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