Line data    Source code

       1             : /*-------------------------------------------------------------------------
       2             :  *
       3             :  * scansup.c
       4             :  *    scanner support routines used by the core lexer
       5             :  *
       6             :  * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
       7             :  * Portions Copyright (c) 1994, Regents of the University of California
       8             :  *
       9             :  *
      10             :  * IDENTIFICATION
      11             :  *    src/backend/parser/scansup.c
      12             :  *
      13             :  *-------------------------------------------------------------------------
      14             :  */
      15             : #include "postgres.h"
      16             : 
      17             : #include <ctype.h>
      18             : 
      19             : #include "mb/pg_wchar.h"
      20             : #include "parser/scansup.h"
      21             : 
      22             : 
      23             : /*
      24             :  * downcase_truncate_identifier() --- do appropriate downcasing and
      25             :  * truncation of an unquoted identifier.  Optionally warn of truncation.
      26             :  *
      27             :  * Returns a palloc'd string containing the adjusted identifier.
      28             :  *
      29             :  * Note: in some usages the passed string is not null-terminated.
      30             :  *
      31             :  * Note: the API of this function is designed to allow for downcasing
      32             :  * transformations that increase the string length, but we don't yet
      33             :  * support that.  If you want to implement it, you'll need to fix
      34             :  * SplitIdentifierString() in utils/adt/varlena.c.
      35             :  */
      36             : char *
      37     5408742 : downcase_truncate_identifier(const char *ident, int len, bool warn)
      38             : {
      39     5408742 :     return downcase_identifier(ident, len, warn, true);
      40             : }
      41             : 
      42             : /*
      43             :  * a workhorse for downcase_truncate_identifier
      44             :  */
      45             : char *
      46     5408844 : downcase_identifier(const char *ident, int len, bool warn, bool truncate)
      47             : {
      48             :     char       *result;
      49             :     int         i;
      50             :     bool        enc_is_single_byte;
      51             : 
      52     5408844 :     result = palloc(len + 1);
      53     5408844 :     enc_is_single_byte = pg_database_encoding_max_length() == 1;
      54             : 
      55             :     /*
      56             :      * SQL99 specifies Unicode-aware case normalization, which we don't yet
      57             :      * have the infrastructure for.  Instead we use tolower() to provide a
      58             :      * locale-aware translation.  However, there are some locales where this
      59             :      * is not right either (eg, Turkish may do strange things with 'i' and
      60             :      * 'I').  Our current compromise is to use tolower() for characters with
      61             :      * the high bit set, as long as they aren't part of a multi-byte
      62             :      * character, and use an ASCII-only downcasing for 7-bit characters.
      63             :      */
      64    42717528 :     for (i = 0; i < len; i++)
      65             :     {
      66    37308684 :         unsigned char ch = (unsigned char) ident[i];
      67             : 
      68    37308684 :         if (ch >= 'A' && ch <= 'Z')
      69      944164 :             ch += 'a' - 'A';
      70    36364520 :         else if (enc_is_single_byte && IS_HIGHBIT_SET(ch) && isupper(ch))
      71           0 :             ch = tolower(ch);
      72    37308684 :         result[i] = (char) ch;
      73             :     }
      74     5408844 :     result[i] = '\0';
      75             : 
      76     5408844 :     if (i >= NAMEDATALEN && truncate)
      77          12 :         truncate_identifier(result, i, warn);
      78             : 
      79     5408844 :     return result;
      80             : }
      81             : 
      82             : 
      83             : /*
      84             :  * truncate_identifier() --- truncate an identifier to NAMEDATALEN-1 bytes.
      85             :  *
      86             :  * The given string is modified in-place, if necessary.  A warning is
      87             :  * issued if requested.
      88             :  *
      89             :  * We require the caller to pass in the string length since this saves a
      90             :  * strlen() call in some common usages.
      91             :  */
      92             : void
      93      276226 : truncate_identifier(char *ident, int len, bool warn)
      94             : {
      95      276226 :     if (len >= NAMEDATALEN)
      96             :     {
      97          14 :         len = pg_mbcliplen(ident, len, NAMEDATALEN - 1);
      98          14 :         if (warn)
      99          14 :             ereport(NOTICE,
     100             :                     (errcode(ERRCODE_NAME_TOO_LONG),
     101             :                      errmsg("identifier \"%s\" will be truncated to \"%.*s\"",
     102             :                             ident, len, ident)));
     103          14 :         ident[len] = '\0';
     104             :     }
     105      276226 : }
     106             : 
     107             : /*
     108             :  * scanner_isspace() --- return true if flex scanner considers char whitespace
     109             :  *
     110             :  * This should be used instead of the potentially locale-dependent isspace()
     111             :  * function when it's important to match the lexer's behavior.
     112             :  *
     113             :  * In principle we might need similar functions for isalnum etc, but for the
     114             :  * moment only isspace seems needed.
     115             :  */
     116             : bool
     117    23808388 : scanner_isspace(char ch)
     118             : {
     119             :     /* This must match scan.l's list of {space} characters */
     120    23808388 :     if (ch == ' ' ||
     121    23719802 :         ch == '\t' ||
     122    23719542 :         ch == '\n' ||
     123    23719524 :         ch == '\r' ||
     124    23719518 :         ch == '\v' ||
     125             :         ch == '\f')
     126       88876 :         return true;
     127    23719512 :     return false;
     128             : }