Sat, 16 Dec 2006 04:59:55 +0000
This is a hand-crafted commit to migrate across subversion revisions
16854:16861, due to some vagaries of the way the original renames were
done. Witness that monotone can do in one revision what svn had to
spread across several.
| 9268 | 1 | /* |
| 2 | * nmrtf.c | |
| 3 | * | |
| 4 | * Copyright (c) 2004 Novell, Inc. All Rights Reserved. | |
| 5 | * | |
| 6 | * This program is free software; you can redistribute it and/or modify | |
| 7 | * it under the terms of the GNU General Public License as published by | |
| 8 | * the Free Software Foundation; version 2 of the License. | |
| 9 | * | |
| 10 | * This program is distributed in the hope that it will be useful, | |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
| 13 | * GNU General Public License for more details. | |
| 14 | * | |
| 15 | * You should have received a copy of the GNU General Public License | |
| 16 | * along with this program; if not, write to the Free Software | |
| 17 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | |
| 18 | * | |
| 19 | */ | |
| 20 | ||
| 21 | /* This code was adapted from the sample RTF reader found here: | |
| 22 | * http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnrtfspec/html/rtfspec.asp | |
| 23 | */ | |
| 24 | ||
| 25 | #include <glib.h> | |
| 26 | #include <stdlib.h> | |
| 27 | #include <stdio.h> | |
| 28 | #include <stddef.h> | |
| 29 | #include <ctype.h> | |
| 30 | #include <string.h> | |
| 31 | #include "nmrtf.h" | |
| 32 | #include "debug.h" | |
| 33 | ||
| 34 | /* Internal RTF parser error codes */ | |
| 35 | #define NMRTF_OK 0 /* Everything's fine! */ | |
| 36 | #define NMRTF_STACK_UNDERFLOW 1 /* Unmatched '}' */ | |
| 37 | #define NMRTF_STACK_OVERFLOW 2 /* Too many '{' -- memory exhausted */ | |
| 38 | #define NMRTF_UNMATCHED_BRACE 3 /* RTF ended during an open group. */ | |
| 39 | #define NMRTF_INVALID_HEX 4 /* invalid hex character found in data */ | |
| 40 | #define NMRTF_BAD_TABLE 5 /* RTF table (sym or prop) invalid */ | |
| 41 | #define NMRTF_ASSERTION 6 /* Assertion failure */ | |
| 42 | #define NMRTF_EOF 7 /* End of file reached while reading RTF */ | |
| 43 | #define NMRTF_CONVERT_ERROR 8 /* Error converting text */ | |
| 44 | ||
| 45 | #define NMRTF_MAX_DEPTH 256 | |
| 46 | ||
| 47 | typedef enum | |
| 48 | { | |
| 49 | NMRTF_STATE_NORMAL, | |
| 50 | NMRTF_STATE_SKIP, | |
| 51 | NMRTF_STATE_FONTTABLE, | |
| 52 | NMRTF_STATE_BIN, | |
| 53 | NMRTF_STATE_HEX | |
| 54 | } NMRtfState; /* Rtf State */ | |
| 55 | ||
| 56 | /* Property types that we care about */ | |
| 57 | typedef enum | |
| 58 | { | |
| 59 | NMRTF_PROP_FONT_IDX, | |
| 60 | NMRTF_PROP_FONT_CHARSET, | |
| 61 | NMRTF_PROP_MAX | |
| 62 | } NMRtfProperty; | |
| 63 | ||
| 64 | typedef enum | |
| 65 | { | |
| 66 | NMRTF_SPECIAL_BIN, | |
| 67 | NMRTF_SPECIAL_HEX, | |
| 68 | NMRTF_SPECIAL_UNICODE, | |
| 69 | NMRTF_SPECIAL_SKIP | |
| 70 | } NMRtfSpecialKwd; | |
| 71 | ||
| 72 | typedef enum | |
| 73 | { | |
| 74 | NMRTF_DEST_FONTTABLE, | |
| 75 | NMRTF_DEST_SKIP | |
| 76 | } NMRtfDestinationType; | |
| 77 | ||
| 78 | typedef enum | |
| 79 | { | |
| 80 | NMRTF_KWD_CHAR, | |
| 81 | NMRTF_KWD_DEST, | |
| 82 | NMRTF_KWD_PROP, | |
| 83 | NMRTF_KWD_SPEC | |
| 84 | } NMRtfKeywordType; | |
| 85 | ||
| 86 | typedef struct _NMRTFCharProp | |
| 87 | { | |
| 88 | /* All we care about for now is the font. | |
| 89 | * bold, italic, underline, etc. should be | |
| 90 | * added here | |
| 91 | */ | |
| 92 | int font_idx; | |
| 93 | int font_charset; | |
| 94 | } NMRtfCharProp; | |
| 95 | ||
| 96 | typedef struct _NMRtfStateSave | |
| 97 | { | |
| 98 | NMRtfCharProp chp; | |
| 99 | NMRtfState rds; | |
| 100 | NMRtfState ris; | |
| 101 | } NMRtfStateSave; | |
| 102 | ||
| 103 | typedef struct _NMRtfSymbol | |
| 104 | { | |
| 105 | char *keyword; /* RTF keyword */ | |
| 106 | int default_val; /* default value to use */ | |
| 107 | gboolean pass_default; /* true to use default value from this table */ | |
| 108 | NMRtfKeywordType kwd_type; /* the type of the keyword */ | |
| 109 | int action; /* property type if the keyword represents a property */ | |
| 110 | /* destination type if the keyword represents a destination */ | |
| 111 | /* character to print if the keyword represents a character */ | |
| 112 | } NMRtfSymbol; | |
| 113 | ||
| 114 | ||
| 115 | typedef struct _NMRtfFont | |
| 116 | { | |
| 117 | int number; | |
| 118 | char *name; | |
| 119 | int charset; | |
| 120 | } NMRtfFont; | |
| 121 | ||
| 122 | /* RTF Context */ | |
| 123 | struct _NMRtfContext | |
| 124 | { | |
| 125 | NMRtfState rds; /* destination state */ | |
| 126 | NMRtfState ris; /* internal state */ | |
| 127 | NMRtfCharProp chp; /* current character properties (ie. font, bold, italic, etc.) */ | |
| 128 | GSList *font_table; /* the font table */ | |
| 129 | GSList *saved; /* saved state stack */ | |
| 130 | int param; /* numeric parameter for the current keyword */ | |
| 131 | long bytes_to_skip; /* number of bytes to skip (after encountering \bin) */ | |
| 132 | int depth; /* how many groups deep are we */ | |
| 133 | gboolean skip_unknown; /* if true, skip any unknown destinations (this is set after encountering '\*') */ | |
| 134 | char *input; /* input string */ | |
| 135 | char nextch; /* next char in input */ | |
| 136 | GString *ansi; /* Temporary ansi text, will be convert/flushed to the output string */ | |
| 137 | GString *output; /* The plain text UTF8 string */ | |
| 138 | }; | |
| 139 | ||
| 140 | static int rtf_parse(NMRtfContext *ctx); | |
| 141 | static int rtf_push_state(NMRtfContext *ctx); | |
| 142 | static int rtf_pop_state(NMRtfContext *ctx); | |
| 143 | static NMRtfFont *rtf_get_font(NMRtfContext *ctx, int index); | |
| 144 | static int rtf_get_char(NMRtfContext *ctx, guchar *ch); | |
| 145 | static int rtf_unget_char(NMRtfContext *ctx, guchar ch); | |
| 146 | static int rtf_flush_data(NMRtfContext *ctx); | |
| 147 | static int rtf_parse_keyword(NMRtfContext *ctx); | |
| 148 | static int rtf_dispatch_control(NMRtfContext *ctx, char *keyword, int param, gboolean param_set); | |
| 149 | static int rtf_dispatch_char(NMRtfContext *ctx, guchar ch); | |
| 150 | static int rtf_dispatch_unicode_char(NMRtfContext *ctx, gunichar ch); | |
| 151 | static int rtf_print_char(NMRtfContext *ctx, guchar ch); | |
| 152 | static int rtf_print_unicode_char(NMRtfContext *ctx, gunichar ch); | |
| 153 | static int rtf_change_destination(NMRtfContext *ctx, NMRtfDestinationType dest); | |
| 154 | static int rtf_dispatch_special(NMRtfContext *ctx, NMRtfSpecialKwd special); | |
| 155 | static int rtf_apply_property(NMRtfContext *ctx, NMRtfProperty prop, int val); | |
| 156 | ||
| 157 | /* RTF parser tables */ | |
| 158 | ||
| 159 | /* Keyword descriptions */ | |
| 160 | NMRtfSymbol rtf_symbols[] = { | |
| 161 | /* keyword, default, pass_default, keyword_type, action */ | |
| 162 | {"fonttbl", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_FONTTABLE}, | |
| 163 | {"f", 0, FALSE, NMRTF_KWD_PROP, NMRTF_PROP_FONT_IDX}, | |
| 164 | {"fcharset", 0, FALSE, NMRTF_KWD_PROP, NMRTF_PROP_FONT_CHARSET}, | |
| 165 | {"par", 0, FALSE, NMRTF_KWD_CHAR, 0x0a}, | |
| 166 | {"line", 0, FALSE, NMRTF_KWD_CHAR, 0x0a}, | |
| 167 | {"\0x0a", 0, FALSE, NMRTF_KWD_CHAR, 0x0a}, | |
| 168 | {"\0x0d", 0, FALSE, NMRTF_KWD_CHAR, 0x0a}, | |
| 169 | {"tab", 0, FALSE, NMRTF_KWD_CHAR, 0x09}, | |
| 170 | {"\r", 0, FALSE, NMRTF_KWD_CHAR, '\r'}, | |
| 171 | {"\n", 0, FALSE, NMRTF_KWD_CHAR, '\n'}, | |
| 172 | {"ldblquote",0, FALSE, NMRTF_KWD_CHAR, '"'}, | |
| 173 | {"rdblquote",0, FALSE, NMRTF_KWD_CHAR, '"'}, | |
| 174 | {"{", 0, FALSE, NMRTF_KWD_CHAR, '{'}, | |
| 175 | {"}", 0, FALSE, NMRTF_KWD_CHAR, '}'}, | |
| 176 | {"\\", 0, FALSE, NMRTF_KWD_CHAR, '\\'}, | |
| 177 | {"bin", 0, FALSE, NMRTF_KWD_SPEC, NMRTF_SPECIAL_BIN}, | |
| 178 | {"*", 0, FALSE, NMRTF_KWD_SPEC, NMRTF_SPECIAL_SKIP}, | |
| 179 | {"'", 0, FALSE, NMRTF_KWD_SPEC, NMRTF_SPECIAL_HEX}, | |
| 180 | {"u", 0, FALSE, NMRTF_KWD_SPEC, NMRTF_SPECIAL_UNICODE}, | |
| 181 | {"colortbl", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 182 | {"author", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 183 | {"buptim", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 184 | {"comment", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 185 | {"creatim", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 186 | {"doccomm", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 187 | {"footer", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 188 | {"footerf", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 189 | {"footerl", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 190 | {"footerr", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 191 | {"footnote", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 192 | {"ftncn", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 193 | {"ftnsep", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 194 | {"ftnsepc", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 195 | {"header", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 196 | {"headerf", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 197 | {"headerl", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 198 | {"headerr", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 199 | {"info", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 200 | {"keywords", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 201 | {"operator", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 202 | {"pict", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 203 | {"printim", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 204 | {"private1", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 205 | {"revtim", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 206 | {"rxe", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 207 | {"stylesheet", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 208 | {"subject", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 209 | {"tc", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 210 | {"title", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 211 | {"txe", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP}, | |
| 212 | {"xe", 0, FALSE, NMRTF_KWD_DEST, NMRTF_DEST_SKIP} | |
| 213 | }; | |
| 214 | int table_size = sizeof(rtf_symbols) / sizeof(NMRtfSymbol); | |
| 215 | ||
| 216 | NMRtfContext * | |
| 217 | nm_rtf_init() | |
| 218 | { | |
| 219 | NMRtfContext *ctx = g_new0(NMRtfContext, 1); | |
| 220 | ctx->nextch = -1; | |
| 221 | ctx->ansi = g_string_new(""); | |
| 222 | ctx->output = g_string_new(""); | |
| 223 | return ctx; | |
| 224 | } | |
| 225 | ||
| 226 | char * | |
| 227 | nm_rtf_strip_formatting(NMRtfContext *ctx, const char *input) | |
| 228 | { | |
| 229 | int status; | |
| 230 | ||
| 231 | ctx->input = (char *)input; | |
| 232 | status = rtf_parse(ctx); | |
| 233 | if (status == NMRTF_OK) | |
| 234 | return g_strdup(ctx->output->str); | |
| 235 | ||
| 236 | gaim_debug_info("novell", "RTF parser failed with error code %d", status); | |
| 237 | return NULL; | |
| 238 | } | |
| 239 | ||
| 240 | void | |
| 241 | nm_rtf_deinit(NMRtfContext *ctx) | |
| 242 | { | |
| 243 | GSList *node; | |
| 244 | NMRtfFont *font; | |
| 245 | NMRtfStateSave *save; | |
| 246 | ||
| 247 | if (ctx) { | |
| 248 | for (node = ctx->font_table; node; node = node->next) { | |
| 249 | font = node->data; | |
| 250 | g_free(font->name); | |
| 251 | g_free(font); | |
| 252 | node->data = NULL; | |
| 253 | } | |
| 254 | g_slist_free(ctx->font_table); | |
| 255 | for (node = ctx->saved; node; node = node->next) { | |
| 256 | save = node->data; | |
| 257 | g_free(save); | |
| 258 | node->data = NULL; | |
| 259 | } | |
| 260 | g_slist_free(ctx->saved); | |
| 261 | g_string_free(ctx->ansi, TRUE); | |
| 262 | g_string_free(ctx->output, TRUE); | |
| 263 | g_free(ctx); | |
| 264 | } | |
| 265 | } | |
| 266 | ||
| 267 | static const char * | |
| 268 | get_current_encoding(NMRtfContext *ctx) | |
| 269 | { | |
| 270 | NMRtfFont *font; | |
| 271 | ||
| 272 | font = rtf_get_font(ctx, ctx->chp.font_idx); | |
| 273 | ||
| 274 | switch (font->charset) { | |
| 275 | case 0: | |
| 276 | return "CP1252"; | |
| 277 | case 77: | |
| 278 | return "MACINTOSH"; | |
| 279 | case 78: | |
| 280 | return "SJIS"; | |
| 281 | case 128: | |
| 282 | return "CP932"; | |
| 283 | case 129: | |
| 284 | return "CP949"; | |
| 285 | case 130: | |
| 286 | return "CP1361"; | |
| 287 | case 134: | |
| 288 | return "CP936"; | |
| 289 | case 136: | |
| 290 | return "CP950"; | |
| 291 | case 161: | |
| 292 | return "CP1253"; | |
| 293 | case 162: | |
| 294 | return "CP1254"; | |
| 295 | case 163: | |
| 296 | return "CP1258"; | |
| 297 | case 181: | |
| 298 | case 177: | |
| 299 | return "CP1255"; | |
| 300 | case 178: | |
| 301 | case 179: | |
| 302 | case 180: | |
| 303 | return "CP1256"; | |
| 304 | case 186: | |
| 305 | return "CP1257"; | |
| 306 | case 204: | |
| 307 | return "CP1251"; | |
| 308 | case 222: | |
| 309 | return "CP874"; | |
| 310 | case 238: | |
| 311 | return "CP1250"; | |
| 312 | case 254: | |
| 313 | return "CP437"; | |
| 314 | default: | |
| 315 | gaim_debug_info("novell", "Unhandled font charset %d\n", font->charset); | |
| 316 | return "CP1252"; | |
| 317 | } | |
| 318 | return "CP1252"; | |
| 319 | } | |
| 320 | ||
| 321 | ||
| 322 | /* | |
| 323 | * Add an entry to the font table | |
| 324 | */ | |
| 325 | static int | |
| 326 | rtf_add_font_entry(NMRtfContext *ctx, int number, const char *name, int charset) | |
| 327 | { | |
| 328 | NMRtfFont *font = g_new0(NMRtfFont, 1); | |
| 329 | ||
| 330 | font->number = number; | |
| 331 | font->name = g_strdup(name); | |
| 332 | font->charset = charset; | |
| 333 | ||
| 334 | gaim_debug_info("novell", "Adding font to table: #%d\t%s\t%d\n", | |
| 335 | font->number, font->name, font->charset); | |
| 336 | ||
| 337 | ctx->font_table = g_slist_append(ctx->font_table, font); | |
| 338 | ||
| 339 | return NMRTF_OK; | |
| 340 | } | |
| 341 | ||
| 342 | /* | |
| 343 | * Return the nth entry in the font table | |
| 344 | */ | |
| 345 | static NMRtfFont * | |
| 346 | rtf_get_font(NMRtfContext *ctx, int nth) | |
| 347 | { | |
| 348 | NMRtfFont *font; | |
| 349 | ||
| 350 | font = g_slist_nth_data(ctx->font_table, nth); | |
| 351 | ||
| 352 | return font; | |
| 353 | } | |
| 354 | ||
| 355 | /* | |
| 356 | * Step 1: | |
| 357 | * Isolate RTF keywords and send them to rtf_parse_keyword; | |
| 358 | * Push and pop state at the start and end of RTF groups; | |
| 359 | * Send text to rtf_dispatch_char for further processing. | |
| 360 | */ | |
| 361 | static int | |
| 362 | rtf_parse(NMRtfContext *ctx) | |
| 363 | { | |
| 364 | int status; | |
| 365 | guchar ch; | |
| 366 | guchar hex_byte = 0; | |
| 367 | int hex_count = 2; | |
| 368 | int len; | |
| 369 | ||
| 370 | if (ctx->input == NULL) | |
| 371 | return NMRTF_OK; | |
| 372 | ||
| 373 | while (rtf_get_char(ctx, &ch) == NMRTF_OK) { | |
| 374 | if (ctx->depth < 0) | |
| 375 | return NMRTF_STACK_UNDERFLOW; | |
| 376 | ||
| 377 | /* if we're parsing binary data, handle it directly */ | |
| 378 | if (ctx->ris == NMRTF_STATE_BIN) { | |
| 379 | if ((status = rtf_dispatch_char(ctx, ch)) != NMRTF_OK) | |
| 380 | return status; | |
| 381 | } else { | |
| 382 | switch (ch) { | |
| 383 | case '{': | |
| 384 | if (ctx->depth > NMRTF_MAX_DEPTH) | |
| 385 | return NMRTF_STACK_OVERFLOW; | |
| 386 | rtf_flush_data(ctx); | |
| 387 | if ((status = rtf_push_state(ctx)) != NMRTF_OK) | |
| 388 | return status; | |
| 389 | break; | |
| 390 | case '}': | |
| 391 | rtf_flush_data(ctx); | |
| 392 | ||
| 393 | /* for some reason there is always an unwanted '\par' at the end */ | |
| 394 | if (ctx->rds == NMRTF_STATE_NORMAL) { | |
| 395 | len = ctx->output->len; | |
| 396 | if (ctx->output->str[len-1] == '\n') | |
| 397 | ctx->output = g_string_truncate(ctx->output, len-1); | |
| 398 | } | |
| 399 | ||
| 400 | if ((status = rtf_pop_state(ctx)) != NMRTF_OK) | |
| 401 | return status; | |
| 402 | ||
| 403 | if (ctx->depth < 0) | |
| 404 | return NMRTF_STACK_OVERFLOW; | |
| 405 | break; | |
| 406 | case '\\': | |
| 407 | if ((status = rtf_parse_keyword(ctx)) != NMRTF_OK) | |
| 408 | return status; | |
| 409 | break; | |
| 410 | case 0x0d: | |
| 411 | case 0x0a: /* cr and lf are noise characters... */ | |
| 412 | break; | |
| 413 | default: | |
| 414 | if (ctx->ris == NMRTF_STATE_NORMAL) { | |
| 415 | if ((status = rtf_dispatch_char(ctx, ch)) != NMRTF_OK) | |
| 416 | return status; | |
| 417 | } else { /* parsing a hex encoded character */ | |
| 418 | if (ctx->ris != NMRTF_STATE_HEX) | |
| 419 | return NMRTF_ASSERTION; | |
| 420 | ||
| 421 | hex_byte = hex_byte << 4; | |
| 422 | if (isdigit(ch)) | |
| 423 | hex_byte += (char) ch - '0'; | |
| 424 | else { | |
| 425 | if (islower(ch)) { | |
| 426 | if (ch < 'a' || ch > 'f') | |
| 427 | return NMRTF_INVALID_HEX; | |
| 428 | hex_byte += (char) ch - 'a' + 10; | |
| 429 | } else { | |
| 430 | if (ch < 'A' || ch > 'F') | |
| 431 | return NMRTF_INVALID_HEX; | |
| 432 | hex_byte += (char) ch - 'A' + 10; | |
| 433 | } | |
| 434 | } | |
| 435 | hex_count--; | |
| 436 | if (hex_count == 0) { | |
| 437 | if ((status = rtf_dispatch_char(ctx, hex_byte)) != NMRTF_OK) | |
| 438 | return status; | |
| 439 | hex_count = 2; | |
| 440 | hex_byte = 0; | |
| 441 | ctx->ris = NMRTF_STATE_NORMAL; | |
| 442 | } | |
| 443 | } | |
| 444 | break; | |
| 445 | } | |
| 446 | } | |
| 447 | } | |
| 448 | if (ctx->depth < 0) | |
| 449 | return NMRTF_STACK_OVERFLOW; | |
| 450 | if (ctx->depth > 0) | |
| 451 | return NMRTF_UNMATCHED_BRACE; | |
| 452 | return NMRTF_OK; | |
| 453 | } | |
| 454 | ||
| 455 | /* | |
| 456 | * Push the current state onto stack | |
| 457 | */ | |
| 458 | static int | |
| 459 | rtf_push_state(NMRtfContext *ctx) | |
| 460 | { | |
| 461 | NMRtfStateSave *save = g_new0(NMRtfStateSave, 1); | |
| 462 | save->chp = ctx->chp; | |
| 463 | save->rds = ctx->rds; | |
| 464 | save->ris = ctx->ris; | |
| 465 | ctx->saved = g_slist_prepend(ctx->saved, save); | |
| 466 | ctx->ris = NMRTF_STATE_NORMAL; | |
| 467 | (ctx->depth)++; | |
| 468 | return NMRTF_OK; | |
| 469 | } | |
| 470 | ||
| 471 | /* | |
| 472 | * Restore the state at the top of the stack | |
| 473 | */ | |
| 474 | static int | |
| 475 | rtf_pop_state(NMRtfContext *ctx) | |
| 476 | { | |
| 477 | NMRtfStateSave *save_old; | |
| 478 | GSList *link_old; | |
| 479 | ||
| 480 | if (ctx->saved == NULL) | |
| 481 | return NMRTF_STACK_UNDERFLOW; | |
| 482 | ||
| 483 | save_old = ctx->saved->data; | |
| 484 | ctx->chp = save_old->chp; | |
| 485 | ctx->rds = save_old->rds; | |
| 486 | ctx->ris = save_old->ris; | |
| 487 | (ctx->depth)--; | |
| 488 | ||
| 489 | g_free(save_old); | |
| 490 | link_old = ctx->saved; | |
| 491 | ctx->saved = g_slist_remove_link(ctx->saved, link_old); | |
| 492 | g_slist_free_1(link_old); | |
| 493 | return NMRTF_OK; | |
| 494 | } | |
| 495 | ||
| 496 | /* | |
| 497 | * Step 2: | |
| 498 | * Get a control word (and its associated value) and | |
| 499 | * dispatch the control. | |
| 500 | */ | |
| 501 | static int | |
| 502 | rtf_parse_keyword(NMRtfContext *ctx) | |
| 503 | { | |
| 504 | int status = NMRTF_OK; | |
| 505 | guchar ch; | |
| 506 | gboolean param_set = FALSE; | |
| 507 | gboolean is_neg = FALSE; | |
| 508 | int param = 0; | |
| 509 | char keyword[30]; | |
| 510 | char parameter[20]; | |
|
9804
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
511 | int i; |
| 9268 | 512 | |
| 513 | keyword[0] = '\0'; | |
| 514 | parameter[0] = '\0'; | |
| 515 | if ((status = rtf_get_char(ctx, &ch)) != NMRTF_OK) | |
| 516 | return status; | |
| 517 | ||
| 518 | if (!isalpha(ch)) { | |
| 519 | /* a control symbol; no delimiter. */ | |
| 520 | keyword[0] = (char) ch; | |
| 521 | keyword[1] = '\0'; | |
| 522 | return rtf_dispatch_control(ctx, keyword, 0, param_set); | |
| 523 | } | |
| 524 | ||
| 525 | /* parse keyword */ | |
|
9804
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
526 | for (i = 0; isalpha(ch) && (i < sizeof(keyword) - 1); rtf_get_char(ctx, &ch)) { |
|
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
527 | keyword[i] = (char) ch; |
|
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
528 | i++; |
| 9268 | 529 | } |
|
9804
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
530 | keyword[i] = '\0'; |
| 9268 | 531 | |
| 532 | /* check for '-' indicated a negative parameter value */ | |
| 533 | if (ch == '-') { | |
| 534 | is_neg = TRUE; | |
| 535 | if ((status = rtf_get_char(ctx, &ch)) != NMRTF_OK) | |
| 536 | return status; | |
| 537 | } | |
| 538 | ||
| 539 | /* check for numerical param */ | |
| 540 | if (isdigit(ch)) { | |
| 541 | ||
| 542 | param_set = TRUE; | |
|
9804
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
543 | for (i = 0; isdigit(ch) && (i < sizeof(parameter) - 1); rtf_get_char(ctx, &ch)) { |
|
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
544 | parameter[i] = (char) ch; |
|
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
545 | i++; |
| 9268 | 546 | } |
|
9804
1fbe747c77aa
[gaim-migrate @ 10672]
Mark Doliner <markdoliner@pidgin.im>
parents:
9268
diff
changeset
|
547 | parameter[i] = '\0'; |
| 9268 | 548 | |
| 549 | ctx->param = param = atoi(parameter); | |
| 550 | if (is_neg) | |
| 551 | ctx->param = param = -param; | |
| 552 | } | |
| 553 | ||
| 554 | /* space after control is optional, put character back if it is not a space */ | |
| 555 | if (ch != ' ') | |
| 556 | rtf_unget_char(ctx, ch); | |
| 557 | ||
| 558 | return rtf_dispatch_control(ctx, keyword, param, param_set); | |
| 559 | } | |
| 560 | ||
| 561 | /* | |
| 562 | * Route the character to the appropriate destination | |
| 563 | */ | |
| 564 | static int | |
| 565 | rtf_dispatch_char(NMRtfContext *ctx, guchar ch) | |
| 566 | { | |
| 567 | if (ctx->ris == NMRTF_STATE_BIN && --(ctx->bytes_to_skip) <= 0) | |
| 568 | ctx->ris = NMRTF_STATE_NORMAL; | |
| 569 | ||
| 570 | switch (ctx->rds) { | |
| 571 | case NMRTF_STATE_SKIP: | |
| 572 | return NMRTF_OK; | |
| 573 | case NMRTF_STATE_NORMAL: | |
| 574 | return rtf_print_char(ctx, ch); | |
| 575 | case NMRTF_STATE_FONTTABLE: | |
| 576 | if (ch == ';') { | |
| 577 | rtf_add_font_entry(ctx, ctx->chp.font_idx, | |
| 578 | ctx->ansi->str, ctx->chp.font_charset); | |
| 579 | g_string_truncate(ctx->ansi, 0); | |
| 580 | } | |
| 581 | else { | |
| 582 | return rtf_print_char(ctx, ch); | |
| 583 | } | |
| 584 | return NMRTF_OK; | |
| 585 | default: | |
| 586 | return NMRTF_OK; | |
| 587 | } | |
| 588 | } | |
| 589 | ||
| 590 | /* Handle a unicode character */ | |
| 591 | static int | |
| 592 | rtf_dispatch_unicode_char(NMRtfContext *ctx, gunichar ch) | |
| 593 | { | |
| 594 | switch (ctx->rds) { | |
| 595 | case NMRTF_STATE_SKIP: | |
| 596 | return NMRTF_OK; | |
| 597 | case NMRTF_STATE_NORMAL: | |
| 598 | case NMRTF_STATE_FONTTABLE: | |
| 599 | return rtf_print_unicode_char(ctx, ch); | |
| 600 | default: | |
| 601 | return NMRTF_OK; | |
| 602 | } | |
| 603 | } | |
| 604 | ||
| 605 | /* | |
| 606 | * Output a character | |
| 607 | */ | |
| 608 | static int | |
| 609 | rtf_print_char(NMRtfContext *ctx, guchar ch) | |
| 610 | { | |
| 611 | ||
| 612 | ctx->ansi = g_string_append_c(ctx->ansi, ch); | |
| 613 | ||
| 614 | return NMRTF_OK; | |
| 615 | } | |
| 616 | ||
| 617 | /* | |
| 618 | * Output a unicode character | |
| 619 | */ | |
| 620 | static int | |
| 621 | rtf_print_unicode_char(NMRtfContext *ctx, gunichar ch) | |
| 622 | { | |
| 623 | char buf[7]; | |
| 624 | int num; | |
| 625 | ||
| 626 | /* convert and flush the ansi buffer to the utf8 buffer */ | |
| 627 | rtf_flush_data(ctx); | |
| 628 | ||
| 629 | /* convert the unicode character to utf8 and add directly to the output buffer */ | |
| 630 | num = g_unichar_to_utf8((gunichar) ch, buf); | |
| 631 | buf[num] = 0; | |
| 632 | gaim_debug_info("novell", "converted unichar 0x%X to utf8 char %s\n", ch, buf); | |
| 633 | ||
| 634 | ctx->output = g_string_append(ctx->output, buf); | |
| 635 | return NMRTF_OK; | |
| 636 | } | |
| 637 | ||
| 638 | /* | |
| 639 | * Flush the output text | |
| 640 | */ | |
| 641 | static int | |
| 642 | rtf_flush_data(NMRtfContext *ctx) | |
| 643 | { | |
| 644 | int status = NMRTF_OK; | |
| 645 | char *conv_data = NULL; | |
| 646 | const char *enc = NULL; | |
| 647 | GError *gerror = NULL; | |
| 648 | ||
| 649 | if (ctx->rds == NMRTF_STATE_NORMAL && ctx->ansi->len > 0) { | |
| 650 | enc = get_current_encoding(ctx); | |
| 651 | conv_data = g_convert(ctx->ansi->str, ctx->ansi->len, "UTF-8", enc, | |
| 652 | NULL, NULL, &gerror); | |
| 653 | if (conv_data) { | |
| 654 | ctx->output = g_string_append(ctx->output, conv_data); | |
| 655 | g_free(conv_data); | |
| 656 | ctx->ansi = g_string_truncate(ctx->ansi, 0); | |
| 657 | } else { | |
| 658 | status = NMRTF_CONVERT_ERROR; | |
| 659 | gaim_debug_info("novell", "failed to convert data! error code = %d msg = %s\n", | |
| 660 | gerror->code, gerror->message); | |
| 661 | g_free(gerror); | |
| 662 | } | |
| 663 | } | |
| 664 | ||
| 665 | return status; | |
| 666 | } | |
| 667 | ||
| 668 | /* | |
| 669 | * Handle a property change | |
| 670 | */ | |
| 671 | static int | |
| 672 | rtf_apply_property(NMRtfContext *ctx, NMRtfProperty prop, int val) | |
| 673 | { | |
| 674 | if (ctx->rds == NMRTF_STATE_SKIP) /* If we're skipping text, */ | |
| 675 | return NMRTF_OK; /* don't do anything. */ | |
| 676 | ||
| 677 | /* Need to flush any temporary data before a property change*/ | |
| 678 | rtf_flush_data(ctx); | |
| 679 | ||
| 680 | switch (prop) { | |
| 681 | case NMRTF_PROP_FONT_IDX: | |
| 682 | ctx->chp.font_idx = val; | |
| 683 | break; | |
| 684 | case NMRTF_PROP_FONT_CHARSET: | |
| 685 | ctx->chp.font_charset = val; | |
| 686 | break; | |
| 687 | default: | |
| 688 | return NMRTF_BAD_TABLE; | |
| 689 | } | |
| 690 | ||
| 691 | return NMRTF_OK; | |
| 692 | } | |
| 693 | ||
| 694 | /* | |
| 695 | * Step 3. | |
| 696 | * Search the table for keyword and evaluate it appropriately. | |
| 697 | * | |
| 698 | * Inputs: | |
| 699 | * keyword: The RTF control to evaluate. | |
| 700 | * param: The parameter of the RTF control. | |
| 701 | * param_set: TRUE if the control had a parameter; (that is, if param is valid) | |
| 702 | * FALSE if it did not. | |
| 703 | */ | |
| 704 | static int | |
| 705 | rtf_dispatch_control(NMRtfContext *ctx, char *keyword, int param, gboolean param_set) | |
| 706 | { | |
| 707 | int idx; | |
| 708 | ||
| 709 | for (idx = 0; idx < table_size; idx++) { | |
| 710 | if (strcmp(keyword, rtf_symbols[idx].keyword) == 0) | |
| 711 | break; | |
| 712 | } | |
| 713 | ||
| 714 | if (idx == table_size) { | |
| 715 | if (ctx->skip_unknown) | |
| 716 | ctx->rds = NMRTF_STATE_SKIP; | |
| 717 | ctx->skip_unknown = FALSE; | |
| 718 | return NMRTF_OK; | |
| 719 | } | |
| 720 | ||
| 721 | /* found it! use kwd_type and action to determine what to do with it. */ | |
| 722 | ctx->skip_unknown = FALSE; | |
| 723 | switch (rtf_symbols[idx].kwd_type) { | |
| 724 | case NMRTF_KWD_PROP: | |
| 725 | if (rtf_symbols[idx].pass_default || !param_set) | |
| 726 | param = rtf_symbols[idx].default_val; | |
| 727 | return rtf_apply_property(ctx, rtf_symbols[idx].action, param); | |
| 728 | case NMRTF_KWD_CHAR: | |
| 729 | return rtf_dispatch_char(ctx, rtf_symbols[idx].action); | |
| 730 | case NMRTF_KWD_DEST: | |
| 731 | return rtf_change_destination(ctx, rtf_symbols[idx].action); | |
| 732 | case NMRTF_KWD_SPEC: | |
| 733 | return rtf_dispatch_special(ctx, rtf_symbols[idx].action); | |
| 734 | default: | |
| 735 | return NMRTF_BAD_TABLE; | |
| 736 | } | |
| 737 | return NMRTF_BAD_TABLE; | |
| 738 | } | |
| 739 | ||
| 740 | /* | |
| 741 | * Change to the destination specified. | |
| 742 | */ | |
| 743 | static int | |
| 744 | rtf_change_destination(NMRtfContext *ctx, NMRtfDestinationType type) | |
| 745 | { | |
| 746 | /* if we're skipping text, don't do anything */ | |
| 747 | if (ctx->rds == NMRTF_STATE_SKIP) | |
| 748 | return NMRTF_OK; | |
| 749 | ||
| 750 | switch (type) { | |
| 751 | case NMRTF_DEST_FONTTABLE: | |
| 752 | ctx->rds = NMRTF_STATE_FONTTABLE; | |
| 753 | g_string_truncate(ctx->ansi, 0); | |
| 754 | break; | |
| 755 | default: | |
| 756 | ctx->rds = NMRTF_STATE_SKIP; /* when in doubt, skip it... */ | |
| 757 | break; | |
| 758 | } | |
| 759 | return NMRTF_OK; | |
| 760 | } | |
| 761 | ||
| 762 | /* | |
| 763 | * Dispatch an RTF control that needs special processing | |
| 764 | */ | |
| 765 | static int | |
| 766 | rtf_dispatch_special(NMRtfContext *ctx, NMRtfSpecialKwd type) | |
| 767 | { | |
| 768 | int status = NMRTF_OK; | |
| 769 | guchar ch; | |
| 770 | ||
| 771 | if (ctx->rds == NMRTF_STATE_SKIP && type != NMRTF_SPECIAL_BIN) /* if we're skipping, and it's not */ | |
| 772 | return NMRTF_OK; /* the \bin keyword, ignore it. */ | |
| 773 | ||
| 774 | switch (type) { | |
| 775 | case NMRTF_SPECIAL_BIN: | |
| 776 | ctx->ris = NMRTF_STATE_BIN; | |
| 777 | ctx->bytes_to_skip = ctx->param; | |
| 778 | break; | |
| 779 | case NMRTF_SPECIAL_SKIP: | |
| 780 | ctx->skip_unknown = TRUE; | |
| 781 | break; | |
| 782 | case NMRTF_SPECIAL_HEX: | |
| 783 | ctx->ris = NMRTF_STATE_HEX; | |
| 784 | break; | |
| 785 | case NMRTF_SPECIAL_UNICODE: | |
| 786 | gaim_debug_info("novell", "parsing unichar\n"); | |
| 787 | status = rtf_dispatch_unicode_char(ctx, ctx->param); | |
| 788 | /* Skip next char */ | |
| 789 | if (status == NMRTF_OK) | |
| 790 | status = rtf_get_char(ctx, &ch); | |
| 791 | break; | |
| 792 | default: | |
| 793 | status = NMRTF_BAD_TABLE; | |
| 794 | break; | |
| 795 | } | |
| 796 | ||
| 797 | return status; | |
| 798 | } | |
| 799 | ||
| 800 | /* | |
| 801 | * Get the next character from the input stream | |
| 802 | */ | |
| 803 | static int | |
| 804 | rtf_get_char(NMRtfContext *ctx, guchar *ch) | |
| 805 | { | |
| 806 | if (ctx->nextch >= 0) { | |
| 807 | *ch = ctx->nextch; | |
| 808 | ctx->nextch = -1; | |
| 809 | } | |
| 810 | else { | |
| 811 | *ch = *(ctx->input); | |
| 812 | ctx->input++; | |
| 813 | } | |
| 814 | ||
| 815 | if (*ch) | |
| 816 | return NMRTF_OK; | |
| 817 | else | |
| 818 | return NMRTF_EOF; | |
| 819 | } | |
| 820 | ||
| 821 | /* | |
| 822 | * Move a character back into the input stream | |
| 823 | */ | |
| 824 | static int | |
| 825 | rtf_unget_char(NMRtfContext *ctx, guchar ch) | |
| 826 | { | |
| 827 | ctx->nextch = ch; | |
| 828 | return NMRTF_OK; | |
| 829 | } |