| 1 |
%top { |
| 2 |
/* This file is part of GNU Dico. |
| 3 |
Copyright (C) 2012-2024 Sergey Poznyakoff |
| 4 |
|
| 5 |
GNU Dico is free software; you can redistribute it and/or modify |
| 6 |
it under the terms of the GNU General Public License as published by |
| 7 |
the Free Software Foundation; either version 3, or (at your option) |
| 8 |
any later version. |
| 9 |
|
| 10 |
GNU Dico 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 GNU Dico. If not, see <http://www.gnu.org/licenses/>. */ |
| 17 |
|
| 18 |
#include <config.h> |
| 19 |
#include <dico.h> |
| 20 |
#include <unistd.h> |
| 21 |
#include <getopt.h> |
| 22 |
#include <stdio.h> |
| 23 |
#include <stdlib.h> |
| 24 |
#include <errno.h> |
| 25 |
#include <sysexits.h> |
| 26 |
#include <setjmp.h> |
| 27 |
#include <ctype.h> |
| 28 |
#include <appi18n.h> |
| 29 |
#include "gcide.h" |
| 30 |
#include "wordsplit.h" |
| 31 |
|
| 32 |
#define yy_create_buffer gcide_markup_yy_create_buffer |
| 33 |
#define yy_delete_buffer gcide_markup_yy_delete_buffer |
| 34 |
#define yy_flex_debug gcide_markup_yy_flex_debug |
| 35 |
#define yy_init_buffer gcide_markup_yy_init_buffer |
| 36 |
#define yy_flush_buffer gcide_markup_yy_flush_buffer |
| 37 |
#define yy_load_buffer_state gcide_markup_yy_load_buffer_state |
| 38 |
#define yy_switch_to_buffer gcide_markup_yy_switch_to_buffer |
| 39 |
#define yyin gcide_markup_yyin |
| 40 |
#define yyleng gcide_markup_yyleng |
| 41 |
#define yylex gcide_markup_yylex |
| 42 |
#define yylineno gcide_markup_yylineno |
| 43 |
#define yyout gcide_markup_yyout |
| 44 |
#define yyrestart gcide_markup_yyrestart |
| 45 |
#define yytext gcide_markup_yytext |
| 46 |
#define yywrap gcide_markup_yywrap |
| 47 |
#define yyalloc gcide_markup_yyalloc |
| 48 |
#define yyrealloc gcide_markup_yyrealloc |
| 49 |
#define yyfree gcide_markup_yyfree |
| 50 |
#define yyunput gcide_markup_yyunput |
| 51 |
|
| 52 |
#define yylex_destroy gcide_markup_yylex_destroy |
| 53 |
#define yyget_debug gcide_markup_yyget_debug |
| 54 |
#define yyset_debug gcide_markup_yyset_debug |
| 55 |
#define yyget_extra gcide_markup_yyget_extra |
| 56 |
#define yyset_extra gcide_markup_yyset_extra |
| 57 |
#define yyget_in gcide_markup_yyget_in |
| 58 |
#define yyset_in gcide_markup_yyset_in |
| 59 |
#define yyget_out gcide_markup_yyget_out |
| 60 |
#define yyset_out gcide_markup_yyset_out |
| 61 |
#define yyget_leng gcide_markup_yyget_leng |
| 62 |
#define yyget_text gcide_markup_yyget_text |
| 63 |
#define yyget_lineno gcide_markup_yyget_lineno |
| 64 |
#define yyset_lineno gcide_markup_yyset_lineno |
| 65 |
|
| 66 |
} |
| 67 |
%{ |
| 68 |
|
| 69 |
static struct gcide_locus *base_locus; |
| 70 |
static char const *input_buf; |
| 71 |
static size_t input_len; |
| 72 |
static unsigned token_beg; |
| 73 |
static unsigned token_end; |
| 74 |
|
| 75 |
/* |
| 76 |
* Dictionary entries in GCIDE corpus files begin with "<p><ent>", but index |
| 77 |
* entries in the GCIDE.IDX file point to the first "<hw>" tag that follows. |
| 78 |
* Thus, the very first "</p>" encountered when parsing an entry has no |
| 79 |
* corresponding "<p>" tag. The first_p_seen variable is set by push_tag |
| 80 |
* when it sees the first opening "<p>" tag and is used in pop_tag to generate |
| 81 |
* the first "p" block tag and avoid issuing a spurious error message when |
| 82 |
* the first "</p>" is encountered. |
| 83 |
*/ |
| 84 |
static int first_p_seen; |
| 85 |
|
| 86 |
#define LOCUS_FILE (base_locus ? base_locus->file : "") |
| 87 |
#define LOCUS_OFF(off) ((base_locus ? base_locus->offset : 0) + off) |
| 88 |
|
| 89 |
static char *textspace; /* Text storage space */ |
| 90 |
static size_t textsize; /* Size of text space */ |
| 91 |
static size_t textpos; /* Current position in the text space */ |
| 92 |
static size_t textstart; /* Start of the current text segment */ |
| 93 |
static size_t textstart_offset; |
| 94 |
|
| 95 |
static dico_list_t tagstk; |
| 96 |
static struct gcide_tag *current_tag; |
| 97 |
|
| 98 |
static jmp_buf errbuf; |
| 99 |
|
| 100 |
static void |
| 101 |
memerr(const char *text) |
| 102 |
{ |
| 103 |
dico_log(L_ERR, ENOMEM, "%s", text); |
| 104 |
longjmp(errbuf, 1); |
| 105 |
} |
| 106 |
|
| 107 |
#define YY_USER_ACTION do { \ |
| 108 |
token_beg = token_end; \ |
| 109 |
token_end += yyleng; \ |
| 110 |
} while (0); |
| 111 |
#define YY_INPUT(buf,result,max_size) do { \ |
| 112 |
size_t __n = (max_size) > input_len ? input_len : (max_size); \ |
| 113 |
if (__n) \ |
| 114 |
memcpy((buf), input_buf, __n); \ |
| 115 |
input_buf += __n; \ |
| 116 |
input_len -= __n; \ |
| 117 |
(result) = __n; \ |
| 118 |
} while(0) |
| 119 |
|
| 120 |
static int retstate; |
| 121 |
static char *endtag; |
| 122 |
static size_t endtaglen; |
| 123 |
#define BEGIN_COMMENT(end) \ |
| 124 |
{ retstate = YYSTATE; endtag = (end); endtaglen = strlen(endtag); BEGIN(COMMENT); } |
| 125 |
|
| 126 |
static void |
| 127 |
text_assert_size(size_t len) |
| 128 |
{ |
| 129 |
while (textsize - textpos < len) { |
| 130 |
size_t nsize; |
| 131 |
char *newp; |
| 132 |
|
| 133 |
if (SIZE_MAX / 2 < textsize) |
| 134 |
memerr("text_add_str"); |
| 135 |
nsize = 2 * textsize; |
| 136 |
newp = realloc(textspace, nsize); |
| 137 |
if (!newp) |
| 138 |
memerr("text_add_str"); |
| 139 |
textspace = newp; |
| 140 |
textsize = nsize; |
| 141 |
} |
| 142 |
} |
| 143 |
|
| 144 |
static void |
| 145 |
text_add_str(char const *s, size_t l) |
| 146 |
{ |
| 147 |
if (textstart == textpos) textstart_offset = token_beg; |
| 148 |
text_assert_size(l); |
| 149 |
memcpy(textspace + textpos, s, l); |
| 150 |
textpos += l; |
| 151 |
} |
| 152 |
|
| 153 |
static void |
| 154 |
text_add_strz(char const *s) |
| 155 |
{ |
| 156 |
text_add_str(s, strlen(s)); |
| 157 |
} |
| 158 |
|
| 159 |
static void |
| 160 |
text_add_chr(int ch) |
| 161 |
{ |
| 162 |
char c = ch; |
| 163 |
text_add_str(&c, 1); |
| 164 |
} |
| 165 |
|
| 166 |
#define text_segment_length() (textpos - textstart) |
| 167 |
|
| 168 |
static size_t |
| 169 |
text_segment_finish() |
| 170 |
{ |
| 171 |
size_t ret = textstart; |
| 172 |
text_add_chr(0); |
| 173 |
textstart = textpos; |
| 174 |
return ret; |
| 175 |
} |
| 176 |
|
| 177 |
static int free_tag(void *item, void *data); |
| 178 |
|
| 179 |
static struct gcide_tag * |
| 180 |
alloc_tag(enum gcide_content_type type) |
| 181 |
{ |
| 182 |
dico_list_t list; |
| 183 |
struct gcide_tag *tag = calloc(1, sizeof(*tag)); |
| 184 |
if (!tag) |
| 185 |
memerr("alloc_tag"); |
| 186 |
tag->tag_type = type; |
| 187 |
switch (type) { |
| 188 |
case gcide_content_top: |
| 189 |
case gcide_content_tag: |
| 190 |
list = dico_list_create(); |
| 191 |
if (!list) { |
| 192 |
free(tag); |
| 193 |
memerr("alloc_tag"); |
| 194 |
} |
| 195 |
dico_list_set_free_item(list, free_tag, NULL); |
| 196 |
tag->v.tag.taglist = list; |
| 197 |
break; |
| 198 |
default: |
| 199 |
break; |
| 200 |
} |
| 201 |
return tag; |
| 202 |
} |
| 203 |
|
| 204 |
struct gcide_tag * |
| 205 |
gcide_tag_alloc(const char *text, size_t len) |
| 206 |
{ |
| 207 |
struct gcide_tag *tag; |
| 208 |
struct wordsplit ws; |
| 209 |
size_t i; |
| 210 |
|
| 211 |
if (wordsplit_len(text, len, &ws, WRDSF_DEFFLAGS & ~WRDSF_CESCAPES)) { |
| 212 |
dico_log(L_ERR, 0, _("cannot parse line %.*s: %s"), |
| 213 |
(int)len, text, wordsplit_strerror(&ws)); |
| 214 |
longjmp(errbuf, 1); |
| 215 |
} |
| 216 |
|
| 217 |
for (i = 0; i < ws.ws_wordc; i++) { |
| 218 |
char *p; |
| 219 |
for (p = ws.ws_wordv[i]; *p; p++) { |
| 220 |
if (*p == '=') |
| 221 |
break; |
| 222 |
*p = tolower(*p); |
| 223 |
} |
| 224 |
} |
| 225 |
tag = alloc_tag(gcide_content_tag); |
| 226 |
tag->v.tag.tag_parmc = ws.ws_wordc; |
| 227 |
tag->v.tag.tag_parmv = ws.ws_wordv; |
| 228 |
tag->offset = token_beg; |
| 229 |
ws.ws_wordc = 0; |
| 230 |
ws.ws_wordv = NULL; |
| 231 |
wordsplit_free(&ws); |
| 232 |
return tag; |
| 233 |
} |
| 234 |
|
| 235 |
#define alloc_taglist gcide_tag_alloc |
| 236 |
|
| 237 |
void |
| 238 |
gcide_tag_free(struct gcide_tag *tag) |
| 239 |
{ |
| 240 |
if (tag) { |
| 241 |
switch (tag->tag_type) { |
| 242 |
default: |
| 243 |
break; |
| 244 |
case gcide_content_top: |
| 245 |
case gcide_content_tag: |
| 246 |
dico_list_destroy(&tag->v.tag.taglist); |
| 247 |
} |
| 248 |
free(tag); |
| 249 |
} |
| 250 |
} |
| 251 |
|
| 252 |
int |
| 253 |
gcide_is_tag(struct gcide_tag *tag, char const *name) |
| 254 |
{ |
| 255 |
return tag != NULL && tag->tag_type == gcide_content_tag && |
| 256 |
strcmp(tag->tag_name, name) == 0; |
| 257 |
} |
| 258 |
|
| 259 |
static int |
| 260 |
free_tag(void *item, void *data) |
| 261 |
{ |
| 262 |
gcide_tag_free(item); |
| 263 |
return 0; |
| 264 |
} |
| 265 |
|
| 266 |
static void |
| 267 |
append_tag(struct gcide_tag *tag) |
| 268 |
{ |
| 269 |
dico_list_t list; |
| 270 |
|
| 271 |
switch (current_tag->tag_type) { |
| 272 |
case gcide_content_text: { |
| 273 |
struct gcide_tag *subtag = alloc_tag(gcide_content_text); |
| 274 |
subtag->offset = current_tag->offset; |
| 275 |
subtag->v.textpos = current_tag->v.textpos; |
| 276 |
list = dico_list_create(); |
| 277 |
if (!list) { |
| 278 |
free(subtag); |
| 279 |
free(tag); |
| 280 |
memerr("append_tag"); |
| 281 |
} |
| 282 |
dico_list_set_free_item(list, free_tag, NULL); |
| 283 |
dico_list_append(list, subtag); |
| 284 |
current_tag->tag_type = gcide_content_tag; |
| 285 |
current_tag->v.tag.taglist = list; |
| 286 |
break; |
| 287 |
} |
| 288 |
|
| 289 |
default: |
| 290 |
break; |
| 291 |
} |
| 292 |
|
| 293 |
dico_list_append(current_tag->v.tag.taglist, tag); |
| 294 |
} |
| 295 |
|
| 296 |
static void push_tag(struct gcide_tag *tag); |
| 297 |
static void pop_tag(const char *tagstr, size_t taglen); |
| 298 |
|
| 299 |
static void |
| 300 |
flush_text_segment(void) |
| 301 |
{ |
| 302 |
if (text_segment_length()) { |
| 303 |
struct gcide_tag *tag = alloc_tag(gcide_content_text); |
| 304 |
tag->v.textpos = text_segment_finish(); |
| 305 |
tag->offset = textstart_offset; |
| 306 |
append_tag(tag); |
| 307 |
} |
| 308 |
} |
| 309 |
%} |
| 310 |
|
| 311 |
%option 8bit |
| 312 |
%option nounput |
| 313 |
%option noinput |
| 314 |
%option nodefault |
| 315 |
|
| 316 |
%x COMMENT PR HW PRE |
| 317 |
|
| 318 |
XD [0-9a-f] |
| 319 |
WS [ \t] |
| 320 |
%% |
| 321 |
<INITIAL,PR,HW,PRE>{ |
| 322 |
"<--" BEGIN_COMMENT("-->"); |
| 323 |
"<!" BEGIN_COMMENT("!>"); |
| 324 |
"<"[a-zA-Z][a-zA-Z0-9._-]*">" { |
| 325 |
flush_text_segment(); |
| 326 |
push_tag(alloc_taglist(yytext + 1, yyleng - 2)); |
| 327 |
} |
| 328 |
"<"[a-zA-Z][a-zA-Z0-9._-]*{WS}[^>]*">" { |
| 329 |
flush_text_segment(); |
| 330 |
push_tag(alloc_taglist(yytext + 1, yyleng - 2)); |
| 331 |
} |
| 332 |
"</"[a-zA-Z][a-zA-Z0-9._-]*">" { |
| 333 |
flush_text_segment(); |
| 334 |
pop_tag(yytext + 2, yyleng - 3); |
| 335 |
} |
| 336 |
"</"[a-zA-Z][^>]*"/" { |
| 337 |
dico_log(L_WARN, 0, _("%s:%zu: malformed entity or tag: %s"), |
| 338 |
LOCUS_FILE, LOCUS_OFF(token_beg), yytext); |
| 339 |
} |
| 340 |
"<br/" { |
| 341 |
flush_text_segment(); |
| 342 |
append_tag(alloc_tag(gcide_content_br)); |
| 343 |
} |
| 344 |
"<"[a-zA-Z?][a-zA-Z0-9]*"/" { |
| 345 |
char const *s = gcide_entity_to_utf8(yytext); |
| 346 |
if (s) |
| 347 |
text_add_str(s, strlen(s)); |
| 348 |
else |
| 349 |
dico_log(L_WARN, 0, _("%s:%zu: unrecognized entity: %s"), |
| 350 |
LOCUS_FILE, LOCUS_OFF(token_beg), yytext); |
| 351 |
} |
| 352 |
"\\'"{XD}{XD} { |
| 353 |
char const *s = gcide_escape_to_utf8(yytext+2); |
| 354 |
|
| 355 |
if (s) |
| 356 |
text_add_str(s, strlen(s)); |
| 357 |
else { |
| 358 |
text_add_str(yytext, yyleng); |
| 359 |
dico_log(L_WARN, 0, |
| 360 |
_("%s:%zu: unknown character sequence %s"), |
| 361 |
LOCUS_FILE, LOCUS_OFF(token_beg), yytext); |
| 362 |
} |
| 363 |
} |
| 364 |
[<\\](.|\n) { |
| 365 |
text_add_chr(yytext[0]); |
| 366 |
yyless(1); |
| 367 |
token_end--; |
| 368 |
} |
| 369 |
\r ; |
| 370 |
} |
| 371 |
<INITIAL,PR,HW>{ |
| 372 |
\n+ { |
| 373 |
flush_text_segment(); |
| 374 |
append_tag(alloc_tag(gcide_content_nl)); |
| 375 |
} |
| 376 |
} |
| 377 |
<INITIAL>{ |
| 378 |
[^<\\\r\n]+ text_add_str(yytext, yyleng); |
| 379 |
} |
| 380 |
<PRE>{ |
| 381 |
[^<\\\r]+ text_add_str(yytext, yyleng); |
| 382 |
} |
| 383 |
<HW>{ |
| 384 |
[""`*] ; |
| 385 |
[^<\\\r""`*]+ text_add_str(yytext, yyleng); |
| 386 |
} |
| 387 |
<PR>{ |
| 388 |
"\"" text_add_strz("ˈ"); /* Main accent */ |
| 389 |
"`" text_add_strz("ˌ"); /* Secondary accent */ |
| 390 |
"*" text_add_strz("‑"); /* Syllable break */ |
| 391 |
[^<\\\r""`*]+ text_add_str(yytext, yyleng); |
| 392 |
} |
| 393 |
<COMMENT>{ |
| 394 |
[^>]+">" { |
| 395 |
if (yyleng >= endtaglen && |
| 396 |
memcmp(yytext + yyleng - endtaglen, endtag, endtaglen) == 0) |
| 397 |
BEGIN(retstate); |
| 398 |
} |
| 399 |
.|\n ; |
| 400 |
} |
| 401 |
|
| 402 |
<INITIAL,PR,HW,PRE>. { |
| 403 |
if (input_len > 0) |
| 404 |
dico_log(L_NOTICE, 0, |
| 405 |
_("%s:%zu: unmatched character %03o; please report"), |
| 406 |
LOCUS_FILE, LOCUS_OFF(token_beg), yytext[0]); |
| 407 |
text_add_chr(yytext[0]); |
| 408 |
} |
| 409 |
%% |
| 410 |
|
| 411 |
int |
| 412 |
yywrap() |
| 413 |
{ |
| 414 |
return 1; |
| 415 |
} |
| 416 |
|
| 417 |
static void |
| 418 |
push_tag(struct gcide_tag *tag) |
| 419 |
{ |
| 420 |
append_tag(tag); |
| 421 |
dico_list_push(tagstk, current_tag); |
| 422 |
current_tag = tag; |
| 423 |
if (strcmp(current_tag->tag_name, "pr") == 0) |
| 424 |
BEGIN(PR); |
| 425 |
else if (strcmp(current_tag->tag_name, "hw") == 0) |
| 426 |
BEGIN(HW); |
| 427 |
else if (strcmp(current_tag->tag_name, "pre") == 0) |
| 428 |
BEGIN(PRE); |
| 429 |
else if (strcmp(current_tag->tag_name, "p") == 0) |
| 430 |
first_p_seen = 1; |
| 431 |
} |
| 432 |
|
| 433 |
static void |
| 434 |
pop_tag(const char *tagstr, size_t taglen) |
| 435 |
{ |
| 436 |
int len; |
| 437 |
char *p; |
| 438 |
|
| 439 |
text_assert_size(taglen); |
| 440 |
p = textspace + textpos; |
| 441 |
for (len = 0; len < taglen; len++) { |
| 442 |
if (tagstr[len] == ' ' || tagstr[len] == '\t') |
| 443 |
break; |
| 444 |
*p++ = tolower(tagstr[len]); |
| 445 |
} |
| 446 |
tagstr = textspace + textpos; |
| 447 |
|
| 448 |
if ((len == 2 && (memcmp(tagstr, "pr", len) == 0 || |
| 449 |
memcmp(tagstr, "hw", len) == 0)) || |
| 450 |
(len == 3 && memcmp(tagstr, "pre", len) == 0)) |
| 451 |
BEGIN(INITIAL); |
| 452 |
|
| 453 |
if (current_tag->tag_type == gcide_content_tag && |
| 454 |
strlen(current_tag->tag_name) == len && |
| 455 |
memcmp(current_tag->tag_name, tagstr, len) == 0) |
| 456 |
current_tag = dico_list_pop(tagstk); |
| 457 |
else if (current_tag->tag_type == gcide_content_top && |
| 458 |
len == 1 && |
| 459 |
tagstr[0] == 'p' && |
| 460 |
!first_p_seen) { |
| 461 |
/* |
| 462 |
* The first instance of </p> has no corresponding open <p> tag |
| 463 |
* because idxgcide stores the offset of the first <hw> tag in the |
| 464 |
* index entry, whereas the opening <p> tag in GCIDE entries appears |
| 465 |
* before the <ent> tag (which precedes the <hw>). |
| 466 |
* |
| 467 |
* Re-create the first <p>...</p> block from the data collected so |
| 468 |
* far and push it as the first element to the newly created top- |
| 469 |
* level tag list. Set first_p_seen flag to make sure this code is |
| 470 |
* executed once. |
| 471 |
* |
| 472 |
* See also the comment to first_p_seen, above. |
| 473 |
*/ |
| 474 |
struct gcide_tag *tag = current_tag; |
| 475 |
tag->tag_type = gcide_content_tag; |
| 476 |
tag->v.tag.tag_parmc = 1; |
| 477 |
tag->v.tag.tag_parmv = calloc(2, sizeof(tag->v.tag.tag_parmv[0])); |
| 478 |
if (!tag->v.tag.tag_parmv) |
| 479 |
memerr("pop_tag"); |
| 480 |
tag->v.tag.tag_parmv[0] = strdup("p"); |
| 481 |
if (!tag->v.tag.tag_parmv[0]) |
| 482 |
memerr("pop_tag"); |
| 483 |
tag->v.tag.tag_parmv[1] = NULL; |
| 484 |
current_tag = alloc_tag(gcide_content_top); |
| 485 |
append_tag(tag); |
| 486 |
first_p_seen = 1; |
| 487 |
} else { |
| 488 |
size_t i, count; |
| 489 |
|
| 490 |
/* Try to find matching entry */ |
| 491 |
count = dico_list_count(tagstk); |
| 492 |
for (i = count; i > 0; i--) { |
| 493 |
struct gcide_tag *p = dico_list_item(tagstk, i-1); |
| 494 |
if (p->tag_type == gcide_content_tag && |
| 495 |
strlen(p->tag_name) == len && |
| 496 |
memcmp(p->tag_name, tagstr, len) == 0) { |
| 497 |
do { |
| 498 |
if (current_tag->tag_type == gcide_content_tag) |
| 499 |
dico_log(L_WARN, 0, "%s:%zu: %s: unclosed tag", |
| 500 |
LOCUS_FILE, LOCUS_OFF(current_tag->offset), |
| 501 |
current_tag->tag_name); |
| 502 |
} while ((current_tag = dico_list_pop(tagstk)) != p); |
| 503 |
current_tag = dico_list_pop(tagstk); |
| 504 |
return; |
| 505 |
} |
| 506 |
} |
| 507 |
dico_log(L_WARN, 0, "%s:%zu: %*.*s: unexpected closing tag", |
| 508 |
LOCUS_FILE, LOCUS_OFF(token_beg), len, len, tagstr); |
| 509 |
} |
| 510 |
} |
| 511 |
|
| 512 |
struct walk_closure { |
| 513 |
int (*fun)(int, struct gcide_tag *, void *); |
| 514 |
void *data; |
| 515 |
}; |
| 516 |
|
| 517 |
static int |
| 518 |
inorder_helper(void *item, void *data) |
| 519 |
{ |
| 520 |
struct gcide_tag *tag = item; |
| 521 |
struct walk_closure *cp = data; |
| 522 |
|
| 523 |
if (cp->fun(0, tag, cp->data)) |
| 524 |
return 1; |
| 525 |
if (tag->tag_type == gcide_content_top || |
| 526 |
tag->tag_type == gcide_content_tag) { |
| 527 |
dico_list_iterate(tag->v.tag.taglist, inorder_helper, data); |
| 528 |
if (cp->fun(1, tag, cp->data)) |
| 529 |
return 1; |
| 530 |
} |
| 531 |
return 0; |
| 532 |
} |
| 533 |
|
| 534 |
int |
| 535 |
gcide_parse_tree_inorder(struct gcide_parse_tree *tp, |
| 536 |
int (*fun)(int, struct gcide_tag *, void *), |
| 537 |
void *data) |
| 538 |
{ |
| 539 |
struct walk_closure clos; |
| 540 |
clos.fun = fun; |
| 541 |
clos.data = data; |
| 542 |
return inorder_helper(tp->root, &clos); |
| 543 |
} |
| 544 |
|
| 545 |
static int |
| 546 |
tag_fixup(void *item, void *data) |
| 547 |
{ |
| 548 |
struct gcide_tag *tag = item; |
| 549 |
char *textspace = data; |
| 550 |
|
| 551 |
switch (tag->tag_type) { |
| 552 |
case gcide_content_text: |
| 553 |
tag->v.text = textspace + tag->v.textpos; |
| 554 |
break; |
| 555 |
case gcide_content_top: |
| 556 |
case gcide_content_tag: |
| 557 |
dico_list_iterate(tag->v.tag.taglist, tag_fixup, textspace); |
| 558 |
break; |
| 559 |
default: |
| 560 |
break; |
| 561 |
} |
| 562 |
return 0; |
| 563 |
} |
| 564 |
|
| 565 |
static char const *block_tags[] = { |
| 566 |
"source", |
| 567 |
"sn", |
| 568 |
"p", |
| 569 |
"q", |
| 570 |
"rj", |
| 571 |
"gcide_quote", |
| 572 |
NULL |
| 573 |
}; |
| 574 |
|
| 575 |
int |
| 576 |
gcide_is_block_tag(struct gcide_tag *tag) |
| 577 |
{ |
| 578 |
if (tag && tag->tag_type == gcide_content_tag) { |
| 579 |
char const **p; |
| 580 |
|
| 581 |
for (p = block_tags; *p; p++) |
| 582 |
if (strcmp(tag->tag_name, *p) == 0) |
| 583 |
return 1; |
| 584 |
} |
| 585 |
return 0; |
| 586 |
} |
| 587 |
|
| 588 |
int |
| 589 |
gcide_is_ws_text(struct gcide_tag *tag) |
| 590 |
{ |
| 591 |
if (tag && tag->tag_type == gcide_content_text) { |
| 592 |
char *p; |
| 593 |
for (p = tag->v.text; *p == '\n' || *p == ' ' || *p == '\t'; p++); |
| 594 |
return *p == 0; |
| 595 |
} |
| 596 |
return 0; |
| 597 |
} |
| 598 |
|
| 599 |
/* |
| 600 |
* Fix up the nl/br usage. See the comment to ws_fixup, below. |
| 601 |
*/ |
| 602 |
static int |
| 603 |
ws_reduce(void *a, void *b, void *data) |
| 604 |
{ |
| 605 |
struct gcide_tag *atag = a; |
| 606 |
struct gcide_tag *btag = b; |
| 607 |
|
| 608 |
if (atag->delete) |
| 609 |
return 1; |
| 610 |
|
| 611 |
switch (atag->tag_type) { |
| 612 |
case gcide_content_tag: |
| 613 |
if (strcmp(atag->tag_name, "source") == 0 && |
| 614 |
btag && |
| 615 |
btag->tag_type == gcide_content_text && |
| 616 |
btag->v.text[0] == ']') |
| 617 |
btag->v.text++; |
| 618 |
dico_list_reduce(atag->v.tag.taglist, ws_reduce, NULL); |
| 619 |
if (gcide_is_block_tag(atag) && gcide_is_ws_text(btag)) |
| 620 |
btag->delete = 1; |
| 621 |
break; |
| 622 |
|
| 623 |
case gcide_content_text: |
| 624 |
if (atag->v.text[0] == 0) |
| 625 |
return 1; |
| 626 |
else if (gcide_is_ws_text(atag) && gcide_is_block_tag(btag)) |
| 627 |
return 1; |
| 628 |
else { |
| 629 |
size_t len = strlen(atag->v.text); |
| 630 |
if (atag->v.text[len-1] == '[' && |
| 631 |
btag && |
| 632 |
btag->tag_type == gcide_content_tag && |
| 633 |
strcmp(btag->tag_name, "source") == 0) { |
| 634 |
if (len == 1) |
| 635 |
return 1; |
| 636 |
atag->v.text[len-1] = 0; |
| 637 |
} |
| 638 |
} |
| 639 |
break; |
| 640 |
|
| 641 |
case gcide_content_nl: |
| 642 |
if (btag == NULL) |
| 643 |
return 1; |
| 644 |
switch (btag->tag_type) { |
| 645 |
case gcide_content_tag: |
| 646 |
return gcide_is_block_tag(btag); |
| 647 |
|
| 648 |
case gcide_content_text: |
| 649 |
if (btag->v.text[0] == ' ' || btag->v.text[0] == '\t') |
| 650 |
return 1; |
| 651 |
break; |
| 652 |
|
| 653 |
case gcide_content_nl: |
| 654 |
case gcide_content_br: |
| 655 |
return 1; |
| 656 |
|
| 657 |
default: |
| 658 |
break; |
| 659 |
} |
| 660 |
break; |
| 661 |
|
| 662 |
case gcide_content_br: |
| 663 |
if (btag != NULL) { |
| 664 |
if (btag->tag_type == gcide_content_nl) |
| 665 |
btag->delete = 1; |
| 666 |
} |
| 667 |
break; |
| 668 |
|
| 669 |
default: |
| 670 |
break; |
| 671 |
} |
| 672 |
return 0; |
| 673 |
} |
| 674 |
|
| 675 |
/* |
| 676 |
* This analyzes the placement of gcide_content_nl and gcide_content_br |
| 677 |
* elements and removes the former where they are superfluous. |
| 678 |
* The gcide_content_nl elements are removed when they appear at either |
| 679 |
* side of a gcide_content_br element, or before a block tag, or a text |
| 680 |
* segment that begins with a whitespace character. |
| 681 |
* |
| 682 |
* Additionally, curly brackets surrounding a <source>...</source> construct |
| 683 |
* are removed as well. |
| 684 |
*/ |
| 685 |
static void |
| 686 |
ws_fixup(struct gcide_tag *tag) |
| 687 |
{ |
| 688 |
if (tag->tag_type == gcide_content_top || |
| 689 |
tag->tag_type == gcide_content_tag) |
| 690 |
dico_list_reduce(tag->v.tag.taglist, ws_reduce, NULL); |
| 691 |
} |
| 692 |
|
| 693 |
static size_t |
| 694 |
greek_translit(size_t n) |
| 695 |
{ |
| 696 |
while (textspace[n]) { |
| 697 |
size_t rd; |
| 698 |
const char *greek = gcide_grk_to_utf8(textspace + n, &rd); |
| 699 |
|
| 700 |
if (greek) { |
| 701 |
text_add_str(greek, strlen(greek)); |
| 702 |
n += rd; |
| 703 |
} else { |
| 704 |
text_add_chr(textspace[n]); |
| 705 |
n++; |
| 706 |
} |
| 707 |
} |
| 708 |
return text_segment_finish(); |
| 709 |
} |
| 710 |
|
| 711 |
static int |
| 712 |
early_fixup(void *item, void *data) |
| 713 |
{ |
| 714 |
struct gcide_tag *tag = item; |
| 715 |
int translate = data ? *(int*)data : 0; |
| 716 |
|
| 717 |
switch (tag->tag_type) { |
| 718 |
case gcide_content_text: |
| 719 |
if (translate) |
| 720 |
tag->v.textpos = greek_translit(tag->v.textpos); |
| 721 |
break; |
| 722 |
case gcide_content_top: |
| 723 |
dico_list_iterate(tag->v.tag.taglist, early_fixup, &translate); |
| 724 |
break; |
| 725 |
case gcide_content_tag: |
| 726 |
translate = gcide_is_tag(tag, "grk"); |
| 727 |
dico_list_iterate(tag->v.tag.taglist, early_fixup, &translate); |
| 728 |
break; |
| 729 |
default: |
| 730 |
break; |
| 731 |
} |
| 732 |
return 0; |
| 733 |
} |
| 734 |
|
| 735 |
/* |
| 736 |
* The <as> blocks in GCIDE corpus often begin with "as, ". This looks strange |
| 737 |
* if another font is used for <as> content or if it is otherwise marked up. |
| 738 |
* To avoid this, move "as, " out of <as>block. |
| 739 |
*/ |
| 740 |
static void |
| 741 |
as_fixup(struct gcide_tag *tag) |
| 742 |
{ |
| 743 |
dico_iterator_t itr; |
| 744 |
|
| 745 |
switch (tag->tag_type) { |
| 746 |
case gcide_content_top: |
| 747 |
case gcide_content_tag: |
| 748 |
itr = dico_list_iterator(tag->v.tag.taglist); |
| 749 |
tag = dico_iterator_first(itr); |
| 750 |
while (tag) { |
| 751 |
if (gcide_is_tag(tag, "as")) { |
| 752 |
struct gcide_tag *t = dico_list_head(tag->v.tag.taglist); |
| 753 |
char *s = textspace + t->v.textpos; |
| 754 |
if (strncmp(s, "as", 2) == 0 && |
| 755 |
(isspace(s[3]) || ispunct(s[3]))) { |
| 756 |
int len; |
| 757 |
struct gcide_tag *t1; |
| 758 |
|
| 759 |
for (len = 3; s[len] && isspace(s[len]); len++); |
| 760 |
text_add_str(s, len); |
| 761 |
t1 = alloc_tag(gcide_content_text); |
| 762 |
t1->v.textpos = text_segment_finish(); |
| 763 |
dico_iterator_insert_before(itr, t1); |
| 764 |
dico_list_remove(tag->v.tag.taglist, t, NULL); |
| 765 |
if (s[len]) { |
| 766 |
t1 = alloc_tag(gcide_content_text); |
| 767 |
text_add_strz(s + len); |
| 768 |
t1->v.textpos = text_segment_finish(); |
| 769 |
dico_list_prepend(tag->v.tag.taglist, t1); |
| 770 |
} |
| 771 |
} |
| 772 |
} else |
| 773 |
as_fixup(tag); |
| 774 |
tag = dico_iterator_next(itr); |
| 775 |
} |
| 776 |
dico_iterator_destroy(&itr); |
| 777 |
break; |
| 778 |
default: |
| 779 |
break; |
| 780 |
} |
| 781 |
} |
| 782 |
|
| 783 |
struct gcide_parse_tree * |
| 784 |
gcide_markup_parse(char const *text, size_t len, int dbg, struct gcide_locus *loc) |
| 785 |
{ |
| 786 |
struct gcide_parse_tree *tp; |
| 787 |
struct gcide_tag *p; |
| 788 |
|
| 789 |
input_buf = text; |
| 790 |
input_len = len; |
| 791 |
token_beg = token_end = 0; |
| 792 |
first_p_seen = 0; |
| 793 |
base_locus = loc; |
| 794 |
|
| 795 |
if (setjmp(errbuf)) { |
| 796 |
free(textspace); |
| 797 |
dico_list_destroy(&tagstk); |
| 798 |
free_tag(current_tag, NULL); |
| 799 |
} |
| 800 |
|
| 801 |
if (SIZE_MAX / 2 < len) |
| 802 |
memerr("gcide_markup_parse"); |
| 803 |
textsize = 2 * len; |
| 804 |
textspace = malloc(textsize); |
| 805 |
if (!textspace) |
| 806 |
memerr("gcide_markup_parse"); |
| 807 |
textpos = textstart = 0; |
| 808 |
|
| 809 |
tagstk = dico_list_create(); |
| 810 |
if (!tagstk) |
| 811 |
memerr("gcide_markup_parse"); |
| 812 |
dico_list_set_free_item(tagstk, free_tag, NULL); |
| 813 |
current_tag = alloc_tag(gcide_content_top); |
| 814 |
|
| 815 |
yy_flex_debug = dbg; |
| 816 |
BEGIN(INITIAL); |
| 817 |
while (yylex()) |
| 818 |
; |
| 819 |
|
| 820 |
/* Append trailing text segment, if any */ |
| 821 |
flush_text_segment(); |
| 822 |
|
| 823 |
if (current_tag->tag_type != gcide_content_top) { |
| 824 |
p = current_tag; |
| 825 |
do { |
| 826 |
if (p->tag_type == gcide_content_tag) |
| 827 |
dico_log(L_WARN, 0, "%s:%zu: unclosed %s tag", |
| 828 |
LOCUS_FILE, LOCUS_OFF(p->offset), p->tag_name); |
| 829 |
current_tag = p; |
| 830 |
} while ((p = dico_list_pop(tagstk))); |
| 831 |
} |
| 832 |
|
| 833 |
dico_list_destroy(&tagstk); |
| 834 |
|
| 835 |
if (!current_tag) { |
| 836 |
free(textspace); |
| 837 |
return NULL; |
| 838 |
} |
| 839 |
|
| 840 |
early_fixup(current_tag, NULL); |
| 841 |
as_fixup(current_tag); |
| 842 |
|
| 843 |
tp = malloc(sizeof(*tp)); |
| 844 |
if (!tp) |
| 845 |
memerr("gcide_markup_parse"); |
| 846 |
|
| 847 |
tp->textspace = textspace; |
| 848 |
tp->textsize = textsize; |
| 849 |
|
| 850 |
tp->root = current_tag; |
| 851 |
|
| 852 |
tag_fixup(tp->root, textspace); |
| 853 |
ws_fixup(tp->root); |
| 854 |
|
| 855 |
return tp; |
| 856 |
} |
| 857 |
|
| 858 |
void |
| 859 |
gcide_parse_tree_free(struct gcide_parse_tree *tp) |
| 860 |
{ |
| 861 |
free(tp->textspace); |
| 862 |
free_tag(tp->root, NULL); |
| 863 |
free(tp); |
| 864 |
} |