Line data    Source code

       1             : /**
       2             :  *    @file  csv.c
       3             :  *   @brief  https://www.rfc-editor.org/rfc/rfc4180
       4             :  *
       5             :  *  <+DETAILED+>
       6             :  *
       7             :  *  @author  François Cerbelle (Fanfan), francois@cerbelle.net
       8             :  *
       9             :  *  @internal
      10             :  *       Created:  29/10/2024
      11             :  *      Revision:  none
      12             :  * Last modified:  2024-12-05 20:17
      13             :  *      Compiler:  gcc
      14             :  *  Organization:  Cerbelle.net
      15             :  *     Copyright:  Copyright (c) 2024, François Cerbelle
      16             :  *
      17             :  *  This source code is released for free distribution under the terms of the
      18             :  *  GNU General Public License as published by the Free Software Foundation.
      19             :  */
      20             : 
      21             : #ifdef HAVE_CONFIG_H
      22             : #include "config.h"
      23             : #endif
      24             : 
      25             : #include "csv.h"
      26             : 
      27             : #include <stdlib.h>
      28             : #include <string.h>
      29             : 
      30             : static unsigned short int _csv_firstfield = 1;
      31             : 
      32          44 : void csv_addline(FILE* reportfile) {
      33          44 :     fprintf(reportfile,"\r\n");
      34          44 :     _csv_firstfield = 1;
      35          44 : }
      36             : 
      37         176 : void csv_addfield(FILE* reportfile, const char* value) {
      38         176 :     char* csv = txt2csv(value);
      39         176 :     if (_csv_firstfield) {
      40          44 :         fprintf(reportfile,"%s",csv);
      41          44 :         _csv_firstfield = 0;
      42             :     } else
      43         132 :         fprintf(reportfile,",%s",csv);
      44         176 :     free(csv);
      45         176 : }
      46             : 
      47             : /* RFC4180 compliant CSV parser, with LF only tolerance when CRLF expected */
      48             : /* https://www.rfc-editor.org/rfc/rfc4180 */
      49             : static char* _csvtok_csv=NULL;
      50         324 : char* csvtok(char* source) {
      51         324 :     size_t csv_idx=0;
      52         324 :     size_t txt_idx=0;
      53         324 :     char* csvtok_txt = NULL;
      54         324 :     unsigned short int quoted = 0;
      55             : 
      56         324 :     if (NULL!=source)
      57          66 :         _csvtok_csv = source;
      58             : 
      59         324 :     if (_csvtok_csv[0]==0)
      60           6 :         return NULL;
      61             : 
      62         318 :     if (NULL==(csvtok_txt = malloc(strlen(_csvtok_csv)+1))) {
      63           0 :         perror("csvtok csvtok_txt malloc");
      64           0 :         exit(EXIT_FAILURE);
      65             :     }
      66         318 :     csvtok_txt[0]=0;
      67             : 
      68        2018 :     while (_csvtok_csv[csv_idx]) {
      69        2009 :         if (!quoted) {
      70        1712 :             if (_csvtok_csv[csv_idx]==',') {
      71         216 :                 if ((_csvtok_csv[csv_idx+1]==0)||
      72         213 :                         (_csvtok_csv[csv_idx+1]=='\n')|| /* Unix style tolerance */
      73         207 :                         ((_csvtok_csv[csv_idx+1]=='\r')&&(_csvtok_csv[csv_idx+2]=='\n'))) {
      74          12 :                     fprintf(stderr,"RFC4180 forbids comma at the end of record %s at %zu.\n",
      75             :                             _csvtok_csv,csv_idx);
      76          12 :                     free(csvtok_txt);
      77          12 :                     _csvtok_csv=NULL;
      78          12 :                     return NULL;
      79             :                 } else
      80             :                     break;
      81        1496 :             } else if ((_csvtok_csv[csv_idx]=='\n')||
      82        1496 :                        ((_csvtok_csv[csv_idx]=='\r')&&(_csvtok_csv[csv_idx+1]=='\n')))
      83             :                 break;
      84        1493 :             else if (_csvtok_csv[csv_idx]=='"') {
      85          93 :                 if (csv_idx==0) {
      86          90 :                     quoted = 1;
      87          90 :                     csv_idx++;
      88          90 :                     continue;
      89             :                 } else {
      90           3 :                     fprintf(stderr,"doublequote in a non quoted value %s at %zu.\n",
      91             :                             _csvtok_csv,csv_idx);
      92           3 :                     free(csvtok_txt);
      93           3 :                     _csvtok_csv=NULL;
      94           3 :                     return NULL;
      95             :                 }
      96             :             }
      97             :         } else { /* Quoted */
      98         297 :             if (_csvtok_csv[csv_idx]=='"') {
      99          93 :                 if (_csvtok_csv[csv_idx+1]=='"') {
     100             :                     /* Skip escaping doublequote and let the copy occur */
     101           6 :                     csv_idx++;
     102          87 :                 } else if ((_csvtok_csv[csv_idx+1]=='0')||
     103          87 :                            (_csvtok_csv[csv_idx+1]==',')||
     104          39 :                            (_csvtok_csv[csv_idx+1]=='\n')||
     105           3 :                            ((_csvtok_csv[csv_idx+1]=='\r')&&(_csvtok_csv[csv_idx+2]=='\n'))) {
     106          84 :                     quoted = 0;
     107          84 :                     csv_idx++;
     108          84 :                     break;
     109             :                 } else {
     110           3 :                     fprintf(stderr,"doublequote should be at the end of field or escaping another doublequote in %s at %zu.\n",
     111             :                             _csvtok_csv,csv_idx);
     112           3 :                     free(csvtok_txt);
     113           3 :                     _csvtok_csv=NULL;
     114           3 :                     return NULL;
     115             :                 }
     116             :             }
     117             :         }
     118        1610 :         csvtok_txt[txt_idx++]=_csvtok_csv[csv_idx++];
     119             :     }
     120             : 
     121             :     /* Close csvtok_txt */
     122         300 :     csvtok_txt[txt_idx]=0;
     123         300 :     if (quoted) {
     124           3 :         fprintf(stderr,"Missing end-of-field doublequote %s\n",csvtok_txt);
     125           3 :         free(csvtok_txt);
     126           3 :         _csvtok_csv=NULL;
     127           3 :         return NULL;
     128         297 :     } else if (_csvtok_csv[csv_idx]==',') /* Next char after end of field should be 0/,/CRLF */
     129         252 :         csv_idx++;
     130          45 :     else if (_csvtok_csv[csv_idx]=='\n')
     131          36 :         csv_idx+=1;
     132           9 :     else if ((_csvtok_csv[csv_idx]=='\r')&&(_csvtok_csv[csv_idx+1]=='\n'))
     133           3 :         csv_idx+=2;
     134           6 :     else if (_csvtok_csv[csv_idx]!=0) {
     135           0 :         fprintf(stderr,"Parsing error after %s\n",csvtok_txt);
     136           0 :         free(csvtok_txt);
     137           0 :         _csvtok_csv=NULL;
     138           0 :         return NULL;
     139             :     }
     140         297 :     _csvtok_csv += csv_idx;
     141             :     /* Shrink overallocated string */
     142         297 :     if (NULL==(csvtok_txt=realloc(csvtok_txt,strlen(csvtok_txt)+1))) {
     143           0 :         fprintf(stderr,"ERROR: csvtok Trimming CSV token\n");
     144           0 :         free(csvtok_txt);
     145           0 :         return NULL;
     146             :     }
     147         297 :     return csvtok_txt;
     148             : }
     149             : 
     150             : /* RFC4180 compliant text to CSV encoder */
     151             : /* https://www.rfc-editor.org/rfc/rfc4180 */
     152         245 : char* txt2csv(const char* text) {
     153             :     char* csv;
     154             :     size_t text_idx;
     155             :     size_t csv_idx;
     156         245 :     unsigned short int need_quotes=0;
     157             : 
     158             :     {
     159             :         /* Check if quotes are needed and how many doublequotes are in the
     160             :          * source text to allocate output buffer. This text iteration could be
     161             :          * avoided but would imply to overallocate for the worst case scenario
     162             :          * and to reallocate at the end with the potentially needed surrounding
     163             :          * quotes */
     164         245 :         size_t extra_chars = 0;
     165        1818 :         for (text_idx=0; text[text_idx]; text_idx++) {
     166        1573 :             if ( (text[text_idx]==',')|| (text[text_idx]=='\r')|| (text[text_idx]=='\n'))
     167          85 :                 need_quotes = 1;
     168        1573 :             if (text[text_idx]=='"') {
     169          53 :                 need_quotes = 1;
     170          53 :                 extra_chars++;
     171             :             }
     172             :         }
     173             :         /* Allocate the right output buffer size */
     174         245 :         if (NULL==(csv=malloc(strlen(text)+(need_quotes?2:0)+extra_chars+1))) {
     175           0 :             perror("txt2csv malloc");
     176           0 :             return NULL;
     177             :         }
     178             :     }
     179             : 
     180         245 :     text_idx = 0;
     181         245 :     csv_idx = 0;
     182             : 
     183             :     /* If quotes are needed add a starting doublequote */
     184         245 :     if (need_quotes)
     185          98 :         csv[csv_idx++] = '"';
     186             : 
     187             :     /* Copy each source char to the destination buffer */
     188        1818 :     while (text[text_idx]) {
     189             :         /* With a doublequote before if the char to copy is a doublequote */
     190        1573 :         if (text[text_idx]=='"')
     191          53 :             csv[csv_idx++] = '"';
     192        1573 :         csv[csv_idx++] = text[text_idx++];
     193             :     }
     194             : 
     195             :     /* If quotes are needed add a closing doublequote */
     196         245 :     if (need_quotes)
     197          98 :         csv[csv_idx++] = '"';
     198             : 
     199             :     /* Properly end the C string */
     200         245 :     csv[csv_idx] = 0;
     201             : 
     202         245 :     return csv;
     203             : }
     204             : /* vim: set tw=80: */