QRScanner/mobile/jni/pdf/pdf-cmap-parse.c

   1 #include "mupdf/pdf.h"
   2
   3 /*
   4  * CMap parser
   5  */
   6
   7 static int
   8 pdf_code_from_string(char *buf, int len)
   9 {
  10         int a = 0;
  11         while (len--)
  12                 a = (a << 8) | *(unsigned char *)buf++;
  13         return a;
  14 }
  15
  16 static void
  17 pdf_parse_cmap_name(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
  18 {
  19         pdf_token tok;
  20
  21         tok = pdf_lex(file, buf);
  22
  23         if (tok == PDF_TOK_NAME)
  24                 fz_strlcpy(cmap->cmap_name, buf->scratch, sizeof(cmap->cmap_name));
  25         else
  26                 fz_warn(ctx, "expected name after CMapName in cmap");
  27 }
  28
  29 static void
  30 pdf_parse_wmode(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
  31 {
  32         pdf_token tok;
  33
  34         tok = pdf_lex(file, buf);
  35
  36         if (tok == PDF_TOK_INT)
  37                 pdf_set_cmap_wmode(ctx, cmap, buf->i);
  38         else
  39                 fz_warn(ctx, "expected integer after WMode in cmap");
  40 }
  41
  42 static void
  43 pdf_parse_codespace_range(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
  44 {
  45         pdf_token tok;
  46         int lo, hi;
  47
  48         while (1)
  49         {
  50                 tok = pdf_lex(file, buf);
  51
  52                 if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "endcodespacerange"))
  53                         return;
  54
  55                 else if (tok == PDF_TOK_STRING)
  56                 {
  57                         lo = pdf_code_from_string(buf->scratch, buf->len);
  58                         tok = pdf_lex(file, buf);
  59                         if (tok == PDF_TOK_STRING)
  60                         {
  61                                 hi = pdf_code_from_string(buf->scratch, buf->len);
  62                                 pdf_add_codespace(ctx, cmap, lo, hi, buf->len);
  63                         }
  64                         else break;
  65                 }
  66
  67                 else break;
  68         }
  69
  70         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or endcodespacerange");
  71 }
  72
  73 static void
  74 pdf_parse_cid_range(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
  75 {
  76         pdf_token tok;
  77         int lo, hi, dst;
  78
  79         while (1)
  80         {
  81                 tok = pdf_lex(file, buf);
  82
  83                 if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "endcidrange"))
  84                         return;
  85
  86                 else if (tok != PDF_TOK_STRING)
  87                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or endcidrange");
  88
  89                 lo = pdf_code_from_string(buf->scratch, buf->len);
  90
  91                 tok = pdf_lex(file, buf);
  92                 if (tok != PDF_TOK_STRING)
  93                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string");
  94
  95                 hi = pdf_code_from_string(buf->scratch, buf->len);
  96
  97                 tok = pdf_lex(file, buf);
  98                 if (tok != PDF_TOK_INT)
  99                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected integer");
 100
 101                 dst = buf->i;
 102
 103                 pdf_map_range_to_range(ctx, cmap, lo, hi, dst);
 104         }
 105 }
 106
 107 static void
 108 pdf_parse_cid_char(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
 109 {
 110         pdf_token tok;
 111         int src, dst;
 112
 113         while (1)
 114         {
 115                 tok = pdf_lex(file, buf);
 116
 117                 if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "endcidchar"))
 118                         return;
 119
 120                 else if (tok != PDF_TOK_STRING)
 121                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or endcidchar");
 122
 123                 src = pdf_code_from_string(buf->scratch, buf->len);
 124
 125                 tok = pdf_lex(file, buf);
 126                 if (tok != PDF_TOK_INT)
 127                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected integer");
 128
 129                 dst = buf->i;
 130
 131                 pdf_map_range_to_range(ctx, cmap, src, src, dst);
 132         }
 133 }
 134
 135 static void
 136 pdf_parse_bf_range_array(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf, int lo, int hi)
 137 {
 138         pdf_token tok;
 139         int dst[256];
 140         int i;
 141
 142         while (1)
 143         {
 144                 tok = pdf_lex(file, buf);
 145
 146                 if (tok == PDF_TOK_CLOSE_ARRAY)
 147                         return;
 148
 149                 /* Note: does not handle [ /Name /Name ... ] */
 150                 else if (tok != PDF_TOK_STRING)
 151                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or ]");
 152
 153                 if (buf->len / 2)
 154                 {
 155                         int len = fz_mini(buf->len / 2, nelem(dst));
 156                         for (i = 0; i < len; i++)
 157                                 dst[i] = pdf_code_from_string(&buf->scratch[i * 2], 2);
 158
 159                         pdf_map_one_to_many(ctx, cmap, lo, dst, buf->len / 2);
 160                 }
 161
 162                 lo ++;
 163         }
 164 }
 165
 166 static void
 167 pdf_parse_bf_range(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
 168 {
 169         pdf_token tok;
 170         int lo, hi, dst;
 171
 172         while (1)
 173         {
 174                 tok = pdf_lex(file, buf);
 175
 176                 if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "endbfrange"))
 177                         return;
 178
 179                 else if (tok != PDF_TOK_STRING)
 180                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or endbfrange");
 181
 182                 lo = pdf_code_from_string(buf->scratch, buf->len);
 183
 184                 tok = pdf_lex(file, buf);
 185                 if (tok != PDF_TOK_STRING)
 186                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string");
 187
 188                 hi = pdf_code_from_string(buf->scratch, buf->len);
 189                 if (lo < 0 || lo > 65535 || hi < 0 || hi > 65535 || lo > hi)
 190                 {
 191                         fz_warn(ctx, "bf_range limits out of range in cmap %s", cmap->cmap_name);
 192                         return;
 193                 }
 194
 195                 tok = pdf_lex(file, buf);
 196
 197                 if (tok == PDF_TOK_STRING)
 198                 {
 199                         if (buf->len == 2)
 200                         {
 201                                 dst = pdf_code_from_string(buf->scratch, buf->len);
 202                                 pdf_map_range_to_range(ctx, cmap, lo, hi, dst);
 203                         }
 204                         else
 205                         {
 206                                 int dststr[256];
 207                                 int i;
 208
 209                                 if (buf->len / 2)
 210                                 {
 211                                         int len = fz_mini(buf->len / 2, nelem(dststr));
 212                                         for (i = 0; i < len; i++)
 213                                                 dststr[i] = pdf_code_from_string(&buf->scratch[i * 2], 2);
 214
 215                                         while (lo <= hi)
 216                                         {
 217                                                 dststr[i-1] ++;
 218                                                 pdf_map_one_to_many(ctx, cmap, lo, dststr, i);
 219                                                 lo ++;
 220                                         }
 221                                 }
 222                         }
 223                 }
 224
 225                 else if (tok == PDF_TOK_OPEN_ARRAY)
 226                 {
 227                         pdf_parse_bf_range_array(ctx, cmap, file, buf, lo, hi);
 228                 }
 229
 230                 else
 231                 {
 232                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or array or endbfrange");
 233                 }
 234         }
 235 }
 236
 237 static void
 238 pdf_parse_bf_char(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, pdf_lexbuf *buf)
 239 {
 240         pdf_token tok;
 241         int dst[256];
 242         int src;
 243         int i;
 244
 245         while (1)
 246         {
 247                 tok = pdf_lex(file, buf);
 248
 249                 if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "endbfchar"))
 250                         return;
 251
 252                 else if (tok != PDF_TOK_STRING)
 253                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string or endbfchar");
 254
 255                 src = pdf_code_from_string(buf->scratch, buf->len);
 256
 257                 tok = pdf_lex(file, buf);
 258                 /* Note: does not handle /dstName */
 259                 if (tok != PDF_TOK_STRING)
 260                         fz_throw(ctx, FZ_ERROR_GENERIC, "expected string");
 261
 262                 if (buf->len / 2)
 263                 {
 264                         int len = fz_mini(buf->len / 2, nelem(dst));
 265                         for (i = 0; i < len; i++)
 266                                 dst[i] = pdf_code_from_string(&buf->scratch[i * 2], 2);
 267                         pdf_map_one_to_many(ctx, cmap, src, dst, i);
 268                 }
 269         }
 270 }
 271
 272 pdf_cmap *
 273 pdf_load_cmap(fz_context *ctx, fz_stream *file)
 274 {
 275         pdf_cmap *cmap;
 276         char key[64];
 277         pdf_lexbuf buf;
 278         pdf_token tok;
 279
 280         pdf_lexbuf_init(ctx, &buf, PDF_LEXBUF_SMALL);
 281         cmap = pdf_new_cmap(ctx);
 282
 283         strcpy(key, ".notdef");
 284
 285         fz_try(ctx)
 286         {
 287                 while (1)
 288                 {
 289                         tok = pdf_lex(file, &buf);
 290
 291                         if (tok == PDF_TOK_EOF)
 292                                 break;
 293
 294                         else if (tok == PDF_TOK_NAME)
 295                         {
 296                                 if (!strcmp(buf.scratch, "CMapName"))
 297                                         pdf_parse_cmap_name(ctx, cmap, file, &buf);
 298                                 else if (!strcmp(buf.scratch, "WMode"))
 299                                         pdf_parse_wmode(ctx, cmap, file, &buf);
 300                                 else
 301                                         fz_strlcpy(key, buf.scratch, sizeof key);
 302                         }
 303
 304                         else if (tok == PDF_TOK_KEYWORD)
 305                         {
 306                                 if (!strcmp(buf.scratch, "endcmap"))
 307                                         break;
 308
 309                                 else if (!strcmp(buf.scratch, "usecmap"))
 310                                         fz_strlcpy(cmap->usecmap_name, key, sizeof(cmap->usecmap_name));
 311
 312                                 else if (!strcmp(buf.scratch, "begincodespacerange"))
 313                                         pdf_parse_codespace_range(ctx, cmap, file, &buf);
 314
 315                                 else if (!strcmp(buf.scratch, "beginbfchar"))
 316                                         pdf_parse_bf_char(ctx, cmap, file, &buf);
 317
 318                                 else if (!strcmp(buf.scratch, "begincidchar"))
 319                                         pdf_parse_cid_char(ctx, cmap, file, &buf);
 320
 321                                 else if (!strcmp(buf.scratch, "beginbfrange"))
 322                                         pdf_parse_bf_range(ctx, cmap, file, &buf);
 323
 324                                 else if (!strcmp(buf.scratch, "begincidrange"))
 325                                         pdf_parse_cid_range(ctx, cmap, file, &buf);
 326                         }
 327
 328                         /* ignore everything else */
 329                 }
 330
 331                 pdf_sort_cmap(ctx, cmap);
 332         }
 333         fz_always(ctx)
 334         {
 335                 pdf_lexbuf_fin(&buf);
 336         }
 337         fz_catch(ctx)
 338         {
 339                 pdf_drop_cmap(ctx, cmap);
 340                 fz_rethrow_message(ctx, "syntaxerror in cmap");
 341         }
 342
 343         return cmap;
 344 }