Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * xml.c
4 : * XML data type support.
5 : *
6 : *
7 : * Portions Copyright (c) 1996-2026, 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 "executor/spi.h"
88 : #include "executor/tablefunc.h"
89 : #include "fmgr.h"
90 : #include "lib/stringinfo.h"
91 : #include "libpq/pqformat.h"
92 : #include "mb/pg_wchar.h"
93 : #include "miscadmin.h"
94 : #include "nodes/execnodes.h"
95 : #include "nodes/miscnodes.h"
96 : #include "nodes/nodeFuncs.h"
97 : #include "utils/array.h"
98 : #include "utils/builtins.h"
99 : #include "utils/date.h"
100 : #include "utils/datetime.h"
101 : #include "utils/lsyscache.h"
102 : #include "utils/rel.h"
103 : #include "utils/syscache.h"
104 : #include "utils/xml.h"
105 :
106 :
107 : /* GUC variables */
108 : int xmlbinary = XMLBINARY_BASE64;
109 : int xmloption = XMLOPTION_CONTENT;
110 :
111 : #ifdef USE_LIBXML
112 :
113 : /* random number to identify PgXmlErrorContext */
114 : #define ERRCXT_MAGIC 68275028
115 :
116 : struct PgXmlErrorContext
117 : {
118 : int magic;
119 : /* strictness argument passed to pg_xml_init */
120 : PgXmlStrictness strictness;
121 : /* current error status and accumulated message, if any */
122 : bool err_occurred;
123 : StringInfoData err_buf;
124 : /* previous libxml error handling state (saved by pg_xml_init) */
125 : xmlStructuredErrorFunc saved_errfunc;
126 : void *saved_errcxt;
127 : /* previous libxml entity handler (saved by pg_xml_init) */
128 : xmlExternalEntityLoader saved_entityfunc;
129 : };
130 :
131 : static xmlParserInputPtr xmlPgEntityLoader(const char *URL, const char *ID,
132 : xmlParserCtxtPtr ctxt);
133 : static void xml_errsave(Node *escontext, PgXmlErrorContext *errcxt,
134 : int sqlcode, const char *msg);
135 : static void xml_errorHandler(void *data, PgXmlErrorPtr error);
136 : static int errdetail_for_xml_code(int code);
137 : static void chopStringInfoNewlines(StringInfo str);
138 : static void appendStringInfoLineSeparator(StringInfo str);
139 :
140 : #ifdef USE_LIBXMLCONTEXT
141 :
142 : static MemoryContext LibxmlContext = NULL;
143 :
144 : static void xml_memory_init(void);
145 : static void *xml_palloc(size_t size);
146 : static void *xml_repalloc(void *ptr, size_t size);
147 : static void xml_pfree(void *ptr);
148 : static char *xml_pstrdup(const char *string);
149 : #endif /* USE_LIBXMLCONTEXT */
150 :
151 : static xmlChar *xml_text2xmlChar(text *in);
152 : static int parse_xml_decl(const xmlChar *str, size_t *lenp,
153 : xmlChar **version, xmlChar **encoding, int *standalone);
154 : static bool print_xml_decl(StringInfo buf, const xmlChar *version,
155 : pg_enc encoding, int standalone);
156 : static bool xml_doctype_in_content(const xmlChar *str);
157 : static xmlDocPtr xml_parse(text *data, XmlOptionType xmloption_arg,
158 : bool preserve_whitespace, int encoding,
159 : XmlOptionType *parsed_xmloptiontype,
160 : xmlNodePtr *parsed_nodes,
161 : Node *escontext);
162 : static text *xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt);
163 : static int xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
164 : ArrayBuildState *astate,
165 : PgXmlErrorContext *xmlerrcxt);
166 : static xmlChar *pg_xmlCharStrndup(const char *str, size_t len);
167 : #endif /* USE_LIBXML */
168 :
169 : static void xmldata_root_element_start(StringInfo result, const char *eltname,
170 : const char *xmlschema, const char *targetns,
171 : bool top_level);
172 : static void xmldata_root_element_end(StringInfo result, const char *eltname);
173 : static StringInfo query_to_xml_internal(const char *query, char *tablename,
174 : const char *xmlschema, bool nulls, bool tableforest,
175 : const char *targetns, bool top_level);
176 : static const char *map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid,
177 : bool nulls, bool tableforest, const char *targetns);
178 : static const char *map_sql_schema_to_xmlschema_types(Oid nspid,
179 : List *relid_list, bool nulls,
180 : bool tableforest, const char *targetns);
181 : static const char *map_sql_catalog_to_xmlschema_types(List *nspid_list,
182 : bool nulls, bool tableforest,
183 : const char *targetns);
184 : static const char *map_sql_type_to_xml_name(Oid typeoid, int typmod);
185 : static const char *map_sql_typecoll_to_xmlschema_types(List *tupdesc_list);
186 : static const char *map_sql_type_to_xmlschema_type(Oid typeoid, int typmod);
187 : static void SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result,
188 : char *tablename, bool nulls, bool tableforest,
189 : const char *targetns, bool top_level);
190 :
191 : /* XMLTABLE support */
192 : #ifdef USE_LIBXML
193 : /* random number to identify XmlTableContext */
194 : #define XMLTABLE_CONTEXT_MAGIC 46922182
195 : typedef struct XmlTableBuilderData
196 : {
197 : int magic;
198 : int natts;
199 : long int row_count;
200 : PgXmlErrorContext *xmlerrcxt;
201 : xmlParserCtxtPtr ctxt;
202 : xmlDocPtr doc;
203 : xmlXPathContextPtr xpathcxt;
204 : xmlXPathCompExprPtr xpathcomp;
205 : xmlXPathObjectPtr xpathobj;
206 : xmlXPathCompExprPtr *xpathscomp;
207 : } XmlTableBuilderData;
208 : #endif
209 :
210 : static void XmlTableInitOpaque(struct TableFuncScanState *state, int natts);
211 : static void XmlTableSetDocument(struct TableFuncScanState *state, Datum value);
212 : static void XmlTableSetNamespace(struct TableFuncScanState *state, const char *name,
213 : const char *uri);
214 : static void XmlTableSetRowFilter(struct TableFuncScanState *state, const char *path);
215 : static void XmlTableSetColumnFilter(struct TableFuncScanState *state,
216 : const char *path, int colnum);
217 : static bool XmlTableFetchRow(struct TableFuncScanState *state);
218 : static Datum XmlTableGetValue(struct TableFuncScanState *state, int colnum,
219 : Oid typid, int32 typmod, bool *isnull);
220 : static void XmlTableDestroyOpaque(struct TableFuncScanState *state);
221 :
222 : const TableFuncRoutine XmlTableRoutine =
223 : {
224 : .InitOpaque = XmlTableInitOpaque,
225 : .SetDocument = XmlTableSetDocument,
226 : .SetNamespace = XmlTableSetNamespace,
227 : .SetRowFilter = XmlTableSetRowFilter,
228 : .SetColumnFilter = XmlTableSetColumnFilter,
229 : .FetchRow = XmlTableFetchRow,
230 : .GetValue = XmlTableGetValue,
231 : .DestroyOpaque = XmlTableDestroyOpaque
232 : };
233 :
234 : #define NO_XML_SUPPORT() \
235 : ereport(ERROR, \
236 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), \
237 : errmsg("unsupported XML feature"), \
238 : errdetail("This functionality requires the server to be built with libxml support.")))
239 :
240 :
241 : /* from SQL/XML:2008 section 4.9 */
242 : #define NAMESPACE_XSD "http://www.w3.org/2001/XMLSchema"
243 : #define NAMESPACE_XSI "http://www.w3.org/2001/XMLSchema-instance"
244 : #define NAMESPACE_SQLXML "http://standards.iso.org/iso/9075/2003/sqlxml"
245 :
246 :
247 : #ifdef USE_LIBXML
248 :
249 : static int
250 0 : xmlChar_to_encoding(const xmlChar *encoding_name)
251 : {
252 0 : int encoding = pg_char_to_encoding((const char *) encoding_name);
253 :
254 0 : if (encoding < 0)
255 0 : ereport(ERROR,
256 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
257 : errmsg("invalid encoding name \"%s\"",
258 : (const char *) encoding_name)));
259 0 : return encoding;
260 : }
261 : #endif
262 :
263 :
264 : /*
265 : * xml_in uses a plain C string to VARDATA conversion, so for the time being
266 : * we use the conversion function for the text datatype.
267 : *
268 : * This is only acceptable so long as xmltype and text use the same
269 : * representation.
270 : */
271 : Datum
272 428 : xml_in(PG_FUNCTION_ARGS)
273 : {
274 : #ifdef USE_LIBXML
275 428 : char *s = PG_GETARG_CSTRING(0);
276 : xmltype *vardata;
277 : xmlDocPtr doc;
278 :
279 : /* Build the result object. */
280 428 : vardata = (xmltype *) cstring_to_text(s);
281 :
282 : /*
283 : * Parse the data to check if it is well-formed XML data.
284 : *
285 : * Note: we don't need to worry about whether a soft error is detected.
286 : */
287 428 : doc = xml_parse(vardata, xmloption, true, GetDatabaseEncoding(),
288 : NULL, NULL, fcinfo->context);
289 404 : if (doc != NULL)
290 398 : xmlFreeDoc(doc);
291 :
292 404 : PG_RETURN_XML_P(vardata);
293 : #else
294 : NO_XML_SUPPORT();
295 : return 0;
296 : #endif
297 : }
298 :
299 :
300 : #define PG_XML_DEFAULT_VERSION "1.0"
301 :
302 :
303 : /*
304 : * xml_out_internal uses a plain VARDATA to C string conversion, so for the
305 : * time being we use the conversion function for the text datatype.
306 : *
307 : * This is only acceptable so long as xmltype and text use the same
308 : * representation.
309 : */
310 : static char *
311 11893 : xml_out_internal(xmltype *x, pg_enc target_encoding)
312 : {
313 11893 : char *str = text_to_cstring((text *) x);
314 :
315 : #ifdef USE_LIBXML
316 11893 : size_t len = strlen(str);
317 : xmlChar *version;
318 : int standalone;
319 : int res_code;
320 :
321 11893 : if ((res_code = parse_xml_decl((xmlChar *) str,
322 : &len, &version, NULL, &standalone)) == 0)
323 : {
324 : StringInfoData buf;
325 :
326 11893 : initStringInfo(&buf);
327 :
328 11893 : if (!print_xml_decl(&buf, version, target_encoding, standalone))
329 : {
330 : /*
331 : * If we are not going to produce an XML declaration, eat a single
332 : * newline in the original string to prevent empty first lines in
333 : * the output.
334 : */
335 11869 : if (*(str + len) == '\n')
336 3 : len += 1;
337 : }
338 11893 : appendStringInfoString(&buf, str + len);
339 :
340 11893 : pfree(str);
341 :
342 11893 : return buf.data;
343 : }
344 :
345 0 : ereport(WARNING,
346 : errcode(ERRCODE_DATA_CORRUPTED),
347 : errmsg_internal("could not parse XML declaration in stored value"),
348 : errdetail_for_xml_code(res_code));
349 : #endif
350 0 : return str;
351 : }
352 :
353 :
354 : Datum
355 11761 : xml_out(PG_FUNCTION_ARGS)
356 : {
357 11761 : xmltype *x = PG_GETARG_XML_P(0);
358 :
359 : /*
360 : * xml_out removes the encoding property in all cases. This is because we
361 : * cannot control from here whether the datum will be converted to a
362 : * different client encoding, so we'd do more harm than good by including
363 : * it.
364 : */
365 11761 : PG_RETURN_CSTRING(xml_out_internal(x, 0));
366 : }
367 :
368 :
369 : Datum
370 0 : xml_recv(PG_FUNCTION_ARGS)
371 : {
372 : #ifdef USE_LIBXML
373 0 : StringInfo buf = (StringInfo) PG_GETARG_POINTER(0);
374 : xmltype *result;
375 : const char *input;
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 : input = 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), input, 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 66 : appendStringInfoText(StringInfo str, const text *t)
460 : {
461 66 : appendBinaryStringInfo(str, VARDATA_ANY(t), VARSIZE_ANY_EXHDR(t));
462 66 : }
463 : #endif
464 :
465 :
466 : static xmltype *
467 11364 : stringinfo_to_xmltype(StringInfo buf)
468 : {
469 11364 : return (xmltype *) cstring_to_text_with_len(buf->data, buf->len);
470 : }
471 :
472 :
473 : static xmltype *
474 39 : cstring_to_xmltype(const char *string)
475 : {
476 39 : return (xmltype *) cstring_to_text(string);
477 : }
478 :
479 :
480 : #ifdef USE_LIBXML
481 : static xmltype *
482 11413 : xmlBuffer_to_xmltype(xmlBufferPtr buf)
483 : {
484 11413 : return (xmltype *) cstring_to_text_with_len((const char *) xmlBufferContent(buf),
485 : xmlBufferLength(buf));
486 : }
487 : #endif
488 :
489 :
490 : Datum
491 21 : xmlcomment(PG_FUNCTION_ARGS)
492 : {
493 : #ifdef USE_LIBXML
494 21 : text *arg = PG_GETARG_TEXT_PP(0);
495 21 : char *argdata = VARDATA_ANY(arg);
496 21 : int len = VARSIZE_ANY_EXHDR(arg);
497 : StringInfoData buf;
498 : int i;
499 :
500 : /* check for "--" in string or "-" at the end */
501 90 : for (i = 1; i < len; i++)
502 : {
503 72 : if (argdata[i] == '-' && argdata[i - 1] == '-')
504 3 : ereport(ERROR,
505 : (errcode(ERRCODE_INVALID_XML_COMMENT),
506 : errmsg("invalid XML comment")));
507 : }
508 18 : if (len > 0 && argdata[len - 1] == '-')
509 3 : ereport(ERROR,
510 : (errcode(ERRCODE_INVALID_XML_COMMENT),
511 : errmsg("invalid XML comment")));
512 :
513 15 : initStringInfo(&buf);
514 15 : appendStringInfoString(&buf, "<!--");
515 15 : appendStringInfoText(&buf, arg);
516 15 : appendStringInfoString(&buf, "-->");
517 :
518 15 : 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 15 : xmltext(PG_FUNCTION_ARGS)
528 : {
529 : #ifdef USE_LIBXML
530 15 : text *arg = PG_GETARG_TEXT_PP(0);
531 : text *result;
532 15 : xmlChar *volatile xmlbuf = NULL;
533 : PgXmlErrorContext *xmlerrcxt;
534 :
535 : /* First we gotta spin up some error handling. */
536 15 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
537 :
538 15 : PG_TRY();
539 : {
540 15 : xmlbuf = xmlEncodeSpecialChars(NULL, xml_text2xmlChar(arg));
541 :
542 15 : if (xmlbuf == NULL || xmlerrcxt->err_occurred)
543 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
544 : "could not allocate xmlChar");
545 :
546 15 : result = cstring_to_text_with_len((const char *) xmlbuf,
547 : xmlStrlen(xmlbuf));
548 : }
549 0 : PG_CATCH();
550 : {
551 0 : if (xmlbuf)
552 0 : xmlFree(xmlbuf);
553 :
554 0 : pg_xml_done(xmlerrcxt, true);
555 0 : PG_RE_THROW();
556 : }
557 15 : PG_END_TRY();
558 :
559 15 : xmlFree(xmlbuf);
560 15 : pg_xml_done(xmlerrcxt, false);
561 :
562 15 : 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 11239 : xmlconcat(List *args)
576 : {
577 : #ifdef USE_LIBXML
578 11239 : int global_standalone = 1;
579 11239 : xmlChar *global_version = NULL;
580 11239 : bool global_version_no_value = false;
581 : StringInfoData buf;
582 : ListCell *v;
583 :
584 11239 : initStringInfo(&buf);
585 33720 : foreach(v, args)
586 : {
587 22481 : xmltype *x = DatumGetXmlP(PointerGetDatum(lfirst(v)));
588 : size_t len;
589 : xmlChar *version;
590 : int standalone;
591 : char *str;
592 :
593 22481 : len = VARSIZE(x) - VARHDRSZ;
594 22481 : str = text_to_cstring((text *) x);
595 :
596 22481 : parse_xml_decl((xmlChar *) str, &len, &version, NULL, &standalone);
597 :
598 22481 : if (standalone == 0 && global_standalone == 1)
599 0 : global_standalone = 0;
600 22481 : if (standalone < 0)
601 22475 : global_standalone = -1;
602 :
603 22481 : if (!version)
604 22472 : global_version_no_value = true;
605 9 : else if (!global_version)
606 6 : global_version = version;
607 3 : else if (xmlStrcmp(version, global_version) != 0)
608 0 : global_version_no_value = true;
609 :
610 22481 : appendStringInfoString(&buf, str + len);
611 22481 : pfree(str);
612 : }
613 :
614 11239 : if (!global_version_no_value || global_standalone >= 0)
615 : {
616 : StringInfoData buf2;
617 :
618 3 : initStringInfo(&buf2);
619 :
620 3 : print_xml_decl(&buf2,
621 3 : (!global_version_no_value) ? global_version : NULL,
622 : 0,
623 : global_standalone);
624 :
625 3 : appendBinaryStringInfo(&buf2, buf.data, buf.len);
626 3 : buf = buf2;
627 : }
628 :
629 11239 : 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 11227 : xmlconcat2(PG_FUNCTION_ARGS)
642 : {
643 11227 : if (PG_ARGISNULL(0))
644 : {
645 9 : if (PG_ARGISNULL(1))
646 0 : PG_RETURN_NULL();
647 : else
648 9 : PG_RETURN_XML_P(PG_GETARG_XML_P(1));
649 : }
650 11218 : else if (PG_ARGISNULL(1))
651 0 : PG_RETURN_XML_P(PG_GETARG_XML_P(0));
652 : else
653 11218 : PG_RETURN_XML_P(xmlconcat(list_make2(PG_GETARG_XML_P(0),
654 : PG_GETARG_XML_P(1))));
655 : }
656 :
657 :
658 : Datum
659 3 : texttoxml(PG_FUNCTION_ARGS)
660 : {
661 3 : text *data = PG_GETARG_TEXT_PP(0);
662 :
663 3 : 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 90 : 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 90 : volatile xmlBufferPtr buf = NULL;
686 90 : volatile xmlSaveCtxtPtr ctxt = NULL;
687 90 : ErrorSaveContext escontext = {T_ErrorSaveContext};
688 90 : PgXmlErrorContext *volatile xmlerrcxt = NULL;
689 : #endif
690 :
691 90 : 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 18 : 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 72 : doc = xml_parse(data, xmloption_arg, !indent, GetDatabaseEncoding(),
710 : &parsed_xmloptiontype, &content_nodes,
711 72 : (Node *) &escontext);
712 72 : if (doc == NULL || escontext.error_occurred)
713 : {
714 15 : if (doc)
715 0 : xmlFreeDoc(doc);
716 : /* A soft error must be failure to conform to XMLOPTION_DOCUMENT */
717 15 : 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 57 : if (!indent)
724 : {
725 9 : xmlFreeDoc(doc);
726 9 : 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 48 : PG_TRY();
736 : {
737 48 : size_t decl_len = 0;
738 :
739 48 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
740 :
741 : /* The serialized data will go into this buffer. */
742 48 : buf = xmlBufferCreate();
743 :
744 48 : 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 48 : 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 48 : if (decl_len == 0)
758 42 : ctxt = xmlSaveToBuffer(buf, NULL,
759 : XML_SAVE_NO_DECL | XML_SAVE_FORMAT);
760 : else
761 6 : ctxt = xmlSaveToBuffer(buf, NULL,
762 : XML_SAVE_FORMAT);
763 :
764 48 : if (ctxt == NULL || xmlerrcxt->err_occurred)
765 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
766 : "could not allocate xmlSaveCtxt");
767 :
768 48 : if (parsed_xmloptiontype == XMLOPTION_DOCUMENT)
769 : {
770 : /* If it's a document, saving is easy. */
771 21 : 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 27 : 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 24 : root = xmlNewNode(NULL, (const xmlChar *) "content-root");
788 24 : 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 24 : oldroot = xmlDocSetRootElement(doc, root);
797 24 : if (oldroot != NULL)
798 0 : xmlFreeNode(oldroot);
799 :
800 24 : if (xmlAddChildList(root, content_nodes) == NULL ||
801 24 : 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 24 : newline = xmlNewDocText(NULL, (const xmlChar *) "\n");
813 24 : if (newline == NULL || xmlerrcxt->err_occurred)
814 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
815 : "could not allocate xml node");
816 :
817 63 : for (xmlNodePtr node = root->children; node; node = node->next)
818 : {
819 : /* insert newlines between nodes */
820 39 : if (node->type != XML_TEXT_NODE && node->prev != NULL)
821 : {
822 12 : 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 39 : 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 24 : xmlFreeNode(newline);
839 : }
840 :
841 48 : 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 48 : if (xmloption_arg == XMLOPTION_DOCUMENT)
852 : {
853 18 : const char *str = (const char *) xmlBufferContent(buf);
854 18 : int len = xmlBufferLength(buf);
855 :
856 36 : while (len > 0 && (str[len - 1] == '\n' ||
857 18 : str[len - 1] == '\r'))
858 18 : len--;
859 :
860 18 : result = cstring_to_text_with_len(str, len);
861 : }
862 : else
863 30 : 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 48 : PG_END_TRY();
879 :
880 48 : xmlBufferFree(buf);
881 48 : xmlFreeDoc(doc);
882 :
883 48 : pg_xml_done(xmlerrcxt, false);
884 :
885 48 : return result;
886 : #else
887 : NO_XML_SUPPORT();
888 : return NULL;
889 : #endif
890 : }
891 :
892 :
893 : xmltype *
894 11305 : xmlelement(XmlExpr *xexpr,
895 : const Datum *named_argvalue, const bool *named_argnull,
896 : const Datum *argvalue, const 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 11305 : volatile xmlBufferPtr buf = NULL;
907 11305 : 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 11305 : named_arg_strings = NIL;
917 11305 : i = 0;
918 11329 : foreach(arg, xexpr->named_args)
919 : {
920 27 : Expr *e = (Expr *) lfirst(arg);
921 : char *str;
922 :
923 27 : if (named_argnull[i])
924 0 : str = NULL;
925 : else
926 27 : str = map_sql_value_to_xml_value(named_argvalue[i],
927 : exprType((Node *) e),
928 : false);
929 24 : named_arg_strings = lappend(named_arg_strings, str);
930 24 : i++;
931 : }
932 :
933 11302 : arg_strings = NIL;
934 11302 : i = 0;
935 22592 : foreach(arg, xexpr->args)
936 : {
937 11290 : Expr *e = (Expr *) lfirst(arg);
938 : char *str;
939 :
940 : /* here we can just forget NULL elements immediately */
941 11290 : if (!argnull[i])
942 : {
943 11290 : str = map_sql_value_to_xml_value(argvalue[i],
944 : exprType((Node *) e),
945 : true);
946 11290 : arg_strings = lappend(arg_strings, str);
947 : }
948 11290 : i++;
949 : }
950 :
951 11302 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
952 :
953 11302 : PG_TRY();
954 : {
955 11302 : buf = xmlBufferCreate();
956 11302 : if (buf == NULL || xmlerrcxt->err_occurred)
957 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
958 : "could not allocate xmlBuffer");
959 11302 : writer = xmlNewTextWriterMemory(buf, 0);
960 11302 : if (writer == NULL || xmlerrcxt->err_occurred)
961 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
962 : "could not allocate xmlTextWriter");
963 :
964 11302 : if (xmlTextWriterStartElement(writer, (xmlChar *) xexpr->name) < 0 ||
965 11302 : xmlerrcxt->err_occurred)
966 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
967 : "could not start xml element");
968 :
969 11326 : forboth(arg, named_arg_strings, narg, xexpr->arg_names)
970 : {
971 24 : char *str = (char *) lfirst(arg);
972 24 : char *argname = strVal(lfirst(narg));
973 :
974 24 : if (str)
975 : {
976 24 : if (xmlTextWriterWriteAttribute(writer,
977 : (xmlChar *) argname,
978 24 : (xmlChar *) str) < 0 ||
979 24 : xmlerrcxt->err_occurred)
980 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
981 : "could not write xml attribute");
982 : }
983 : }
984 :
985 22592 : foreach(arg, arg_strings)
986 : {
987 11290 : char *str = (char *) lfirst(arg);
988 :
989 11290 : if (xmlTextWriterWriteRaw(writer, (xmlChar *) str) < 0 ||
990 11290 : xmlerrcxt->err_occurred)
991 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
992 : "could not write raw xml text");
993 : }
994 :
995 11302 : if (xmlTextWriterEndElement(writer) < 0 ||
996 11302 : 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 11302 : xmlFreeTextWriter(writer);
1002 11302 : writer = NULL;
1003 :
1004 11302 : 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 11302 : PG_END_TRY();
1018 :
1019 11302 : xmlBufferFree(buf);
1020 :
1021 11302 : pg_xml_done(xmlerrcxt, false);
1022 :
1023 11302 : return result;
1024 : #else
1025 : NO_XML_SUPPORT();
1026 : return NULL;
1027 : #endif
1028 : }
1029 :
1030 :
1031 : xmltype *
1032 69 : xmlparse(text *data, XmlOptionType xmloption_arg, bool preserve_whitespace)
1033 : {
1034 : #ifdef USE_LIBXML
1035 : xmlDocPtr doc;
1036 :
1037 69 : doc = xml_parse(data, xmloption_arg, preserve_whitespace,
1038 : GetDatabaseEncoding(), NULL, NULL, NULL);
1039 45 : xmlFreeDoc(doc);
1040 :
1041 45 : return (xmltype *) data;
1042 : #else
1043 : NO_XML_SUPPORT();
1044 : return NULL;
1045 : #endif
1046 : }
1047 :
1048 :
1049 : xmltype *
1050 36 : 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 36 : if (pg_strcasecmp(target, "xml") == 0)
1057 6 : 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 30 : *result_is_null = arg_is_null;
1067 30 : if (*result_is_null)
1068 6 : return NULL;
1069 :
1070 24 : initStringInfo(&buf);
1071 :
1072 24 : appendStringInfo(&buf, "<?%s", target);
1073 :
1074 24 : if (arg != NULL)
1075 : {
1076 : char *string;
1077 :
1078 12 : string = text_to_cstring(arg);
1079 12 : if (strstr(string, "?>") != NULL)
1080 3 : ereport(ERROR,
1081 : (errcode(ERRCODE_INVALID_XML_PROCESSING_INSTRUCTION),
1082 : errmsg("invalid XML processing instruction"),
1083 : errdetail("XML processing instruction cannot contain \"?>\".")));
1084 :
1085 9 : appendStringInfoChar(&buf, ' ');
1086 9 : appendStringInfoString(&buf, string + strspn(string, " "));
1087 9 : pfree(string);
1088 : }
1089 21 : appendStringInfoString(&buf, "?>");
1090 :
1091 21 : result = stringinfo_to_xmltype(&buf);
1092 21 : pfree(buf.data);
1093 21 : return result;
1094 : #else
1095 : NO_XML_SUPPORT();
1096 : return NULL;
1097 : #endif
1098 : }
1099 :
1100 :
1101 : xmltype *
1102 30 : 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 30 : len = VARSIZE(data) - VARHDRSZ;
1112 30 : str = text_to_cstring((text *) data);
1113 :
1114 30 : parse_xml_decl((xmlChar *) str, &len, &orig_version, NULL, &orig_standalone);
1115 :
1116 30 : if (version)
1117 12 : orig_version = xml_text2xmlChar(version);
1118 : else
1119 18 : orig_version = NULL;
1120 :
1121 30 : switch (standalone)
1122 : {
1123 9 : case XML_STANDALONE_YES:
1124 9 : orig_standalone = 1;
1125 9 : break;
1126 6 : case XML_STANDALONE_NO:
1127 6 : orig_standalone = 0;
1128 6 : break;
1129 6 : case XML_STANDALONE_NO_VALUE:
1130 6 : orig_standalone = -1;
1131 6 : break;
1132 9 : case XML_STANDALONE_OMITTED:
1133 : /* leave original value */
1134 9 : break;
1135 : }
1136 :
1137 30 : initStringInfo(&buf);
1138 30 : print_xml_decl(&buf, orig_version, 0, orig_standalone);
1139 30 : appendStringInfoString(&buf, str + len);
1140 :
1141 30 : 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 12 : xml_is_document(xmltype *arg)
1169 : {
1170 : #ifdef USE_LIBXML
1171 : xmlDocPtr doc;
1172 12 : ErrorSaveContext escontext = {T_ErrorSaveContext};
1173 :
1174 : /*
1175 : * We'll report "true" if no soft error is reported by xml_parse().
1176 : */
1177 12 : doc = xml_parse((text *) arg, XMLOPTION_DOCUMENT, true,
1178 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
1179 12 : if (doc)
1180 6 : xmlFreeDoc(doc);
1181 :
1182 12 : 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 47669 : pg_xml_init_library(void)
1205 : {
1206 : static bool first_time = true;
1207 :
1208 47669 : 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 13 : LIBXML_TEST_VERSION;
1229 :
1230 13 : first_time = false;
1231 : }
1232 47669 : }
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 12450 : pg_xml_init(PgXmlStrictness strictness)
1251 : {
1252 : PgXmlErrorContext *errcxt;
1253 : void *new_errcxt;
1254 :
1255 : /* Do one-time setup if needed */
1256 12450 : pg_xml_init_library();
1257 :
1258 : /* Create error handling context structure */
1259 12450 : errcxt = palloc_object(PgXmlErrorContext);
1260 12450 : errcxt->magic = ERRCXT_MAGIC;
1261 12450 : errcxt->strictness = strictness;
1262 12450 : errcxt->err_occurred = false;
1263 12450 : 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 12450 : errcxt->saved_errfunc = xmlStructuredError;
1273 :
1274 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1275 12450 : errcxt->saved_errcxt = xmlStructuredErrorContext;
1276 : #else
1277 : errcxt->saved_errcxt = xmlGenericErrorContext;
1278 : #endif
1279 :
1280 12450 : 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 12450 : new_errcxt = xmlStructuredErrorContext;
1298 : #else
1299 : new_errcxt = xmlGenericErrorContext;
1300 : #endif
1301 :
1302 12450 : 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 12450 : errcxt->saved_entityfunc = xmlGetExternalEntityLoader();
1315 12450 : xmlSetExternalEntityLoader(xmlPgEntityLoader);
1316 :
1317 12450 : 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 12450 : 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 12450 : cur_errcxt = xmlStructuredErrorContext;
1352 : #else
1353 : cur_errcxt = xmlGenericErrorContext;
1354 : #endif
1355 :
1356 12450 : 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 12450 : xmlSetStructuredErrorFunc(errcxt->saved_errcxt, errcxt->saved_errfunc);
1361 12450 : 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 12450 : errcxt->magic = 0;
1368 :
1369 : /* Release memory */
1370 12450 : pfree(errcxt->err_buf.data);
1371 12450 : pfree(errcxt);
1372 12450 : }
1373 :
1374 :
1375 : /*
1376 : * pg_xml_error_occurred() --- test the error flag
1377 : */
1378 : bool
1379 39 : pg_xml_error_occurred(PgXmlErrorContext *errcxt)
1380 : {
1381 39 : 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 96 : xml_pnstrdup(const xmlChar *str, size_t len)
1415 : {
1416 : xmlChar *result;
1417 :
1418 96 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1419 96 : memcpy(result, str, len * sizeof(xmlChar));
1420 96 : result[len] = 0;
1421 96 : return result;
1422 : }
1423 :
1424 : /* Ditto, except input is char* */
1425 : static xmlChar *
1426 1212 : pg_xmlCharStrndup(const char *str, size_t len)
1427 : {
1428 : xmlChar *result;
1429 :
1430 1212 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1431 1212 : memcpy(result, str, len);
1432 1212 : result[len] = '\0';
1433 :
1434 1212 : 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 56838 : xml_pstrdup_and_free(xmlChar *str)
1444 : {
1445 : char *result;
1446 :
1447 56838 : if (str)
1448 : {
1449 56838 : PG_TRY();
1450 : {
1451 56838 : result = pstrdup((char *) str);
1452 : }
1453 0 : PG_FINALLY();
1454 : {
1455 56838 : xmlFree(str);
1456 : }
1457 56838 : PG_END_TRY();
1458 : }
1459 : else
1460 0 : result = NULL;
1461 :
1462 56838 : 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 35219 : 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 35219 : pg_xml_init_library();
1488 :
1489 : /* Initialize output arguments to "not present" */
1490 35219 : if (version)
1491 34895 : *version = NULL;
1492 35219 : if (encoding)
1493 0 : *encoding = NULL;
1494 35219 : if (standalone)
1495 34895 : *standalone = -1;
1496 :
1497 35219 : p = str;
1498 :
1499 35219 : if (xmlStrncmp(p, (xmlChar *) "<?xml", 5) != 0)
1500 35108 : 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 111 : utf8len = strnlen((const char *) (p + 5), MAX_MULTIBYTE_CHAR_LEN);
1511 111 : utf8char = xmlGetUTF8Char(p + 5, &utf8len);
1512 111 : if (PG_XMLISNAMECHAR(utf8char))
1513 6 : goto finished;
1514 :
1515 105 : p += 5;
1516 :
1517 : /* version */
1518 105 : CHECK_XML_SPACE(p);
1519 210 : SKIP_XML_SPACE(p);
1520 105 : if (xmlStrncmp(p, (xmlChar *) "version", 7) != 0)
1521 0 : return XML_ERR_VERSION_MISSING;
1522 105 : p += 7;
1523 105 : SKIP_XML_SPACE(p);
1524 105 : if (*p != '=')
1525 0 : return XML_ERR_VERSION_MISSING;
1526 105 : p += 1;
1527 105 : SKIP_XML_SPACE(p);
1528 :
1529 105 : if (*p == '\'' || *p == '"')
1530 105 : {
1531 : const xmlChar *q;
1532 :
1533 105 : q = xmlStrchr(p + 1, *p);
1534 105 : if (!q)
1535 0 : return XML_ERR_VERSION_MISSING;
1536 :
1537 105 : if (version)
1538 96 : *version = xml_pnstrdup(p + 1, q - p - 1);
1539 105 : p = q + 1;
1540 : }
1541 : else
1542 0 : return XML_ERR_VERSION_MISSING;
1543 :
1544 : /* encoding */
1545 105 : save_p = p;
1546 186 : SKIP_XML_SPACE(p);
1547 105 : if (xmlStrncmp(p, (xmlChar *) "encoding", 8) == 0)
1548 : {
1549 27 : CHECK_XML_SPACE(save_p);
1550 27 : p += 8;
1551 27 : SKIP_XML_SPACE(p);
1552 27 : if (*p != '=')
1553 0 : return XML_ERR_MISSING_ENCODING;
1554 27 : p += 1;
1555 27 : SKIP_XML_SPACE(p);
1556 :
1557 27 : if (*p == '\'' || *p == '"')
1558 27 : {
1559 : const xmlChar *q;
1560 :
1561 27 : q = xmlStrchr(p + 1, *p);
1562 27 : if (!q)
1563 0 : return XML_ERR_MISSING_ENCODING;
1564 :
1565 27 : if (encoding)
1566 0 : *encoding = xml_pnstrdup(p + 1, q - p - 1);
1567 27 : p = q + 1;
1568 : }
1569 : else
1570 0 : return XML_ERR_MISSING_ENCODING;
1571 : }
1572 : else
1573 : {
1574 78 : p = save_p;
1575 : }
1576 :
1577 : /* standalone */
1578 105 : save_p = p;
1579 159 : SKIP_XML_SPACE(p);
1580 105 : if (xmlStrncmp(p, (xmlChar *) "standalone", 10) == 0)
1581 : {
1582 54 : CHECK_XML_SPACE(save_p);
1583 54 : p += 10;
1584 54 : SKIP_XML_SPACE(p);
1585 54 : if (*p != '=')
1586 0 : return XML_ERR_STANDALONE_VALUE;
1587 54 : p += 1;
1588 54 : SKIP_XML_SPACE(p);
1589 108 : if (xmlStrncmp(p, (xmlChar *) "'yes'", 5) == 0 ||
1590 54 : xmlStrncmp(p, (xmlChar *) "\"yes\"", 5) == 0)
1591 : {
1592 30 : if (standalone)
1593 30 : *standalone = 1;
1594 30 : p += 5;
1595 : }
1596 48 : else if (xmlStrncmp(p, (xmlChar *) "'no'", 4) == 0 ||
1597 24 : xmlStrncmp(p, (xmlChar *) "\"no\"", 4) == 0)
1598 : {
1599 18 : if (standalone)
1600 18 : *standalone = 0;
1601 18 : p += 4;
1602 : }
1603 : else
1604 6 : return XML_ERR_STANDALONE_VALUE;
1605 : }
1606 : else
1607 : {
1608 51 : p = save_p;
1609 : }
1610 :
1611 99 : SKIP_XML_SPACE(p);
1612 99 : if (xmlStrncmp(p, (xmlChar *) "?>", 2) != 0)
1613 0 : return XML_ERR_XMLDECL_NOT_FINISHED;
1614 99 : p += 2;
1615 :
1616 35213 : finished:
1617 35213 : len = p - str;
1618 :
1619 38579 : for (p = str; p < str + len; p++)
1620 3366 : if (*p > 127)
1621 0 : return XML_ERR_INVALID_CHAR;
1622 :
1623 35213 : if (lenp)
1624 35213 : *lenp = len;
1625 :
1626 35213 : 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 11926 : print_xml_decl(StringInfo buf, const xmlChar *version,
1646 : pg_enc encoding, int standalone)
1647 : {
1648 11926 : if ((version && strcmp((const char *) version, PG_XML_DEFAULT_VERSION) != 0)
1649 11908 : || (encoding && encoding != PG_UTF8)
1650 11908 : || standalone != -1)
1651 : {
1652 48 : appendStringInfoString(buf, "<?xml");
1653 :
1654 48 : if (version)
1655 36 : appendStringInfo(buf, " version=\"%s\"", version);
1656 : else
1657 12 : appendStringInfo(buf, " version=\"%s\"", PG_XML_DEFAULT_VERSION);
1658 :
1659 48 : 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 48 : if (standalone == 1)
1670 24 : appendStringInfoString(buf, " standalone=\"yes\"");
1671 24 : else if (standalone == 0)
1672 12 : appendStringInfoString(buf, " standalone=\"no\"");
1673 48 : appendStringInfoString(buf, "?>");
1674 :
1675 48 : return true;
1676 : }
1677 : else
1678 11878 : 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 485 : xml_doctype_in_content(const xmlChar *str)
1712 : {
1713 485 : const xmlChar *p = str;
1714 :
1715 : for (;;)
1716 18 : {
1717 : const xmlChar *e;
1718 :
1719 548 : SKIP_XML_SPACE(p);
1720 503 : if (*p != '<')
1721 103 : return false;
1722 400 : p++;
1723 :
1724 400 : if (*p == '!')
1725 : {
1726 36 : p++;
1727 :
1728 : /* if we see <!DOCTYPE, we can return true */
1729 36 : if (xmlStrncmp(p, (xmlChar *) "DOCTYPE", 7) == 0)
1730 21 : return true;
1731 :
1732 : /* otherwise, if it's not a comment, fail */
1733 15 : if (xmlStrncmp(p, (xmlChar *) "--", 2) != 0)
1734 0 : return false;
1735 : /* find end of comment: find -- and a > must follow */
1736 15 : p = xmlStrstr(p + 2, (xmlChar *) "--");
1737 15 : if (!p || p[2] != '>')
1738 0 : return false;
1739 : /* advance over comment, and keep scanning */
1740 15 : p += 3;
1741 15 : continue;
1742 : }
1743 :
1744 : /* otherwise, if it's not a PI <?target something?>, fail */
1745 364 : if (*p != '?')
1746 361 : return false;
1747 3 : p++;
1748 :
1749 : /* find end of PI (the string ?> is forbidden within a PI) */
1750 3 : e = xmlStrstr(p, (xmlChar *) "?>");
1751 3 : if (!e)
1752 0 : return false;
1753 :
1754 : /* advance over PI, keep scanning */
1755 3 : 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 638 : 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 638 : volatile xmlParserCtxtPtr ctxt = NULL;
1797 638 : volatile xmlDocPtr doc = NULL;
1798 638 : 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 638 : len = VARSIZE_ANY_EXHDR(data); /* will be useful later */
1805 638 : 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 638 : utf8string = pg_do_encoding_conversion(string,
1815 : len,
1816 : encoding,
1817 : PG_UTF8);
1818 :
1819 : /* Start up libxml and its parser */
1820 638 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_WELLFORMED);
1821 :
1822 : /* Use a TRY block to ensure we clean up correctly */
1823 638 : PG_TRY();
1824 : {
1825 638 : bool parse_as_document = false;
1826 : int res_code;
1827 638 : size_t count = 0;
1828 638 : xmlChar *version = NULL;
1829 638 : int standalone = 0;
1830 :
1831 : /* Any errors here are reported as hard ereport's */
1832 638 : xmlInitParser();
1833 :
1834 : /* Decide whether to parse as document or content */
1835 638 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1836 147 : parse_as_document = true;
1837 : else
1838 : {
1839 : /* Parse and skip over the XML declaration, if any */
1840 491 : res_code = parse_xml_decl(utf8string,
1841 : &count, &version, NULL, &standalone);
1842 491 : if (res_code != 0)
1843 : {
1844 6 : 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 6 : goto fail;
1849 : }
1850 :
1851 : /* Is there a DOCTYPE element? */
1852 485 : if (xml_doctype_in_content(utf8string + count))
1853 21 : parse_as_document = true;
1854 : }
1855 :
1856 : /* initialize output parameters */
1857 632 : if (parsed_xmloptiontype != NULL)
1858 72 : *parsed_xmloptiontype = parse_as_document ? XMLOPTION_DOCUMENT :
1859 : XMLOPTION_CONTENT;
1860 632 : if (parsed_nodes != NULL)
1861 72 : *parsed_nodes = NULL;
1862 :
1863 632 : if (parse_as_document)
1864 : {
1865 : int options;
1866 :
1867 : /* set up parser context used by xmlCtxtReadDoc */
1868 168 : ctxt = xmlNewParserCtxt();
1869 168 : 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 168 : options = XML_PARSE_NOENT | XML_PARSE_DTDATTR
1883 168 : | (preserve_whitespace ? 0 : XML_PARSE_NOBLANKS);
1884 :
1885 168 : doc = xmlCtxtReadDoc(ctxt, utf8string,
1886 : NULL, /* no URL */
1887 : "UTF-8",
1888 : options);
1889 :
1890 168 : if (doc == NULL || xmlerrcxt->err_occurred)
1891 : {
1892 : /* Use original option to decide which error code to report */
1893 72 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1894 69 : xml_errsave(escontext, xmlerrcxt,
1895 : ERRCODE_INVALID_XML_DOCUMENT,
1896 : "invalid XML document");
1897 : else
1898 3 : xml_errsave(escontext, xmlerrcxt,
1899 : ERRCODE_INVALID_XML_CONTENT,
1900 : "invalid XML content");
1901 48 : goto fail;
1902 : }
1903 : }
1904 : else
1905 : {
1906 : /* set up document that xmlParseBalancedChunkMemory will add to */
1907 464 : doc = xmlNewDoc(version);
1908 464 : 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 464 : doc->encoding = xmlStrdup((const xmlChar *) "UTF-8");
1914 464 : if (doc->encoding == NULL || xmlerrcxt->err_occurred)
1915 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1916 : "could not allocate XML document");
1917 464 : doc->standalone = standalone;
1918 :
1919 : /* set parse options --- have to do this the ugly way */
1920 464 : save_keep_blanks = xmlKeepBlanksDefault(preserve_whitespace ? 1 : 0);
1921 :
1922 : /* allow empty content */
1923 464 : if (*(utf8string + count))
1924 : {
1925 904 : res_code = xmlParseBalancedChunkMemory(doc, NULL, NULL, 0,
1926 452 : utf8string + count,
1927 : parsed_nodes);
1928 452 : if (res_code != 0 || xmlerrcxt->err_occurred)
1929 : {
1930 30 : xml_errsave(escontext, xmlerrcxt,
1931 : ERRCODE_INVALID_XML_CONTENT,
1932 : "invalid XML content");
1933 6 : goto fail;
1934 : }
1935 : }
1936 : }
1937 :
1938 590 : fail:
1939 : ;
1940 : }
1941 48 : PG_CATCH();
1942 : {
1943 48 : if (save_keep_blanks != -1)
1944 24 : xmlKeepBlanksDefault(save_keep_blanks);
1945 48 : if (doc != NULL)
1946 24 : xmlFreeDoc(doc);
1947 48 : if (ctxt != NULL)
1948 24 : xmlFreeParserCtxt(ctxt);
1949 :
1950 48 : pg_xml_done(xmlerrcxt, true);
1951 :
1952 48 : PG_RE_THROW();
1953 : }
1954 590 : PG_END_TRY();
1955 :
1956 590 : if (save_keep_blanks != -1)
1957 440 : xmlKeepBlanksDefault(save_keep_blanks);
1958 :
1959 590 : if (ctxt != NULL)
1960 144 : xmlFreeParserCtxt(ctxt);
1961 :
1962 590 : pg_xml_done(xmlerrcxt, false);
1963 :
1964 590 : return doc;
1965 : }
1966 :
1967 :
1968 : /*
1969 : * xmlChar<->text conversions
1970 : */
1971 : static xmlChar *
1972 713 : xml_text2xmlChar(text *in)
1973 : {
1974 713 : 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 9 : xmlPgEntityLoader(const char *URL, const char *ID,
2044 : xmlParserCtxtPtr ctxt)
2045 : {
2046 9 : 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 6 : 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 6 : 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 6 : errcxt->err_occurred = false;
2071 :
2072 : /* Include detail only if we have some text from libxml */
2073 6 : if (errcxt->err_buf.len > 0)
2074 6 : detail = errcxt->err_buf.data;
2075 : else
2076 0 : detail = NULL;
2077 :
2078 6 : 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 102 : 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 102 : 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 102 : errcxt->err_occurred = false;
2109 :
2110 : /* Include detail only if we have some text from libxml */
2111 102 : if (errcxt->err_buf.len > 0)
2112 102 : detail = errcxt->err_buf.data;
2113 : else
2114 0 : detail = NULL;
2115 :
2116 102 : errsave(escontext,
2117 : (errcode(sqlcode),
2118 : errmsg_internal("%s", msg),
2119 : detail ? errdetail_internal("%s", detail) : 0));
2120 54 : }
2121 :
2122 :
2123 : /*
2124 : * Error handler for libxml errors and warnings
2125 : */
2126 : static void
2127 199 : xml_errorHandler(void *data, PgXmlErrorPtr error)
2128 : {
2129 199 : PgXmlErrorContext *xmlerrcxt = (PgXmlErrorContext *) data;
2130 199 : xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) error->ctxt;
2131 199 : xmlParserInputPtr input = (ctxt != NULL) ? ctxt->input : NULL;
2132 199 : xmlNodePtr node = error->node;
2133 199 : const xmlChar *name = (node != NULL &&
2134 199 : node->type == XML_ELEMENT_NODE) ? node->name : NULL;
2135 199 : int domain = error->domain;
2136 199 : int level = error->level;
2137 : StringInfoData 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 199 : 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 199 : switch (error->code)
2157 : {
2158 15 : case XML_WAR_NS_URI:
2159 15 : level = XML_ERR_ERROR;
2160 15 : domain = XML_FROM_NAMESPACE;
2161 15 : break;
2162 :
2163 27 : 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 27 : domain = XML_FROM_NAMESPACE;
2172 27 : break;
2173 : }
2174 :
2175 : /* Decide whether to act on the error or not */
2176 199 : switch (domain)
2177 : {
2178 157 : 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 157 : if (error->code == XML_ERR_NOT_WELL_BALANCED &&
2188 15 : xmlerrcxt->err_occurred)
2189 52 : return;
2190 : pg_fallthrough;
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 142 : if (error->code == XML_WAR_UNDECLARED_ENTITY)
2201 3 : return;
2202 :
2203 : /* Otherwise, accept error regardless of the parsing purpose */
2204 139 : break;
2205 :
2206 42 : default:
2207 : /* Ignore error if only doing well-formedness check */
2208 42 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_WELLFORMED)
2209 33 : return;
2210 9 : break;
2211 : }
2212 :
2213 : /* Prepare error message in errorBuf */
2214 148 : initStringInfo(&errorBuf);
2215 :
2216 148 : if (error->line > 0)
2217 148 : appendStringInfo(&errorBuf, "line %d: ", error->line);
2218 148 : if (name != NULL)
2219 0 : appendStringInfo(&errorBuf, "element %s: ", name);
2220 148 : if (error->message != NULL)
2221 148 : 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 148 : if (input != NULL)
2237 : {
2238 148 : xmlGenericErrorFunc errFuncSaved = xmlGenericError;
2239 148 : void *errCtxSaved = xmlGenericErrorContext;
2240 :
2241 148 : xmlSetGenericErrorFunc(&errorBuf,
2242 : (xmlGenericErrorFunc) appendStringInfo);
2243 :
2244 : /* Add context information to errorBuf */
2245 148 : appendStringInfoLineSeparator(&errorBuf);
2246 :
2247 148 : xmlParserPrintFileContext(input);
2248 :
2249 : /* Restore generic error func */
2250 148 : xmlSetGenericErrorFunc(errCtxSaved, errFuncSaved);
2251 : }
2252 :
2253 : /* Get rid of any trailing newlines in errorBuf */
2254 148 : 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 148 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_LEGACY)
2265 : {
2266 1 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2267 1 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf.data,
2268 : errorBuf.len);
2269 :
2270 1 : pfree(errorBuf.data);
2271 1 : 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 147 : if (level >= XML_ERR_ERROR)
2283 : {
2284 144 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2285 144 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf.data,
2286 : errorBuf.len);
2287 :
2288 144 : xmlerrcxt->err_occurred = true;
2289 : }
2290 3 : else if (level >= XML_ERR_WARNING)
2291 : {
2292 3 : 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 147 : pfree(errorBuf.data);
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 3 : errdetail_for_xml_code(int code)
2316 : {
2317 : const char *det;
2318 :
2319 3 : 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 3 : case XML_ERR_STANDALONE_VALUE:
2328 3 : det = gettext_noop("standalone accepts only 'yes' or 'no'.");
2329 3 : 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 3 : return errdetail(det, code);
2345 : }
2346 :
2347 :
2348 : /*
2349 : * Remove all trailing newlines from a StringInfo string
2350 : */
2351 : static void
2352 441 : chopStringInfoNewlines(StringInfo str)
2353 : {
2354 737 : while (str->len > 0 && str->data[str->len - 1] == '\n')
2355 296 : str->data[--str->len] = '\0';
2356 441 : }
2357 :
2358 :
2359 : /*
2360 : * Append a newline after removing any existing trailing newlines
2361 : */
2362 : static void
2363 293 : appendStringInfoLineSeparator(StringInfo str)
2364 : {
2365 293 : chopStringInfoNewlines(str);
2366 293 : if (str->len > 0)
2367 184 : appendStringInfoChar(str, '\n');
2368 293 : }
2369 :
2370 :
2371 : /*
2372 : * Convert one char in the current server encoding to a Unicode codepoint.
2373 : */
2374 : static pg_wchar
2375 9140 : sqlchar_to_unicode(const char *s)
2376 : {
2377 : char *utf8string;
2378 : pg_wchar ret[2]; /* need space for trailing zero */
2379 :
2380 9140 : utf8string = pg_server_to_any(s, pg_mblen_cstr(s), PG_UTF8);
2381 :
2382 9140 : pg_encoding_mb2wchar_with_len(PG_UTF8, utf8string, ret,
2383 : pg_encoding_mblen(PG_UTF8, utf8string));
2384 :
2385 9140 : if (utf8string != s)
2386 0 : pfree(utf8string);
2387 :
2388 9140 : return ret[0];
2389 : }
2390 :
2391 :
2392 : static bool
2393 1819 : is_valid_xml_namefirst(pg_wchar c)
2394 : {
2395 : /* (Letter | '_' | ':') */
2396 1822 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2397 3641 : || c == '_' || c == ':');
2398 : }
2399 :
2400 :
2401 : static bool
2402 7321 : is_valid_xml_namechar(pg_wchar c)
2403 : {
2404 : /* Letter | Digit | '.' | '-' | '_' | ':' | CombiningChar | Extender */
2405 7766 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2406 445 : || xmlIsDigitQ(c)
2407 127 : || c == '.' || c == '-' || c == '_' || c == ':'
2408 6 : || xmlIsCombiningQ(c)
2409 15532 : || xmlIsExtenderQ(c));
2410 : }
2411 : #endif /* USE_LIBXML */
2412 :
2413 :
2414 : /*
2415 : * Map SQL identifier to XML name; see SQL/XML:2008 section 9.1.
2416 : */
2417 : char *
2418 1826 : map_sql_identifier_to_xml_name(const char *ident, bool fully_escaped,
2419 : bool escape_period)
2420 : {
2421 : #ifdef USE_LIBXML
2422 : StringInfoData buf;
2423 : const char *p;
2424 :
2425 : /*
2426 : * SQL/XML doesn't make use of this case anywhere, so it's probably a
2427 : * mistake.
2428 : */
2429 : Assert(fully_escaped || !escape_period);
2430 :
2431 1826 : initStringInfo(&buf);
2432 :
2433 10976 : for (p = ident; *p; p += pg_mblen_cstr(p))
2434 : {
2435 9150 : if (*p == ':' && (p == ident || fully_escaped))
2436 7 : appendStringInfoString(&buf, "_x003A_");
2437 9143 : else if (*p == '_' && *(p + 1) == 'x')
2438 3 : appendStringInfoString(&buf, "_x005F_");
2439 10784 : else if (fully_escaped && p == ident &&
2440 1644 : pg_strncasecmp(p, "xml", 3) == 0)
2441 : {
2442 0 : if (*p == 'x')
2443 0 : appendStringInfoString(&buf, "_x0078_");
2444 : else
2445 0 : appendStringInfoString(&buf, "_x0058_");
2446 : }
2447 9140 : else if (escape_period && *p == '.')
2448 0 : appendStringInfoString(&buf, "_x002E_");
2449 : else
2450 : {
2451 9140 : pg_wchar u = sqlchar_to_unicode(p);
2452 :
2453 18280 : if ((p == ident)
2454 1819 : ? !is_valid_xml_namefirst(u)
2455 7321 : : !is_valid_xml_namechar(u))
2456 9 : appendStringInfo(&buf, "_x%04X_", (unsigned int) u);
2457 : else
2458 9131 : appendBinaryStringInfo(&buf, p, pg_mblen_cstr(p));
2459 : }
2460 : }
2461 :
2462 1826 : return buf.data;
2463 : #else /* not USE_LIBXML */
2464 : NO_XML_SUPPORT();
2465 : return NULL;
2466 : #endif /* not USE_LIBXML */
2467 : }
2468 :
2469 :
2470 : /*
2471 : * Map XML name to SQL identifier; see SQL/XML:2008 section 9.3.
2472 : */
2473 : char *
2474 64 : map_xml_name_to_sql_identifier(const char *name)
2475 : {
2476 : StringInfoData buf;
2477 : const char *p;
2478 :
2479 64 : initStringInfo(&buf);
2480 :
2481 352 : for (p = name; *p; p += pg_mblen_cstr(p))
2482 : {
2483 288 : if (*p == '_' && *(p + 1) == 'x'
2484 8 : && isxdigit((unsigned char) *(p + 2))
2485 8 : && isxdigit((unsigned char) *(p + 3))
2486 8 : && isxdigit((unsigned char) *(p + 4))
2487 8 : && isxdigit((unsigned char) *(p + 5))
2488 8 : && *(p + 6) == '_')
2489 8 : {
2490 : char cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
2491 : unsigned int u;
2492 :
2493 8 : sscanf(p + 2, "%X", &u);
2494 8 : pg_unicode_to_server(u, (unsigned char *) cbuf);
2495 8 : appendStringInfoString(&buf, cbuf);
2496 8 : p += 6;
2497 : }
2498 : else
2499 280 : appendBinaryStringInfo(&buf, p, pg_mblen_cstr(p));
2500 : }
2501 :
2502 64 : return buf.data;
2503 : }
2504 :
2505 : /*
2506 : * Map SQL value to XML value; see SQL/XML:2008 section 9.8.
2507 : *
2508 : * When xml_escape_strings is true, then certain characters in string
2509 : * values are replaced by entity references (< etc.), as specified
2510 : * in SQL/XML:2008 section 9.8 GR 9) a) iii). This is normally what is
2511 : * wanted. The false case is mainly useful when the resulting value
2512 : * is used with xmlTextWriterWriteAttribute() to write out an
2513 : * attribute, because that function does the escaping itself.
2514 : */
2515 : char *
2516 67963 : map_sql_value_to_xml_value(Datum value, Oid type, bool xml_escape_strings)
2517 : {
2518 67963 : if (type_is_array_domain(type))
2519 : {
2520 : ArrayType *array;
2521 : Oid elmtype;
2522 : int16 elmlen;
2523 : bool elmbyval;
2524 : char elmalign;
2525 : int num_elems;
2526 : Datum *elem_values;
2527 : bool *elem_nulls;
2528 : StringInfoData buf;
2529 : int i;
2530 :
2531 3 : array = DatumGetArrayTypeP(value);
2532 3 : elmtype = ARR_ELEMTYPE(array);
2533 3 : get_typlenbyvalalign(elmtype, &elmlen, &elmbyval, &elmalign);
2534 :
2535 3 : deconstruct_array(array, elmtype,
2536 : elmlen, elmbyval, elmalign,
2537 : &elem_values, &elem_nulls,
2538 : &num_elems);
2539 :
2540 3 : initStringInfo(&buf);
2541 :
2542 12 : for (i = 0; i < num_elems; i++)
2543 : {
2544 9 : if (elem_nulls[i])
2545 0 : continue;
2546 9 : appendStringInfoString(&buf, "<element>");
2547 9 : appendStringInfoString(&buf,
2548 9 : map_sql_value_to_xml_value(elem_values[i],
2549 : elmtype, true));
2550 9 : appendStringInfoString(&buf, "</element>");
2551 : }
2552 :
2553 3 : pfree(elem_values);
2554 3 : pfree(elem_nulls);
2555 :
2556 3 : return buf.data;
2557 : }
2558 : else
2559 : {
2560 : Oid typeOut;
2561 : bool isvarlena;
2562 : char *str;
2563 :
2564 : /*
2565 : * Flatten domains; the special-case treatments below should apply to,
2566 : * eg, domains over boolean not just boolean.
2567 : */
2568 67960 : type = getBaseType(type);
2569 :
2570 : /*
2571 : * Special XSD formatting for some data types
2572 : */
2573 67960 : switch (type)
2574 : {
2575 33 : case BOOLOID:
2576 33 : if (DatumGetBool(value))
2577 30 : return "true";
2578 : else
2579 3 : return "false";
2580 :
2581 24 : case DATEOID:
2582 : {
2583 : DateADT date;
2584 : struct pg_tm tm;
2585 : char buf[MAXDATELEN + 1];
2586 :
2587 24 : date = DatumGetDateADT(value);
2588 : /* XSD doesn't support infinite values */
2589 24 : if (DATE_NOT_FINITE(date))
2590 0 : ereport(ERROR,
2591 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2592 : errmsg("date out of range"),
2593 : errdetail("XML does not support infinite date values.")));
2594 24 : j2date(date + POSTGRES_EPOCH_JDATE,
2595 : &(tm.tm_year), &(tm.tm_mon), &(tm.tm_mday));
2596 24 : EncodeDateOnly(&tm, USE_XSD_DATES, buf);
2597 :
2598 24 : return pstrdup(buf);
2599 : }
2600 :
2601 18 : case TIMESTAMPOID:
2602 : {
2603 : Timestamp timestamp;
2604 : struct pg_tm tm;
2605 : fsec_t fsec;
2606 : char buf[MAXDATELEN + 1];
2607 :
2608 18 : timestamp = DatumGetTimestamp(value);
2609 :
2610 : /* XSD doesn't support infinite values */
2611 18 : if (TIMESTAMP_NOT_FINITE(timestamp))
2612 3 : ereport(ERROR,
2613 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2614 : errmsg("timestamp out of range"),
2615 : errdetail("XML does not support infinite timestamp values.")));
2616 15 : else if (timestamp2tm(timestamp, NULL, &tm, &fsec, NULL, NULL) == 0)
2617 15 : EncodeDateTime(&tm, fsec, false, 0, NULL, USE_XSD_DATES, buf);
2618 : else
2619 0 : ereport(ERROR,
2620 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2621 : errmsg("timestamp out of range")));
2622 :
2623 15 : return pstrdup(buf);
2624 : }
2625 :
2626 12 : case TIMESTAMPTZOID:
2627 : {
2628 : TimestampTz timestamp;
2629 : struct pg_tm tm;
2630 : int tz;
2631 : fsec_t fsec;
2632 12 : const char *tzn = NULL;
2633 : char buf[MAXDATELEN + 1];
2634 :
2635 12 : timestamp = DatumGetTimestamp(value);
2636 :
2637 : /* XSD doesn't support infinite values */
2638 12 : if (TIMESTAMP_NOT_FINITE(timestamp))
2639 0 : ereport(ERROR,
2640 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2641 : errmsg("timestamp out of range"),
2642 : errdetail("XML does not support infinite timestamp values.")));
2643 12 : else if (timestamp2tm(timestamp, &tz, &tm, &fsec, &tzn, NULL) == 0)
2644 12 : EncodeDateTime(&tm, fsec, true, tz, tzn, USE_XSD_DATES, buf);
2645 : else
2646 0 : ereport(ERROR,
2647 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2648 : errmsg("timestamp out of range")));
2649 :
2650 12 : return pstrdup(buf);
2651 : }
2652 :
2653 : #ifdef USE_LIBXML
2654 18 : case BYTEAOID:
2655 : {
2656 18 : bytea *bstr = DatumGetByteaPP(value);
2657 : PgXmlErrorContext *xmlerrcxt;
2658 18 : volatile xmlBufferPtr buf = NULL;
2659 18 : volatile xmlTextWriterPtr writer = NULL;
2660 : char *result;
2661 :
2662 18 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
2663 :
2664 18 : PG_TRY();
2665 : {
2666 18 : buf = xmlBufferCreate();
2667 18 : if (buf == NULL || xmlerrcxt->err_occurred)
2668 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2669 : "could not allocate xmlBuffer");
2670 18 : writer = xmlNewTextWriterMemory(buf, 0);
2671 18 : if (writer == NULL || xmlerrcxt->err_occurred)
2672 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2673 : "could not allocate xmlTextWriter");
2674 :
2675 18 : if (xmlbinary == XMLBINARY_BASE64)
2676 15 : xmlTextWriterWriteBase64(writer, VARDATA_ANY(bstr),
2677 15 : 0, VARSIZE_ANY_EXHDR(bstr));
2678 : else
2679 3 : xmlTextWriterWriteBinHex(writer, VARDATA_ANY(bstr),
2680 3 : 0, VARSIZE_ANY_EXHDR(bstr));
2681 :
2682 : /* we MUST do this now to flush data out to the buffer */
2683 18 : xmlFreeTextWriter(writer);
2684 18 : writer = NULL;
2685 :
2686 18 : result = pstrdup((const char *) xmlBufferContent(buf));
2687 : }
2688 0 : PG_CATCH();
2689 : {
2690 0 : if (writer)
2691 0 : xmlFreeTextWriter(writer);
2692 0 : if (buf)
2693 0 : xmlBufferFree(buf);
2694 :
2695 0 : pg_xml_done(xmlerrcxt, true);
2696 :
2697 0 : PG_RE_THROW();
2698 : }
2699 18 : PG_END_TRY();
2700 :
2701 18 : xmlBufferFree(buf);
2702 :
2703 18 : pg_xml_done(xmlerrcxt, false);
2704 :
2705 18 : return result;
2706 : }
2707 : #endif /* USE_LIBXML */
2708 :
2709 : }
2710 :
2711 : /*
2712 : * otherwise, just use the type's native text representation
2713 : */
2714 67855 : getTypeOutputInfo(type, &typeOut, &isvarlena);
2715 67855 : str = OidOutputFunctionCall(typeOut, value);
2716 :
2717 : /* ... exactly as-is for XML, and when escaping is not wanted */
2718 67855 : if (type == XMLOID || !xml_escape_strings)
2719 11260 : return str;
2720 :
2721 : /* otherwise, translate special characters as needed */
2722 56595 : return escape_xml(str);
2723 : }
2724 : }
2725 :
2726 :
2727 : /*
2728 : * Escape characters in text that have special meanings in XML.
2729 : *
2730 : * Returns a palloc'd string.
2731 : *
2732 : * NB: this is intentionally not dependent on libxml.
2733 : */
2734 : char *
2735 56832 : escape_xml(const char *str)
2736 : {
2737 : StringInfoData buf;
2738 : const char *p;
2739 :
2740 56832 : initStringInfo(&buf);
2741 358497 : for (p = str; *p; p++)
2742 : {
2743 301665 : switch (*p)
2744 : {
2745 0 : case '&':
2746 0 : appendStringInfoString(&buf, "&");
2747 0 : break;
2748 18 : case '<':
2749 18 : appendStringInfoString(&buf, "<");
2750 18 : break;
2751 12 : case '>':
2752 12 : appendStringInfoString(&buf, ">");
2753 12 : break;
2754 0 : case '\r':
2755 0 : appendStringInfoString(&buf, "
");
2756 0 : break;
2757 301635 : default:
2758 301635 : appendStringInfoCharMacro(&buf, *p);
2759 301635 : break;
2760 : }
2761 : }
2762 56832 : return buf.data;
2763 : }
2764 :
2765 :
2766 : static char *
2767 12 : _SPI_strdup(const char *s)
2768 : {
2769 12 : size_t len = strlen(s) + 1;
2770 12 : char *ret = SPI_palloc(len);
2771 :
2772 12 : memcpy(ret, s, len);
2773 12 : return ret;
2774 : }
2775 :
2776 :
2777 : /*
2778 : * SQL to XML mapping functions
2779 : *
2780 : * What follows below was at one point intentionally organized so that
2781 : * you can read along in the SQL/XML standard. The functions are
2782 : * mostly split up the way the clauses lay out in the standards
2783 : * document, and the identifiers are also aligned with the standard
2784 : * text. Unfortunately, SQL/XML:2006 reordered the clauses
2785 : * differently than SQL/XML:2003, so the order below doesn't make much
2786 : * sense anymore.
2787 : *
2788 : * There are many things going on there:
2789 : *
2790 : * There are two kinds of mappings: Mapping SQL data (table contents)
2791 : * to XML documents, and mapping SQL structure (the "schema") to XML
2792 : * Schema. And there are functions that do both at the same time.
2793 : *
2794 : * Then you can map a database, a schema, or a table, each in both
2795 : * ways. This breaks down recursively: Mapping a database invokes
2796 : * mapping schemas, which invokes mapping tables, which invokes
2797 : * mapping rows, which invokes mapping columns, although you can't
2798 : * call the last two from the outside. Because of this, there are a
2799 : * number of xyz_internal() functions which are to be called both from
2800 : * the function manager wrapper and from some upper layer in a
2801 : * recursive call.
2802 : *
2803 : * See the documentation about what the common function arguments
2804 : * nulls, tableforest, and targetns mean.
2805 : *
2806 : * Some style guidelines for XML output: Use double quotes for quoting
2807 : * XML attributes. Indent XML elements by two spaces, but remember
2808 : * that a lot of code is called recursively at different levels, so
2809 : * it's better not to indent rather than create output that indents
2810 : * and outdents weirdly. Add newlines to make the output look nice.
2811 : */
2812 :
2813 :
2814 : /*
2815 : * Visibility of objects for XML mappings; see SQL/XML:2008 section
2816 : * 4.10.8.
2817 : */
2818 :
2819 : /*
2820 : * Given a query, which must return type oid as first column, produce
2821 : * a list of Oids with the query results.
2822 : */
2823 : static List *
2824 18 : query_to_oid_list(const char *query)
2825 : {
2826 : uint64 i;
2827 18 : List *list = NIL;
2828 : int spi_result;
2829 :
2830 18 : spi_result = SPI_execute(query, true, 0);
2831 18 : if (spi_result != SPI_OK_SELECT)
2832 0 : elog(ERROR, "SPI_execute returned %s for %s",
2833 : SPI_result_code_string(spi_result), query);
2834 :
2835 54 : for (i = 0; i < SPI_processed; i++)
2836 : {
2837 : Datum oid;
2838 : bool isnull;
2839 :
2840 36 : oid = SPI_getbinval(SPI_tuptable->vals[i],
2841 36 : SPI_tuptable->tupdesc,
2842 : 1,
2843 : &isnull);
2844 36 : if (!isnull)
2845 36 : list = lappend_oid(list, DatumGetObjectId(oid));
2846 : }
2847 :
2848 18 : return list;
2849 : }
2850 :
2851 :
2852 : static List *
2853 18 : schema_get_xml_visible_tables(Oid nspid)
2854 : {
2855 : StringInfoData query;
2856 :
2857 18 : initStringInfo(&query);
2858 18 : appendStringInfo(&query, "SELECT oid FROM pg_catalog.pg_class"
2859 : " WHERE relnamespace = %u AND relkind IN ("
2860 : CppAsString2(RELKIND_RELATION) ","
2861 : CppAsString2(RELKIND_MATVIEW) ","
2862 : CppAsString2(RELKIND_VIEW) ")"
2863 : " AND pg_catalog.has_table_privilege (oid, 'SELECT')"
2864 : " ORDER BY relname;", nspid);
2865 :
2866 18 : return query_to_oid_list(query.data);
2867 : }
2868 :
2869 :
2870 : /*
2871 : * Including the system schemas is probably not useful for a database
2872 : * mapping.
2873 : */
2874 : #define XML_VISIBLE_SCHEMAS_EXCLUDE "(nspname ~ '^pg_' OR nspname = 'information_schema')"
2875 :
2876 : #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
2877 :
2878 :
2879 : static List *
2880 0 : database_get_xml_visible_schemas(void)
2881 : {
2882 0 : return query_to_oid_list(XML_VISIBLE_SCHEMAS " ORDER BY nspname;");
2883 : }
2884 :
2885 :
2886 : static List *
2887 0 : database_get_xml_visible_tables(void)
2888 : {
2889 : /* At the moment there is no order required here. */
2890 0 : return query_to_oid_list("SELECT oid FROM pg_catalog.pg_class"
2891 : " WHERE relkind IN ("
2892 : CppAsString2(RELKIND_RELATION) ","
2893 : CppAsString2(RELKIND_MATVIEW) ","
2894 : CppAsString2(RELKIND_VIEW) ")"
2895 : " AND pg_catalog.has_table_privilege(pg_class.oid, 'SELECT')"
2896 : " AND relnamespace IN (" XML_VISIBLE_SCHEMAS ");");
2897 : }
2898 :
2899 :
2900 : /*
2901 : * Map SQL table to XML and/or XML Schema document; see SQL/XML:2008
2902 : * section 9.11.
2903 : */
2904 :
2905 : static StringInfo
2906 48 : table_to_xml_internal(Oid relid,
2907 : const char *xmlschema, bool nulls, bool tableforest,
2908 : const char *targetns, bool top_level)
2909 : {
2910 : StringInfoData query;
2911 :
2912 48 : initStringInfo(&query);
2913 48 : appendStringInfo(&query, "SELECT * FROM %s",
2914 : DatumGetCString(DirectFunctionCall1(regclassout,
2915 : ObjectIdGetDatum(relid))));
2916 48 : return query_to_xml_internal(query.data, get_rel_name(relid),
2917 : xmlschema, nulls, tableforest,
2918 : targetns, top_level);
2919 : }
2920 :
2921 :
2922 : Datum
2923 18 : table_to_xml(PG_FUNCTION_ARGS)
2924 : {
2925 18 : Oid relid = PG_GETARG_OID(0);
2926 18 : bool nulls = PG_GETARG_BOOL(1);
2927 18 : bool tableforest = PG_GETARG_BOOL(2);
2928 18 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2929 :
2930 18 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid, NULL,
2931 : nulls, tableforest,
2932 : targetns, true)));
2933 : }
2934 :
2935 :
2936 : Datum
2937 5 : query_to_xml(PG_FUNCTION_ARGS)
2938 : {
2939 5 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
2940 5 : bool nulls = PG_GETARG_BOOL(1);
2941 5 : bool tableforest = PG_GETARG_BOOL(2);
2942 5 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2943 :
2944 5 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
2945 : NULL, nulls, tableforest,
2946 : targetns, true)));
2947 : }
2948 :
2949 :
2950 : Datum
2951 6 : cursor_to_xml(PG_FUNCTION_ARGS)
2952 : {
2953 6 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
2954 6 : int32 count = PG_GETARG_INT32(1);
2955 6 : bool nulls = PG_GETARG_BOOL(2);
2956 6 : bool tableforest = PG_GETARG_BOOL(3);
2957 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(4));
2958 :
2959 : StringInfoData result;
2960 : Portal portal;
2961 : uint64 i;
2962 :
2963 6 : initStringInfo(&result);
2964 :
2965 6 : if (!tableforest)
2966 : {
2967 3 : xmldata_root_element_start(&result, "table", NULL, targetns, true);
2968 3 : appendStringInfoChar(&result, '\n');
2969 : }
2970 :
2971 6 : SPI_connect();
2972 6 : portal = SPI_cursor_find(name);
2973 6 : if (portal == NULL)
2974 0 : ereport(ERROR,
2975 : (errcode(ERRCODE_UNDEFINED_CURSOR),
2976 : errmsg("cursor \"%s\" does not exist", name)));
2977 :
2978 6 : SPI_cursor_fetch(portal, true, count);
2979 24 : for (i = 0; i < SPI_processed; i++)
2980 18 : SPI_sql_row_to_xmlelement(i, &result, NULL, nulls,
2981 : tableforest, targetns, true);
2982 :
2983 6 : SPI_finish();
2984 :
2985 6 : if (!tableforest)
2986 3 : xmldata_root_element_end(&result, "table");
2987 :
2988 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(&result));
2989 : }
2990 :
2991 :
2992 : /*
2993 : * Write the start tag of the root element of a data mapping.
2994 : *
2995 : * top_level means that this is the very top level of the eventual
2996 : * output. For example, when the user calls table_to_xml, then a call
2997 : * with a table name to this function is the top level. When the user
2998 : * calls database_to_xml, then a call with a schema name to this
2999 : * function is not the top level. If top_level is false, then the XML
3000 : * namespace declarations are omitted, because they supposedly already
3001 : * appeared earlier in the output. Repeating them is not wrong, but
3002 : * it looks ugly.
3003 : */
3004 : static void
3005 119 : xmldata_root_element_start(StringInfo result, const char *eltname,
3006 : const char *xmlschema, const char *targetns,
3007 : bool top_level)
3008 : {
3009 : /* This isn't really wrong but currently makes no sense. */
3010 : Assert(top_level || !xmlschema);
3011 :
3012 119 : appendStringInfo(result, "<%s", eltname);
3013 119 : if (top_level)
3014 : {
3015 89 : appendStringInfoString(result, " xmlns:xsi=\"" NAMESPACE_XSI "\"");
3016 89 : if (strlen(targetns) > 0)
3017 15 : appendStringInfo(result, " xmlns=\"%s\"", targetns);
3018 : }
3019 119 : if (xmlschema)
3020 : {
3021 : /* FIXME: better targets */
3022 9 : if (strlen(targetns) > 0)
3023 3 : appendStringInfo(result, " xsi:schemaLocation=\"%s #\"", targetns);
3024 : else
3025 6 : appendStringInfoString(result, " xsi:noNamespaceSchemaLocation=\"#\"");
3026 : }
3027 119 : appendStringInfoString(result, ">\n");
3028 119 : }
3029 :
3030 :
3031 : static void
3032 119 : xmldata_root_element_end(StringInfo result, const char *eltname)
3033 : {
3034 119 : appendStringInfo(result, "</%s>\n", eltname);
3035 119 : }
3036 :
3037 :
3038 : static StringInfo
3039 56 : query_to_xml_internal(const char *query, char *tablename,
3040 : const char *xmlschema, bool nulls, bool tableforest,
3041 : const char *targetns, bool top_level)
3042 : {
3043 : StringInfo result;
3044 : char *xmltn;
3045 : uint64 i;
3046 :
3047 56 : if (tablename)
3048 48 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
3049 : else
3050 8 : xmltn = "table";
3051 :
3052 56 : result = makeStringInfo();
3053 :
3054 56 : SPI_connect();
3055 56 : if (SPI_execute(query, true, 0) != SPI_OK_SELECT)
3056 0 : ereport(ERROR,
3057 : (errcode(ERRCODE_DATA_EXCEPTION),
3058 : errmsg("invalid query")));
3059 :
3060 56 : if (!tableforest)
3061 : {
3062 26 : xmldata_root_element_start(result, xmltn, xmlschema,
3063 : targetns, top_level);
3064 26 : appendStringInfoChar(result, '\n');
3065 : }
3066 :
3067 56 : if (xmlschema)
3068 15 : appendStringInfo(result, "%s\n\n", xmlschema);
3069 :
3070 194 : for (i = 0; i < SPI_processed; i++)
3071 138 : SPI_sql_row_to_xmlelement(i, result, tablename, nulls,
3072 : tableforest, targetns, top_level);
3073 :
3074 56 : if (!tableforest)
3075 26 : xmldata_root_element_end(result, xmltn);
3076 :
3077 56 : SPI_finish();
3078 :
3079 56 : return result;
3080 : }
3081 :
3082 :
3083 : Datum
3084 15 : table_to_xmlschema(PG_FUNCTION_ARGS)
3085 : {
3086 15 : Oid relid = PG_GETARG_OID(0);
3087 15 : bool nulls = PG_GETARG_BOOL(1);
3088 15 : bool tableforest = PG_GETARG_BOOL(2);
3089 15 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3090 : const char *result;
3091 : Relation rel;
3092 :
3093 15 : rel = table_open(relid, AccessShareLock);
3094 15 : result = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
3095 : tableforest, targetns);
3096 15 : table_close(rel, NoLock);
3097 :
3098 15 : PG_RETURN_XML_P(cstring_to_xmltype(result));
3099 : }
3100 :
3101 :
3102 : Datum
3103 3 : query_to_xmlschema(PG_FUNCTION_ARGS)
3104 : {
3105 3 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
3106 3 : bool nulls = PG_GETARG_BOOL(1);
3107 3 : bool tableforest = PG_GETARG_BOOL(2);
3108 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3109 : const char *result;
3110 : SPIPlanPtr plan;
3111 : Portal portal;
3112 :
3113 3 : SPI_connect();
3114 :
3115 3 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
3116 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
3117 :
3118 3 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
3119 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
3120 :
3121 3 : result = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3122 : InvalidOid, nulls,
3123 : tableforest, targetns));
3124 3 : SPI_cursor_close(portal);
3125 3 : SPI_finish();
3126 :
3127 3 : PG_RETURN_XML_P(cstring_to_xmltype(result));
3128 : }
3129 :
3130 :
3131 : Datum
3132 6 : cursor_to_xmlschema(PG_FUNCTION_ARGS)
3133 : {
3134 6 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
3135 6 : bool nulls = PG_GETARG_BOOL(1);
3136 6 : bool tableforest = PG_GETARG_BOOL(2);
3137 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3138 : const char *xmlschema;
3139 : Portal portal;
3140 :
3141 6 : SPI_connect();
3142 6 : portal = SPI_cursor_find(name);
3143 6 : if (portal == NULL)
3144 0 : ereport(ERROR,
3145 : (errcode(ERRCODE_UNDEFINED_CURSOR),
3146 : errmsg("cursor \"%s\" does not exist", name)));
3147 6 : if (portal->tupDesc == NULL)
3148 0 : ereport(ERROR,
3149 : (errcode(ERRCODE_INVALID_CURSOR_STATE),
3150 : errmsg("portal \"%s\" does not return tuples", name)));
3151 :
3152 6 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3153 : InvalidOid, nulls,
3154 : tableforest, targetns));
3155 6 : SPI_finish();
3156 :
3157 6 : PG_RETURN_XML_P(cstring_to_xmltype(xmlschema));
3158 : }
3159 :
3160 :
3161 : Datum
3162 12 : table_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3163 : {
3164 12 : Oid relid = PG_GETARG_OID(0);
3165 12 : bool nulls = PG_GETARG_BOOL(1);
3166 12 : bool tableforest = PG_GETARG_BOOL(2);
3167 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3168 : Relation rel;
3169 : const char *xmlschema;
3170 :
3171 12 : rel = table_open(relid, AccessShareLock);
3172 12 : xmlschema = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
3173 : tableforest, targetns);
3174 12 : table_close(rel, NoLock);
3175 :
3176 12 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid,
3177 : xmlschema, nulls, tableforest,
3178 : targetns, true)));
3179 : }
3180 :
3181 :
3182 : Datum
3183 3 : query_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3184 : {
3185 3 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
3186 3 : bool nulls = PG_GETARG_BOOL(1);
3187 3 : bool tableforest = PG_GETARG_BOOL(2);
3188 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3189 :
3190 : const char *xmlschema;
3191 : SPIPlanPtr plan;
3192 : Portal portal;
3193 :
3194 3 : SPI_connect();
3195 :
3196 3 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
3197 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
3198 :
3199 3 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
3200 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
3201 :
3202 3 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3203 : InvalidOid, nulls, tableforest, targetns));
3204 3 : SPI_cursor_close(portal);
3205 3 : SPI_finish();
3206 :
3207 3 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
3208 : xmlschema, nulls, tableforest,
3209 : targetns, true)));
3210 : }
3211 :
3212 :
3213 : /*
3214 : * Map SQL schema to XML and/or XML Schema document; see SQL/XML:2008
3215 : * sections 9.13, 9.14.
3216 : */
3217 :
3218 : static StringInfo
3219 9 : schema_to_xml_internal(Oid nspid, const char *xmlschema, bool nulls,
3220 : bool tableforest, const char *targetns, bool top_level)
3221 : {
3222 : StringInfo result;
3223 : char *xmlsn;
3224 : List *relid_list;
3225 : ListCell *cell;
3226 :
3227 9 : xmlsn = map_sql_identifier_to_xml_name(get_namespace_name(nspid),
3228 : true, false);
3229 9 : result = makeStringInfo();
3230 :
3231 9 : xmldata_root_element_start(result, xmlsn, xmlschema, targetns, top_level);
3232 9 : appendStringInfoChar(result, '\n');
3233 :
3234 9 : if (xmlschema)
3235 3 : appendStringInfo(result, "%s\n\n", xmlschema);
3236 :
3237 9 : SPI_connect();
3238 :
3239 9 : relid_list = schema_get_xml_visible_tables(nspid);
3240 :
3241 27 : foreach(cell, relid_list)
3242 : {
3243 18 : Oid relid = lfirst_oid(cell);
3244 : StringInfo subres;
3245 :
3246 18 : subres = table_to_xml_internal(relid, NULL, nulls, tableforest,
3247 : targetns, false);
3248 :
3249 18 : appendBinaryStringInfo(result, subres->data, subres->len);
3250 18 : appendStringInfoChar(result, '\n');
3251 : }
3252 :
3253 9 : SPI_finish();
3254 :
3255 9 : xmldata_root_element_end(result, xmlsn);
3256 :
3257 9 : return result;
3258 : }
3259 :
3260 :
3261 : Datum
3262 6 : schema_to_xml(PG_FUNCTION_ARGS)
3263 : {
3264 6 : Name name = PG_GETARG_NAME(0);
3265 6 : bool nulls = PG_GETARG_BOOL(1);
3266 6 : bool tableforest = PG_GETARG_BOOL(2);
3267 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3268 :
3269 : char *schemaname;
3270 : Oid nspid;
3271 :
3272 6 : schemaname = NameStr(*name);
3273 6 : nspid = LookupExplicitNamespace(schemaname, false);
3274 :
3275 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid, NULL,
3276 : nulls, tableforest, targetns, true)));
3277 : }
3278 :
3279 :
3280 : /*
3281 : * Write the start element of the root element of an XML Schema mapping.
3282 : */
3283 : static void
3284 48 : xsd_schema_element_start(StringInfo result, const char *targetns)
3285 : {
3286 48 : appendStringInfoString(result,
3287 : "<xsd:schema\n"
3288 : " xmlns:xsd=\"" NAMESPACE_XSD "\"");
3289 48 : if (strlen(targetns) > 0)
3290 9 : appendStringInfo(result,
3291 : "\n"
3292 : " targetNamespace=\"%s\"\n"
3293 : " elementFormDefault=\"qualified\"",
3294 : targetns);
3295 48 : appendStringInfoString(result,
3296 : ">\n\n");
3297 48 : }
3298 :
3299 :
3300 : static void
3301 48 : xsd_schema_element_end(StringInfo result)
3302 : {
3303 48 : appendStringInfoString(result, "</xsd:schema>");
3304 48 : }
3305 :
3306 :
3307 : static StringInfo
3308 9 : schema_to_xmlschema_internal(const char *schemaname, bool nulls,
3309 : bool tableforest, const char *targetns)
3310 : {
3311 : Oid nspid;
3312 : List *relid_list;
3313 : List *tupdesc_list;
3314 : ListCell *cell;
3315 : StringInfo result;
3316 :
3317 9 : result = makeStringInfo();
3318 :
3319 9 : nspid = LookupExplicitNamespace(schemaname, false);
3320 :
3321 9 : xsd_schema_element_start(result, targetns);
3322 :
3323 9 : SPI_connect();
3324 :
3325 9 : relid_list = schema_get_xml_visible_tables(nspid);
3326 :
3327 9 : tupdesc_list = NIL;
3328 27 : foreach(cell, relid_list)
3329 : {
3330 : Relation rel;
3331 :
3332 18 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3333 18 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3334 18 : table_close(rel, NoLock);
3335 : }
3336 :
3337 9 : appendStringInfoString(result,
3338 : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3339 :
3340 9 : appendStringInfoString(result,
3341 : map_sql_schema_to_xmlschema_types(nspid, relid_list,
3342 : nulls, tableforest, targetns));
3343 :
3344 9 : xsd_schema_element_end(result);
3345 :
3346 9 : SPI_finish();
3347 :
3348 9 : return result;
3349 : }
3350 :
3351 :
3352 : Datum
3353 6 : schema_to_xmlschema(PG_FUNCTION_ARGS)
3354 : {
3355 6 : Name name = PG_GETARG_NAME(0);
3356 6 : bool nulls = PG_GETARG_BOOL(1);
3357 6 : bool tableforest = PG_GETARG_BOOL(2);
3358 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3359 :
3360 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xmlschema_internal(NameStr(*name),
3361 : nulls, tableforest, targetns)));
3362 : }
3363 :
3364 :
3365 : Datum
3366 3 : schema_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3367 : {
3368 3 : Name name = PG_GETARG_NAME(0);
3369 3 : bool nulls = PG_GETARG_BOOL(1);
3370 3 : bool tableforest = PG_GETARG_BOOL(2);
3371 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3372 : char *schemaname;
3373 : Oid nspid;
3374 : StringInfo xmlschema;
3375 :
3376 3 : schemaname = NameStr(*name);
3377 3 : nspid = LookupExplicitNamespace(schemaname, false);
3378 :
3379 3 : xmlschema = schema_to_xmlschema_internal(schemaname, nulls,
3380 : tableforest, targetns);
3381 :
3382 3 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid,
3383 : xmlschema->data, nulls,
3384 : tableforest, targetns, true)));
3385 : }
3386 :
3387 :
3388 : /*
3389 : * Map SQL database to XML and/or XML Schema document; see SQL/XML:2008
3390 : * sections 9.16, 9.17.
3391 : */
3392 :
3393 : static StringInfo
3394 0 : database_to_xml_internal(const char *xmlschema, bool nulls,
3395 : bool tableforest, const char *targetns)
3396 : {
3397 : StringInfo result;
3398 : List *nspid_list;
3399 : ListCell *cell;
3400 : char *xmlcn;
3401 :
3402 0 : xmlcn = map_sql_identifier_to_xml_name(get_database_name(MyDatabaseId),
3403 : true, false);
3404 0 : result = makeStringInfo();
3405 :
3406 0 : xmldata_root_element_start(result, xmlcn, xmlschema, targetns, true);
3407 0 : appendStringInfoChar(result, '\n');
3408 :
3409 0 : if (xmlschema)
3410 0 : appendStringInfo(result, "%s\n\n", xmlschema);
3411 :
3412 0 : SPI_connect();
3413 :
3414 0 : nspid_list = database_get_xml_visible_schemas();
3415 :
3416 0 : foreach(cell, nspid_list)
3417 : {
3418 0 : Oid nspid = lfirst_oid(cell);
3419 : StringInfo subres;
3420 :
3421 0 : subres = schema_to_xml_internal(nspid, NULL, nulls,
3422 : tableforest, targetns, false);
3423 :
3424 0 : appendBinaryStringInfo(result, subres->data, subres->len);
3425 0 : appendStringInfoChar(result, '\n');
3426 : }
3427 :
3428 0 : SPI_finish();
3429 :
3430 0 : xmldata_root_element_end(result, xmlcn);
3431 :
3432 0 : return result;
3433 : }
3434 :
3435 :
3436 : Datum
3437 0 : database_to_xml(PG_FUNCTION_ARGS)
3438 : {
3439 0 : bool nulls = PG_GETARG_BOOL(0);
3440 0 : bool tableforest = PG_GETARG_BOOL(1);
3441 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3442 :
3443 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(NULL, nulls,
3444 : tableforest, targetns)));
3445 : }
3446 :
3447 :
3448 : static StringInfo
3449 0 : database_to_xmlschema_internal(bool nulls, bool tableforest,
3450 : const char *targetns)
3451 : {
3452 : List *relid_list;
3453 : List *nspid_list;
3454 : List *tupdesc_list;
3455 : ListCell *cell;
3456 : StringInfo result;
3457 :
3458 0 : result = makeStringInfo();
3459 :
3460 0 : xsd_schema_element_start(result, targetns);
3461 :
3462 0 : SPI_connect();
3463 :
3464 0 : relid_list = database_get_xml_visible_tables();
3465 0 : nspid_list = database_get_xml_visible_schemas();
3466 :
3467 0 : tupdesc_list = NIL;
3468 0 : foreach(cell, relid_list)
3469 : {
3470 : Relation rel;
3471 :
3472 0 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3473 0 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3474 0 : table_close(rel, NoLock);
3475 : }
3476 :
3477 0 : appendStringInfoString(result,
3478 : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3479 :
3480 0 : appendStringInfoString(result,
3481 : map_sql_catalog_to_xmlschema_types(nspid_list, nulls, tableforest, targetns));
3482 :
3483 0 : xsd_schema_element_end(result);
3484 :
3485 0 : SPI_finish();
3486 :
3487 0 : return result;
3488 : }
3489 :
3490 :
3491 : Datum
3492 0 : database_to_xmlschema(PG_FUNCTION_ARGS)
3493 : {
3494 0 : bool nulls = PG_GETARG_BOOL(0);
3495 0 : bool tableforest = PG_GETARG_BOOL(1);
3496 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3497 :
3498 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xmlschema_internal(nulls,
3499 : tableforest, targetns)));
3500 : }
3501 :
3502 :
3503 : Datum
3504 0 : database_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3505 : {
3506 0 : bool nulls = PG_GETARG_BOOL(0);
3507 0 : bool tableforest = PG_GETARG_BOOL(1);
3508 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3509 : StringInfo xmlschema;
3510 :
3511 0 : xmlschema = database_to_xmlschema_internal(nulls, tableforest, targetns);
3512 :
3513 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(xmlschema->data,
3514 : nulls, tableforest, targetns)));
3515 : }
3516 :
3517 :
3518 : /*
3519 : * Map a multi-part SQL name to an XML name; see SQL/XML:2008 section
3520 : * 9.2.
3521 : */
3522 : static char *
3523 192 : map_multipart_sql_identifier_to_xml_name(const char *a, const char *b, const char *c, const char *d)
3524 : {
3525 : StringInfoData result;
3526 :
3527 192 : initStringInfo(&result);
3528 :
3529 192 : if (a)
3530 192 : appendStringInfoString(&result,
3531 192 : map_sql_identifier_to_xml_name(a, true, true));
3532 192 : if (b)
3533 192 : appendStringInfo(&result, ".%s",
3534 : map_sql_identifier_to_xml_name(b, true, true));
3535 192 : if (c)
3536 192 : appendStringInfo(&result, ".%s",
3537 : map_sql_identifier_to_xml_name(c, true, true));
3538 192 : if (d)
3539 183 : appendStringInfo(&result, ".%s",
3540 : map_sql_identifier_to_xml_name(d, true, true));
3541 :
3542 192 : return result.data;
3543 : }
3544 :
3545 :
3546 : /*
3547 : * Map an SQL table to an XML Schema document; see SQL/XML:2008
3548 : * section 9.11.
3549 : *
3550 : * Map an SQL table to XML Schema data types; see SQL/XML:2008 section
3551 : * 9.9.
3552 : */
3553 : static const char *
3554 39 : map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid, bool nulls,
3555 : bool tableforest, const char *targetns)
3556 : {
3557 : int i;
3558 : char *xmltn;
3559 : char *tabletypename;
3560 : char *rowtypename;
3561 : StringInfoData result;
3562 :
3563 39 : initStringInfo(&result);
3564 :
3565 39 : if (OidIsValid(relid))
3566 : {
3567 : HeapTuple tuple;
3568 : Form_pg_class reltuple;
3569 :
3570 27 : tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relid));
3571 27 : if (!HeapTupleIsValid(tuple))
3572 0 : elog(ERROR, "cache lookup failed for relation %u", relid);
3573 27 : reltuple = (Form_pg_class) GETSTRUCT(tuple);
3574 :
3575 27 : xmltn = map_sql_identifier_to_xml_name(NameStr(reltuple->relname),
3576 : true, false);
3577 :
3578 27 : tabletypename = map_multipart_sql_identifier_to_xml_name("TableType",
3579 27 : get_database_name(MyDatabaseId),
3580 27 : get_namespace_name(reltuple->relnamespace),
3581 27 : NameStr(reltuple->relname));
3582 :
3583 27 : rowtypename = map_multipart_sql_identifier_to_xml_name("RowType",
3584 27 : get_database_name(MyDatabaseId),
3585 27 : get_namespace_name(reltuple->relnamespace),
3586 27 : NameStr(reltuple->relname));
3587 :
3588 27 : ReleaseSysCache(tuple);
3589 : }
3590 : else
3591 : {
3592 12 : if (tableforest)
3593 6 : xmltn = "row";
3594 : else
3595 6 : xmltn = "table";
3596 :
3597 12 : tabletypename = "TableType";
3598 12 : rowtypename = "RowType";
3599 : }
3600 :
3601 39 : xsd_schema_element_start(&result, targetns);
3602 :
3603 39 : appendStringInfoString(&result,
3604 39 : map_sql_typecoll_to_xmlschema_types(list_make1(tupdesc)));
3605 :
3606 39 : appendStringInfo(&result,
3607 : "<xsd:complexType name=\"%s\">\n"
3608 : " <xsd:sequence>\n",
3609 : rowtypename);
3610 :
3611 162 : for (i = 0; i < tupdesc->natts; i++)
3612 : {
3613 123 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3614 :
3615 123 : if (att->attisdropped)
3616 3 : continue;
3617 240 : appendStringInfo(&result,
3618 : " <xsd:element name=\"%s\" type=\"%s\"%s></xsd:element>\n",
3619 120 : map_sql_identifier_to_xml_name(NameStr(att->attname),
3620 : true, false),
3621 : map_sql_type_to_xml_name(att->atttypid, -1),
3622 : nulls ? " nillable=\"true\"" : " minOccurs=\"0\"");
3623 : }
3624 :
3625 39 : appendStringInfoString(&result,
3626 : " </xsd:sequence>\n"
3627 : "</xsd:complexType>\n\n");
3628 :
3629 39 : if (!tableforest)
3630 : {
3631 21 : appendStringInfo(&result,
3632 : "<xsd:complexType name=\"%s\">\n"
3633 : " <xsd:sequence>\n"
3634 : " <xsd:element name=\"row\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n"
3635 : " </xsd:sequence>\n"
3636 : "</xsd:complexType>\n\n",
3637 : tabletypename, rowtypename);
3638 :
3639 21 : appendStringInfo(&result,
3640 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3641 : xmltn, tabletypename);
3642 : }
3643 : else
3644 18 : appendStringInfo(&result,
3645 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3646 : xmltn, rowtypename);
3647 :
3648 39 : xsd_schema_element_end(&result);
3649 :
3650 39 : return result.data;
3651 : }
3652 :
3653 :
3654 : /*
3655 : * Map an SQL schema to XML Schema data types; see SQL/XML:2008
3656 : * section 9.12.
3657 : */
3658 : static const char *
3659 9 : map_sql_schema_to_xmlschema_types(Oid nspid, List *relid_list, bool nulls,
3660 : bool tableforest, const char *targetns)
3661 : {
3662 : char *dbname;
3663 : char *nspname;
3664 : char *xmlsn;
3665 : char *schematypename;
3666 : StringInfoData result;
3667 : ListCell *cell;
3668 :
3669 9 : dbname = get_database_name(MyDatabaseId);
3670 9 : nspname = get_namespace_name(nspid);
3671 :
3672 9 : initStringInfo(&result);
3673 :
3674 9 : xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3675 :
3676 9 : schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3677 : dbname,
3678 : nspname,
3679 : NULL);
3680 :
3681 9 : appendStringInfo(&result,
3682 : "<xsd:complexType name=\"%s\">\n", schematypename);
3683 9 : if (!tableforest)
3684 3 : appendStringInfoString(&result,
3685 : " <xsd:all>\n");
3686 : else
3687 6 : appendStringInfoString(&result,
3688 : " <xsd:sequence>\n");
3689 :
3690 27 : foreach(cell, relid_list)
3691 : {
3692 18 : Oid relid = lfirst_oid(cell);
3693 18 : char *relname = get_rel_name(relid);
3694 18 : char *xmltn = map_sql_identifier_to_xml_name(relname, true, false);
3695 18 : char *tabletypename = map_multipart_sql_identifier_to_xml_name(tableforest ? "RowType" : "TableType",
3696 : dbname,
3697 : nspname,
3698 : relname);
3699 :
3700 18 : if (!tableforest)
3701 6 : appendStringInfo(&result,
3702 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3703 : xmltn, tabletypename);
3704 : else
3705 12 : appendStringInfo(&result,
3706 : " <xsd:element name=\"%s\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n",
3707 : xmltn, tabletypename);
3708 : }
3709 :
3710 9 : if (!tableforest)
3711 3 : appendStringInfoString(&result,
3712 : " </xsd:all>\n");
3713 : else
3714 6 : appendStringInfoString(&result,
3715 : " </xsd:sequence>\n");
3716 9 : appendStringInfoString(&result,
3717 : "</xsd:complexType>\n\n");
3718 :
3719 9 : appendStringInfo(&result,
3720 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3721 : xmlsn, schematypename);
3722 :
3723 9 : return result.data;
3724 : }
3725 :
3726 :
3727 : /*
3728 : * Map an SQL catalog to XML Schema data types; see SQL/XML:2008
3729 : * section 9.15.
3730 : */
3731 : static const char *
3732 0 : map_sql_catalog_to_xmlschema_types(List *nspid_list, bool nulls,
3733 : bool tableforest, const char *targetns)
3734 : {
3735 : char *dbname;
3736 : char *xmlcn;
3737 : char *catalogtypename;
3738 : StringInfoData result;
3739 : ListCell *cell;
3740 :
3741 0 : dbname = get_database_name(MyDatabaseId);
3742 :
3743 0 : initStringInfo(&result);
3744 :
3745 0 : xmlcn = map_sql_identifier_to_xml_name(dbname, true, false);
3746 :
3747 0 : catalogtypename = map_multipart_sql_identifier_to_xml_name("CatalogType",
3748 : dbname,
3749 : NULL,
3750 : NULL);
3751 :
3752 0 : appendStringInfo(&result,
3753 : "<xsd:complexType name=\"%s\">\n", catalogtypename);
3754 0 : appendStringInfoString(&result,
3755 : " <xsd:all>\n");
3756 :
3757 0 : foreach(cell, nspid_list)
3758 : {
3759 0 : Oid nspid = lfirst_oid(cell);
3760 0 : char *nspname = get_namespace_name(nspid);
3761 0 : char *xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3762 0 : char *schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3763 : dbname,
3764 : nspname,
3765 : NULL);
3766 :
3767 0 : appendStringInfo(&result,
3768 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3769 : xmlsn, schematypename);
3770 : }
3771 :
3772 0 : appendStringInfoString(&result,
3773 : " </xsd:all>\n");
3774 0 : appendStringInfoString(&result,
3775 : "</xsd:complexType>\n\n");
3776 :
3777 0 : appendStringInfo(&result,
3778 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3779 : xmlcn, catalogtypename);
3780 :
3781 0 : return result.data;
3782 : }
3783 :
3784 :
3785 : /*
3786 : * Map an SQL data type to an XML name; see SQL/XML:2008 section 9.4.
3787 : */
3788 : static const char *
3789 405 : map_sql_type_to_xml_name(Oid typeoid, int typmod)
3790 : {
3791 : StringInfoData result;
3792 :
3793 405 : initStringInfo(&result);
3794 :
3795 405 : switch (typeoid)
3796 : {
3797 15 : case BPCHAROID:
3798 15 : if (typmod == -1)
3799 15 : appendStringInfoString(&result, "CHAR");
3800 : else
3801 0 : appendStringInfo(&result, "CHAR_%d", typmod - VARHDRSZ);
3802 15 : break;
3803 27 : case VARCHAROID:
3804 27 : if (typmod == -1)
3805 27 : appendStringInfoString(&result, "VARCHAR");
3806 : else
3807 0 : appendStringInfo(&result, "VARCHAR_%d", typmod - VARHDRSZ);
3808 27 : break;
3809 15 : case NUMERICOID:
3810 15 : if (typmod == -1)
3811 15 : appendStringInfoString(&result, "NUMERIC");
3812 : else
3813 0 : appendStringInfo(&result, "NUMERIC_%d_%d",
3814 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
3815 0 : (typmod - VARHDRSZ) & 0xffff);
3816 15 : break;
3817 87 : case INT4OID:
3818 87 : appendStringInfoString(&result, "INTEGER");
3819 87 : break;
3820 15 : case INT2OID:
3821 15 : appendStringInfoString(&result, "SMALLINT");
3822 15 : break;
3823 15 : case INT8OID:
3824 15 : appendStringInfoString(&result, "BIGINT");
3825 15 : break;
3826 15 : case FLOAT4OID:
3827 15 : appendStringInfoString(&result, "REAL");
3828 15 : break;
3829 0 : case FLOAT8OID:
3830 0 : appendStringInfoString(&result, "DOUBLE");
3831 0 : break;
3832 15 : case BOOLOID:
3833 15 : appendStringInfoString(&result, "BOOLEAN");
3834 15 : break;
3835 15 : case TIMEOID:
3836 15 : if (typmod == -1)
3837 15 : appendStringInfoString(&result, "TIME");
3838 : else
3839 0 : appendStringInfo(&result, "TIME_%d", typmod);
3840 15 : break;
3841 15 : case TIMETZOID:
3842 15 : if (typmod == -1)
3843 15 : appendStringInfoString(&result, "TIME_WTZ");
3844 : else
3845 0 : appendStringInfo(&result, "TIME_WTZ_%d", typmod);
3846 15 : break;
3847 15 : case TIMESTAMPOID:
3848 15 : if (typmod == -1)
3849 15 : appendStringInfoString(&result, "TIMESTAMP");
3850 : else
3851 0 : appendStringInfo(&result, "TIMESTAMP_%d", typmod);
3852 15 : break;
3853 15 : case TIMESTAMPTZOID:
3854 15 : if (typmod == -1)
3855 15 : appendStringInfoString(&result, "TIMESTAMP_WTZ");
3856 : else
3857 0 : appendStringInfo(&result, "TIMESTAMP_WTZ_%d", typmod);
3858 15 : break;
3859 15 : case DATEOID:
3860 15 : appendStringInfoString(&result, "DATE");
3861 15 : break;
3862 15 : case XMLOID:
3863 15 : appendStringInfoString(&result, "XML");
3864 15 : break;
3865 111 : default:
3866 : {
3867 : HeapTuple tuple;
3868 : Form_pg_type typtuple;
3869 :
3870 111 : tuple = SearchSysCache1(TYPEOID, ObjectIdGetDatum(typeoid));
3871 111 : if (!HeapTupleIsValid(tuple))
3872 0 : elog(ERROR, "cache lookup failed for type %u", typeoid);
3873 111 : typtuple = (Form_pg_type) GETSTRUCT(tuple);
3874 :
3875 111 : appendStringInfoString(&result,
3876 111 : map_multipart_sql_identifier_to_xml_name((typtuple->typtype == TYPTYPE_DOMAIN) ? "Domain" : "UDT",
3877 111 : get_database_name(MyDatabaseId),
3878 111 : get_namespace_name(typtuple->typnamespace),
3879 111 : NameStr(typtuple->typname)));
3880 :
3881 111 : ReleaseSysCache(tuple);
3882 : }
3883 : }
3884 :
3885 405 : return result.data;
3886 : }
3887 :
3888 :
3889 : /*
3890 : * Map a collection of SQL data types to XML Schema data types; see
3891 : * SQL/XML:2008 section 9.7.
3892 : */
3893 : static const char *
3894 48 : map_sql_typecoll_to_xmlschema_types(List *tupdesc_list)
3895 : {
3896 48 : List *uniquetypes = NIL;
3897 : int i;
3898 : StringInfoData result;
3899 : ListCell *cell0;
3900 :
3901 : /* extract all column types used in the set of TupleDescs */
3902 105 : foreach(cell0, tupdesc_list)
3903 : {
3904 57 : TupleDesc tupdesc = (TupleDesc) lfirst(cell0);
3905 :
3906 351 : for (i = 0; i < tupdesc->natts; i++)
3907 : {
3908 294 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3909 :
3910 294 : if (att->attisdropped)
3911 12 : continue;
3912 282 : uniquetypes = list_append_unique_oid(uniquetypes, att->atttypid);
3913 : }
3914 : }
3915 :
3916 : /* add base types of domains */
3917 321 : foreach(cell0, uniquetypes)
3918 : {
3919 273 : Oid typid = lfirst_oid(cell0);
3920 273 : Oid basetypid = getBaseType(typid);
3921 :
3922 273 : if (basetypid != typid)
3923 12 : uniquetypes = list_append_unique_oid(uniquetypes, basetypid);
3924 : }
3925 :
3926 : /* Convert to textual form */
3927 48 : initStringInfo(&result);
3928 :
3929 321 : foreach(cell0, uniquetypes)
3930 : {
3931 273 : appendStringInfo(&result, "%s\n",
3932 : map_sql_type_to_xmlschema_type(lfirst_oid(cell0),
3933 : -1));
3934 : }
3935 :
3936 48 : return result.data;
3937 : }
3938 :
3939 :
3940 : /*
3941 : * Map an SQL data type to a named XML Schema data type; see
3942 : * SQL/XML:2008 sections 9.5 and 9.6.
3943 : *
3944 : * (The distinction between 9.5 and 9.6 is basically that 9.6 adds
3945 : * a name attribute, which this function does. The name-less version
3946 : * 9.5 doesn't appear to be required anywhere.)
3947 : */
3948 : static const char *
3949 273 : map_sql_type_to_xmlschema_type(Oid typeoid, int typmod)
3950 : {
3951 : StringInfoData result;
3952 273 : const char *typename = map_sql_type_to_xml_name(typeoid, typmod);
3953 :
3954 273 : initStringInfo(&result);
3955 :
3956 273 : if (typeoid == XMLOID)
3957 : {
3958 12 : appendStringInfoString(&result,
3959 : "<xsd:complexType mixed=\"true\">\n"
3960 : " <xsd:sequence>\n"
3961 : " <xsd:any name=\"element\" minOccurs=\"0\" maxOccurs=\"unbounded\" processContents=\"skip\"/>\n"
3962 : " </xsd:sequence>\n"
3963 : "</xsd:complexType>\n");
3964 : }
3965 : else
3966 : {
3967 261 : appendStringInfo(&result,
3968 : "<xsd:simpleType name=\"%s\">\n", typename);
3969 :
3970 261 : switch (typeoid)
3971 : {
3972 69 : case BPCHAROID:
3973 : case VARCHAROID:
3974 : case TEXTOID:
3975 69 : appendStringInfoString(&result,
3976 : " <xsd:restriction base=\"xsd:string\">\n");
3977 69 : if (typmod != -1)
3978 0 : appendStringInfo(&result,
3979 : " <xsd:maxLength value=\"%d\"/>\n",
3980 : typmod - VARHDRSZ);
3981 69 : appendStringInfoString(&result, " </xsd:restriction>\n");
3982 69 : break;
3983 :
3984 12 : case BYTEAOID:
3985 12 : appendStringInfo(&result,
3986 : " <xsd:restriction base=\"xsd:%s\">\n"
3987 : " </xsd:restriction>\n",
3988 12 : xmlbinary == XMLBINARY_BASE64 ? "base64Binary" : "hexBinary");
3989 12 : break;
3990 :
3991 12 : case NUMERICOID:
3992 12 : if (typmod != -1)
3993 0 : appendStringInfo(&result,
3994 : " <xsd:restriction base=\"xsd:decimal\">\n"
3995 : " <xsd:totalDigits value=\"%d\"/>\n"
3996 : " <xsd:fractionDigits value=\"%d\"/>\n"
3997 : " </xsd:restriction>\n",
3998 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
3999 0 : (typmod - VARHDRSZ) & 0xffff);
4000 12 : break;
4001 :
4002 12 : case INT2OID:
4003 12 : appendStringInfo(&result,
4004 : " <xsd:restriction base=\"xsd:short\">\n"
4005 : " <xsd:maxInclusive value=\"%d\"/>\n"
4006 : " <xsd:minInclusive value=\"%d\"/>\n"
4007 : " </xsd:restriction>\n",
4008 : SHRT_MAX, SHRT_MIN);
4009 12 : break;
4010 :
4011 48 : case INT4OID:
4012 48 : appendStringInfo(&result,
4013 : " <xsd:restriction base=\"xsd:int\">\n"
4014 : " <xsd:maxInclusive value=\"%d\"/>\n"
4015 : " <xsd:minInclusive value=\"%d\"/>\n"
4016 : " </xsd:restriction>\n",
4017 : INT_MAX, INT_MIN);
4018 48 : break;
4019 :
4020 12 : case INT8OID:
4021 12 : appendStringInfo(&result,
4022 : " <xsd:restriction base=\"xsd:long\">\n"
4023 : " <xsd:maxInclusive value=\"" INT64_FORMAT "\"/>\n"
4024 : " <xsd:minInclusive value=\"" INT64_FORMAT "\"/>\n"
4025 : " </xsd:restriction>\n",
4026 : PG_INT64_MAX,
4027 : PG_INT64_MIN);
4028 12 : break;
4029 :
4030 12 : case FLOAT4OID:
4031 12 : appendStringInfoString(&result,
4032 : " <xsd:restriction base=\"xsd:float\"></xsd:restriction>\n");
4033 12 : break;
4034 :
4035 0 : case FLOAT8OID:
4036 0 : appendStringInfoString(&result,
4037 : " <xsd:restriction base=\"xsd:double\"></xsd:restriction>\n");
4038 0 : break;
4039 :
4040 12 : case BOOLOID:
4041 12 : appendStringInfoString(&result,
4042 : " <xsd:restriction base=\"xsd:boolean\"></xsd:restriction>\n");
4043 12 : break;
4044 :
4045 24 : case TIMEOID:
4046 : case TIMETZOID:
4047 : {
4048 24 : const char *tz = (typeoid == TIMETZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
4049 :
4050 24 : if (typmod == -1)
4051 24 : appendStringInfo(&result,
4052 : " <xsd:restriction base=\"xsd:time\">\n"
4053 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}(.\\p{Nd}+)?%s\"/>\n"
4054 : " </xsd:restriction>\n", tz);
4055 0 : else if (typmod == 0)
4056 0 : appendStringInfo(&result,
4057 : " <xsd:restriction base=\"xsd:time\">\n"
4058 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
4059 : " </xsd:restriction>\n", tz);
4060 : else
4061 0 : appendStringInfo(&result,
4062 : " <xsd:restriction base=\"xsd:time\">\n"
4063 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}.\\p{Nd}{%d}%s\"/>\n"
4064 : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
4065 24 : break;
4066 : }
4067 :
4068 24 : case TIMESTAMPOID:
4069 : case TIMESTAMPTZOID:
4070 : {
4071 24 : const char *tz = (typeoid == TIMESTAMPTZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
4072 :
4073 24 : if (typmod == -1)
4074 24 : appendStringInfo(&result,
4075 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4076 : " <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"
4077 : " </xsd:restriction>\n", tz);
4078 0 : else if (typmod == 0)
4079 0 : appendStringInfo(&result,
4080 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4081 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
4082 : " </xsd:restriction>\n", tz);
4083 : else
4084 0 : appendStringInfo(&result,
4085 : " <xsd:restriction base=\"xsd:dateTime\">\n"
4086 : " <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"
4087 : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
4088 24 : break;
4089 : }
4090 :
4091 12 : case DATEOID:
4092 12 : appendStringInfoString(&result,
4093 : " <xsd:restriction base=\"xsd:date\">\n"
4094 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}\"/>\n"
4095 : " </xsd:restriction>\n");
4096 12 : break;
4097 :
4098 12 : default:
4099 12 : if (get_typtype(typeoid) == TYPTYPE_DOMAIN)
4100 : {
4101 : Oid base_typeoid;
4102 12 : int32 base_typmod = -1;
4103 :
4104 12 : base_typeoid = getBaseTypeAndTypmod(typeoid, &base_typmod);
4105 :
4106 12 : appendStringInfo(&result,
4107 : " <xsd:restriction base=\"%s\"/>\n",
4108 : map_sql_type_to_xml_name(base_typeoid, base_typmod));
4109 : }
4110 12 : break;
4111 : }
4112 261 : appendStringInfoString(&result, "</xsd:simpleType>\n");
4113 : }
4114 :
4115 273 : return result.data;
4116 : }
4117 :
4118 :
4119 : /*
4120 : * Map an SQL row to an XML element, taking the row from the active
4121 : * SPI cursor. See also SQL/XML:2008 section 9.10.
4122 : */
4123 : static void
4124 156 : SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result, char *tablename,
4125 : bool nulls, bool tableforest,
4126 : const char *targetns, bool top_level)
4127 : {
4128 : int i;
4129 : char *xmltn;
4130 :
4131 156 : if (tablename)
4132 114 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
4133 : else
4134 : {
4135 42 : if (tableforest)
4136 18 : xmltn = "row";
4137 : else
4138 24 : xmltn = "table";
4139 : }
4140 :
4141 156 : if (tableforest)
4142 81 : xmldata_root_element_start(result, xmltn, NULL, targetns, top_level);
4143 : else
4144 75 : appendStringInfoString(result, "<row>\n");
4145 :
4146 636 : for (i = 1; i <= SPI_tuptable->tupdesc->natts; i++)
4147 : {
4148 : char *colname;
4149 : Datum colval;
4150 : bool isnull;
4151 :
4152 480 : colname = map_sql_identifier_to_xml_name(SPI_fname(SPI_tuptable->tupdesc, i),
4153 : true, false);
4154 480 : colval = SPI_getbinval(SPI_tuptable->vals[rownum],
4155 480 : SPI_tuptable->tupdesc,
4156 : i,
4157 : &isnull);
4158 480 : if (isnull)
4159 : {
4160 57 : if (nulls)
4161 30 : appendStringInfo(result, " <%s xsi:nil=\"true\"/>\n", colname);
4162 : }
4163 : else
4164 423 : appendStringInfo(result, " <%s>%s</%s>\n",
4165 : colname,
4166 : map_sql_value_to_xml_value(colval,
4167 423 : SPI_gettypeid(SPI_tuptable->tupdesc, i), true),
4168 : colname);
4169 : }
4170 :
4171 156 : if (tableforest)
4172 : {
4173 81 : xmldata_root_element_end(result, xmltn);
4174 81 : appendStringInfoChar(result, '\n');
4175 : }
4176 : else
4177 75 : appendStringInfoString(result, "</row>\n\n");
4178 156 : }
4179 :
4180 :
4181 : /*
4182 : * XPath related functions
4183 : */
4184 :
4185 : #ifdef USE_LIBXML
4186 :
4187 : /*
4188 : * Convert XML node to text.
4189 : *
4190 : * For attribute and text nodes, return the escaped text. For anything else,
4191 : * dump the whole subtree.
4192 : */
4193 : static text *
4194 96 : xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt)
4195 : {
4196 96 : xmltype *result = NULL;
4197 :
4198 96 : if (cur->type != XML_ATTRIBUTE_NODE && cur->type != XML_TEXT_NODE)
4199 81 : {
4200 81 : void (*volatile nodefree) (xmlNodePtr) = NULL;
4201 81 : volatile xmlBufferPtr buf = NULL;
4202 81 : volatile xmlNodePtr cur_copy = NULL;
4203 :
4204 81 : PG_TRY();
4205 : {
4206 : int bytes;
4207 :
4208 81 : buf = xmlBufferCreate();
4209 81 : if (buf == NULL || xmlerrcxt->err_occurred)
4210 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4211 : "could not allocate xmlBuffer");
4212 :
4213 : /*
4214 : * Produce a dump of the node that we can serialize. xmlNodeDump
4215 : * does that, but the result of that function won't contain
4216 : * namespace definitions from ancestor nodes, so we first do a
4217 : * xmlCopyNode() which duplicates the node along with its required
4218 : * namespace definitions.
4219 : *
4220 : * Some old libxml2 versions such as 2.7.6 produce partially
4221 : * broken XML_DOCUMENT_NODE nodes (unset content field) when
4222 : * copying them. xmlNodeDump of such a node works fine, but
4223 : * xmlFreeNode crashes; set us up to call xmlFreeDoc instead.
4224 : */
4225 81 : cur_copy = xmlCopyNode(cur, 1);
4226 81 : if (cur_copy == NULL || xmlerrcxt->err_occurred)
4227 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4228 : "could not copy node");
4229 162 : nodefree = (cur_copy->type == XML_DOCUMENT_NODE) ?
4230 81 : (void (*) (xmlNodePtr)) xmlFreeDoc : xmlFreeNode;
4231 :
4232 81 : bytes = xmlNodeDump(buf, NULL, cur_copy, 0, 0);
4233 81 : if (bytes == -1 || xmlerrcxt->err_occurred)
4234 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4235 : "could not dump node");
4236 :
4237 81 : result = xmlBuffer_to_xmltype(buf);
4238 : }
4239 0 : PG_FINALLY();
4240 : {
4241 81 : if (nodefree)
4242 81 : nodefree(cur_copy);
4243 81 : if (buf)
4244 81 : xmlBufferFree(buf);
4245 : }
4246 81 : PG_END_TRY();
4247 : }
4248 : else
4249 : {
4250 15 : xmlChar *volatile str = NULL;
4251 :
4252 15 : PG_TRY();
4253 : {
4254 : char *escaped;
4255 :
4256 15 : str = xmlXPathCastNodeToString(cur);
4257 15 : if (str == NULL || xmlerrcxt->err_occurred)
4258 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4259 : "could not allocate xmlChar");
4260 :
4261 : /* Here we rely on XML having the same representation as TEXT */
4262 15 : escaped = escape_xml((char *) str);
4263 :
4264 15 : result = (xmltype *) cstring_to_text(escaped);
4265 15 : pfree(escaped);
4266 : }
4267 0 : PG_FINALLY();
4268 : {
4269 15 : if (str)
4270 15 : xmlFree(str);
4271 : }
4272 15 : PG_END_TRY();
4273 : }
4274 :
4275 96 : return result;
4276 : }
4277 :
4278 : /*
4279 : * Convert an XML XPath object (the result of evaluating an XPath expression)
4280 : * to an array of xml values, which are appended to astate. The function
4281 : * result value is the number of elements in the array.
4282 : *
4283 : * If "astate" is NULL then we don't generate the array value, but we still
4284 : * return the number of elements it would have had.
4285 : *
4286 : * Nodesets are converted to an array containing the nodes' textual
4287 : * representations. Primitive values (float, double, string) are converted
4288 : * to a single-element array containing the value's string representation.
4289 : */
4290 : static int
4291 270 : xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
4292 : ArrayBuildState *astate,
4293 : PgXmlErrorContext *xmlerrcxt)
4294 : {
4295 270 : int result = 0;
4296 : Datum datum;
4297 : Oid datumtype;
4298 : char *result_str;
4299 :
4300 270 : switch (xpathobj->type)
4301 : {
4302 249 : case XPATH_NODESET:
4303 249 : if (xpathobj->nodesetval != NULL)
4304 : {
4305 177 : result = xpathobj->nodesetval->nodeNr;
4306 177 : if (astate != NULL)
4307 : {
4308 : int i;
4309 :
4310 84 : for (i = 0; i < result; i++)
4311 : {
4312 45 : datum = PointerGetDatum(xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
4313 : xmlerrcxt));
4314 45 : (void) accumArrayResult(astate, datum, false,
4315 : XMLOID, CurrentMemoryContext);
4316 : }
4317 : }
4318 : }
4319 249 : return result;
4320 :
4321 6 : case XPATH_BOOLEAN:
4322 6 : if (astate == NULL)
4323 0 : return 1;
4324 6 : datum = BoolGetDatum(xpathobj->boolval);
4325 6 : datumtype = BOOLOID;
4326 6 : break;
4327 :
4328 9 : case XPATH_NUMBER:
4329 9 : if (astate == NULL)
4330 6 : return 1;
4331 3 : datum = Float8GetDatum(xpathobj->floatval);
4332 3 : datumtype = FLOAT8OID;
4333 3 : break;
4334 :
4335 6 : case XPATH_STRING:
4336 6 : if (astate == NULL)
4337 0 : return 1;
4338 6 : datum = CStringGetDatum((char *) xpathobj->stringval);
4339 6 : datumtype = CSTRINGOID;
4340 6 : break;
4341 :
4342 0 : default:
4343 0 : elog(ERROR, "xpath expression result type %d is unsupported",
4344 : xpathobj->type);
4345 : return 0; /* keep compiler quiet */
4346 : }
4347 :
4348 : /* Common code for scalar-value cases */
4349 15 : result_str = map_sql_value_to_xml_value(datum, datumtype, true);
4350 15 : datum = PointerGetDatum(cstring_to_xmltype(result_str));
4351 15 : (void) accumArrayResult(astate, datum, false,
4352 : XMLOID, CurrentMemoryContext);
4353 15 : return 1;
4354 : }
4355 :
4356 :
4357 : /*
4358 : * Common code for xpath() and xmlexists()
4359 : *
4360 : * Evaluate XPath expression and return number of nodes in res_nitems
4361 : * and array of XML values in astate. Either of those pointers can be
4362 : * NULL if the corresponding result isn't wanted.
4363 : *
4364 : * It is up to the user to ensure that the XML passed is in fact
4365 : * an XML document - XPath doesn't work easily on fragments without
4366 : * a context node being known.
4367 : */
4368 : static void
4369 279 : xpath_internal(text *xpath_expr_text, xmltype *data, ArrayType *namespaces,
4370 : int *res_nitems, ArrayBuildState *astate)
4371 : {
4372 : PgXmlErrorContext *xmlerrcxt;
4373 279 : volatile xmlParserCtxtPtr ctxt = NULL;
4374 279 : volatile xmlDocPtr doc = NULL;
4375 279 : volatile xmlXPathContextPtr xpathctx = NULL;
4376 279 : volatile xmlXPathCompExprPtr xpathcomp = NULL;
4377 279 : volatile xmlXPathObjectPtr xpathobj = NULL;
4378 : char *datastr;
4379 : int32 len;
4380 : int32 xpath_len;
4381 : xmlChar *string;
4382 : xmlChar *xpath_expr;
4383 279 : size_t xmldecl_len = 0;
4384 : int i;
4385 : int ndim;
4386 : Datum *ns_names_uris;
4387 : bool *ns_names_uris_nulls;
4388 : int ns_count;
4389 :
4390 : /*
4391 : * Namespace mappings are passed as text[]. If an empty array is passed
4392 : * (ndim = 0, "0-dimensional"), then there are no namespace mappings.
4393 : * Else, a 2-dimensional array with length of the second axis being equal
4394 : * to 2 should be passed, i.e., every subarray contains 2 elements, the
4395 : * first element defining the name, the second one the URI. Example:
4396 : * ARRAY[ARRAY['myns', 'http://example.com'], ARRAY['myns2',
4397 : * 'http://example2.com']].
4398 : */
4399 279 : ndim = namespaces ? ARR_NDIM(namespaces) : 0;
4400 279 : if (ndim != 0)
4401 : {
4402 : int *dims;
4403 :
4404 63 : dims = ARR_DIMS(namespaces);
4405 :
4406 63 : if (ndim != 2 || dims[1] != 2)
4407 0 : ereport(ERROR,
4408 : (errcode(ERRCODE_DATA_EXCEPTION),
4409 : errmsg("invalid array for XML namespace mapping"),
4410 : errdetail("The array must be two-dimensional with length of the second axis equal to 2.")));
4411 :
4412 : Assert(ARR_ELEMTYPE(namespaces) == TEXTOID);
4413 :
4414 63 : deconstruct_array_builtin(namespaces, TEXTOID,
4415 : &ns_names_uris, &ns_names_uris_nulls,
4416 : &ns_count);
4417 :
4418 : Assert((ns_count % 2) == 0); /* checked above */
4419 63 : ns_count /= 2; /* count pairs only */
4420 : }
4421 : else
4422 : {
4423 216 : ns_names_uris = NULL;
4424 216 : ns_names_uris_nulls = NULL;
4425 216 : ns_count = 0;
4426 : }
4427 :
4428 279 : datastr = VARDATA(data);
4429 279 : len = VARSIZE(data) - VARHDRSZ;
4430 279 : xpath_len = VARSIZE_ANY_EXHDR(xpath_expr_text);
4431 279 : if (xpath_len == 0)
4432 3 : ereport(ERROR,
4433 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4434 : errmsg("empty XPath expression")));
4435 :
4436 276 : string = pg_xmlCharStrndup(datastr, len);
4437 276 : xpath_expr = pg_xmlCharStrndup(VARDATA_ANY(xpath_expr_text), xpath_len);
4438 :
4439 : /*
4440 : * In a UTF8 database, skip any xml declaration, which might assert
4441 : * another encoding. Ignore parse_xml_decl() failure, letting
4442 : * xmlCtxtReadMemory() report parse errors. Documentation disclaims
4443 : * xpath() support for non-ASCII data in non-UTF8 databases, so leave
4444 : * those scenarios bug-compatible with historical behavior.
4445 : */
4446 276 : if (GetDatabaseEncoding() == PG_UTF8)
4447 276 : parse_xml_decl(string, &xmldecl_len, NULL, NULL, NULL);
4448 :
4449 276 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4450 :
4451 276 : PG_TRY();
4452 : {
4453 276 : xmlInitParser();
4454 :
4455 : /*
4456 : * redundant XML parsing (two parsings for the same value during one
4457 : * command execution are possible)
4458 : */
4459 276 : ctxt = xmlNewParserCtxt();
4460 276 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4461 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4462 : "could not allocate parser context");
4463 552 : doc = xmlCtxtReadMemory(ctxt, (char *) string + xmldecl_len,
4464 276 : len - xmldecl_len, NULL, NULL, 0);
4465 276 : if (doc == NULL || xmlerrcxt->err_occurred)
4466 6 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4467 : "could not parse XML document");
4468 270 : xpathctx = xmlXPathNewContext(doc);
4469 270 : if (xpathctx == NULL || xmlerrcxt->err_occurred)
4470 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4471 : "could not allocate XPath context");
4472 270 : xpathctx->node = (xmlNodePtr) doc;
4473 :
4474 : /* register namespaces, if any */
4475 270 : if (ns_count > 0)
4476 : {
4477 126 : for (i = 0; i < ns_count; i++)
4478 : {
4479 : char *ns_name;
4480 : char *ns_uri;
4481 :
4482 63 : if (ns_names_uris_nulls[i * 2] ||
4483 63 : ns_names_uris_nulls[i * 2 + 1])
4484 0 : ereport(ERROR,
4485 : (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
4486 : errmsg("neither namespace name nor URI may be null")));
4487 63 : ns_name = TextDatumGetCString(ns_names_uris[i * 2]);
4488 63 : ns_uri = TextDatumGetCString(ns_names_uris[i * 2 + 1]);
4489 63 : if (xmlXPathRegisterNs(xpathctx,
4490 : (xmlChar *) ns_name,
4491 : (xmlChar *) ns_uri) != 0)
4492 0 : ereport(ERROR, /* is this an internal error??? */
4493 : (errmsg("could not register XML namespace with name \"%s\" and URI \"%s\"",
4494 : ns_name, ns_uri)));
4495 : }
4496 : }
4497 :
4498 : /*
4499 : * Note: here and elsewhere, be careful to use xmlXPathCtxtCompile not
4500 : * xmlXPathCompile. In libxml2 2.13.3 and older, the latter function
4501 : * fails to defend itself against recursion-to-stack-overflow. See
4502 : * https://gitlab.gnome.org/GNOME/libxml2/-/issues/799
4503 : */
4504 270 : xpathcomp = xmlXPathCtxtCompile(xpathctx, xpath_expr);
4505 270 : if (xpathcomp == NULL || xmlerrcxt->err_occurred)
4506 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4507 : "invalid XPath expression");
4508 :
4509 : /*
4510 : * Version 2.6.27 introduces a function named
4511 : * xmlXPathCompiledEvalToBoolean, which would be enough for xmlexists,
4512 : * but we can derive the existence by whether any nodes are returned,
4513 : * thereby preventing a library version upgrade and keeping the code
4514 : * the same.
4515 : */
4516 270 : xpathobj = xmlXPathCompiledEval(xpathcomp, xpathctx);
4517 270 : if (xpathobj == NULL || xmlerrcxt->err_occurred)
4518 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4519 : "could not create XPath object");
4520 :
4521 : /*
4522 : * Extract the results as requested.
4523 : */
4524 270 : if (res_nitems != NULL)
4525 216 : *res_nitems = xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4526 : else
4527 54 : (void) xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4528 : }
4529 6 : PG_CATCH();
4530 : {
4531 6 : if (xpathobj)
4532 0 : xmlXPathFreeObject(xpathobj);
4533 6 : if (xpathcomp)
4534 0 : xmlXPathFreeCompExpr(xpathcomp);
4535 6 : if (xpathctx)
4536 0 : xmlXPathFreeContext(xpathctx);
4537 6 : if (doc)
4538 6 : xmlFreeDoc(doc);
4539 6 : if (ctxt)
4540 6 : xmlFreeParserCtxt(ctxt);
4541 :
4542 6 : pg_xml_done(xmlerrcxt, true);
4543 :
4544 6 : PG_RE_THROW();
4545 : }
4546 270 : PG_END_TRY();
4547 :
4548 270 : xmlXPathFreeObject(xpathobj);
4549 270 : xmlXPathFreeCompExpr(xpathcomp);
4550 270 : xmlXPathFreeContext(xpathctx);
4551 270 : xmlFreeDoc(doc);
4552 270 : xmlFreeParserCtxt(ctxt);
4553 :
4554 270 : pg_xml_done(xmlerrcxt, false);
4555 270 : }
4556 : #endif /* USE_LIBXML */
4557 :
4558 : /*
4559 : * Evaluate XPath expression and return array of XML values.
4560 : *
4561 : * As we have no support of XQuery sequences yet, this function seems
4562 : * to be the most useful one (array of XML functions plays a role of
4563 : * some kind of substitution for XQuery sequences).
4564 : */
4565 : Datum
4566 63 : xpath(PG_FUNCTION_ARGS)
4567 : {
4568 : #ifdef USE_LIBXML
4569 63 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4570 63 : xmltype *data = PG_GETARG_XML_P(1);
4571 63 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4572 : ArrayBuildState *astate;
4573 :
4574 63 : astate = initArrayResult(XMLOID, CurrentMemoryContext, true);
4575 63 : xpath_internal(xpath_expr_text, data, namespaces,
4576 : NULL, astate);
4577 54 : PG_RETURN_DATUM(makeArrayResult(astate, CurrentMemoryContext));
4578 : #else
4579 : NO_XML_SUPPORT();
4580 : return 0;
4581 : #endif
4582 : }
4583 :
4584 : /*
4585 : * Determines if the node specified by the supplied XPath exists
4586 : * in a given XML document, returning a boolean.
4587 : */
4588 : Datum
4589 99 : xmlexists(PG_FUNCTION_ARGS)
4590 : {
4591 : #ifdef USE_LIBXML
4592 99 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4593 99 : xmltype *data = PG_GETARG_XML_P(1);
4594 : int res_nitems;
4595 :
4596 99 : xpath_internal(xpath_expr_text, data, NULL,
4597 : &res_nitems, NULL);
4598 :
4599 99 : PG_RETURN_BOOL(res_nitems > 0);
4600 : #else
4601 : NO_XML_SUPPORT();
4602 : return 0;
4603 : #endif
4604 : }
4605 :
4606 : /*
4607 : * Determines if the node specified by the supplied XPath exists
4608 : * in a given XML document, returning a boolean. Differs from
4609 : * xmlexists as it supports namespaces and is not defined in SQL/XML.
4610 : */
4611 : Datum
4612 117 : xpath_exists(PG_FUNCTION_ARGS)
4613 : {
4614 : #ifdef USE_LIBXML
4615 117 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4616 117 : xmltype *data = PG_GETARG_XML_P(1);
4617 117 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4618 : int res_nitems;
4619 :
4620 117 : xpath_internal(xpath_expr_text, data, namespaces,
4621 : &res_nitems, NULL);
4622 :
4623 117 : PG_RETURN_BOOL(res_nitems > 0);
4624 : #else
4625 : NO_XML_SUPPORT();
4626 : return 0;
4627 : #endif
4628 : }
4629 :
4630 : /*
4631 : * Functions for checking well-formed-ness
4632 : */
4633 :
4634 : #ifdef USE_LIBXML
4635 : static bool
4636 57 : wellformed_xml(text *data, XmlOptionType xmloption_arg)
4637 : {
4638 : xmlDocPtr doc;
4639 57 : ErrorSaveContext escontext = {T_ErrorSaveContext};
4640 :
4641 : /*
4642 : * We'll report "true" if no soft error is reported by xml_parse().
4643 : */
4644 57 : doc = xml_parse(data, xmloption_arg, true,
4645 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
4646 57 : if (doc)
4647 30 : xmlFreeDoc(doc);
4648 :
4649 57 : return !escontext.error_occurred;
4650 : }
4651 : #endif
4652 :
4653 : Datum
4654 45 : xml_is_well_formed(PG_FUNCTION_ARGS)
4655 : {
4656 : #ifdef USE_LIBXML
4657 45 : text *data = PG_GETARG_TEXT_PP(0);
4658 :
4659 45 : PG_RETURN_BOOL(wellformed_xml(data, xmloption));
4660 : #else
4661 : NO_XML_SUPPORT();
4662 : return 0;
4663 : #endif /* not USE_LIBXML */
4664 : }
4665 :
4666 : Datum
4667 6 : xml_is_well_formed_document(PG_FUNCTION_ARGS)
4668 : {
4669 : #ifdef USE_LIBXML
4670 6 : text *data = PG_GETARG_TEXT_PP(0);
4671 :
4672 6 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_DOCUMENT));
4673 : #else
4674 : NO_XML_SUPPORT();
4675 : return 0;
4676 : #endif /* not USE_LIBXML */
4677 : }
4678 :
4679 : Datum
4680 6 : xml_is_well_formed_content(PG_FUNCTION_ARGS)
4681 : {
4682 : #ifdef USE_LIBXML
4683 6 : text *data = PG_GETARG_TEXT_PP(0);
4684 :
4685 6 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_CONTENT));
4686 : #else
4687 : NO_XML_SUPPORT();
4688 : return 0;
4689 : #endif /* not USE_LIBXML */
4690 : }
4691 :
4692 : /*
4693 : * support functions for XMLTABLE
4694 : *
4695 : */
4696 : #ifdef USE_LIBXML
4697 :
4698 : /*
4699 : * Returns private data from executor state. Ensure validity by check with
4700 : * MAGIC number.
4701 : */
4702 : static inline XmlTableBuilderData *
4703 80695 : GetXmlTableBuilderPrivateData(TableFuncScanState *state, const char *fname)
4704 : {
4705 : XmlTableBuilderData *result;
4706 :
4707 80695 : if (!IsA(state, TableFuncScanState))
4708 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4709 80695 : result = (XmlTableBuilderData *) state->opaque;
4710 80695 : if (result->magic != XMLTABLE_CONTEXT_MAGIC)
4711 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4712 :
4713 80695 : return result;
4714 : }
4715 : #endif
4716 :
4717 : /*
4718 : * XmlTableInitOpaque
4719 : * Fill in TableFuncScanState->opaque for XmlTable processor; initialize
4720 : * the XML parser.
4721 : *
4722 : * Note: Because we call pg_xml_init() here and pg_xml_done() in
4723 : * XmlTableDestroyOpaque, it is critical for robustness that no other
4724 : * executor nodes run until this node is processed to completion. Caller
4725 : * must execute this to completion (probably filling a tuplestore to exhaust
4726 : * this node in a single pass) instead of using row-per-call mode.
4727 : */
4728 : static void
4729 132 : XmlTableInitOpaque(TableFuncScanState *state, int natts)
4730 : {
4731 : #ifdef USE_LIBXML
4732 132 : volatile xmlParserCtxtPtr ctxt = NULL;
4733 : XmlTableBuilderData *xtCxt;
4734 : PgXmlErrorContext *xmlerrcxt;
4735 :
4736 132 : xtCxt = palloc0_object(XmlTableBuilderData);
4737 132 : xtCxt->magic = XMLTABLE_CONTEXT_MAGIC;
4738 132 : xtCxt->natts = natts;
4739 132 : xtCxt->xpathscomp = palloc0_array(xmlXPathCompExprPtr, natts);
4740 :
4741 132 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4742 :
4743 132 : PG_TRY();
4744 : {
4745 132 : xmlInitParser();
4746 :
4747 132 : ctxt = xmlNewParserCtxt();
4748 132 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4749 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4750 : "could not allocate parser context");
4751 : }
4752 0 : PG_CATCH();
4753 : {
4754 0 : if (ctxt != NULL)
4755 0 : xmlFreeParserCtxt(ctxt);
4756 :
4757 0 : pg_xml_done(xmlerrcxt, true);
4758 :
4759 0 : PG_RE_THROW();
4760 : }
4761 132 : PG_END_TRY();
4762 :
4763 132 : xtCxt->xmlerrcxt = xmlerrcxt;
4764 132 : xtCxt->ctxt = ctxt;
4765 :
4766 132 : state->opaque = xtCxt;
4767 : #else
4768 : NO_XML_SUPPORT();
4769 : #endif /* not USE_LIBXML */
4770 132 : }
4771 :
4772 : /*
4773 : * XmlTableSetDocument
4774 : * Install the input document
4775 : */
4776 : static void
4777 132 : XmlTableSetDocument(TableFuncScanState *state, Datum value)
4778 : {
4779 : #ifdef USE_LIBXML
4780 : XmlTableBuilderData *xtCxt;
4781 132 : xmltype *xmlval = DatumGetXmlP(value);
4782 : char *str;
4783 : xmlChar *xstr;
4784 : int length;
4785 132 : volatile xmlDocPtr doc = NULL;
4786 132 : volatile xmlXPathContextPtr xpathcxt = NULL;
4787 :
4788 132 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetDocument");
4789 :
4790 : /*
4791 : * Use out function for casting to string (remove encoding property). See
4792 : * comment in xml_out.
4793 : */
4794 132 : str = xml_out_internal(xmlval, 0);
4795 :
4796 132 : length = strlen(str);
4797 132 : xstr = pg_xmlCharStrndup(str, length);
4798 :
4799 132 : PG_TRY();
4800 : {
4801 132 : doc = xmlCtxtReadMemory(xtCxt->ctxt, (char *) xstr, length, NULL, NULL, 0);
4802 132 : if (doc == NULL || xtCxt->xmlerrcxt->err_occurred)
4803 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4804 : "could not parse XML document");
4805 132 : xpathcxt = xmlXPathNewContext(doc);
4806 132 : if (xpathcxt == NULL || xtCxt->xmlerrcxt->err_occurred)
4807 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4808 : "could not allocate XPath context");
4809 132 : xpathcxt->node = (xmlNodePtr) doc;
4810 : }
4811 0 : PG_CATCH();
4812 : {
4813 0 : if (xpathcxt != NULL)
4814 0 : xmlXPathFreeContext(xpathcxt);
4815 0 : if (doc != NULL)
4816 0 : xmlFreeDoc(doc);
4817 :
4818 0 : PG_RE_THROW();
4819 : }
4820 132 : PG_END_TRY();
4821 :
4822 132 : xtCxt->doc = doc;
4823 132 : xtCxt->xpathcxt = xpathcxt;
4824 : #else
4825 : NO_XML_SUPPORT();
4826 : #endif /* not USE_LIBXML */
4827 132 : }
4828 :
4829 : /*
4830 : * XmlTableSetNamespace
4831 : * Add a namespace declaration
4832 : */
4833 : static void
4834 9 : XmlTableSetNamespace(TableFuncScanState *state, const char *name, const char *uri)
4835 : {
4836 : #ifdef USE_LIBXML
4837 : XmlTableBuilderData *xtCxt;
4838 :
4839 9 : if (name == NULL)
4840 3 : ereport(ERROR,
4841 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
4842 : errmsg("DEFAULT namespace is not supported")));
4843 6 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetNamespace");
4844 :
4845 6 : if (xmlXPathRegisterNs(xtCxt->xpathcxt,
4846 6 : pg_xmlCharStrndup(name, strlen(name)),
4847 6 : pg_xmlCharStrndup(uri, strlen(uri))))
4848 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4849 : "could not set XML namespace");
4850 : #else
4851 : NO_XML_SUPPORT();
4852 : #endif /* not USE_LIBXML */
4853 6 : }
4854 :
4855 : /*
4856 : * XmlTableSetRowFilter
4857 : * Install the row-filter Xpath expression.
4858 : */
4859 : static void
4860 129 : XmlTableSetRowFilter(TableFuncScanState *state, const char *path)
4861 : {
4862 : #ifdef USE_LIBXML
4863 : XmlTableBuilderData *xtCxt;
4864 : xmlChar *xstr;
4865 :
4866 129 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetRowFilter");
4867 :
4868 129 : if (*path == '\0')
4869 0 : ereport(ERROR,
4870 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4871 : errmsg("row path filter must not be empty string")));
4872 :
4873 129 : xstr = pg_xmlCharStrndup(path, strlen(path));
4874 :
4875 : /* We require XmlTableSetDocument to have been done already */
4876 : Assert(xtCxt->xpathcxt != NULL);
4877 :
4878 129 : xtCxt->xpathcomp = xmlXPathCtxtCompile(xtCxt->xpathcxt, xstr);
4879 129 : if (xtCxt->xpathcomp == NULL || xtCxt->xmlerrcxt->err_occurred)
4880 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4881 : "invalid XPath expression");
4882 : #else
4883 : NO_XML_SUPPORT();
4884 : #endif /* not USE_LIBXML */
4885 129 : }
4886 :
4887 : /*
4888 : * XmlTableSetColumnFilter
4889 : * Install the column-filter Xpath expression, for the given column.
4890 : */
4891 : static void
4892 387 : XmlTableSetColumnFilter(TableFuncScanState *state, const char *path, int colnum)
4893 : {
4894 : #ifdef USE_LIBXML
4895 : XmlTableBuilderData *xtCxt;
4896 : xmlChar *xstr;
4897 :
4898 : Assert(path);
4899 :
4900 387 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetColumnFilter");
4901 :
4902 387 : if (*path == '\0')
4903 0 : ereport(ERROR,
4904 : (errcode(ERRCODE_INVALID_ARGUMENT_FOR_XQUERY),
4905 : errmsg("column path filter must not be empty string")));
4906 :
4907 387 : xstr = pg_xmlCharStrndup(path, strlen(path));
4908 :
4909 : /* We require XmlTableSetDocument to have been done already */
4910 : Assert(xtCxt->xpathcxt != NULL);
4911 :
4912 387 : xtCxt->xpathscomp[colnum] = xmlXPathCtxtCompile(xtCxt->xpathcxt, xstr);
4913 387 : if (xtCxt->xpathscomp[colnum] == NULL || xtCxt->xmlerrcxt->err_occurred)
4914 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4915 : "invalid XPath expression");
4916 : #else
4917 : NO_XML_SUPPORT();
4918 : #endif /* not USE_LIBXML */
4919 387 : }
4920 :
4921 : /*
4922 : * XmlTableFetchRow
4923 : * Prepare the next "current" tuple for upcoming GetValue calls.
4924 : * Returns false if the row-filter expression returned no more rows.
4925 : */
4926 : static bool
4927 11596 : XmlTableFetchRow(TableFuncScanState *state)
4928 : {
4929 : #ifdef USE_LIBXML
4930 : XmlTableBuilderData *xtCxt;
4931 :
4932 11596 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableFetchRow");
4933 :
4934 : /* Propagate our own error context to libxml2 */
4935 11596 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
4936 :
4937 11596 : if (xtCxt->xpathobj == NULL)
4938 : {
4939 129 : xtCxt->xpathobj = xmlXPathCompiledEval(xtCxt->xpathcomp, xtCxt->xpathcxt);
4940 129 : if (xtCxt->xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
4941 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
4942 : "could not create XPath object");
4943 :
4944 129 : xtCxt->row_count = 0;
4945 : }
4946 :
4947 11596 : if (xtCxt->xpathobj->type == XPATH_NODESET)
4948 : {
4949 11596 : if (xtCxt->xpathobj->nodesetval != NULL)
4950 : {
4951 11596 : if (xtCxt->row_count++ < xtCxt->xpathobj->nodesetval->nodeNr)
4952 11473 : return true;
4953 : }
4954 : }
4955 :
4956 123 : return false;
4957 : #else
4958 : NO_XML_SUPPORT();
4959 : return false;
4960 : #endif /* not USE_LIBXML */
4961 : }
4962 :
4963 : /*
4964 : * XmlTableGetValue
4965 : * Return the value for column number 'colnum' for the current row. If
4966 : * column -1 is requested, return representation of the whole row.
4967 : *
4968 : * This leaks memory, so be sure to reset often the context in which it's
4969 : * called.
4970 : */
4971 : static Datum
4972 68313 : XmlTableGetValue(TableFuncScanState *state, int colnum,
4973 : Oid typid, int32 typmod, bool *isnull)
4974 : {
4975 : #ifdef USE_LIBXML
4976 68313 : Datum result = (Datum) 0;
4977 : XmlTableBuilderData *xtCxt;
4978 68313 : volatile xmlXPathObjectPtr xpathobj = NULL;
4979 :
4980 68313 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableGetValue");
4981 :
4982 : Assert(xtCxt->xpathobj &&
4983 : xtCxt->xpathobj->type == XPATH_NODESET &&
4984 : xtCxt->xpathobj->nodesetval != NULL);
4985 :
4986 : /* Propagate our own error context to libxml2 */
4987 68313 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
4988 :
4989 68313 : *isnull = false;
4990 :
4991 : Assert(xtCxt->xpathscomp[colnum] != NULL);
4992 :
4993 68313 : PG_TRY();
4994 : {
4995 : xmlNodePtr cur;
4996 68313 : char *cstr = NULL;
4997 :
4998 : /* Set current node as entry point for XPath evaluation */
4999 68313 : cur = xtCxt->xpathobj->nodesetval->nodeTab[xtCxt->row_count - 1];
5000 68313 : xtCxt->xpathcxt->node = cur;
5001 :
5002 : /* Evaluate column path */
5003 68313 : xpathobj = xmlXPathCompiledEval(xtCxt->xpathscomp[colnum], xtCxt->xpathcxt);
5004 68313 : if (xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
5005 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_ARGUMENT_FOR_XQUERY,
5006 : "could not create XPath object");
5007 :
5008 : /*
5009 : * There are four possible cases, depending on the number of nodes
5010 : * returned by the XPath expression and the type of the target column:
5011 : * a) XPath returns no nodes. b) The target type is XML (return all
5012 : * as XML). For non-XML return types: c) One node (return content).
5013 : * d) Multiple nodes (error).
5014 : */
5015 68313 : if (xpathobj->type == XPATH_NODESET)
5016 : {
5017 68298 : int count = 0;
5018 :
5019 68298 : if (xpathobj->nodesetval != NULL)
5020 68193 : count = xpathobj->nodesetval->nodeNr;
5021 :
5022 68298 : if (xpathobj->nodesetval == NULL || count == 0)
5023 : {
5024 11427 : *isnull = true;
5025 : }
5026 : else
5027 : {
5028 56871 : if (typid == XMLOID)
5029 : {
5030 : text *textstr;
5031 : StringInfoData str;
5032 :
5033 : /* Concatenate serialized values */
5034 36 : initStringInfo(&str);
5035 87 : for (int i = 0; i < count; i++)
5036 : {
5037 : textstr =
5038 51 : xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
5039 : xtCxt->xmlerrcxt);
5040 :
5041 51 : appendStringInfoText(&str, textstr);
5042 : }
5043 36 : cstr = str.data;
5044 : }
5045 : else
5046 : {
5047 : xmlChar *str;
5048 :
5049 56835 : if (count > 1)
5050 3 : ereport(ERROR,
5051 : (errcode(ERRCODE_CARDINALITY_VIOLATION),
5052 : errmsg("more than one value returned by column XPath expression")));
5053 :
5054 56832 : str = xmlXPathCastNodeSetToString(xpathobj->nodesetval);
5055 56832 : cstr = str ? xml_pstrdup_and_free(str) : "";
5056 : }
5057 : }
5058 : }
5059 15 : else if (xpathobj->type == XPATH_STRING)
5060 : {
5061 : /* Content should be escaped when target will be XML */
5062 9 : if (typid == XMLOID)
5063 3 : cstr = escape_xml((char *) xpathobj->stringval);
5064 : else
5065 6 : cstr = (char *) xpathobj->stringval;
5066 : }
5067 6 : else if (xpathobj->type == XPATH_BOOLEAN)
5068 : {
5069 : char typcategory;
5070 : bool typispreferred;
5071 : xmlChar *str;
5072 :
5073 : /* Allow implicit casting from boolean to numbers */
5074 3 : get_type_category_preferred(typid, &typcategory, &typispreferred);
5075 :
5076 3 : if (typcategory != TYPCATEGORY_NUMERIC)
5077 3 : str = xmlXPathCastBooleanToString(xpathobj->boolval);
5078 : else
5079 0 : str = xmlXPathCastNumberToString(xmlXPathCastBooleanToNumber(xpathobj->boolval));
5080 :
5081 3 : cstr = xml_pstrdup_and_free(str);
5082 : }
5083 3 : else if (xpathobj->type == XPATH_NUMBER)
5084 : {
5085 : xmlChar *str;
5086 :
5087 3 : str = xmlXPathCastNumberToString(xpathobj->floatval);
5088 3 : cstr = xml_pstrdup_and_free(str);
5089 : }
5090 : else
5091 0 : elog(ERROR, "unexpected XPath object type %u", xpathobj->type);
5092 :
5093 : /*
5094 : * By here, either cstr contains the result value, or the isnull flag
5095 : * has been set.
5096 : */
5097 : Assert(cstr || *isnull);
5098 :
5099 68310 : if (!*isnull)
5100 56883 : result = InputFunctionCall(&state->in_functions[colnum],
5101 : cstr,
5102 56883 : state->typioparams[colnum],
5103 : typmod);
5104 : }
5105 3 : PG_FINALLY();
5106 : {
5107 68313 : if (xpathobj != NULL)
5108 68313 : xmlXPathFreeObject(xpathobj);
5109 : }
5110 68313 : PG_END_TRY();
5111 :
5112 68310 : return result;
5113 : #else
5114 : NO_XML_SUPPORT();
5115 : return 0;
5116 : #endif /* not USE_LIBXML */
5117 : }
5118 :
5119 : /*
5120 : * XmlTableDestroyOpaque
5121 : * Release all libxml2 resources
5122 : */
5123 : static void
5124 132 : XmlTableDestroyOpaque(TableFuncScanState *state)
5125 : {
5126 : #ifdef USE_LIBXML
5127 : XmlTableBuilderData *xtCxt;
5128 :
5129 132 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableDestroyOpaque");
5130 :
5131 : /* Propagate our own error context to libxml2 */
5132 132 : xmlSetStructuredErrorFunc(xtCxt->xmlerrcxt, xml_errorHandler);
5133 :
5134 132 : if (xtCxt->xpathscomp != NULL)
5135 : {
5136 : int i;
5137 :
5138 558 : for (i = 0; i < xtCxt->natts; i++)
5139 426 : if (xtCxt->xpathscomp[i] != NULL)
5140 387 : xmlXPathFreeCompExpr(xtCxt->xpathscomp[i]);
5141 : }
5142 :
5143 132 : if (xtCxt->xpathobj != NULL)
5144 129 : xmlXPathFreeObject(xtCxt->xpathobj);
5145 132 : if (xtCxt->xpathcomp != NULL)
5146 129 : xmlXPathFreeCompExpr(xtCxt->xpathcomp);
5147 132 : if (xtCxt->xpathcxt != NULL)
5148 132 : xmlXPathFreeContext(xtCxt->xpathcxt);
5149 132 : if (xtCxt->doc != NULL)
5150 132 : xmlFreeDoc(xtCxt->doc);
5151 132 : if (xtCxt->ctxt != NULL)
5152 132 : xmlFreeParserCtxt(xtCxt->ctxt);
5153 :
5154 132 : pg_xml_done(xtCxt->xmlerrcxt, true);
5155 :
5156 : /* not valid anymore */
5157 132 : xtCxt->magic = 0;
5158 132 : state->opaque = NULL;
5159 :
5160 : #else
5161 : NO_XML_SUPPORT();
5162 : #endif /* not USE_LIBXML */
5163 132 : }
|
