889 : {
2890 : /* At the moment there is no order required here. */
2891 0 : return query_to_oid_list("SELECT oid FROM pg_catalog.pg_class"
2892 : " WHERE relkind IN ("
2893 : CppAsString2(RELKIND_RELATION) ","
2894 : CppAsString2(RELKIND_MATVIEW) ","
2895 : CppAsString2(RELKIND_VIEW) ")"
2896 : " AND pg_catalog.has_table_privilege(pg_class.oid, 'SELECT')"
2897 : " AND relnamespace IN (" XML_VISIBLE_SCHEMAS ");");
2898 : }
2899 :
2900 :
2901 : /*
2902 : * Map SQL table to XML and/or XML Schema document; see SQL/XML:2008
2903 : * section 9.11.
2904 : */
2905 :
2906 : static StringInfo
2907 96 : table_to_xml_internal(Oid relid,
2908 : const char *xmlschema, bool nulls, bool tableforest,
2909 : const char *targetns, bool top_level)
2910 : {
2911 : StringInfoData query;
2912 :
2913 96 : initStringInfo(&query);
2914 96 : appendStringInfo(&query, "SELECT * FROM %s",
2915 : DatumGetCString(DirectFunctionCall1(regclassout,
2916 : ObjectIdGetDatum(relid))));
2917 96 : return query_to_xml_internal(query.data, get_rel_name(relid),
2918 : xmlschema, nulls, tableforest,
2919 : targetns, top_level);
2920 : }
2921 :
2922 :
2923 : Datum
2924 36 : table_to_xml(PG_FUNCTION_ARGS)
2925 : {
2926 36 : Oid relid = PG_GETARG_OID(0);
2927 36 : bool nulls = PG_GETARG_BOOL(1);
2928 36 : bool tableforest = PG_GETARG_BOOL(2);
2929 36 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2930 :
2931 36 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid, NULL,
2932 : nulls, tableforest,
2933 : targetns, true)));
2934 : }
2935 :
2936 :
2937 : Datum
2938 10 : query_to_xml(PG_FUNCTION_ARGS)
2939 : {
2940 10 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
2941 10 : bool nulls = PG_GETARG_BOOL(1);
2942 10 : bool tableforest = PG_GETARG_BOOL(2);
2943 10 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2944 :
2945 10 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
2946 : NULL, nulls, tableforest,
2947 : targetns, true)));
2948 : }
2949 :
2950 :
2951 : Datum
2952 12 : cursor_to_xml(PG_FUNCTION_ARGS)
2953 : {
2954 12 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
2955 12 : int32 count = PG_GETARG_INT32(1);
2956 12 : bool nulls = PG_GETARG_BOOL(2);
2957 12 : bool tableforest = PG_GETARG_BOOL(3);
2958 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(4));
2959 :
2960 : StringInfoData result;
2961 : Portal portal;
2962 : uint64 i;
2963 :
2964 12 : initStringInfo(&result);
2965 :
2966 12 : if (!tableforest)
2967 : {
2968 6 : xmldata_root_element_start(&result, "table", NULL, targetns, true);
2969 6 : appendStringInfoChar(&result, '\n');
2970 : }
2971 :
2972 12 : SPI_connect();
2973 12 : portal = SPI_cursor_find(name);
2974 12 : if (portal == NULL)
2975 0 : ereport(ERROR,
2976 : (errcode(ERRCODE_UNDEFINED_CURSOR),
2977 : errmsg("cursor \"%s\" does not exist", name)));
2978 :
2979 12 : SPI_cursor_fetch(portal, true, count);
2980 48 : for (i = 0; i < SPI_processed; i++)
2981 36 : SPI_sql_row_to_xmlelement(i, &result, NULL, nulls,
2982 : tableforest, targetns, true);
2983 :
2984 12 : SPI_finish();
2985 :
2986 12 : if (!tableforest)
2987 6 : xmldata_root_element_end(&result, "table");
2988 :
2989 12 : PG_RETURN_XML_P(stringinfo_to_xmltype(&result));
2990 : }
2991 :
2992 :
2993 : /*
2994 : * Write the start tag of the root element of a data mapping.
2995 : *
2996 : * top_level means that this is the very top level of the eventual
2997 : * output. For example, when the user calls table_to_xml, then a call
2998 : * with a table name to this function is the top level. When the user
2999 : * calls database_to_xml, then a call with a schema name to this
3000 : * function is not the top level. If top_level is false, then the XML
3001 : * namespace declarations are omitted, because they supposedly already
3002 : * appeared earlier in the output. Repeating them is not wrong, but
3003 : * it looks ugly.
3004 : */
3005 : static void
3006 238 : xmldata_root_element_start(StringInfo result, const char *eltname,
3007 : const char *xmlschema, const char *targetns,
3008 : bool top_level)
3009 : {
3010 : /* This isn't really wrong but currently makes no sense. */
3011 : Assert(top_level || !xmlschema);
3012 :
3013 238 : appendStringInfo(result, "<%s", eltname);
3014 238 : if (top_level)
3015 : {
3016 178 : appendStringInfoString(result, " xmlns:xsi=\"" NAMESPACE_XSI "\"");
3017 178 : if (strlen(targetns) > 0)
3018 30 : appendStringInfo(result, " xmlns=\"%s\"", targetns);
3019 : }
3020 238 : if (xmlschema)
3021 : {
3022 : /* FIXME: better targets */
3023 18 : if (strlen(targetns) > 0)
3024 6 : appendStringInfo(result, " xsi:schemaLocation=\"%s #\"", targetns);
3025 : else
3026 12 : appendStringInfoString(result, " xsi:noNamespaceSchemaLocation=\"#\"");
3027 : }
3028 238 : appendStringInfoString(result, ">\n");
3029 238 : }
3030 :
3031 :
3032 : static void
3033 238 : xmldata_root_element_end(StringInfo result, const char *eltname)
3034 : {
3035 238 : appendStringInfo(result, "</%s>\n", eltname);
3036 238 : }
3037 :
3038 :
3039 : static StringInfo
3040 112 : query_to_xml_internal(const char *query, char *tablename,
3041 : const char *xmlschema, bool nulls, bool tableforest,
3042 : const char *targetns, bool top_level)
3043 : {
3044 : StringInfo result;
3045 : char *xmltn;
3046 : uint64 i;
3047 :
3048 112 : if (tablename)
3049 96 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
3050 : else
3051 16 : xmltn = "table";
3052 :
3053 112 : result = makeStringInfo();
3054 :
3055 112 : SPI_connect();
3056 112 : if (SPI_execute(query, true, 0) != SPI_OK_SELECT)
3057 0 : ereport(ERROR,
3058 : (errcode(ERRCODE_DATA_EXCEPTION),
3059 : errmsg("invalid query")));
3060 :
3061 112 : if (!tableforest)
3062 : {
3063 52 : xmldata_root_element_start(result, xmltn, xmlschema,
3064 : targetns, top_level);
3065 52 : appendStringInfoChar(result, '\n');
3066 : }
3067 :
3068 112 : if (xmlschema)
3069 30 : appendStringInfo(result, "%s\n\n", xmlschema);
3070 :
3071 388 : for (i = 0; i < SPI_processed; i++)
3072 276 : SPI_sql_row_to_xmlelement(i, result, tablename, nulls,
3073 : tableforest, targetns, top_level);
3074 :
3075 112 : if (!tableforest)
3076 52 : xmldata_root_element_end(result, xmltn);
3077 :
3078 112 : SPI_finish();
3079 :
3080 112 : return result;
3081 : }
3082 :
3083 :
3084 : Datum
3085 30 : table_to_xmlschema(PG_FUNCTION_ARGS)
3086 : {
3087 30 : Oid relid = PG_GETARG_OID(0);
3088 30 : bool nulls = PG_GETARG_BOOL(1);
3089 30 : bool tableforest = PG_GETARG_BOOL(2);
3090 30 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3091 : const char *result;
3092 : Relation rel;
3093 :
3094 30 : rel = table_open(relid, AccessShareLock);
3095 30 : result = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
3096 : tableforest, targetns);
3097 30 : table_close(rel, NoLock);
3098 :
3099 30 : PG_RETURN_XML_P(cstring_to_xmltype(result));
3100 : }
3101 :
3102 :
3103 : Datum
3104 6 : query_to_xmlschema(PG_FUNCTION_ARGS)
3105 : {
3106 6 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
3107 6 : bool nulls = PG_GETARG_BOOL(1);
3108 6 : bool tableforest = PG_GETARG_BOOL(2);
3109 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3110 : const char *result;
3111 : SPIPlanPtr plan;
3112 : Portal portal;
3113 :
3114 6 : SPI_connect();
3115 :
3116 6 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
3117 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
3118 :
3119 6 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
3120 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
3121 :
3122 6 : result = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3123 : InvalidOid, nulls,
3124 : tableforest, targetns));
3125 6 : SPI_cursor_close(portal);
3126 6 : SPI_finish();
3127 :
3128 6 : PG_RETURN_XML_P(cstring_to_xmltype(result));
3129 : }
3130 :
3131 :
3132 : Datum
3133 12 : cursor_to_xmlschema(PG_FUNCTION_ARGS)
3134 : {
3135 12 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
3136 12 : bool nulls = PG_GETARG_BOOL(1);
3137 12 : bool tableforest = PG_GETARG_BOOL(2);
3138 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3139 : const char *xmlschema;
3140 : Portal portal;
3141 :
3142 12 : SPI_connect();
3143 12 : portal = SPI_cursor_find(name);
3144 12 : if (portal == NULL)
3145 0 : ereport(ERROR,
3146 : (errcode(ERRCODE_UNDEFINED_CURSOR),
3147 : errmsg("cursor \"%s\" does not exist", name)));
3148 12 : if (portal->tupDesc == NULL)
3149 0 : ereport(ERROR,
3150 : (errcode(ERRCODE_INVALID_CURSOR_STATE),
3151 : errmsg("portal \"%s\" does not return tuples", name)));
3152 :
3153 12 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3154 : InvalidOid, nulls,
3155 : tableforest, targetns));
3156 12 : SPI_finish();
3157 :
3158 12 : PG_RETURN_XML_P(cstring_to_xmltype(xmlschema));
3159 : }
3160 :
3161 :
3162 : Datum
3163 24 : table_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3164 : {
3165 24 : Oid relid = PG_GETARG_OID(0);
3166 24 : bool nulls = PG_GETARG_BOOL(1);
3167 24 : bool tableforest = PG_GETARG_BOOL(2);
3168 24 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3169 : Relation rel;
3170 : const char *xmlschema;
3171 :
3172 24 : rel = table_open(relid, AccessShareLock);
3173 24 : xmlschema = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
3174 : tableforest, targetns);
3175 24 : table_close(rel, NoLock);
3176 :
3177 24 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid,
3178 : xmlschema, nulls, tableforest,
3179 : targetns, true)));
3180 : }
3181 :
3182 :
3183 : Datum
3184 6 : query_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3185 : {
3186 6 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
3187 6 : bool nulls = PG_GETARG_BOOL(1);
3188 6 : bool tableforest = PG_GETARG_BOOL(2);
3189 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3190 :
3191 : const char *xmlschema;
3192 : SPIPlanPtr plan;
3193 : Portal portal;
3194 :
3195 6 : SPI_connect();
3196 :
3197 6 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
3198 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
3199 :
3200 6 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
3201 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
3202 :
3203 6 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3204 : InvalidOid, nulls, tableforest, targetns));
3205 6 : SPI_cursor_close(portal);
3206 6 : SPI_finish();
3207 :
3208 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
3209 : xmlschema, nulls, tableforest,
3210 : targetns, true)));
3211 : }
3212 :
3213 :
3214 : /*
3215 : * Map SQL schema to XML and/or XML Schema document; see SQL/XML:2008
3216 : * sections 9.13, 9.14.
3217 : */
3218 :
3219 : static StringInfo
3220 18 : schema_to_xml_internal(Oid nspid, const char *xmlschema, bool nulls,
3221 : bool tableforest, const char *targetns, bool top_level)
3222 : {
3223 : StringInfo result;
3224 : char *xmlsn;
3225 : List *relid_list;
3226 : ListCell *cell;
3227 :
3228 18 : xmlsn = map_sql_identifier_to_xml_name(get_namespace_name(nspid),
3229 : true, false);
3230 18 : result = makeStringInfo();
3231 :
3232 18 : xmldata_root_element_start(result, xmlsn, xmlschema, targetns, top_level);
3233 18 : appendStringInfoChar(result, '\n');
3234 :
3235 18 : if (xmlschema)
3236 6 : appendStringInfo(result, "%s\n\n", xmlschema);
3237 :
3238 18 : SPI_connect();
3239 :
3240 18 : relid_list = schema_get_xml_visible_tables(nspid);
3241 :
3242 54 : foreach(cell, relid_list)
3243 : {
3244 36 : Oid relid = lfirst_oid(cell);
3245 : StringInfo subres;
3246 :
3247 36 : subres = table_to_xml_internal(relid, NULL, nulls, tableforest,
3248 : targetns, false);
3249 :
3250 36 : appendBinaryStringInfo(result, subres->data, subres->len);
3251 36 : appendStringInfoChar(result, '\n');
3252 : }
3253 :
3254 18 : SPI_finish();
3255 :
3256 18 : xmldata_root_element_end(result, xmlsn);
3257 :
3258 18 : return result;
3259 : }
3260 :
3261 :
3262 : Datum
3263 12 : schema_to_xml(PG_FUNCTION_ARGS)
3264 : {
3265 12 : Name name = PG_GETARG_NAME(0);
3266 12 : bool nulls = PG_GETARG_BOOL(1);
3267 12 : bool tableforest = PG_GETARG_BOOL(2);
3268 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3269 :
3270 : char *schemaname;
3271 : Oid nspid;
3272 :
3273 12 : schemaname = NameStr(*name);
3274 12 : nspid = LookupExplicitNamespace(schemaname, false);
3275 :
3276 12 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid, NULL,
3277 : nulls, tableforest, targetns, true)));
3278 : }
3279 :
3280 :
3281 : /*
3282 : * Write the start element of the root element of an XML Schema mapping.
3283 : */
3284 : static void
3285 96 : xsd_schema_element_start(StringInfo result, const char *targetns)
3286 : {
3287 96 : appendStringInfoString(result,
3288 : "<xsd:schema\n"
3289 : " xmlns:xsd=\"" NAMESPACE_XSD "\"");
3290 96 : if (strlen(targetns) > 0)
3291 18 : appendStringInfo(result,
3292 : "\n"
3293 : " targetNamespace=\"%s\"\n"
3294 : " elementFormDefault=\"qualified\"",
3295 : targetns);
3296 96 : appendStringInfoString(result,
3297 : ">\n\n");
3298 96 : }
3299 :
3300 :
3301 : static void
3302 96 : xsd_schema_element_end(StringInfo result)
3303 : {
3304 96 : appendStringInfoString(result, "</xsd:schema>");
3305 96 : }
3306 :
3307 :
3308 : static StringInfo
3309 18 : schema_to_xmlschema_internal(const char *schemaname, bool nulls,
3310 : bool tableforest, const char *targetns)
3311 : {
3312 : Oid nspid;
3313 : List *relid_list;
3314 : List *tupdesc_list;
3315 : ListCell *cell;
3316 : StringInfo result;
3317 :
3318 18 : result = makeStringInfo();
3319 :
3320 18 : nspid = LookupExplicitNamespace(schemaname, false);
3321 :
3322 18 : xsd_schema_element_start(result, targetns);
3323 :
3324 18 : SPI_connect();
3325 :
3326 18 : relid_list = schema_get_xml_visible_tables(nspid);
3327 :
3328 18 : tupdesc_list = NIL;
3329 54 : foreach(cell, relid_list)
3330 : {
3331 : Relation rel;
3332 :
3333 36 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3334 36 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3335 36 : table_close(rel, NoLock);
3336 : }
3337 :
3338 18 : appendStringInfoString(result,
3339 : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3340 :
3341 18 : appendStringInfoString(result,
3342 : map_sql_schema_to_xmlschema_types(nspid, relid_list,
3343 : nulls, tableforest, targetns));
3344 :
3345 18 : xsd_schema_element_end(result);
3346 :
3347 18 : SPI_finish();
3348 :
3349 18 : return result;
3350 : }
3351 :
3352 :
3353 : Datum
3354 12 : schema_to_xmlschema(PG_FUNCTION_ARGS)
3355 : {
3356 12 : Name name = PG_GETARG_NAME(0);
3357 12 : bool nulls = PG_GETARG_BOOL(1);
3358 12 : bool tableforest = PG_GETARG_BOOL(2);
3359 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3360 :
3361 12 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xmlschema_internal(NameStr(*name),
3362 : nulls, tableforest, targetns)));
3363 : }
3364 :
3365 :
3366 : Datum
3367 6 : schema_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3368 : {
3369 6 : Name name = PG_GETARG_NAME(0);
3370 6 : bool nulls = PG_GETARG_BOOL(1);
3371 6 : bool tableforest = PG_GETARG_BOOL(2);
3372 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3373 : char *schemaname;
3374 : Oid nspid;
3375 : StringInfo xmlschema;
3376 :
3377 6 : schemaname = NameStr(*name);
3378 6 : nspid = LookupExplicitNamespace(schemaname, false);
3379 :
3380 6 : xmlschema = schema_to_xmlschema_internal(schemaname, nulls,
3381 : tableforest, targetns);
3382 :
3383 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid,
3384 : xmlschema->data, nulls,
3385 : tableforest, targetns, true)));
3386 : }
3387 :
3388 :
3389 : /*
3390 : * Map SQL database to XML and/or XML Schema document; see SQL/XML:2008
3391 : * sections 9.16, 9.17.
3392 : */
3393 :
3394 : static StringInfo
3395 0 : database_to_xml_internal(const char *xmlschema, bool nulls,
3396 : bool tableforest, const char *targetns)
3397 : {
3398 : StringInfo result;
3399 : List *nspid_list;
3400 : ListCell *cell;
3401 : char *xmlcn;
3402 :
3403 0 : xmlcn = map_sql_identifier_to_xml_name(get_database_name(MyDatabaseId),
3404 : true, false);
3405 0 : result = makeStringInfo();
3406 :
3407 0 : xmldata_root_element_start(result, xmlcn, xmlschema, targetns, true);
3408 0 : appendStringInfoChar(result, '\n');
3409 :
3410 0 : if (xmlschema)
3411 0 : appendStringInfo(result, "%s\n\n", xmlschema);
3412 :
3413 0 : SPI_connect();
3414 :
3415 0 : nspid_list = database_get_xml_visible_schemas();
3416 :
3417 0 : foreach(cell, nspid_list)
3418 : {
3419 0 : Oid nspid = lfirst_oid(cell);
3420 : StringInfo subres;
3421 :
3422 0 : subres = schema_to_xml_internal(nspid, NULL, nulls,
3423 : tableforest, targetns, false);
3424 :
3425 0 : appendBinaryStringInfo(result, subres->data, subres->len);
3426 0 : appendStringInfoChar(result, '\n');
3427 : }
3428 :
3429 0 : SPI_finish();
3430 :
3431 0 : xmldata_root_element_end(result, xmlcn);
3432 :
3433 0 : return result;
3434 : }
3435 :
3436 :
3437 : Datum
3438 0 : database_to_xml(PG_FUNCTION_ARGS)
3439 : {
3440 0 : bool nulls = PG_GETARG_BOOL(0);
3441 0 : bool tableforest = PG_GETARG_BOOL(1);
3442 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3443 :
3444 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(NULL, nulls,
3445 : tableforest, targetns)));
3446 : }
3447 :
3448 :
3449 : static StringInfo
3450 0 : database_to_xmlschema_internal(bool nulls, bool tableforest,
3451 : const char *targetns)
3452 : {
3453 : List *relid_list;
3454 : List *nspid_list;
3455 : List *tupdesc_list;
3456 : ListCell *cell;
3457 : StringInfo result;
3458 :
3459 0 : result = makeStringInfo();
3460 :
3461 0 : xsd_schema_element_start(result, targetns);
3462 :
3463 0 : SPI_connect();
3464 :
3465 0 : relid_list = database_get_xml_visible_tables();
3466 0 : nspid_list = database_get_xml_visible_schemas();
3467 :
3468 0 : tupdesc_list = NIL;
3469 0 : foreach(cell, relid_list)
3470 : {
3471 : Relation rel;
3472 :
3473 0 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3474 0 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3475 0 : table_close(rel, NoLock);
3476 : }
3477 :
3478 0 : appendStringInfoString(result,
3479 : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3480 :
3481 0 : appendStringInfoString(result,
3482 : map_sql_catalog_to_xmlschema_types(nspid_list, nulls, tableforest, targetns));
3483 :
3484 0 : xsd_schema_element_end(result);
3485 :
3486 0 : SPI_finish();
3487 :
3488 0 : return result;
3489 : }
3490 :
3491 :
3492 : Datum
3493 0 : database_to_xmlschema(PG_FUNCTION_ARGS)
3494 : {
3495 0 : bool nulls = PG_GETARG_BOOL(0);
3496 0 : bool tableforest = PG_GETARG_BOOL(1);
3497 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3498 :
3499 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xmlschema_internal(nulls,
3500 : tableforest, targetns)));
3501 : }
3502 :
3503 :
3504 : Datum
3505 0 : database_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3506 : {
3507 0 : bool nulls = PG_GETARG_BOOL(0);
3508 0 : bool tableforest = PG_GETARG_BOOL(1);
3509 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3510 : StringInfo xmlschema;
3511 :
3512 0 : xmlschema = database_to_xmlschema_internal(nulls, tableforest, targetns);
3513 :
3514 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(xmlschema->data,
3515 : nulls, tableforest, targetns)));
3516 : }
3517 :
3518 :
3519 : /*
3520 : * Map a multi-part SQL name to an XML name; see SQL/XML:2008 section
3521 : * 9.2.
3522 : */
3523 : static char *
3524 384 : map_multipart_sql_identifier_to_xml_name(const char *a, const char *b, const char *c, const char *d)
3525 : {
3526 : StringInfoData result;
3527 :
3528 384 : initStringInfo(&result);
3529 :
3530 384 : if (a)
3531 384 : appendStringInfoString(&result,
3532 384 : map_sql_identifier_to_xml_name(a, true, true));
3533 384 : if (b)
3534 384 : appendStringInfo(&result, ".%s",
3535 : map_sql_identifier_to_xml_name(b, true, true));
3536 384 : if (c)
3537 384 : appendStringInfo(&result, ".%s",
3538 : map_sql_identifier_to_xml_name(c, true, true));
3539 384 : if (d)
3540 366 : appendStringInfo(&result, ".%s",
3541 : map_sql_identifier_to_xml_name(d, true, true));
3542 :
3543 384 : return result.data;
3544 : }
3545 :
3546 :
3547 : /*
3548 : * Map an SQL table to an XML Schema document; see SQL/XML:2008
3549 : * section 9.11.
3550 : *
3551 : * Map an SQL table to XML Schema data types; see SQL/XML:2008 section
3552 : * 9.9.
3553 : */
3554 : static const char *
3555 78 : map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid, bool nulls,
3556 : bool tableforest, const char *targetns)
3557 : {
3558 : int i;
3559 : char *xmltn;
3560 : char *tabletypename;
3561 : char *rowtypename;
3562 : StringInfoData result;
3563 :
3564 78 : initStringInfo(&result);
3565 :
3566 78 : if (OidIsValid(relid))
3567 : {
3568 : HeapTuple tuple;
3569 : Form_pg_class reltuple;
3570 :
3571 54 : tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relid));
3572 54 : if (!HeapTupleIsValid(tuple))
3573 0 : elog(ERROR, "cache lookup failed for relation %u", relid);
3574 54 : reltuple = (Form_pg_class) GETSTRUCT(tuple);
3575 :
3576 54 : xmltn = map_sql_identifier_to_xml_name(NameStr(reltuple->relname),
3577 : true, false);
3578 :
3579 54 : tabletypename = map_multipart_sql_identifier_to_xml_name("TableType",
3580 54 : get_database_name(MyDatabaseId),
3581 54 : get_namespace_name(reltuple->relnamespace),
3582 54 : NameStr(reltuple->relname));
3583 :
3584 54 : rowtypename = map_multipart_sql_identifier_to_xml_name("RowType",
3585 54 : get_database_name(MyDatabaseId),
3586 54 : get_namespace_name(reltuple->relnamespace),
3587 54 : NameStr(reltuple->relname));
3588 :
3589 54 : ReleaseSysCache(tuple);
3590 : }
3591 : else
3592 : {
3593 24 : if (tableforest)
3594 12 : xmltn = "row";
3595 : else
3596 12 : xmltn = "table";
3597 :
3598 24 : tabletypename = "TableType";
3599 24 : rowtypename = "RowType";
3600 : }
3601 :
3602 78 : xsd_schema_element_start(&result, targetns);
3603 :
3604 78 : appendStringInfoString(&result,
3605 78 : map_sql_typecoll_to_xmlschema_types(list_make1(tupdesc)));
3606 :
3607 78 : appendStringInfo(&result,
3608 : "<xsd:complexType name=\"%s\">\n"
3609 : " <xsd:sequence>\n",
3610 : rowtypename);
3611 :
3612 324 : for (i = 0; i < tupdesc->natts; i++)
3613 : {
3614 246 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3615 :
3616 246 : if (att->attisdropped)
3617 6 : continue;
3618 480 : appendStringInfo(&result,
3619 : " <xsd:element name=\"%s\" type=\"%s\"%s></xsd:element>\n",
3620 240 : map_sql_identifier_to_xml_name(NameStr(att->attname),
3621 : true, false),
3622 : map_sql_type_to_xml_name(att->atttypid, -1),
3623 : nulls ? " nillable=\"true\"" : " minOccurs=\"0\"");
3624 : }
3625 :
3626 78 : appendStringInfoString(&result,
3627 : " </xsd:sequence>\n"
3628 : "</xsd:complexType>\n\n");
3629 :
3630 78 : if (!tableforest)
3631 : {
3632 42 : appendStringInfo(&result,
3633 : "<xsd:complexType name=\"%s\">\n"
3634 : " <xsd:sequence>\n"
3635 : " <xsd:element name=\"row\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n"
3636 : " </xsd:sequence>\n"
3637 : "</xsd:complexType>\n\n",
3638 : tabletypename, rowtypename);
3639 :
3640 42 : appendStringInfo(&result,
3641 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3642 : xmltn, tabletypename);
3643 : }
3644 : else
3645 36 : appendStringInfo(&result,
3646 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3647 : xmltn, rowtypename);
3648 :
3649 78 : xsd_schema_element_end(&result);
3650 :
3651 78 : return result.data;
3652 : }
3653 :
3654 :
3655 : /*
3656 : * Map an SQL schema to XML Schema data types; see SQL/XML:2008
3657 : * section 9.12.
3658 : */
3659 : static const char *
3660 18 : map_sql_schema_to_xmlschema_types(Oid nspid, List *relid_list, bool nulls,
3661 : bool tableforest, const char *targetns)
3662 : {
3663 : char *dbname;
3664 : char *nspname;
3665 : char *xmlsn;
3666 : char *schematypename;
3667 : StringInfoData result;
3668 : ListCell *cell;
3669 :
3670 18 : dbname = get_database_name(MyDatabaseId);
3671 18 : nspname = get_namespace_name(nspid);
3672 :
3673 18 : initStringInfo(&result);
3674 :
3675 18 : xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3676 :
3677 18 : schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3678 : dbname,
3679 : nspname,
3680 : NULL);
3681 :
3682 18 : appendStringInfo(&result,
3683 : "<xsd:complexType name=\"%s\">\n", schematypename);
3684 18 : if (!tableforest)
3685 6 : appendStringInfoString(&result,
3686 : " <xsd:all>\n");
3687 : else
3688 12 : appendStringInfoString(&result,
3689 : " <xsd:sequence>\n");
3690 :
3691 54 : foreach(cell, relid_list)
3692 : {
3693 36 : Oid relid = lfirst_oid(cell);
3694 36 : char *relname = get_rel_name(relid);
3695 36 : char *xmltn = map_sql_identifier_to_xml_name(relname, true, false);
3696 36 : char *tabletypename = map_multipart_sql_identifier_to_xml_name(tableforest ? "RowType" : "TableType",
3697 : dbname,
3698 : nspname,
3699 : relname);
3700 :
3701 36 : if (!tableforest)
3702 12 : appendStringInfo(&result,
3703 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3704 : xmltn, tabletypename);
3705 : else
3706 24 : appendStringInfo(&result,
3707 : " <xsd:element name=\"%s\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n",
3708 : xmltn, tabletypename);
3709 : }
3710 :
3711 18 : if (!tableforest)
3712 6 : appendStringInfoString(&result,
3713 : " </xsd:all>\n");
3714 : else
3715 12 : appendStringInfoString(&result,
3716 : " </xsd:sequence>\n");
3717 18 : appendStringInfoString(&result,
3718 : "</xsd:complexType>\n\n");
3719 :
3720 18 : appendStringInfo(&result,
3721 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3722 : xmlsn, schematypename);
3723 :
3724 18 : return result.data;
3725 : }
3726 :
3727 :
3728 : /*
3729 : * Map an SQL catalog to XML Schema data types; see SQL/XML:2008
3730 : * section 9.15.
3731 : */
3732 : static const char *
3733 0 : map_sql_catalog_to_xmlschema_types(List *nspid_list, bool nulls,
3734 : bool tableforest, const char *targetns)
3735 : {
3736 : char *dbname;
3737 : char *xmlcn;
3738 : char *catalogtypename;
3739 : StringInfoData result;
3740 : ListCell *cell;
3741 :
3742 0 : dbname = get_database_name(MyDatabaseId);
3743 :
3744 0 : initStringInfo(&result);
3745 :
3746 0 : xmlcn = map_sql_identifier_to_xml_name(dbname, true, false);
3747 :
3748 0 : catalogtypename = map_multipart_sql_identifier_to_xml_name("CatalogType",
3749 : dbname,
3750 : NULL,
3751 : NULL);
3752 :
3753 0 : appendStringInfo(&result,
3754 : "<xsd:complexType name=\"%s\">\n", catalogtypename);
3755 0 : appendStringInfoString(&result,
3756 : " <xsd:all>\n");
3757 :
3758 0 : foreach(cell, nspid_list)
3759 : {
3760 0 : Oid nspid = lfirst_oid(cell);
3761 0 : char *nspname = get_namespace_name(nspid);
3762 0 : char *xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3763 0 : char *schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3764 : dbname,
3765 : nspname,
3766 : NULL);
3767 :
3768 0 : appendStringInfo(&result,
3769 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3770 : xmlsn, schematypename);
3771 : }
3772 :
3773 0 : appendStringInfoString(&result,
3774 : " </xsd:all>\n");
3775 0 : appendStringInfoString(&result,
3776 : "</xsd:complexType>\n\n");
3777 :
3778 0 : appendStringInfo(&result,
3779 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3780 : xmlcn, catalogtypename);
3781 :
3782 0 : return result.data;
3783 : }
3784 :
3785 :
3786 : /*
3787 : * Map an SQL data type to an XML name; see SQL/XML:2008 section 9.4.
3788 : */
3789 : static const char *
3790 810 : map_sql_type_to_xml_name(Oid typeoid, int typmod)
3791 : {
3792 : StringInfoData result;
3793 :
3794 810 : initStringInfo(&result);
3795 :
3796 810 : switch (typeoid)
3797 : {
3798 30 : case BPCHAROID:
3799 30 : if (typmod == -1)
3800 30 : appendStringInfoString(&result, "CHAR");
3801 : else
3802 0 : appendStringInfo(&result, "CHAR_%d", typmod - VARHDRSZ);
3803 30 : break;
3804 54 : case VARCHAROID:
3805 54 : if (typmod == -1)
3806 54 : appendStringInfoString(&result, "VARCHAR");
3807 : else
3808 0 : appendStringInfo(&result, "VARCHAR_%d", typmod - VARHDRSZ);
3809 54 : break;
3810 30 : case NUMERICOID:
3811 30 : if (typmod == -1)
3812 30 : appendStringInfoString(&result, "NUMERIC");
3813 : else
3814 0 : appendStringInfo(&result, "NUMERIC_%d_%d",
3815 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
3816 0 : (typmod - VARHDRSZ) & 0xffff);
3817 30 : break;
3818 174 : case INT4OID:
3819 174 : appendStringInfoString(&result, "INTEGER");
3820 174 : break;
3821 30 : case INT2OID:
3822 30 : appendStringInfoString(&result, "SMALLINT");
3823 30 : break;
3824 30 : case INT8OID:
3825 30 : appendStringInfoString(&result, "BIGINT");
3826 30 : break;
3827 30 : case FLOAT4OID:
3828 30 : appendStringInfoString(&result, "REAL");
3829 30 : break;
3830 0 : case FLOAT8OID:
3831 0 : appendStringInfoString(&result, "DOUBLE");
3832 0 : break;
3833 30 : case BOOLOID:
3834 30 : appendStringInfoString(&result, "BOOLEAN");
3835 30 : break;
3836 30 : case TIMEOID:
3837 30 : if (typmod == -1)
3838 30 : appendStringInfoString(&result, "TIME");
3839 : else
3840 0 : appendStringInfo(&result, "TIME_%d", typmod);
3841 30 : break;
3842 30 : case TIMETZOID:
3843 30 : if (typmod == -1)
3844 30 : appendStringInfoString(&result, "TIME_WTZ");
3845 : else
3846 0 : appendStringInfo(&result, "TIME_WTZ_%d", typmod);
3847 30 : break;
3848 30 : case TIMESTAMPOID:
3849 30 : if (typmod == -1)
3850 30 : appendStringInfoString(&result, "TIMESTAMP");
3851 : else
3852 0 : appendStringInfo(&result, "TIMESTAMP_%d", typmod);
3853 30 : break;
3854 30 : case TIMESTAMPTZOID:
3855 30 : if (typmod == -1)
3856 30 : appendStringInfoString(&result, "TIMESTAMP_WTZ");
3857 : else
3858 0 : appendStringInfo(&result, "TIMESTAMP_WTZ_%d", typmod);
3859 30 : break;
3860 30 : case DATEOID:
3861 30 : appendStringInfoString(&result, "DATE");
3862 30 : break;
3863 30 : case XMLOID:
3864 30 : appendStringInfoString(&result, "XML");
3865 30 : break;
3866 222 : default:
3867 : {
3868 : HeapTuple tuple;
3869 : Form_pg_type typtuple;
3870 :
3871 222 : tuple = SearchSysCache1(TYPEOID, ObjectIdGetDatum(typeoid));
3872 222 : if (!HeapTupleIsValid(tuple))
3873 0 : elog(ERROR, "cache lookup failed for type %u", typeoid);
3874 222 : typtuple = (Form_pg_type) GETSTRUCT(tuple);
3875 :
3876 222 : appendStringInfoString(&result,
3877 222 : map_multipart_sql_identifier_to_xml_name((typtuple->typtype == TYPTYPE_DOMAIN) ? "Domain" : "UDT",
3878 222 : get_database_name(MyDatabaseId),
3879 222 : get_namespace_name(typtuple->typnamespace),
3880 222 : NameStr(typtuple->typname)));
3881 :
3882 222 : ReleaseSysCache(tuple);
3883 : }
3884 : }
3885 :
3886 810 : return result.data;
3887 : }
3888 :
3889 :
3890 : /*
3891 : * Map a collection of SQL data types to XML Schema data types; see
3892 : * SQL/XML:2008 section 9.7.
3893 : */
3894 : static const char *
3895 96 : map_sql_typecoll_to_xmlschema_types(List *tupdesc_list)
3896 : {
3897 96 : List *uniquetypes = NIL;
3898 : int i;
3899 : StringInfoData result;
3900 : ListCell *cell0;
3901 :
3902 : /* extract all column types used in the set of TupleDescs */
3903 210 : foreach(cell0, tupdesc_list)
3904 : {
3905 114 : TupleDesc tupdesc = (TupleDesc) lfirst(cell0);
3906 :
3907 702 : for (i = 0; i < tupdesc->natts; i++)
3908 : {
3909 588 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3910 :
3911 588 : if (att->attisdropped)
3912 24 : continue;
3913 564 : uniquetypes = list_append_unique_oid(uniquetypes, att->atttypid);
3914 : }
3915 : }
3916 :
3917 : /* add base types of domains */
3918 642 : foreach(cell0, uniquetypes)
3919 : {
3920 546 : Oid typid = lfirst_oid(cell0);
3921 546 : Oid basetypid = getBaseType(typid);
3922 :
3923 546 : if (basetypid != typid)
3924 24 : uniquetypes = list_append_unique_oid(uniquetypes, basetypid);
3925 : }
3926 :
3927 : /* Convert to textual form */
3928 96 : initStringInfo(&result);
3929 :
3930 642 : foreach(cell0, uniquetypes)
3931 : {
3932 546 : appendStringInfo(&result, "%s\n",
3933 : map_sql_type_to_xmlschema_type(lfirst_oid(cell0),
3934 : -1));
3935 : }
3936 :
3937 96 : return result.data;
3938 : }
3939 :
3940 :
3941 : /*
3942 : * Map an SQL data type to a named XML Schema data type; see
3943 : * SQL/XML:2008 sections 9.5 and 9.6.
3944 : *
3945 : * (The distinction between 9.5 and 9.6 is basically that 9.6 adds
3946 : * a name attribute, which this function does. The name-less version
3947 : * 9.5 doesn't appear to be required anywhere.)
3948 : */
3949 : static const char *
3950 546 : map_sql_type_to_xmlschema_type(Oid typeoid, int typmod)
3951 : {
3952 : StringInfoData result;
3953 546 : const char *typename = map_sql_type_to_xml_name(typeoid, typmod);
3954 :
3955 546 : initStringInfo(&result);
3956 :
3957 546 : if (typeoid == XMLOID)
3958 : {
3959 24 : appendStringInfoString(&result,
3960 : "<xsd:complexType mixed=\"true\">\n"
3961 : " <xsd:sequence>\n"
3962 : " <xsd:any name=\"element\" minOccurs=\"0\" maxOccurs=\"unbounded\" processContents=\"skip\"/>\n"
3963 : " </xsd:sequence>\n"
3964 : "</xsd:complexType>\n");
3965 : }
3966 : else
3967 : {
3968 522 : appendStringInfo(&result,
3969 : "<xsd:simpleType name=\"%s\">\n", typename);
3970 :
3971 522 : switch (typeoid)
3972 : {
3973 138 : case BPCHAROID:
3974 : case VARCHAROID:
3975 : case TEXTOID:
3976 138 : appendStringInfoString(&result,
3977 : " <xsd:restriction base=\"xsd:string\">\n");
3978 138 : if (typmod != -1)
3979 0 : appendStringInfo(&result,
3980 : " <xsd:maxLength value=\"%d\"/>\n",
3981 : typmod - VARHDRSZ);
3982 138 : appendStringInfoString(&result, " </xsd:restriction>\n");
3983 138 : break;
3984 :
3985 24 : case BYTEAOID:
3986 24 : appendStringInfo(&result,
3987 : " <xsd:restriction base=\"xsd:%s\">\n"
3988 : " </xsd:restriction>\n",
3989 24 : xmlbinary == XMLBINARY_BASE64 ? "base64Binary" : "hexBinary");
3990 24 : break;
3991 :
3992 24 : case NUMERICOID:
3993 24 : if (typmod != -1)
3994 0 : appendStringInfo(&result,
3995 : " <xsd:restriction base=\"xsd:decimal\">\n"
3996 : " <xsd:totalDigits value=\"%d\"/>\n"
3997 : " <xsd:fractionDigits value=\"%d\"/>\n"
3998 : " </xsd:restriction>\n",
3999 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
4000 0 : (typmod - VARHDRSZ) & 0xffff);
4001 24 : break;
4002 :
4003 24 : case INT2OID:
4004 24 : appendStringInfo(&result,
4005 : " <xsd:restriction base=\"xsd:short\">\n"
4006 : " <xsd:maxInclusive value=\"%d\"/>\n"
4007 : " <xsd:minInclusive value=\"%d\"/>\n"
4008 : " </xsd:restriction>\n",
4009 : SHRT_MAX, SHRT_MIN);
4010 24 : break;
4011 :
4012 96 : case INT4OID:
4013 96 : appendStringInfo(&result,
4014 : " <xsd:restriction base=\"xsd:int\">\n"
4015 : " <xsd:maxInclusive value=\"%d\"/>\n"
4016 : " <xsd:minInclusive value=\"%d\"/>\n"
4017 : " </xsd:restriction>\n",
4018 : INT_MAX, INT_MIN);
4019 96 : break;
4020 :
4021 24 : case INT8OID:
4022 24 : appendStringInfo(&result,
4023 : " <xsd:restriction base=\"xsd:long\">\n"
4024 : " <xsd:maxInclusive value=\"" INT64_FORMAT "\"/>\n"
4025 : " <xsd:minInclusive value=\"" INT64_FORMAT "\"/>\n"
4026 : " </xsd:restriction>\n",
4027 : PG_INT64_MAX,
4028 : PG_INT64_MIN);
4029 24 : break;
4030 :
4031 24 : case FLOAT4OID:
4032 24 : appendStringInfoString(&result,
4033 : " <xsd:restriction base=\"xsd:float\"></xsd:restriction>\n");
4034 24 : break;
4035 :
4036 0 : case FLOAT8OID:
4037 0 : appendStringInfoString(&result,
4038 : " <xsd:restriction base=\"xsd:double\"></xsd:restriction>\n");
4039 0 : break;
4040 :
4041 24 : case BOOLOID:
4042 24 : appendStringInfoString(&result,
4043 : " <xsd:restriction base=\"xsd:boolean\"></xsd:restriction>\n");
4044 24 : break;
4045 :
4046 48 : case TIMEOID:
4047 : case TIMETZOID:
4048 : {
4049 48 : const char *tz = (typeoid == TIMETZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
4050 :
4051 48 : if (typmod == -1)
4052 48 : appendStringInfo(&result,
4053 : " <xsd:restriction base=\"xsd:time\">\n"
4054 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}(.\\p{Nd}+)?%s\"/>\n"
4055 : " </xsd:restriction>\n", tz);
4056 0 : else if (typmod == 0)
4057 0 : appendStringInfo(&result,
4058 : " <xsd:restriction base=\"xsd:time\">\n"
4059 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
4060 : " </xsd:restriction>\n", tz);
4061 : else
4062 0 : appendStringInfo(&result,
4063 : " <xsd:restriction base=\"xsd:time\">\n"
4064 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}.\\p{Nd}{%d}%s\"/>\n"
4065 : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
4066 48 : break;
4067 : }
4068 :
4069 48 : case TIMESTAMPOID:
4070 : case TIMESTAMPTZOID:
4071 : {
4072 48 : const char *tz = (typeoid == TIMESTAMPTZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
4073 :
4074 48 : if (typmod == -1)
4075 48 : appendStringInfo(&result,
4076 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4077 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}(.\\p{Nd}+)?%s\"/>\n"
4078 : " </xsd:restriction>\n", tz);
4079 0 : else if (typmod == 0)
4080 0 : appendStringInfo(&result,
4081 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4082 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
4083 : " </xsd:restriction>\n", tz);
4084 : else
4085 0 : appendStringInfo(&result,
4086 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4087 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}.\\p{Nd}{%d}%s\"/>\n"
4088 : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
4089 48 : break;
4090 : }
4091 :
4092 24 : case DATEOID:
4093 24 : appendStringInfoString(&result,
4094 : " <xsd:restriction base=\"xsd:date\">\n"
4095 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}\"/>\n"
4096 : " </xsd:restriction>\n");
4097 24 : break;
4098 :
4099 24 : default:
4100 24 : if (get_typtype(typeoid) == TYPTYPE_DOMAIN)
4101 : {
4102 : Oid base_typeoid;
4103 24 : int32 base_typmod = -1;
4104 :
4105 24 : base_typeoid = getBaseTypeAndTypmod(typeoid, &base_typmod);
4106 :
4107 24 : appendStringInfo(&result,
4108 : " <xsd:restriction base=\"%s\"/>\n",
4109 : map_sql_type_to_xml_name(base_typeoid, base_typmod));
4110 : }
4111 24 : break;
4112 : }
4113 522 : appendStringInfoString(&result, "</xsd:simpleType>\n");
4114 : }
4115 :
4116 546 : return result.data;
4117 : }
4118 :
4119 :
4120 : /*
4121 : * Map an SQL row to an XML element, taking the row from the active
4122 : * SPI cursor. See also SQL/XML:2008 section 9.10.
4123 : */
4124 : static void
4125 312 : SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result, char *tablename,
4126 : bool nulls, bool tableforest,
4127 : const char *targetns, bool top_level)
4128 : {
4129 : int i;
4130 : char *xmltn;
4131 :
4132 312 : if (tablename)
4133 228 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
4134 : else
4135 : {
4136 84 : if (tableforest)
4137 36 : xmltn = "row";
4138 : else
4139 48 : xmltn = "table";
4140 : }
4141 :
4142 312 : if (tableforest)
4143 162 : xmldata_root_element_start(result, xmltn, NULL, targetns, top_level);
4144 : else
4145 150 : appendStringInfoString(result, "<row>\n");
4146 :
4147 1272 : for (i = 1; i <= SPI_tuptable->tupdesc->natts; i++)
4148 : {
4149 : char *colname;
4150 : Datum colval;
4151 : bool isnull;
4152 :
4153 960 : colname = map_sql_identifier_to_xml_name(SPI_fname(SPI_tuptable->tupdesc, i),
4154 : true, false);
4155 960 : colval = SPI_getbinval(SPI_tuptable->vals[rownum],
4156 960 : SPI_tuptable->tupdesc,
4157 : i,
4158 : &isnull);
4159 960 : if (isnull)
4160 : {
4161 114 : if (nulls)
4162 60 : appendStringInfo(result, " <%s xsi:nil=\"true\"/>\n", colname);
4163 : }
4164 : else
4165 846 : appendStringInfo(result, " <%s>%s</%s>\n",
4166 : colname,
4167 : map_sql_value_to_xml_value(colval,
4168 846 : SPI_gettypeid(SPI_tuptable->tupdesc, i), true),
4169 : colname);
4170 : }
4171 :
4172 312 : if (tableforest)
4173 : {
4174 162 : xmldata_root_element_end(result, xmltn);
4175 162 : appendStringInfoChar(result, '\n');
4176 : }
4177 : else
4178 150 : appendStringInfoString(result, "</row>\n\n");
4179 312 : }
4180 :
4181 :
4182 : /*
4183 : * XPath related functions
4184 : */
4185 :
4186 : #ifdef USE_LIBXML
4187 :
4188 : /*
4189 : * Convert XML node to text.
4190 : *
4191 : * For attribute and text nodes, return the escaped text. For anything else,
4192 : * dump the whole subtree.
4193 : */
4194 : static text *
4195 192 : xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt)
4196 : {
4197 192 : xmltype *result = NULL;
4198 :
4199 192 : if (cur->type != XML_ATTRIBUTE_NODE && cur->type != XML_TEXT_NODE)
4200 162 : {
4201 162 : void (*volatile nodefree) (xmlNodePtr) = NULL;
4202 162 : volatile xmlBufferPtr buf = NULL;
4203 162 : volatile xmlNodePtr cur_copy = NULL;
4204 :
4205 162 : PG_TRY();
4206 : {
4207 : int bytes;
4208 :
4209 162 : buf = xmlBufferCreate();
4210 162 : if (buf == NULL || xmlerrcxt->err_occurred)
4211 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4212 : "could not allocate xmlBuffer");
4213 :
4214 : /*
4215 : * Produce a dump of the node that we can serialize. xmlNodeDump
4216 : * does that, but the result of that function won't contain
4217 : * namespace definitions from ancestor nodes, so we first do a
4218 : * xmlCopyNode() which duplicates the node along with its required
4219 : * namespace definitions.
4220 : *
4221 : * Some old libxml2 versions such as 2.7.6 produce partially
4222 : * broken XML_DOCUMENT_NODE nodes (unset content field) when
4223 : * copying them. xmlNodeDump of such a node works fine, but
4224 : * xmlFreeNode crashes; set us up to call xmlFreeDoc instead.
4225 : */
4226 162 : cur_copy = xmlCopyNode(cur, 1);
4227 162 : if (cur_copy == NULL || xmlerrcxt->err_occurred)
4228 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4229 : "could not copy node");
4230 324 : nodefree = (cur_copy->type == XML_DOCUMENT_NODE) ?
4231 162 : (void (*) (xmlNodePtr)) xmlFreeDoc : xmlFreeNode;
4232 :
4233 162 : bytes = xmlNodeDump(buf, NULL, cur_copy, 0, 0);
4234 162 : if (bytes == -1 || xmlerrcxt->err_occurred)
4235 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4236 : "could not dump node");
4237 :
4238 162 : result = xmlBuffer_to_xmltype(buf);
4239 : }
4240 0 : PG_FINALLY();
4241 : {
4242 162 : if (nodefree)
4243 162 : nodefree(cur_copy);
4244 162 : if (buf)
4245 162 : xmlBufferFree(buf);
4246 : }
4247 162 : PG_END_TRY();
4248 : }
4249 : else
4250 : {
4251 30 : volatile xmlChar *str = NULL;
4252 :
4253 30 : PG_TRY();
4254 : {
4255 : char *escaped;
4256 :
4257 30 : str = xmlXPathCastNodeToString(cur);
4258 30 : if (str == NULL || xmlerrcxt->err_occurred)
4259 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4260 : "could not allocate xmlChar");
4261 :
4262 : /* Here we rely on XML having the same representation as TEXT */
4263 30 : escaped = escape_xml((char *) str);
4264 :
4265 30 : result = (xmltype *) cstring_to_text(escaped);
4266 30 : pfree(escaped);
4267 : }
4268 0 : PG_FINALLY();
4269 : {
4270 30 : if (str)
4271 30 : xmlFree((xmlChar *) str);
4272 : }
4273 30 : PG_END_TRY();
4274 : }
4275 :
4276 192 : return result;
4277 : }
4278 :
4279 : /*
4280 : * Convert an XML XPath object (the result of evaluating an XPath expression)
4281 : * to an array of xml values, which are appended to astate. The function
4282 : * result value is the number of elements in the array.
4283 : *
4284 : * If "astate" is NULL then we don't generate the array value, but we still
4285 : * return the number of elements it would have had.
4286 : *
4287 : * Nodesets are converted to an array containing the nodes' textual
4288 : * representations. Primitive values (float, double, string) are converted
4289 : * to a single-element array containing the value's string representation.
4290 : */
4291 : static int
4292 540 : xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
4293 : ArrayBuildState *astate,
4294 : PgXmlErrorContext *xmlerrcxt)
4295 : {
4296 540 : int result = 0;
4297 : Datum datum;
4298 : Oid datumtype;
4299 : char *result_str;
4300 :
4301 540 : switch (xpathobj->type)
4302 : {
4303 498 : case XPATH_NODESET:
4304 498 : if (xpathobj->nodesetval != NULL)
4305 : {
4306 354 : result = xpathobj->nodesetval->nodeNr;
4307 354 : if (astate != NULL)
4308 : {
4309 : int i;
4310 :
4311 168 : for (i = 0; i < result; i++)
4312 : {
4313 90 : datum = PointerGetDatum(xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
4314 : xmlerrcxt));
4315 90 : (void) accumArrayResult(astate, datum, false,
4316 : XMLOID, CurrentMemoryContext);
4317 : }
4318 : }
4319 : }
4320 498 : return result;
4321 :
4322 12 : case XPATH_BOOLEAN:
4323 12 : if (astate == NULL)
4324 0 : return 1;
4325 12 : datum = BoolGetDatum(xpathobj->boolval);
4326 12 : datumtype = BOOLOID;
4327 12 : break;
4328 :
4329 18 : case XPATH_NUMBER:
4330 18 : if (astate == NULL)
4331 12 : return 1;
4332 6 : datum = Float8GetDatum(xpathobj->floatval);
4333 6 : datumtype = FLOAT8OID;
4334 6 : break;
4335 :
4336 12 : case XPATH_STRING:
4337 12 : if (astate == NULL)
4338 0 : return 1;
4339 12 : datum = CStringGetDatum((char *) xpathobj->stringval);
4340 12 : datumtype = CSTRINGOID;
4341 12 : break;
4342 :
4343 0 : default:
4344 0 : elog(ERROR, "xpath expression result type %d is unsupported",
4345 : xpathobj->type);
4346 : return 0; /* keep compiler quiet */
4347 : }
4348 :
4349 : /* Common code for scalar-value cases */
4350 30 : result_str = map_sql_value_to_xml_value(datum, datumtype, true);
4351 30 : datum = PointerGetDatum(cstring_to_xmltype(result_str));
4352 30 : (void) accumArrayResult(astate, datum, false,
4353 : XMLOID, CurrentMemoryContext);
4354 30 : return 1;
4355 : }
4356 :
4357 :
4358 : /*
4359 : * Common code for xpath() and xmlexists()
4360 : *
4361 : * Evaluate XPath expression and return number of nodes in res_nitems
4362 : * and array of XML values in astate. Either of those pointers can be
4363 : * NULL if the corresponding result isn't wanted.
4364 : *
4365 : * It is up to the user to ensure that the XML passed is in fact
4366 : * an XML document - XPath doesn't work easily on fragments without
4367 : * a context node being known.
4368 : */
4369 : static void
4370 558 : xpath_internal(text *xpath_expr_text, xmltype *data, ArrayType *namespaces,
4371 : int *res_nitems, ArrayBuildState *astate)
4372 : {
4373 : PgXmlErrorContext *xmlerrcxt;
4374 558 : volatile xmlParserCtxtPtr ctxt = NULL;
4375 558 : volatile xmlDocPtr doc = NULL;
4376 558 : volatile xmlXPathContextPtr xpathctx = NULL;
4377 558 : volatile xmlXPathCompExprPtr xpathcomp = NULL;
4378 558 : volatile xmlXPathObjectPtr xpathobj = NULL;
4379 : char *datastr;
4380 : int32 len;
4381 : int32 xpath_len;
4382 : xmlChar *string;
4383 : xmlChar *xpath_expr;
4384 558 : size_t xmldecl_len = 0;
4385 : int i;
4386 : int ndim;
4387 : Datum *ns_names_uris;
4388 : bool *ns_names_uris_nulls;
4389 : int ns_count;
4390 :
4391 : /*
4392 : * Namespace mappings are passed as text[]. If an empty array is passed
4393 : * (ndim = 0, "0-dimensional"), then there are no namespace mappings.
4394 : * Else, a 2-dimensional array with length of the second axis being equal
4395 : * to 2 should be passed, i.e., every subarray contains 2 elements, the
4396 : * first element defining the name, the second one the URI. Example:
4397 : * ARRAY[ARRAY['myns', 'http://example.com'], ARRAY['myns2',
4398 : * 'http://example2.com']].
4399 : */
4400 558 : ndim = namespaces ? ARR_NDIM(namespaces) : 0;
4401 558 : if (ndim != 0)
4402 : {
4403 : int *dims;
4404 :
4405 126 : dims = ARR_DIMS(namespaces);
4406 :
4407 126 : if (ndim != 2 || dims[1] != 2)
4408 0 : ereport(ERROR,
4409 : (errcode(ERRCODE_DATA_EXCEPTION),
4410 : errmsg("invalid array for XML namespace mapping"),
4411 : errdetail("The array must be two-dimensional with length of the second axis equal to 2.")));
4412 :
4413 : Assert(ARR_ELEMTYPE(namespaces) == TEXTOID);
4414 :
4415 126 : deconstruct_array_builtin(namespaces, TEXTOID,
4416 : &ns_names_uris, &ns_names_uris_nulls,
4417 : &ns_count);
4418 :
4419 : Assert((ns_count % 2) == 0); /* checked above */
4420 126 : ns_count /= 2; /* count pairs only */
4421 : }
4422 : else
4423 : {
4424 432 : ns_names_uris = NULL;
4425 432 : ns_names_uris_nulls = NULL;
4426 432 : ns_count = 0;
4427 : }
4428 :
4429 558 : datastr = VARDATA(data);
4430 558 : len = VARSIZE(data) - VARHDRSZ;
4431 558 : xpath_len = VARSIZE_ANY_EXHDR(xpath_expr_text);
4432 558 : if (xpath_len == 0)
4433 6 : ereport(ERROR,
4434 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4435 : errmsg("empty XPath expression")));
4436 :
4437 552 : string = pg_xmlCharStrndup(datastr, len);
4438 552 : xpath_expr = pg_xmlCharStrndup(VARDATA_ANY(xpath_expr_text), xpath_len);
4439 :
4440 : /*
4441 : * In a UTF8 database, skip any xml declaration, which might assert
4442 : * another encoding. Ignore parse_xml_decl() failure, letting
4443 : * xmlCtxtReadMemory() report parse errors. Documentation disclaims
4444 : * xpath() support for non-ASCII data in non-UTF8 databases, so leave
4445 : * those scenarios bug-compatible with historical behavior.
4446 : */
4447 552 : if (GetDatabaseEncoding() == PG_UTF8)
4448 552 : parse_xml_decl(string, &xmldecl_len, NULL, NULL, NULL);
4449 :
4450 552 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4451 :
4452 552 : PG_TRY();
4453 : {
4454 552 : xmlInitParser();
4455 :
4456 : /*
4457 : * redundant XML parsing (two parsings for the same value during one
4458 : * command execution are possible)
4459 : */
4460 552 : ctxt = xmlNewParserCtxt();
4461 552 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4462 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4463 : "could not allocate parser context");
4464 1104 : doc = xmlCtxtReadMemory(ctxt, (char *) string + xmldecl_len,
4465 552 : len - xmldecl_len, NULL, NULL, 0);
4466 552 : if (doc == NULL || xmlerrcxt->err_occurred)
4467 12 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4468 : "could not parse XML document");
4469 540 : xpathctx = xmlXPathNewContext(doc);
4470 540 : if (xpathctx == NULL || xmlerrcxt->err_occurred)
4471 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4472 : "could not allocate XPath context");
4473 540 : xpathctx->node = (xmlNodePtr) doc;
4474 :
4475 : /* register namespaces, if any */
4476 540 : if (ns_count > 0)
4477 : {
4478 252 : for (i = 0; i < ns_count; i++)
4479 : {
4480 : char *ns_name;
4481 : char *ns_uri;
4482 :
4483 126 : if (ns_names_uris_nulls[i * 2] ||
4484 126 : ns_names_uris_nulls[i * 2 + 1])
4485 0 : ereport(ERROR,
4486 : (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
4487 : errmsg("neither namespace name nor URI may be null")));
4488 126 : ns_name = TextDatumGetCString(ns_names_uris[i * 2]);
4489 126 : ns_uri = TextDatumGetCString(ns_names_uris[i * 2 + 1]);
4490 126 : if (xmlXPathRegisterNs(xpathctx,
4491 : (xmlChar *) ns_name,
4492 : (xmlChar *) ns_uri) != 0)
4493 0 : ereport(ERROR, /* is this an internal error??? */
4494 : (errmsg("could not register XML namespace with name \"%s\" and URI \"%s\"",
4495 : ns_name, ns_uri)));
4496 : }
4497 : }
4498 :
4499 : /*
4500 : * Note: here and elsewhere, be careful to use xmlXPathCtxtCompile not
4501 : * xmlXPathCompile. In libxml2 2.13.3 and older, the latter function
4502 : * fails to defend itself against recursion-to-stack-overflow. See
4503 : * https://gitlab.gnome.org/GNOME/libxml2/-/issues/799
4504 : */
4505 540 : xpathcomp = xmlXPathCtxtCompile(xpathctx, xpath_expr);
4506 540 : if (xpathcomp == NULL || xmlerrcxt->err_occurred)
4507 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4508 : "invalid XPath expression");
4509 :
4510 : /*
4511 : * Version 2.6.27 introduces a function named
4512 : * xmlXPathCompiledEvalToBoolean, which would be enough for xmlexists,
4513 : * but we can derive the existence by whether any nodes are returned,
4514 : * thereby preventing a library version upgrade and keeping the code
4515 : * the same.
4516 : */
4517 540 : xpathobj = xmlXPathCompiledEval(xpathcomp, xpathctx);
4518 540 : if (xpathobj == NULL || xmlerrcxt->err_occurred)
4519 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4520 : "could not create XPath object");
4521 :
4522 : /*
4523 : * Extract the results as requested.
4524 : */
4525 540 : if (res_nitems != NULL)
4526 432 : *res_nitems = xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4527 : else
4528 108 : (void) xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4529 : }
4530 12 : PG_CATCH();
4531 : {
4532 12 : if (xpathobj)
4533 0 : xmlXPathFreeObject(xpathobj);
4534 12 : if (xpathcomp)
4535 0 : xmlXPathFreeCompExpr(xpathcomp);
4536 12 : if (xpathctx)
4537 0 : xmlXPathFreeContext(xpathctx);
4538 12 : if (doc)
4539 12 : xmlFreeDoc(doc);
4540 12 : if (ctxt)
4541 12 : xmlFreeParserCtxt(ctxt);
4542 :
4543 12 : pg_xml_done(xmlerrcxt, true);
4544 :
4545 12 : PG_RE_THROW();
4546 : }
4547 540 : PG_END_TRY();
4548 :
4549 540 : xmlXPathFreeObject(xpathobj);
4550 540 : xmlXPathFreeCompExpr(xpathcomp);
4551 540 : xmlXPathFreeContext(xpathctx);
4552 540 : xmlFreeDoc(doc);
4553 540 : xmlFreeParserCtxt(ctxt);
4554 :
4555 540 : pg_xml_done(xmlerrcxt, false);
4556 540 : }
4557 : #endif /* USE_LIBXML */
4558 :
4559 : /*
4560 : * Evaluate XPath expression and return array of XML values.
4561 : *
4562 : * As we have no support of XQuery sequences yet, this function seems
4563 : * to be the most useful one (array of XML functions plays a role of
4564 : * some kind of substitution for XQuery sequences).
4565 : */
4566 : Datum
4567 126 : xpath(PG_FUNCTION_ARGS)
4568 : {
4569 : #ifdef USE_LIBXML
4570 126 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4571 126 : xmltype *data = PG_GETARG_XML_P(1);
4572 126 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4573 : ArrayBuildState *astate;
4574 :
4575 126 : astate = initArrayResult(XMLOID, CurrentMemoryContext, true);
4576 126 : xpath_internal(xpath_expr_text, data, namespaces,
4577 : NULL, astate);
4578 108 : PG_RETURN_DATUM(makeArrayResult(astate, CurrentMemoryContext));
4579 : #else
4580 : NO_XML_SUPPORT();
4581 : return 0;
4582 : #endif
4583 : }
4584 :
4585 : /*
4586 : * Determines if the node specified by the supplied XPath exists
4587 : * in a given XML document, returning a boolean.
4588 : */
4589 : Datum
4590 198 : xmlexists(PG_FUNCTION_ARGS)
4591 : {
4592 : #ifdef USE_LIBXML
4593 198 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4594 198 : xmltype *data = PG_GETARG_XML_P(1);
4595 : int res_nitems;
4596 :
4597 198 : xpath_internal(xpath_expr_text, data, NULL,
4598 : &res_nitems, NULL);
4599 :
4600 198 : PG_RETURN_BOOL(res_nitems > 0);
4601 : #else
4602 : NO_XML_SUPPORT();
4603 : return 0;
4604 : #endif
4605 : }
4606 :
4607 : /*
4608 : * Determines if the node specified by the supplied XPath exists
4609 : * in a given XML document, returning a boolean. Differs from
4610 : * xmlexists as it supports namespaces and is not defined in SQL/XML.
4611 : */
4612 : Datum
4613 234 : xpath_exists(PG_FUNCTION_ARGS)
4614 : {
4615 : #ifdef USE_LIBXML
4616 234 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4617 234 : xmltype *data = PG_GETARG_XML_P(1);
4618 234 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4619 : int res_nitems;
4620 :
4621 234 : xpath_internal(xpath_expr_text, data, namespaces,
4622 : &res_nitems, NULL);
4623 :
4624 234 : PG_RETURN_BOOL(res_nitems > 0);
4625 : #else
4626 : NO_XML_SUPPORT();
4627 : return 0;
4628 : #endif
4629 : }
4630 :
4631 : /*
4632 : * Functions for checking well-formed-ness
4633 : */
4634 :
4635 : #ifdef USE_LIBXML
4636 : static bool
4637 114 : wellformed_xml(text *data, XmlOptionType xmloption_arg)
4638 : {
4639 : xmlDocPtr doc;
4640 114 : ErrorSaveContext escontext = {T_ErrorSaveContext};
4641 :
4642 : /*
4643 : * We'll report "true" if no soft error is reported by xml_parse().
4644 : */
4645 114 : doc = xml_parse(data, xmloption_arg, true,
4646 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
4647 114 : if (doc)
4648 60 : xmlFreeDoc(doc);
4649 :
4650 114 : return !escontext.error_occurred;
4651 : }
4652 : #endif
4653 :
4654 : Datum
4655 90 : xml_is_well_formed(PG_FUNCTION_ARGS)
4656 : {
4657 : #ifdef USE_LIBXML
4658 90 : text *data = PG_GETARG_TEXT_PP(0);
4659 :
4660 90 : PG_RETURN_BOOL(wellformed_xml(data, xmloption));
4661 : #else
4662 : NO_XML_SUPPORT();
4663 : return 0;
4664 : #endif /* not USE_LIBXML */
4665 : }
4666 :
4667 : Datum
4668 12 : xml_is_well_formed_document(PG_FUNCTION_ARGS)
4669 : {
4670 : #ifdef USE_LIBXML
4671 12 : text *data = PG_GETARG_TEXT_PP(0);
4672 :
4673 12 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_DOCUMENT));
4674 : #else
4675 : NO_XML_SUPPORT();
4676 : return 0;
4677 : #endif /* not USE_LIBXML */
4678 : }
4679 :
4680 : Datum
4681 12 : xml_is_well_formed_content(PG_FUNCTION_ARGS)
4682 : {
4683 : #ifdef USE_LIBXML
4684 12 : text *data = PG_GETARG_TEXT_PP(0);
4685 :
4686 12 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_CONTENT));
4687 : #else
4688 : NO_XML_SUPPORT();
4689 : return 0;
4690 : #endif /* not USE_LIBXML */
4691 : }
4692 :
4693 : /*
4694 : * support functions for XMLTABLE
4695 : *
4696 : */
4697 : #ifdef USE_LIBXML
4698 :
4699 : /*
4700 : * Returns private data from executor state. Ensure validity by check with
4701 : * MAGIC number.
4702 : */
4703 : static inline XmlTableBuilderData *
4704 160256 : GetXmlTableBuilderPrivateData(TableFuncScanState *state, const char *fname)
4705 : {
4706 : XmlTableBuilderData *result;
4707 :
4708 160256 : if (!IsA(state, TableFuncScanState))
4709 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4710 160256 : result = (XmlTableBuilderData *) state->opaque;
4711 160256 : if (result->magic != XMLTABLE_CONTEXT_MAGIC)
4712 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4713 :
4714 160256 : return result;
4715 : }
4716 : #endif
4717 :
4718 : /*
4719 : * XmlTableInitOpaque
4720 : * Fill in TableFuncScanState->opaque for XmlTable processor; initialize
4721 : * the XML parser.
4722 : *
4723 : * Note: Because we call pg_xml_init() here and pg_xml_done() in
4724 : * XmlTableDestroyOpaque, it is critical for robustness that no other
4725 : * executor nodes run until this node is processed to completion. Caller
4726 : * must execute this to completion (probably filling a tuplestore to exhaust
4727 : * this node in a single pass) instead of using row-per-call mode.
4728 : */
4729 : static void
4730 264 : XmlTableInitOpaque(TableFuncScanState *state, int natts)
4731 : {
4732 : #ifdef USE_LIBXML
4733 264 : volatile xmlParserCtxtPtr ctxt = NULL;
4734 : XmlTableBuilderData *xtCxt;
4735 : PgXmlErrorContext *xmlerrcxt;
4736 :
4737 264 : xtCxt = palloc0(sizeof(XmlTableBuilderData));
4738 264 : xtCxt->magic = XMLTABLE_CONTEXT_MAGIC;
4739 264 : xtCxt->natts = natts;
4740 264 : xtCxt->xpathscomp = palloc0(sizeof(xmlXPathCompExprPtr) * natts);
4741 :
4742 264 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4743 :
4744 264 : PG_TRY();
4745 : {
4746 264 : xmlInitParser();
4747 :
4748 264 : ctxt = xmlNewParserCtxt();
4749 264 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4750 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4751 : "could not allocate parser context");
4752 : }
4753 0 : PG_CATCH();
4754 : {
4755 0 : if (ctxt != NULL)
4756 0 : xmlFreeParserCtxt(ctxt);
4757 :
4758 0 : pg_xml_done(xmlerrcxt, true);
4759 :
4760 0 : PG_RE_THROW();
4761 : }
4762 264 : PG_END_TRY();
4763 :
4764 264 : xtCxt->xmlerrcxt = xmlerrcxt;
4765 264 : xtCxt->ctxt = ctxt;
4766 :
4767 264 : state->opaque = xtCxt;
4768 : #else
4769 : NO_XML_SUPPORT();
4770 : #endif /* not USE_LIBXML */
4771 264 : }
4772 :
4773 : /*
4774 : * XmlTableSetDocument
4775 : * Install the input document
4776 : */
4777 : static void
4778 264 : XmlTableSetDocument(TableFuncScanState *state, Datum value)
4779 : {
4780 : #ifdef USE_LIBXML
4781 : XmlTableBuilderData *xtCxt;
4782 264 : xmltype *xmlval = DatumGetXmlP(value);
4783 : char *str;
4784 : xmlChar *xstr;
4785 : int length;
4786 264 : volatile xmlDocPtr doc = NULL;
4787 264 : volatile xmlXPathContextPtr xpathcxt = NULL;
4788 :
4789 264 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetDocument");
4790 :
4791 : /*
4792 : * Use out function for casting to string (remove encoding property). See
4793 : * comment in xml_out.
4794 : */
4795 264 : str = xml_out_internal(xmlval, 0);
4796 :
4797 264 : length = strlen(str);
4798 264 : xstr = pg_xmlCharStrndup(str, length);
4799 :
4800 264 : PG_TRY();
4801 : {
4802 264 : doc = xmlCtxtReadMemory(xtCxt->ctxt, (char *) xstr, length, NULL, NULL, 0);
4803 264 : if (doc == NULL || xtCxt->xmlerrcxt->err_occurred)
4804 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4805 : "could not parse XML document");
4806 264 : xpathcxt = xmlXPathNewContext(doc);
4807 264 : if (xpathcxt == NULL || xtCxt->xmlerrcxt->err_occurred)
4808 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4809 : "could not allocate XPath context");
4810 264 : xpathcxt->node = (xmlNodePtr) doc;
4811 : }
4812 0 : PG_CATCH();
4813 : {
4814 0 : if (xpathcxt != NULL)
4815 0 : xmlXPathFreeContext(xpathcxt);
4816 0 : if (doc != NULL)
4817 0 : xmlFreeDoc(doc);
4818 :
4819 0 : PG_RE_THROW();
4820 : }
4821 264 : PG_END_TRY();
4822 :
4823 264 : xtCxt->doc = doc;
4824 264 : xtCxt->xpathcxt = xpathcxt;
4825 : #else
4826 : NO_XML_SUPPORT();
4827 : #endif /* not USE_LIBXML */
4828 264 : }
4829 :
4830 : /*
4831 : * XmlTableSetNamespace
4832 : * Add a namespace declaration
4833 : */
4834 : static void
4835 18 : XmlTableSetNamespace(TableFuncScanState *state, const char *name, const char *uri)
4836 : {
4837 : #ifdef USE_LIBXML
4838 : XmlTableBuilderData *xtCxt;
4839 :
4840 18 : if (name == NULL)
4841 6 : ereport(ERROR,
4842 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
4843 : errmsg("DEFAULT namespace is not supported")));
4844 12 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetNamespace");
4845 :
4846 12 : if (xmlXPathRegisterNs(xtCxt->xpathcxt,
4847 12 : pg_xmlCharStrndup(name, strlen(name)),
4848 12 : pg_xmlCharStrndup(uri, strlen(uri))))
4849 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4850 : "could not set XML namespace");
4851 : #else
4852 : NO_XML_SUPPORT();
4853 : #endif /* not USE_LIBXML */
4854 12 : }
4855 :
4856 : /*
4857 : * XmlTableSetRowFilter
4858 : * Install the row-filter Xpath expression.
4859 : */
4860 : static void
4861 258 : XmlTableSetRowFilter(TableFuncScanState *state, const char *path)
4862 : {
4863 : #ifdef USE_LIBXML
4864 : XmlTableBuilderData *xtCxt;
4865 : xmlChar *xstr;
4866 :
4867 258 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetRowFilter");
4868 :
4869 258 : if (*path == '\0')
4870 0 : ereport(ERROR,
4871 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4872 : errmsg("row path filter must not be empty string")));
4873 :
4874 258 : xstr = pg_xmlCharStrndup(path, strlen(path));
4875 :
4876 : /* We require XmlTableSetDocument to have been done already */
4877 : Assert(xtCxt->xpathcxt != NULL);
4878 :
4879 258 : xtCxt->xpathcomp = xmlXPathCtxtCompile(xtCxt->xpathcxt, xstr);
4880 258 : if (xtCxt->xpathcomp == NULL || xtCxt->xmlerrcxt->err_occurred)
4881 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4882 : "invalid XPath expression");
4883 : #else
4884 : NO_XML_SUPPORT();
4885 : #endif /* not USE_LIBXML */
4886 258 : }
4887 :
4888 : /*
4889 : * XmlTableSetColumnFilter
4890 : * Install the column-filter Xpath expression, for the given column.
4891 : */
4892 : static void
4893 774 : XmlTableSetColumnFilter(TableFuncScanState *state, const char *path, int colnum)
4894 : {
4895 : #ifdef USE_LIBXML
4896 : XmlTableBuilderData *xtCxt;
4897 : xmlChar *xstr;
4898 :
4899 : Assert(PointerIsValid(path));
4900 :
4901 774 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetColumnFilter");
4902 :
4903 774 : if (*path == '\0')
4904 0 : ereport(ERROR,
4905 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4906 : errmsg("column path filter must not be empty string")));
4907 :
4908 774 : xstr = pg_xmlCharStrndup(path, strlen(path));
4909 :
4910 : /* We require XmlTableSetDocument to have been done already */
4911 : Assert(xtCxt->xpathcxt != NULL);
4912 :
4913 774 : xtCxt->xpathscomp[colnum] = xmlXPathCtxtCompile(xtCxt->xpathcxt, xstr);
4914 774 : if (xtCxt->xpathscomp[colnum] == NULL || xtCxt->xmlerrcxt->err_occurred)
4915 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4916 : "invalid XPath expression");
4917 : #else
4918 : NO_XML_SUPPORT();
4919 : #endif /* not USE_LIBXML */
4920 774 : }
4921 :
4922 : /*
4923 : * XmlTableFetchRow
4924 : * Prepare the next "current" tuple for upcoming GetValue calls.
4925 : * Returns false if the row-filter expression returned no more rows.
4926 : */
4927 : static bool
4928 23030 : XmlTableFetchRow(TableFuncScanState *state)
4929 : {
4930 : #ifdef USE_LIBXML
4931 : XmlTableBuilderData *xtCxt;
4932 :
4933 23030 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableFetchRow");
4934 :
4935 : /* Propagate our own error context to libxml2 */
4936 23030 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
4937 :
4938 23030 : if (xtCxt->xpathobj == NULL)
4939 : {
4940 258 : xtCxt->xpathobj = xmlXPathCompiledEval(xtCxt->xpathcomp, xtCxt->xpathcxt);
4941 258 : if (xtCxt->xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
4942 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4943 : "could not create XPath object");
4944 :
4945 258 : xtCxt->row_count = 0;
4946 : }
4947 :
4948 23030 : if (xtCxt->xpathobj->type == XPATH_NODESET)
4949 : {
4950 23030 : if (xtCxt->xpathobj->nodesetval != NULL)
4951 : {
4952 23030 : if (xtCxt->row_count++ < xtCxt->xpathobj->nodesetval->nodeNr)
4953 22784 : return true;
4954 : }
4955 : }
4956 :
4957 246 : return false;
4958 : #else
4959 : NO_XML_SUPPORT();
4960 : return false;
4961 : #endif /* not USE_LIBXML */
4962 : }
4963 :
4964 : /*
4965 : * XmlTableGetValue
4966 : * Return the value for column number 'colnum' for the current row. If
4967 : * column -1 is requested, return representation of the whole row.
4968 : *
4969 : * This leaks memory, so be sure to reset often the context in which it's
4970 : * called.
4971 : */
4972 : static Datum
4973 135654 : XmlTableGetValue(TableFuncScanState *state, int colnum,
4974 : Oid typid, int32 typmod, bool *isnull)
4975 : {
4976 : #ifdef USE_LIBXML
4977 135654 : Datum result = (Datum) 0;
4978 : XmlTableBuilderData *xtCxt;
4979 135654 : volatile xmlXPathObjectPtr xpathobj = NULL;
4980 :
4981 135654 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableGetValue");
4982 :
4983 : Assert(xtCxt->xpathobj &&
4984 : xtCxt->xpathobj->type == XPATH_NODESET &&
4985 : xtCxt->xpathobj->nodesetval != NULL);
4986 :
4987 : /* Propagate our own error context to libxml2 */
4988 135654 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
4989 :
4990 135654 : *isnull = false;
4991 :
4992 : Assert(xtCxt->xpathscomp[colnum] != NULL);
4993 :
4994 135654 : PG_TRY();
4995 : {
4996 : xmlNodePtr cur;
4997 135654 : char *cstr = NULL;
4998 :
4999 : /* Set current node as entry point for XPath evaluation */
5000 135654 : cur = xtCxt->xpathobj->nodesetval->nodeTab[xtCxt->row_count - 1];
5001 135654 : xtCxt->xpathcxt->node = cur;
5002 :
5003 : /* Evaluate column path */
5004 135654 : xpathobj = xmlXPathCompiledEval(xtCxt->xpathscomp[colnum], xtCxt->xpathcxt);
5005 135654 : if (xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
5006 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
5007 : "could not create XPath object");
5008 :
5009 : /*
5010 : * There are four possible cases, depending on the number of nodes
5011 : * returned by the XPath expression and the type of the target column:
5012 : * a) XPath returns no nodes. b) The target type is XML (return all
5013 : * as XML). For non-XML return types: c) One node (return content).
5014 : * d) Multiple nodes (error).
5015 : */
5016 135654 : if (xpathobj->type == XPATH_NODESET)
5017 : {
5018 135624 : int count = 0;
5019 :
5020 135624 : if (xpathobj->nodesetval != NULL)
5021 135414 : count = xpathobj->nodesetval->nodeNr;
5022 :
5023 135624 : if (xpathobj->nodesetval == NULL || count == 0)
5024 : {
5025 22730 : *isnull = true;
5026 : }
5027 : else
5028 : {
5029 112894 : if (typid == XMLOID)
5030 : {
5031 : text *textstr;
5032 : StringInfoData str;
5033 :
5034 : /* Concatenate serialized values */
5035 72 : initStringInfo(&str);
5036 174 : for (int i = 0; i < count; i++)
5037 : {
5038 : textstr =
5039 102 : xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
5040 : xtCxt->xmlerrcxt);
5041 :
5042 102 : appendStringInfoText(&str, textstr);
5043 : }
5044 72 : cstr = str.data;
5045 : }
5046 : else
5047 : {
5048 : xmlChar *str;
5049 :
5050 112822 : if (count > 1)
5051 6 : ereport(ERROR,
5052 : (errcode(ERRCODE_CARDINALITY_VIOLATION),
5053 : errmsg("more than one value returned by column XPath expression")));
5054 :
5055 112816 : str = xmlXPathCastNodeSetToString(xpathobj->nodesetval);
5056 112816 : cstr = str ? xml_pstrdup_and_free(str) : "";
5057 : }
5058 : }
5059 : }
5060 30 : else if (xpathobj->type == XPATH_STRING)
5061 : {
5062 : /* Content should be escaped when target will be XML */
5063 18 : if (typid == XMLOID)
5064 6 : cstr = escape_xml((char *) xpathobj->stringval);
5065 : else
5066 12 : cstr = (char *) xpathobj->stringval;
5067 : }
5068 12 : else if (xpathobj->type == XPATH_BOOLEAN)
5069 : {
5070 : char typcategory;
5071 : bool typispreferred;
5072 : xmlChar *str;
5073 :
5074 : /* Allow implicit casting from boolean to numbers */
5075 6 : get_type_category_preferred(typid, &typcategory, &typispreferred);
5076 :
5077 6 : if (typcategory != TYPCATEGORY_NUMERIC)
5078 6 : str = xmlXPathCastBooleanToString(xpathobj->boolval);
5079 : else
5080 0 : str = xmlXPathCastNumberToString(xmlXPathCastBooleanToNumber(xpathobj->boolval));
5081 :
5082 6 : cstr = xml_pstrdup_and_free(str);
5083 : }
5084 6 : else if (xpathobj->type == XPATH_NUMBER)
5085 : {
5086 : xmlChar *str;
5087 :
5088 6 : str = xmlXPathCastNumberToString(xpathobj->floatval);
5089 6 : cstr = xml_pstrdup_and_free(str);
5090 : }
5091 : else
5092 0 : elog(ERROR, "unexpected XPath object type %u", xpathobj->type);
5093 :
5094 : /*
5095 : * By here, either cstr contains the result value, or the isnull flag
5096 : * has been set.
5097 : */
5098 : Assert(cstr || *isnull);
5099 :
5100 135648 : if (!*isnull)
5101 112918 : result = InputFunctionCall(&state->in_functions[colnum],
5102 : cstr,
5103 112918 : state->typioparams[colnum],
5104 : typmod);
5105 : }
5106 6 : PG_FINALLY();
5107 : {
5108 135654 : if (xpathobj != NULL)
5109 135654 : xmlXPathFreeObject(xpathobj);
5110 : }
5111 135654 : PG_END_TRY();
5112 :
5113 135648 : return result;
5114 : #else
5115 : NO_XML_SUPPORT();
5116 : return 0;
5117 : #endif /* not USE_LIBXML */
5118 : }
5119 :
5120 : /*
5121 : * XmlTableDestroyOpaque
5122 : * Release all libxml2 resources
5123 : */
5124 : static void
5125 264 : XmlTableDestroyOpaque(TableFuncScanState *state)
5126 : {
5127 : #ifdef USE_LIBXML
5128 : XmlTableBuilderData *xtCxt;
5129 :
5130 264 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableDestroyOpaque");
5131 :
5132 : /* Propagate our own error context to libxml2 */
5133 264 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
5134 :
5135 264 : if (xtCxt->xpathscomp != NULL)
5136 : {
5137 : int i;
5138 :
5139 1116 : for (i = 0; i < xtCxt->natts; i++)
5140 852 : if (xtCxt->xpathscomp[i] != NULL)
5141 774 : xmlXPathFreeCompExpr(xtCxt->xpathscomp[i]);
5142 : }
5143 :
5144 264 : if (xtCxt->xpathobj != NULL)
5145 258 : xmlXPathFreeObject(xtCxt->xpathobj);
5146 264 : if (xtCxt->xpathcomp != NULL)
5147 258 : xmlXPathFreeCompExpr(xtCxt->xpathcomp);
5148 264 : if (xtCxt->xpathcxt != NULL)
5149 264 : xmlXPathFreeContext(xtCxt->xpathcxt);
5150 264 : if (xtCxt->doc != NULL)
5151 264 : xmlFreeDoc(xtCxt->doc);
5152 264 : if (xtCxt->ctxt != NULL)
5153 264 : xmlFreeParserCtxt(xtCxt->ctxt);
5154 :
5155 264 : pg_xml_done(xtCxt->xmlerrcxt, true);
5156 :
5157 : /* not valid anymore */
5158 264 : xtCxt->magic = 0;
5159 264 : state->opaque = NULL;
5160 :
5161 : #else
5162 : NO_XML_SUPPORT();
5163 : #endif /* not USE_LIBXML */
5164 264 : }
|
