granicus.if.org Git - postgresql/blob - src/backend/utils/adt/ascii.c

   1 /*-----------------------------------------------------------------------
   2  * ascii.c
   3  *       The PostgreSQL routine for string to ascii conversion.
   4  *
   5  *       Portions Copyright (c) 1999-2014, PostgreSQL Global Development Group
   6  *
   7  * IDENTIFICATION
   8  *        src/backend/utils/adt/ascii.c
   9  *
  10  *-----------------------------------------------------------------------
  11  */
  12 #include "postgres.h"
  13
  14 #include "mb/pg_wchar.h"
  15 #include "utils/ascii.h"
  16
  17 static void pg_to_ascii(unsigned char *src, unsigned char *src_end,
  18                         unsigned char *dest, int enc);
  19 static text *encode_to_ascii(text *data, int enc);
  20
  21
  22 /* ----------
  23  * to_ascii
  24  * ----------
  25  */
  26 static void
  27 pg_to_ascii(unsigned char *src, unsigned char *src_end, unsigned char *dest, int enc)
  28 {
  29         unsigned char *x;
  30         const unsigned char *ascii;
  31         int                     range;
  32
  33         /*
  34          * relevant start for an encoding
  35          */
  36 #define RANGE_128       128
  37 #define RANGE_160       160
  38
  39         if (enc == PG_LATIN1)
  40         {
  41                 /*
  42                  * ISO-8859-1 <range: 160 -- 255>
  43                  */
  44                 ascii = (const unsigned char *) "  cL Y  \"Ca  -R     'u .,      ?AAAAAAACEEEEIIII NOOOOOxOUUUUYTBaaaaaaaceeeeiiii nooooo/ouuuuyty";
  45                 range = RANGE_160;
  46         }
  47         else if (enc == PG_LATIN2)
  48         {
  49                 /*
  50                  * ISO-8859-2 <range: 160 -- 255>
  51                  */
  52                 ascii = (const unsigned char *) " A L LS \"SSTZ-ZZ a,l'ls ,sstz\"zzRAAAALCCCEEEEIIDDNNOOOOxRUUUUYTBraaaalccceeeeiiddnnoooo/ruuuuyt.";
  53                 range = RANGE_160;
  54         }
  55         else if (enc == PG_LATIN9)
  56         {
  57                 /*
  58                  * ISO-8859-15 <range: 160 -- 255>
  59                  */
  60                 ascii = (const unsigned char *) "  cL YS sCa  -R     Zu .z   EeY?AAAAAAACEEEEIIII NOOOOOxOUUUUYTBaaaaaaaceeeeiiii nooooo/ouuuuyty";
  61                 range = RANGE_160;
  62         }
  63         else if (enc == PG_WIN1250)
  64         {
  65                 /*
  66                  * Window CP1250 <range: 128 -- 255>
  67                  */
  68                 ascii = (const unsigned char *) "  ' \"    %S<STZZ `'\"\".--  s>stzz   L A  \"CS  -RZ  ,l'u .,as L\"lzRAAAALCCCEEEEIIDDNNOOOOxRUUUUYTBraaaalccceeeeiiddnnoooo/ruuuuyt ";
  69                 range = RANGE_128;
  70         }
  71         else
  72         {
  73                 ereport(ERROR,
  74                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
  75                                  errmsg("encoding conversion from %s to ASCII not supported",
  76                                                 pg_encoding_to_char(enc))));
  77                 return;                                 /* keep compiler quiet */
  78         }
  79
  80         /*
  81          * Encode
  82          */
  83         for (x = src; x < src_end; x++)
  84         {
  85                 if (*x < 128)
  86                         *dest++ = *x;
  87                 else if (*x < range)
  88                         *dest++ = ' ';          /* bogus 128 to 'range' */
  89                 else
  90                         *dest++ = ascii[*x - range];
  91         }
  92 }
  93
  94 /* ----------
  95  * encode text
  96  *
  97  * The text datum is overwritten in-place, therefore this coding method
  98  * cannot support conversions that change the string length!
  99  * ----------
 100  */
 101 static text *
 102 encode_to_ascii(text *data, int enc)
 103 {
 104         pg_to_ascii((unsigned char *) VARDATA(data),            /* src */
 105                                 (unsigned char *) (data) + VARSIZE(data),               /* src end */
 106                                 (unsigned char *) VARDATA(data),                /* dest */
 107                                 enc);                   /* encoding */
 108
 109         return data;
 110 }
 111
 112 /* ----------
 113  * convert to ASCII - enc is set as 'name' arg.
 114  * ----------
 115  */
 116 Datum
 117 to_ascii_encname(PG_FUNCTION_ARGS)
 118 {
 119         text       *data = PG_GETARG_TEXT_P_COPY(0);
 120         char       *encname = NameStr(*PG_GETARG_NAME(1));
 121         int                     enc = pg_char_to_encoding(encname);
 122
 123         if (enc < 0)
 124                 ereport(ERROR,
 125                                 (errcode(ERRCODE_UNDEFINED_OBJECT),
 126                                  errmsg("%s is not a valid encoding name", encname)));
 127
 128         PG_RETURN_TEXT_P(encode_to_ascii(data, enc));
 129 }
 130
 131 /* ----------
 132  * convert to ASCII - enc is set as int4
 133  * ----------
 134  */
 135 Datum
 136 to_ascii_enc(PG_FUNCTION_ARGS)
 137 {
 138         text       *data = PG_GETARG_TEXT_P_COPY(0);
 139         int                     enc = PG_GETARG_INT32(1);
 140
 141         if (!PG_VALID_ENCODING(enc))
 142                 ereport(ERROR,
 143                                 (errcode(ERRCODE_UNDEFINED_OBJECT),
 144                                  errmsg("%d is not a valid encoding code", enc)));
 145
 146         PG_RETURN_TEXT_P(encode_to_ascii(data, enc));
 147 }
 148
 149 /* ----------
 150  * convert to ASCII - current enc is DatabaseEncoding
 151  * ----------
 152  */
 153 Datum
 154 to_ascii_default(PG_FUNCTION_ARGS)
 155 {
 156         text       *data = PG_GETARG_TEXT_P_COPY(0);
 157         int                     enc = GetDatabaseEncoding();
 158
 159         PG_RETURN_TEXT_P(encode_to_ascii(data, enc));
 160 }
 161
 162 /* ----------
 163  * Copy a string in an arbitrary backend-safe encoding, converting it to a
 164  * valid ASCII string by replacing non-ASCII bytes with '?'.  Otherwise the
 165  * behavior is identical to strlcpy(), except that we don't bother with a
 166  * return value.
 167  *
 168  * This must not trigger ereport(ERROR), as it is called in postmaster.
 169  * ----------
 170  */
 171 void
 172 ascii_safe_strlcpy(char *dest, const char *src, size_t destsiz)
 173 {
 174         if (destsiz == 0)                       /* corner case: no room for trailing nul */
 175                 return;
 176
 177         while (--destsiz > 0)
 178         {
 179                 /* use unsigned char here to avoid compiler warning */
 180                 unsigned char ch = *src++;
 181
 182                 if (ch == '\0')
 183                         break;
 184                 /* Keep printable ASCII characters */
 185                 if (32 <= ch && ch <= 127)
 186                         *dest = ch;
 187                 /* White-space is also OK */
 188                 else if (ch == '\n' || ch == '\r' || ch == '\t')
 189                         *dest = ch;
 190                 /* Everything else is replaced with '?' */
 191                 else
 192                         *dest = '?';
 193                 dest++;
 194         }
 195
 196         *dest = '\0';
 197 }