begriffs open source - sa-parse/blob - src/csv.l

   1 %{
   2 #include <stdio.h>
   3 #include <string.h>
   4 #include <stdlib.h>
   5 #include <unistd.h>
   6 #include "csv.tab.h"
   7
   8 /* Fix for bison-bridge with prefix */
   9 #define YYSTYPE CSV_STYPE
  10
  11 /* Buffer for building field content */
  12 static char field_buffer[8192];
  13 static size_t field_pos = 0;
  14
  15 void reset_field_buffer() {
  16     field_pos = 0;
  17     field_buffer[0] = '\0';
  18 }
  19
  20 void append_to_field(char c) {
  21     if (field_pos < sizeof(field_buffer) - 1) {  /* Reserve space for null terminator */
  22         field_buffer[field_pos++] = c;
  23         field_buffer[field_pos] = '\0';
  24     }
  25 }
  26
  27 void append_string_to_field(const char* s) {
  28     while (*s && field_pos < sizeof(field_buffer) - 1) {  /* Reserve space for null terminator */
  29         field_buffer[field_pos++] = *s++;
  30     }
  31     field_buffer[field_pos] = '\0';
  32 }
  33
  34 char* get_field_content() {
  35     char* result = strdup(field_buffer);  /* NOTE: Caller must free() this memory */
  36     if (!result) {
  37         fprintf(stderr, "Memory allocation failed in get_field_content()\n");
  38         exit(1);  /* Conservative approach: exit on memory failure */
  39     }
  40     return result;
  41 }
  42 %}
  43
  44 %option noyywrap
  45 %option yylineno
  46 %option nounput
  47 %option noinput
  48 %option prefix="csv_"
  49 %option bison-bridge
  50 %option reentrant
  51
  52 %x ESCAPED_FIELD
  53
  54 TEXTDATA    [\x20-\x21\x23-\x2B\x2D-\x7E]
  55 COMMA       \x2C
  56 CR          \x0D
  57 LF          \x0A
  58 DQUOTE      \x22
  59 CRLF        {CR}{LF}
  60
  61 %%
  62
  63 {COMMA}                 {
  64                           /* Check if we need to emit an empty field before the comma */
  65                           /* This will be handled by the parser grammar instead */
  66                           return COMMA_TOK;
  67                         }
  68
  69 {CRLF}                  { return CRLF_TOK; }
  70
  71 {LF}                    { return CRLF_TOK; }
  72
  73 {DQUOTE}                {
  74                           reset_field_buffer();
  75                           BEGIN(ESCAPED_FIELD);
  76                         }
  77
  78 <ESCAPED_FIELD>{DQUOTE}{DQUOTE}  { append_to_field('"'); }
  79
  80 <ESCAPED_FIELD>{DQUOTE}          {
  81                                    yylval->str = get_field_content();
  82                                    BEGIN(INITIAL);
  83                                    return FIELD_TOK;
  84                                  }
  85
  86 <ESCAPED_FIELD>{TEXTDATA}        { append_to_field(yytext[0]); }
  87
  88 <ESCAPED_FIELD>{COMMA}           { append_to_field(','); }
  89
  90 <ESCAPED_FIELD>{CR}              { append_to_field('\r'); }
  91
  92 <ESCAPED_FIELD>{LF}              { append_to_field('\n'); }
  93
  94 <ESCAPED_FIELD>.                 { append_to_field(yytext[0]); }
  95
  96 <ESCAPED_FIELD><<EOF>>           {
  97                                    /* Handle unterminated quoted field */
  98                                    yylval->str = get_field_content();
  99                                    BEGIN(INITIAL);
 100                                    return FIELD_TOK;
 101                                  }
 102
 103 {TEXTDATA}+             {
 104                           yylval->str = strdup(yytext);  /* NOTE: Caller must free() this memory */
 105                           if (!yylval->str) {
 106                               fprintf(stderr, "Memory allocation failed in TEXTDATA rule\n");
 107                               exit(1);  /* Conservative approach: exit on memory failure */
 108                           }
 109                           return FIELD_TOK;
 110                         }
 111
 112 [ \t]                   { /* ignore whitespace outside of fields */ }
 113
 114 .                       { return yytext[0]; }
 115
 116 %%