test/parsecsv.c

   1 /* $Id: parsecsv.c $
   2
   3    This file is part of the HandBrake source code.
   4    Homepage: <http://handbrake.m0k.org/>.
   5    It may be used under the terms of the GNU General Public License. */
   6
   7 #include <fcntl.h>
   8 #include "hb.h"
   9 #include "parsecsv.h"
  10
  11 /* Internal declarations */
  12 #define is_newline(_x)      ( (_x) == 13 || \
  13                               (_x) == 11 || \
  14                               (_x) == 10 )
  15
  16 #define is_white(_x)        ( (_x) == '\t' || \
  17                               (_x) == ' '  || \
  18                               is_newline(_x) )
  19
  20 #define is_sep(_x)          ( (_x) == ',' )
  21
  22 #define is_esc(_x)          ( (_x) == '\\' )
  23
  24 #define CSV_CHAR_ERROR      0x8000
  25 #define CSV_CHAR_EOF        0x4000
  26 #define CSV_CHAR_ROWSEP     0x2000
  27 #define CSV_CHAR_COLSEP     0x1000
  28
  29 #define CSV_PARSE_NORMAL    0x0000
  30 #define CSV_PARSE_SEEK      0x0001
  31 #define CSV_PARSE_ESC       0x0002
  32
  33 static uint16_t hb_parse_character( hb_csv_file_t * file );
  34 static void hb_trim_end( char *text );
  35
  36 /* Open a CSV File */
  37 hb_csv_file_t *hb_open_csv_file( const char *filepath )
  38 {
  39     hb_csv_file_t *file = NULL;
  40     FILE * fileref;
  41
  42     if( filepath == NULL )
  43     {
  44         return file;
  45     }
  46
  47     fileref = fopen( filepath, "r" );
  48     if( fileref == NULL )
  49     {
  50         return file;
  51     }
  52
  53     file = malloc( sizeof( hb_csv_file_t ) );
  54     file->fileref       = fileref;
  55     file->eof           = 0;
  56     file->parse_state   = CSV_PARSE_SEEK;
  57     file->curr_col      = 0;
  58     file->curr_row      = 0;
  59     return file;
  60 }
  61
  62 void hb_close_csv_file( hb_csv_file_t *file )
  63 {
  64     if( file == NULL )
  65     {
  66         return;
  67     }
  68
  69     fclose( file->fileref );
  70     free( file );
  71 }
  72
  73 /* Parse CSV Cells */
  74 hb_csv_cell_t *hb_read_next_cell( hb_csv_file_t *file )
  75 {
  76     hb_csv_cell_t *cell = NULL;
  77     uint16_t c;
  78     int index;
  79
  80     if( file == NULL  )
  81     {
  82         return cell;
  83     }
  84
  85     if( file->eof )
  86     {
  87         return cell;
  88     }
  89
  90     cell = malloc( sizeof( hb_csv_cell_t ) );
  91     cell->cell_row = file->curr_row;
  92     cell->cell_col = file->curr_col;
  93     index = 0;
  94     while( CSV_CHAR_EOF != (c = hb_parse_character( file ) ) )
  95     {
  96         if( c == CSV_CHAR_ROWSEP )
  97         {
  98             file->curr_row++;
  99             file->curr_col = 0;
 100             break;
 101         }
 102         else if( c == CSV_CHAR_COLSEP )
 103         {
 104             file->curr_col++;
 105             break;
 106         }
 107         else
 108         {
 109             if( index < 1023 )
 110             {
 111                 cell->cell_text[index] = (char)c;
 112                 index++;
 113             }
 114         }
 115     }
 116
 117     if( c == CSV_CHAR_EOF )
 118     {
 119         file->eof = 1;
 120     }
 121
 122     /* Terminate the cell text */
 123     cell->cell_text[index] = '\0';
 124     hb_trim_end( cell->cell_text );
 125     return cell;
 126 }
 127
 128 void hb_dispose_cell( hb_csv_cell_t *cell )
 129 {
 130     if( cell == NULL )
 131     {
 132         return;
 133     }
 134
 135     free( cell );
 136 }
 137
 138 /* Raw parsing */
 139 static uint16_t hb_parse_character( hb_csv_file_t * file )
 140 {
 141     int byte;
 142     uint16_t c;
 143     int need_char = 1;
 144
 145     if( file == NULL )
 146     {
 147         return CSV_CHAR_ERROR;
 148     }
 149
 150     while( need_char )
 151     {
 152         byte = fgetc( file->fileref );
 153         if( feof( file->fileref ) )
 154         {
 155             return CSV_CHAR_EOF;
 156         }
 157         if( ferror( file->fileref ) )
 158         {
 159             return CSV_CHAR_ERROR;
 160         }
 161
 162         if( file->parse_state == CSV_PARSE_SEEK && is_white(byte) )
 163         {
 164             continue;
 165         }
 166         else if( file->parse_state != CSV_PARSE_ESC && is_esc(byte) )
 167         {
 168             file->parse_state = CSV_PARSE_ESC;
 169             continue;
 170         }
 171         else if( file->parse_state != CSV_PARSE_ESC && is_sep(byte) )
 172         {
 173             file->parse_state = CSV_PARSE_SEEK;
 174             need_char = 0;
 175             c = CSV_CHAR_COLSEP;
 176         }
 177         else if( file->parse_state == CSV_PARSE_ESC )
 178         {
 179             file->parse_state = CSV_PARSE_NORMAL;
 180             need_char = 0;
 181             c = (uint16_t)byte;
 182         }
 183         else if( is_newline(byte) )
 184         {
 185             file->parse_state = CSV_PARSE_SEEK;
 186             need_char = 0;
 187             c = CSV_CHAR_ROWSEP;
 188         }
 189         else
 190         {
 191             file->parse_state = CSV_PARSE_NORMAL;
 192             need_char = 0;
 193             c = (uint16_t)byte;
 194         }
 195     }
 196
 197     return c;
 198 }
 199
 200 static void hb_trim_end( char *text )
 201 {
 202     if( text == NULL )
 203     {
 204         return;
 205     }
 206
 207     int i = strlen(text) - 1;
 208
 209     for( i = strlen(text) - 1; i >= 0 && is_white(text[i]) ; i-- )
 210     {
 211         text[i] = '\0';
 212     }
 213 }