50
|
1
|
|
2 /*
|
|
3 * Copyright (C) Igor Sysoev
|
|
4 */
|
|
5
|
|
6
|
|
7 #include <ngx_config.h>
|
|
8 #include <ngx_core.h>
|
|
9 #include <ngx_http.h>
|
|
10
|
|
11
|
505
|
12 #define NGX_HTTP_CHARSET_OFF -2
|
|
13 #define NGX_HTTP_NO_CHARSET -3
|
|
14 #define NGX_HTTP_CHARSET_VAR 0x10000
|
206
|
15
|
|
16 /* 1 byte length and up to 3 bytes for the UTF-8 encoding of the UCS-2 */
|
|
17 #define NGX_UTF_LEN 4
|
|
18
|
|
19 #define NGX_HTML_ENTITY_LEN (sizeof("") - 1)
|
78
|
20
|
|
21
|
50
|
22 typedef struct {
|
206
|
23 u_char **tables;
|
|
24 ngx_str_t name;
|
72
|
25
|
206
|
26 unsigned length:16;
|
|
27 unsigned utf8:1;
|
50
|
28 } ngx_http_charset_t;
|
|
29
|
|
30
|
|
31 typedef struct {
|
206
|
32 ngx_int_t src;
|
|
33 ngx_int_t dst;
|
78
|
34 } ngx_http_charset_recode_t;
|
|
35
|
|
36
|
|
37 typedef struct {
|
206
|
38 ngx_int_t src;
|
|
39 ngx_int_t dst;
|
|
40 u_char *src2dst;
|
|
41 u_char *dst2src;
|
50
|
42 } ngx_http_charset_tables_t;
|
|
43
|
|
44
|
|
45 typedef struct {
|
206
|
46 ngx_array_t charsets; /* ngx_http_charset_t */
|
|
47 ngx_array_t tables; /* ngx_http_charset_tables_t */
|
|
48 ngx_array_t recodes; /* ngx_http_charset_recode_t */
|
50
|
49 } ngx_http_charset_main_conf_t;
|
|
50
|
|
51
|
|
52 typedef struct {
|
206
|
53 ngx_int_t charset;
|
|
54 ngx_int_t source_charset;
|
|
55 ngx_flag_t override_charset;
|
397
|
56
|
|
57 ngx_hash_t types;
|
|
58 ngx_array_t *types_keys;
|
50
|
59 } ngx_http_charset_loc_conf_t;
|
|
60
|
|
61
|
|
62 typedef struct {
|
206
|
63 u_char *table;
|
|
64 ngx_int_t charset;
|
505
|
65 ngx_str_t charset_name;
|
206
|
66
|
|
67 ngx_chain_t *busy;
|
|
68 ngx_chain_t *free_bufs;
|
|
69 ngx_chain_t *free_buffers;
|
|
70
|
|
71 size_t saved_len;
|
|
72 u_char saved[NGX_UTF_LEN];
|
|
73
|
|
74 unsigned length:16;
|
|
75 unsigned from_utf8:1;
|
|
76 unsigned to_utf8:1;
|
50
|
77 } ngx_http_charset_ctx_t;
|
|
78
|
|
79
|
206
|
80 typedef struct {
|
|
81 ngx_http_charset_tables_t *table;
|
|
82 ngx_http_charset_t *charset;
|
|
83 ngx_uint_t characters;
|
|
84 } ngx_http_charset_conf_ctx_t;
|
|
85
|
50
|
86
|
505
|
87 static ngx_int_t ngx_http_destination_charset(ngx_http_request_t *r,
|
|
88 ngx_str_t *name);
|
|
89 static ngx_int_t ngx_http_main_request_charset(ngx_http_request_t *r,
|
|
90 ngx_str_t *name);
|
|
91 static ngx_int_t ngx_http_source_charset(ngx_http_request_t *r,
|
|
92 ngx_str_t *name);
|
|
93 static ngx_int_t ngx_http_get_charset(ngx_http_request_t *r, ngx_str_t *name);
|
|
94 static ngx_inline void ngx_http_set_charset(ngx_http_request_t *r,
|
|
95 ngx_str_t *charset);
|
|
96 static ngx_int_t ngx_http_charset_ctx(ngx_http_request_t *r,
|
206
|
97 ngx_http_charset_t *charsets, ngx_int_t charset, ngx_int_t source_charset);
|
|
98 static ngx_uint_t ngx_http_charset_recode(ngx_buf_t *b, u_char *table);
|
|
99 static ngx_chain_t *ngx_http_charset_recode_from_utf8(ngx_pool_t *pool,
|
|
100 ngx_buf_t *buf, ngx_http_charset_ctx_t *ctx);
|
|
101 static ngx_chain_t *ngx_http_charset_recode_to_utf8(ngx_pool_t *pool,
|
|
102 ngx_buf_t *buf, ngx_http_charset_ctx_t *ctx);
|
|
103
|
|
104 static ngx_chain_t *ngx_http_charset_get_buf(ngx_pool_t *pool,
|
|
105 ngx_http_charset_ctx_t *ctx);
|
|
106 static ngx_chain_t *ngx_http_charset_get_buffer(ngx_pool_t *pool,
|
|
107 ngx_http_charset_ctx_t *ctx, size_t size);
|
|
108
|
|
109 static char *ngx_http_charset_map_block(ngx_conf_t *cf, ngx_command_t *cmd,
|
50
|
110 void *conf);
|
206
|
111 static char *ngx_http_charset_map(ngx_conf_t *cf, ngx_command_t *dummy,
|
|
112 void *conf);
|
50
|
113
|
|
114 static char *ngx_http_set_charset_slot(ngx_conf_t *cf, ngx_command_t *cmd,
|
|
115 void *conf);
|
|
116 static ngx_int_t ngx_http_add_charset(ngx_array_t *charsets, ngx_str_t *name);
|
|
117
|
|
118 static void *ngx_http_charset_create_main_conf(ngx_conf_t *cf);
|
|
119 static void *ngx_http_charset_create_loc_conf(ngx_conf_t *cf);
|
|
120 static char *ngx_http_charset_merge_loc_conf(ngx_conf_t *cf,
|
|
121 void *parent, void *child);
|
78
|
122 static ngx_int_t ngx_http_charset_postconfiguration(ngx_conf_t *cf);
|
50
|
123
|
|
124
|
397
|
125 ngx_str_t ngx_http_charset_default_types[] = {
|
|
126 ngx_string("text/html"),
|
|
127 ngx_string("text/xml"),
|
|
128 ngx_string("text/plain"),
|
|
129 ngx_string("text/vnd.wap.wml"),
|
|
130 ngx_string("application/x-javascript"),
|
|
131 ngx_string("application/rss+xml"),
|
|
132 ngx_null_string
|
|
133 };
|
|
134
|
|
135
|
50
|
136 static ngx_command_t ngx_http_charset_filter_commands[] = {
|
|
137
|
78
|
138 { ngx_string("charset"),
|
108
|
139 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
140 |NGX_HTTP_LIF_CONF|NGX_CONF_TAKE1,
|
78
|
141 ngx_http_set_charset_slot,
|
|
142 NGX_HTTP_LOC_CONF_OFFSET,
|
|
143 offsetof(ngx_http_charset_loc_conf_t, charset),
|
|
144 NULL },
|
|
145
|
|
146 { ngx_string("source_charset"),
|
108
|
147 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
148 |NGX_HTTP_LIF_CONF|NGX_CONF_TAKE1,
|
78
|
149 ngx_http_set_charset_slot,
|
|
150 NGX_HTTP_LOC_CONF_OFFSET,
|
|
151 offsetof(ngx_http_charset_loc_conf_t, source_charset),
|
|
152 NULL },
|
|
153
|
184
|
154 { ngx_string("override_charset"),
|
|
155 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
156 |NGX_HTTP_LIF_CONF|NGX_CONF_FLAG,
|
|
157 ngx_conf_set_flag_slot,
|
|
158 NGX_HTTP_LOC_CONF_OFFSET,
|
|
159 offsetof(ngx_http_charset_loc_conf_t, override_charset),
|
|
160 NULL },
|
|
161
|
397
|
162 { ngx_string("charset_types"),
|
|
163 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF|NGX_CONF_1MORE,
|
|
164 ngx_http_types_slot,
|
|
165 NGX_HTTP_LOC_CONF_OFFSET,
|
|
166 offsetof(ngx_http_charset_loc_conf_t, types_keys),
|
|
167 &ngx_http_charset_default_types[0] },
|
|
168
|
50
|
169 { ngx_string("charset_map"),
|
|
170 NGX_HTTP_MAIN_CONF|NGX_CONF_BLOCK|NGX_CONF_TAKE2,
|
206
|
171 ngx_http_charset_map_block,
|
50
|
172 NGX_HTTP_MAIN_CONF_OFFSET,
|
|
173 0,
|
|
174 NULL },
|
|
175
|
|
176 ngx_null_command
|
|
177 };
|
|
178
|
|
179
|
|
180 static ngx_http_module_t ngx_http_charset_filter_module_ctx = {
|
58
|
181 NULL, /* preconfiguration */
|
78
|
182 ngx_http_charset_postconfiguration, /* postconfiguration */
|
50
|
183
|
|
184 ngx_http_charset_create_main_conf, /* create main configuration */
|
78
|
185 NULL, /* init main configuration */
|
50
|
186
|
|
187 NULL, /* create server configuration */
|
|
188 NULL, /* merge server configuration */
|
|
189
|
|
190 ngx_http_charset_create_loc_conf, /* create location configuration */
|
|
191 ngx_http_charset_merge_loc_conf /* merge location configuration */
|
|
192 };
|
|
193
|
|
194
|
|
195 ngx_module_t ngx_http_charset_filter_module = {
|
58
|
196 NGX_MODULE_V1,
|
50
|
197 &ngx_http_charset_filter_module_ctx, /* module context */
|
|
198 ngx_http_charset_filter_commands, /* module directives */
|
|
199 NGX_HTTP_MODULE, /* module type */
|
90
|
200 NULL, /* init master */
|
230
|
201 NULL, /* init module */
|
90
|
202 NULL, /* init process */
|
|
203 NULL, /* init thread */
|
|
204 NULL, /* exit thread */
|
|
205 NULL, /* exit process */
|
|
206 NULL, /* exit master */
|
|
207 NGX_MODULE_V1_PADDING
|
50
|
208 };
|
|
209
|
|
210
|
|
211 static ngx_http_output_header_filter_pt ngx_http_next_header_filter;
|
|
212 static ngx_http_output_body_filter_pt ngx_http_next_body_filter;
|
|
213
|
|
214
|
|
215 static ngx_int_t
|
|
216 ngx_http_charset_header_filter(ngx_http_request_t *r)
|
|
217 {
|
184
|
218 ngx_int_t charset, source_charset;
|
505
|
219 ngx_str_t dst, src;
|
50
|
220 ngx_http_charset_t *charsets;
|
|
221 ngx_http_charset_main_conf_t *mcf;
|
|
222
|
202
|
223 if (r == r->main) {
|
505
|
224 charset = ngx_http_destination_charset(r, &dst);
|
202
|
225
|
|
226 } else {
|
505
|
227 charset = ngx_http_main_request_charset(r, &dst);
|
|
228 }
|
206
|
229
|
505
|
230 if (charset == NGX_ERROR) {
|
|
231 return NGX_ERROR;
|
|
232 }
|
202
|
233
|
505
|
234 if (charset == NGX_DECLINED) {
|
|
235 return ngx_http_next_header_filter(r);
|
50
|
236 }
|
|
237
|
505
|
238 /* charset: charset index or NGX_HTTP_NO_CHARSET */
|
250
|
239
|
505
|
240 source_charset = ngx_http_source_charset(r, &src);
|
244
|
241
|
505
|
242 if (source_charset == NGX_ERROR) {
|
|
243 return NGX_ERROR;
|
206
|
244 }
|
184
|
245
|
505
|
246 /*
|
|
247 * source_charset: charset index, NGX_HTTP_NO_CHARSET,
|
|
248 * or NGX_HTTP_CHARSET_OFF
|
|
249 */
|
|
250
|
|
251 ngx_log_debug2(NGX_LOG_DEBUG_HTTP, r->connection->log, 0,
|
|
252 "charset: \"%V\" > \"%V\"", &src, &dst);
|
|
253
|
|
254 if (source_charset == NGX_HTTP_CHARSET_OFF) {
|
|
255 ngx_http_set_charset(r, &dst);
|
|
256
|
|
257 return ngx_http_next_header_filter(r);
|
|
258 }
|
184
|
259
|
210
|
260 if (charset == NGX_HTTP_NO_CHARSET
|
|
261 || source_charset == NGX_HTTP_NO_CHARSET)
|
|
262 {
|
505
|
263 if (source_charset != charset
|
|
264 || ngx_strncasecmp(dst.data, src.data, dst.len) != 0)
|
210
|
265 {
|
212
|
266 goto no_charset_map;
|
210
|
267 }
|
184
|
268
|
505
|
269 ngx_http_set_charset(r, &dst);
|
|
270
|
206
|
271 return ngx_http_next_header_filter(r);
|
|
272 }
|
184
|
273
|
505
|
274 mcf = ngx_http_get_module_main_conf(r, ngx_http_charset_filter_module);
|
|
275 charsets = mcf->charsets.elts;
|
|
276
|
206
|
277 if (source_charset != charset
|
|
278 && (charsets[source_charset].tables == NULL
|
|
279 || charsets[source_charset].tables[charset] == NULL))
|
|
280 {
|
212
|
281 goto no_charset_map;
|
50
|
282 }
|
|
283
|
206
|
284 r->headers_out.content_type.len = r->headers_out.content_type_len;
|
|
285
|
505
|
286 ngx_http_set_charset(r, &dst);
|
|
287
|
|
288 if (source_charset != charset) {
|
|
289 return ngx_http_charset_ctx(r, charsets, charset, source_charset);
|
|
290 }
|
|
291
|
|
292 return ngx_http_next_header_filter(r);
|
212
|
293
|
|
294 no_charset_map:
|
|
295
|
|
296 ngx_log_error(NGX_LOG_ERR, r->connection->log, 0,
|
342
|
297 "no \"charset_map\" between the charsets \"%V\" and \"%V\"",
|
505
|
298 &src, &dst);
|
212
|
299
|
|
300 return ngx_http_next_header_filter(r);
|
206
|
301 }
|
|
302
|
|
303
|
|
304 static ngx_int_t
|
505
|
305 ngx_http_destination_charset(ngx_http_request_t *r, ngx_str_t *name)
|
|
306 {
|
|
307 ngx_int_t charset;
|
|
308 ngx_http_charset_t *charsets;
|
|
309 ngx_http_variable_value_t *vv;
|
|
310 ngx_http_charset_loc_conf_t *mlcf;
|
|
311 ngx_http_charset_main_conf_t *mcf;
|
|
312
|
|
313 if (!r->ignore_content_encoding
|
|
314 && r->headers_out.content_encoding
|
|
315 && r->headers_out.content_encoding->value.len)
|
|
316 {
|
|
317 return NGX_DECLINED;
|
|
318 }
|
|
319
|
|
320 if (r->headers_out.content_type.len == 0) {
|
|
321 return NGX_DECLINED;
|
|
322 }
|
|
323
|
|
324 if (r->headers_out.override_charset
|
|
325 && r->headers_out.override_charset->len)
|
|
326 {
|
|
327 *name = *r->headers_out.override_charset;
|
|
328
|
|
329 charset = ngx_http_get_charset(r, name);
|
|
330
|
|
331 if (charset != NGX_HTTP_NO_CHARSET) {
|
|
332 return charset;
|
|
333 }
|
|
334
|
|
335 ngx_log_error(NGX_LOG_ERR, r->connection->log, 0,
|
|
336 "unknown charset \"%V\" to override", name);
|
|
337
|
|
338 return NGX_DECLINED;
|
|
339 }
|
|
340
|
|
341 mlcf = ngx_http_get_module_loc_conf(r, ngx_http_charset_filter_module);
|
|
342 charset = mlcf->charset;
|
|
343
|
|
344 if (charset == NGX_HTTP_CHARSET_OFF) {
|
|
345 return NGX_DECLINED;
|
|
346 }
|
|
347
|
|
348 if (r->headers_out.charset.len) {
|
|
349 if (mlcf->override_charset == 0) {
|
|
350 return NGX_DECLINED;
|
|
351 }
|
|
352
|
|
353 } else {
|
|
354 if (ngx_http_test_content_type(r, &mlcf->types) == NULL) {
|
|
355 return NGX_DECLINED;
|
|
356 }
|
|
357 }
|
|
358
|
|
359 if (charset < NGX_HTTP_CHARSET_VAR) {
|
|
360 mcf = ngx_http_get_module_main_conf(r, ngx_http_charset_filter_module);
|
|
361 charsets = mcf->charsets.elts;
|
|
362 *name = charsets[charset].name;
|
|
363 return charset;
|
|
364 }
|
|
365
|
|
366 vv = ngx_http_get_indexed_variable(r, charset - NGX_HTTP_CHARSET_VAR);
|
|
367
|
|
368 if (vv == NULL || vv->not_found) {
|
|
369 return NGX_ERROR;
|
|
370 }
|
|
371
|
|
372 name->len = vv->len;
|
|
373 name->data = vv->data;
|
|
374
|
|
375 return ngx_http_get_charset(r, name);
|
|
376 }
|
|
377
|
|
378
|
|
379 static ngx_int_t
|
|
380 ngx_http_main_request_charset(ngx_http_request_t *r, ngx_str_t *src)
|
206
|
381 {
|
505
|
382 ngx_int_t charset;
|
|
383 ngx_str_t *main_charset;
|
|
384 ngx_http_charset_ctx_t *ctx;
|
|
385
|
|
386 ctx = ngx_http_get_module_ctx(r->main, ngx_http_charset_filter_module);
|
|
387
|
|
388 if (ctx) {
|
|
389 *src = ctx->charset_name;
|
|
390 return ctx->charset;
|
|
391 }
|
|
392
|
|
393 main_charset = &r->main->headers_out.charset;
|
|
394
|
|
395 if (main_charset->len == 0) {
|
|
396 return NGX_DECLINED;
|
|
397 }
|
|
398
|
|
399 ctx = ngx_pcalloc(r->pool, sizeof(ngx_http_charset_ctx_t));
|
|
400 if (ctx == NULL) {
|
|
401 return NGX_ERROR;
|
|
402 }
|
|
403
|
|
404 ngx_http_set_ctx(r->main, ctx, ngx_http_charset_filter_module);
|
|
405
|
|
406 charset = ngx_http_get_charset(r, main_charset);
|
|
407
|
|
408 ctx->charset = charset;
|
|
409 ctx->charset_name = *main_charset;
|
|
410 *src = *main_charset;
|
|
411
|
|
412 return charset;
|
|
413 }
|
|
414
|
|
415
|
|
416 static ngx_int_t
|
|
417 ngx_http_source_charset(ngx_http_request_t *r, ngx_str_t *name)
|
|
418 {
|
|
419 ngx_int_t charset;
|
|
420 ngx_http_charset_t *charsets;
|
|
421 ngx_http_variable_value_t *vv;
|
|
422 ngx_http_charset_loc_conf_t *lcf;
|
|
423 ngx_http_charset_main_conf_t *mcf;
|
|
424
|
|
425 if (r->headers_out.charset.len) {
|
|
426 *name = r->headers_out.charset;
|
|
427 return ngx_http_get_charset(r, name);
|
|
428 }
|
|
429
|
|
430 lcf = ngx_http_get_module_loc_conf(r, ngx_http_charset_filter_module);
|
|
431
|
|
432 charset = lcf->source_charset;
|
|
433
|
|
434 if (charset == NGX_HTTP_CHARSET_OFF) {
|
|
435 name->len = 0;
|
|
436 return charset;
|
|
437 }
|
|
438
|
|
439 if (charset < NGX_HTTP_CHARSET_VAR) {
|
|
440 mcf = ngx_http_get_module_main_conf(r, ngx_http_charset_filter_module);
|
|
441 charsets = mcf->charsets.elts;
|
|
442 *name = charsets[charset].name;
|
|
443 return charset;
|
|
444 }
|
|
445
|
|
446 vv = ngx_http_get_indexed_variable(r, charset - NGX_HTTP_CHARSET_VAR);
|
|
447
|
|
448 if (vv == NULL || vv->not_found) {
|
|
449 return NGX_ERROR;
|
|
450 }
|
|
451
|
|
452 name->len = vv->len;
|
|
453 name->data = vv->data;
|
|
454
|
|
455 return ngx_http_get_charset(r, name);
|
|
456 }
|
|
457
|
|
458
|
|
459 static ngx_int_t
|
|
460 ngx_http_get_charset(ngx_http_request_t *r, ngx_str_t *name)
|
|
461 {
|
|
462 ngx_uint_t i, n;
|
|
463 ngx_http_charset_t *charset;
|
|
464 ngx_http_charset_main_conf_t *mcf;
|
|
465
|
|
466 mcf = ngx_http_get_module_main_conf(r, ngx_http_charset_filter_module);
|
|
467
|
|
468 charset = mcf->charsets.elts;
|
|
469 n = mcf->charsets.nelts;
|
206
|
470
|
|
471 for (i = 0; i < n; i++) {
|
505
|
472 if (charset[i].name.len != name->len) {
|
244
|
473 continue;
|
|
474 }
|
|
475
|
505
|
476 if (ngx_strncasecmp(charset[i].name.data, name->data, name->len) == 0) {
|
206
|
477 return i;
|
|
478 }
|
|
479 }
|
|
480
|
|
481 return NGX_HTTP_NO_CHARSET;
|
|
482 }
|
|
483
|
|
484
|
505
|
485 static ngx_inline void
|
|
486 ngx_http_set_charset(ngx_http_request_t *r, ngx_str_t *charset)
|
206
|
487 {
|
505
|
488 if (r != r->main) {
|
|
489 return;
|
|
490 }
|
206
|
491
|
50
|
492 if (r->headers_out.status == NGX_HTTP_MOVED_PERMANENTLY
|
|
493 || r->headers_out.status == NGX_HTTP_MOVED_TEMPORARILY)
|
|
494 {
|
|
495 /*
|
184
|
496 * do not set charset for the redirect because NN 4.x
|
|
497 * use this charset instead of the next page charset
|
50
|
498 */
|
|
499
|
|
500 r->headers_out.charset.len = 0;
|
505
|
501 return;
|
50
|
502 }
|
|
503
|
505
|
504 r->headers_out.charset = *charset;
|
|
505 }
|
|
506
|
184
|
507
|
505
|
508 static ngx_int_t
|
|
509 ngx_http_charset_ctx(ngx_http_request_t *r, ngx_http_charset_t *charsets,
|
|
510 ngx_int_t charset, ngx_int_t source_charset)
|
|
511 {
|
|
512 ngx_http_charset_ctx_t *ctx;
|
50
|
513
|
|
514 ctx = ngx_pcalloc(r->pool, sizeof(ngx_http_charset_ctx_t));
|
|
515 if (ctx == NULL) {
|
|
516 return NGX_ERROR;
|
|
517 }
|
|
518
|
|
519 ngx_http_set_ctx(r, ctx, ngx_http_charset_filter_module);
|
|
520
|
184
|
521 ctx->table = charsets[source_charset].tables[charset];
|
|
522 ctx->charset = charset;
|
505
|
523 ctx->charset_name = charsets[charset].name;
|
206
|
524 ctx->length = charsets[charset].length;
|
|
525 ctx->from_utf8 = charsets[source_charset].utf8;
|
|
526 ctx->to_utf8 = charsets[charset].utf8;
|
|
527
|
212
|
528 r->filter_need_in_memory = 1;
|
|
529
|
206
|
530 if ((ctx->to_utf8 || ctx->from_utf8) && r == r->main) {
|
|
531 ngx_http_clear_content_length(r);
|
212
|
532
|
|
533 } else {
|
|
534 r->filter_need_temporary = 1;
|
206
|
535 }
|
50
|
536
|
|
537 return ngx_http_next_header_filter(r);
|
|
538 }
|
|
539
|
|
540
|
|
541 static ngx_int_t
|
|
542 ngx_http_charset_body_filter(ngx_http_request_t *r, ngx_chain_t *in)
|
|
543 {
|
206
|
544 ngx_int_t rc;
|
|
545 ngx_buf_t *b;
|
|
546 ngx_chain_t *cl, *out, **ll;
|
184
|
547 ngx_http_charset_ctx_t *ctx;
|
50
|
548
|
|
549 ctx = ngx_http_get_module_ctx(r, ngx_http_charset_filter_module);
|
|
550
|
184
|
551 if (ctx == NULL || ctx->table == NULL) {
|
50
|
552 return ngx_http_next_body_filter(r, in);
|
|
553 }
|
|
554
|
206
|
555 if ((ctx->to_utf8 || ctx->from_utf8) || ctx->busy) {
|
|
556
|
|
557 out = NULL;
|
|
558 ll = &out;
|
|
559
|
|
560 for (cl = in; cl; cl = cl->next) {
|
|
561 b = cl->buf;
|
|
562
|
|
563 if (ngx_buf_size(b) == 0) {
|
214
|
564
|
|
565 *ll = ngx_alloc_chain_link(r->pool);
|
|
566 if (*ll == NULL) {
|
|
567 return NGX_ERROR;
|
|
568 }
|
|
569
|
|
570 (*ll)->buf = b;
|
|
571 (*ll)->next = NULL;
|
|
572
|
|
573 ll = &(*ll)->next;
|
|
574
|
206
|
575 continue;
|
|
576 }
|
|
577
|
|
578 if (ctx->to_utf8) {
|
|
579 *ll = ngx_http_charset_recode_to_utf8(r->pool, b, ctx);
|
|
580
|
|
581 } else {
|
|
582 *ll = ngx_http_charset_recode_from_utf8(r->pool, b, ctx);
|
|
583 }
|
|
584
|
|
585 if (*ll == NULL) {
|
|
586 return NGX_ERROR;
|
|
587 }
|
|
588
|
|
589 while (*ll) {
|
|
590 ll = &(*ll)->next;
|
|
591 }
|
|
592 }
|
|
593
|
|
594 rc = ngx_http_next_body_filter(r, out);
|
|
595
|
|
596 if (out) {
|
|
597 if (ctx->busy == NULL) {
|
|
598 ctx->busy = out;
|
|
599
|
|
600 } else {
|
|
601 for (cl = ctx->busy; cl->next; cl = cl->next) { /* void */ }
|
|
602 cl->next = out;
|
|
603 }
|
|
604 }
|
|
605
|
|
606 while (ctx->busy) {
|
|
607
|
|
608 cl = ctx->busy;
|
|
609 b = cl->buf;
|
|
610
|
|
611 if (ngx_buf_size(b) != 0) {
|
|
612 break;
|
|
613 }
|
|
614
|
|
615 ctx->busy = cl->next;
|
|
616
|
|
617 if (b->tag != (ngx_buf_tag_t) &ngx_http_charset_filter_module) {
|
|
618 continue;
|
|
619 }
|
|
620
|
|
621 if (b->shadow) {
|
|
622 b->shadow->pos = b->shadow->last;
|
|
623 }
|
|
624
|
|
625 if (b->pos) {
|
|
626 cl->next = ctx->free_buffers;
|
|
627 ctx->free_buffers = cl;
|
|
628 continue;
|
|
629 }
|
|
630
|
|
631 cl->next = ctx->free_bufs;
|
|
632 ctx->free_bufs = cl;
|
|
633 }
|
|
634
|
|
635 return rc;
|
|
636 }
|
|
637
|
50
|
638 for (cl = in; cl; cl = cl->next) {
|
184
|
639 (void) ngx_http_charset_recode(cl->buf, ctx->table);
|
50
|
640 }
|
|
641
|
|
642 return ngx_http_next_body_filter(r, in);
|
|
643 }
|
|
644
|
|
645
|
|
646 static ngx_uint_t
|
184
|
647 ngx_http_charset_recode(ngx_buf_t *b, u_char *table)
|
50
|
648 {
|
370
|
649 u_char *p, *last;
|
184
|
650
|
370
|
651 last = b->last;
|
50
|
652
|
370
|
653 for (p = b->pos; p < last; p++) {
|
|
654
|
|
655 if (*p != table[*p]) {
|
|
656 goto recode;
|
50
|
657 }
|
|
658 }
|
|
659
|
184
|
660 return 0;
|
370
|
661
|
|
662 recode:
|
|
663
|
|
664 do {
|
|
665 if (*p != table[*p]) {
|
|
666 *p = table[*p];
|
|
667 }
|
|
668
|
|
669 p++;
|
|
670
|
|
671 } while (p < last);
|
|
672
|
|
673 b->in_file = 0;
|
|
674
|
|
675 return 1;
|
50
|
676 }
|
|
677
|
|
678
|
206
|
679 static ngx_chain_t *
|
|
680 ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
|
|
681 ngx_http_charset_ctx_t *ctx)
|
|
682 {
|
|
683 size_t len, size;
|
|
684 u_char c, *p, *src, *dst, *saved, **table;
|
|
685 uint32_t n;
|
|
686 ngx_buf_t *b;
|
|
687 ngx_uint_t i;
|
|
688 ngx_chain_t *out, *cl, **ll;
|
|
689
|
|
690 src = buf->pos;
|
|
691
|
|
692 if (ctx->saved_len == 0) {
|
|
693
|
|
694 for ( /* void */ ; src < buf->last; src++) {
|
|
695
|
|
696 if (*src < 0x80) {
|
|
697 continue;
|
|
698 }
|
|
699
|
|
700 len = src - buf->pos;
|
|
701
|
|
702 if (len > 512) {
|
|
703 out = ngx_http_charset_get_buf(pool, ctx);
|
|
704 if (out == NULL) {
|
|
705 return NULL;
|
|
706 }
|
|
707
|
|
708 b = out->buf;
|
|
709
|
|
710 b->temporary = buf->temporary;
|
|
711 b->memory = buf->memory;
|
|
712 b->mmap = buf->mmap;
|
|
713 b->flush = buf->flush;
|
|
714
|
|
715 b->pos = buf->pos;
|
|
716 b->last = src;
|
|
717
|
|
718 out->buf = b;
|
|
719 out->next = NULL;
|
|
720
|
|
721 size = buf->last - src;
|
|
722
|
|
723 saved = src;
|
392
|
724 n = ngx_utf8_decode(&saved, size);
|
206
|
725
|
|
726 if (n == 0xfffffffe) {
|
|
727 /* incomplete UTF-8 symbol */
|
|
728
|
|
729 ngx_memcpy(ctx->saved, src, size);
|
|
730 ctx->saved_len = size;
|
|
731
|
|
732 b->shadow = buf;
|
|
733
|
|
734 return out;
|
|
735 }
|
|
736
|
|
737 } else {
|
|
738 out = NULL;
|
|
739 size = len + buf->last - src;
|
|
740 src = buf->pos;
|
|
741 }
|
|
742
|
|
743 if (size < NGX_HTML_ENTITY_LEN) {
|
|
744 size += NGX_HTML_ENTITY_LEN;
|
|
745 }
|
|
746
|
|
747 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
748 if (cl == NULL) {
|
|
749 return NULL;
|
|
750 }
|
|
751
|
|
752 if (out) {
|
|
753 out->next = cl;
|
|
754
|
|
755 } else {
|
|
756 out = cl;
|
|
757 }
|
|
758
|
|
759 b = cl->buf;
|
|
760 dst = b->pos;
|
|
761
|
|
762 goto recode;
|
|
763 }
|
|
764
|
|
765 out = ngx_alloc_chain_link(pool);
|
|
766 if (out == NULL) {
|
|
767 return NULL;
|
|
768 }
|
|
769
|
|
770 out->buf = buf;
|
|
771 out->next = NULL;
|
|
772
|
|
773 return out;
|
|
774 }
|
|
775
|
|
776 /* process incomplete UTF sequence from previous buffer */
|
|
777
|
|
778 ngx_log_debug1(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
779 "http charset utf saved: %z", ctx->saved_len);
|
|
780
|
|
781 p = src;
|
|
782
|
|
783 for (i = ctx->saved_len; i < NGX_UTF_LEN; i++) {
|
|
784 ctx->saved[i] = *p++;
|
|
785
|
|
786 if (p == buf->last) {
|
|
787 break;
|
|
788 }
|
|
789 }
|
|
790
|
|
791 saved = ctx->saved;
|
392
|
792 n = ngx_utf8_decode(&saved, i);
|
206
|
793
|
|
794 c = '\0';
|
|
795
|
|
796 if (n < 0x10000) {
|
|
797 table = (u_char **) ctx->table;
|
|
798 p = table[n >> 8];
|
|
799
|
|
800 if (p) {
|
|
801 c = p[n & 0xff];
|
|
802 }
|
|
803
|
|
804 } else if (n == 0xfffffffe) {
|
|
805
|
|
806 /* incomplete UTF-8 symbol */
|
|
807
|
|
808 if (i < NGX_UTF_LEN) {
|
|
809 out = ngx_http_charset_get_buf(pool, ctx);
|
|
810 if (out == NULL) {
|
|
811 return NULL;
|
|
812 }
|
|
813
|
|
814 b = out->buf;
|
|
815
|
|
816 b->pos = buf->pos;
|
|
817 b->last = buf->last;
|
|
818 b->sync = 1;
|
|
819 b->shadow = buf;
|
|
820
|
|
821 ngx_memcpy(&ctx->saved[ctx->saved_len], src, i);
|
|
822 ctx->saved_len += i;
|
|
823
|
|
824 return out;
|
|
825 }
|
|
826 }
|
|
827
|
|
828 size = buf->last - buf->pos;
|
|
829
|
|
830 if (size < NGX_HTML_ENTITY_LEN) {
|
|
831 size += NGX_HTML_ENTITY_LEN;
|
|
832 }
|
|
833
|
|
834 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
835 if (cl == NULL) {
|
|
836 return NULL;
|
|
837 }
|
|
838
|
|
839 out = cl;
|
|
840
|
|
841 b = cl->buf;
|
|
842 dst = b->pos;
|
|
843
|
|
844 if (c) {
|
|
845 *dst++ = c;
|
|
846
|
|
847 } else if (n == 0xfffffffe) {
|
|
848 *dst++ = '?';
|
|
849
|
|
850 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
851 "http charset invalid utf 0");
|
|
852
|
|
853 saved = &ctx->saved[NGX_UTF_LEN];
|
|
854
|
|
855 } else if (n > 0x10ffff) {
|
|
856 *dst++ = '?';
|
|
857
|
|
858 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
859 "http charset invalid utf 1");
|
|
860
|
|
861 } else {
|
|
862 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
863 }
|
|
864
|
|
865 src += (saved - ctx->saved) - ctx->saved_len;
|
|
866 ctx->saved_len = 0;
|
|
867
|
|
868 recode:
|
|
869
|
|
870 ll = &cl->next;
|
|
871
|
|
872 table = (u_char **) ctx->table;
|
|
873
|
|
874 while (src < buf->last) {
|
|
875
|
|
876 if ((size_t) (b->end - dst) < NGX_HTML_ENTITY_LEN) {
|
|
877 b->last = dst;
|
|
878
|
|
879 size = buf->last - src + NGX_HTML_ENTITY_LEN;
|
|
880
|
|
881 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
882 if (cl == NULL) {
|
|
883 return NULL;
|
|
884 }
|
|
885
|
|
886 *ll = cl;
|
|
887 ll = &cl->next;
|
|
888
|
|
889 b = cl->buf;
|
|
890 dst = b->pos;
|
|
891 }
|
|
892
|
|
893 if (*src < 0x80) {
|
|
894 *dst++ = *src++;
|
|
895 continue;
|
|
896 }
|
|
897
|
|
898 len = buf->last - src;
|
|
899
|
392
|
900 n = ngx_utf8_decode(&src, len);
|
206
|
901
|
|
902 if (n < 0x10000) {
|
|
903
|
|
904 p = table[n >> 8];
|
|
905
|
|
906 if (p) {
|
|
907 c = p[n & 0xff];
|
|
908
|
|
909 if (c) {
|
|
910 *dst++ = c;
|
|
911 continue;
|
|
912 }
|
|
913 }
|
|
914
|
|
915 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
916
|
|
917 continue;
|
|
918 }
|
|
919
|
|
920 if (n == 0xfffffffe) {
|
|
921 /* incomplete UTF-8 symbol */
|
|
922
|
|
923 ngx_memcpy(ctx->saved, src, len);
|
|
924 ctx->saved_len = len;
|
|
925
|
|
926 if (b->pos == dst) {
|
|
927 b->sync = 1;
|
|
928 b->temporary = 0;
|
|
929 }
|
|
930
|
|
931 break;
|
|
932 }
|
|
933
|
|
934 if (n > 0x10ffff) {
|
|
935 *dst++ = '?';
|
|
936
|
|
937 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
938 "http charset invalid utf 2");
|
|
939
|
|
940 continue;
|
|
941 }
|
|
942
|
|
943 /* n > 0xffff */
|
|
944
|
|
945 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
946 }
|
|
947
|
|
948 b->last = dst;
|
|
949
|
|
950 b->last_buf = buf->last_buf;
|
|
951 b->last_in_chain = buf->last_in_chain;
|
|
952 b->flush = buf->flush;
|
|
953
|
|
954 b->shadow = buf;
|
|
955
|
|
956 return out;
|
|
957 }
|
|
958
|
|
959
|
|
960 static ngx_chain_t *
|
|
961 ngx_http_charset_recode_to_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
|
|
962 ngx_http_charset_ctx_t *ctx)
|
|
963 {
|
|
964 size_t len, size;
|
|
965 u_char *p, *src, *dst, *table;
|
|
966 ngx_buf_t *b;
|
|
967 ngx_chain_t *out, *cl, **ll;
|
|
968
|
|
969 table = ctx->table;
|
|
970
|
|
971 for (src = buf->pos; src < buf->last; src++) {
|
|
972 if (table[*src * NGX_UTF_LEN] == '\1') {
|
|
973 continue;
|
|
974 }
|
|
975
|
|
976 goto recode;
|
|
977 }
|
|
978
|
|
979 out = ngx_alloc_chain_link(pool);
|
|
980 if (out == NULL) {
|
|
981 return NULL;
|
|
982 }
|
|
983
|
|
984 out->buf = buf;
|
|
985 out->next = NULL;
|
|
986
|
|
987 return out;
|
|
988
|
|
989 recode:
|
|
990
|
|
991 /*
|
|
992 * we assume that there are about half of characters to be recoded,
|
|
993 * so we preallocate "size / 2 + size / 2 * ctx->length"
|
|
994 */
|
|
995
|
|
996 len = src - buf->pos;
|
|
997
|
|
998 if (len > 512) {
|
|
999 out = ngx_http_charset_get_buf(pool, ctx);
|
|
1000 if (out == NULL) {
|
|
1001 return NULL;
|
|
1002 }
|
|
1003
|
|
1004 b = out->buf;
|
|
1005
|
|
1006 b->temporary = buf->temporary;
|
|
1007 b->memory = buf->memory;
|
|
1008 b->mmap = buf->mmap;
|
|
1009 b->flush = buf->flush;
|
|
1010
|
|
1011 b->pos = buf->pos;
|
|
1012 b->last = src;
|
|
1013
|
|
1014 out->buf = b;
|
|
1015 out->next = NULL;
|
|
1016
|
|
1017 size = buf->last - src;
|
|
1018 size = size / 2 + size / 2 * ctx->length;
|
|
1019
|
|
1020 } else {
|
|
1021 out = NULL;
|
|
1022
|
|
1023 size = buf->last - src;
|
|
1024 size = len + size / 2 + size / 2 * ctx->length;
|
|
1025
|
|
1026 src = buf->pos;
|
|
1027 }
|
|
1028
|
|
1029 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
1030 if (cl == NULL) {
|
|
1031 return NULL;
|
|
1032 }
|
|
1033
|
|
1034 if (out) {
|
|
1035 out->next = cl;
|
|
1036
|
|
1037 } else {
|
|
1038 out = cl;
|
|
1039 }
|
|
1040
|
|
1041 ll = &cl->next;
|
|
1042
|
|
1043 b = cl->buf;
|
|
1044 dst = b->pos;
|
|
1045
|
|
1046 while (src < buf->last) {
|
|
1047
|
|
1048 p = &table[*src++ * NGX_UTF_LEN];
|
|
1049 len = *p++;
|
|
1050
|
|
1051 if ((size_t) (b->end - dst) < len) {
|
|
1052 b->last = dst;
|
|
1053
|
|
1054 size = buf->last - src;
|
|
1055 size = len + size / 2 + size / 2 * ctx->length;
|
|
1056
|
|
1057 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
1058 if (cl == NULL) {
|
|
1059 return NULL;
|
|
1060 }
|
|
1061
|
|
1062 *ll = cl;
|
|
1063 ll = &cl->next;
|
|
1064
|
|
1065 b = cl->buf;
|
|
1066 dst = b->pos;
|
|
1067 }
|
|
1068
|
|
1069 while (len) {
|
|
1070 *dst++ = *p++;
|
|
1071 len--;
|
|
1072 }
|
|
1073 }
|
|
1074
|
|
1075 b->last = dst;
|
|
1076
|
|
1077 b->last_buf = buf->last_buf;
|
|
1078 b->last_in_chain = buf->last_in_chain;
|
|
1079 b->flush = buf->flush;
|
|
1080
|
|
1081 b->shadow = buf;
|
|
1082
|
|
1083 return out;
|
|
1084 }
|
|
1085
|
|
1086
|
|
1087 static ngx_chain_t *
|
|
1088 ngx_http_charset_get_buf(ngx_pool_t *pool, ngx_http_charset_ctx_t *ctx)
|
|
1089 {
|
|
1090 ngx_chain_t *cl;
|
|
1091
|
|
1092 cl = ctx->free_bufs;
|
|
1093
|
|
1094 if (cl) {
|
|
1095 ctx->free_bufs = cl->next;
|
|
1096
|
|
1097 cl->buf->shadow = NULL;
|
|
1098 cl->next = NULL;
|
|
1099
|
|
1100 return cl;
|
|
1101 }
|
|
1102
|
|
1103 cl = ngx_alloc_chain_link(pool);
|
|
1104 if (cl == NULL) {
|
|
1105 return NULL;
|
|
1106 }
|
|
1107
|
|
1108 cl->buf = ngx_calloc_buf(pool);
|
|
1109 if (cl->buf == NULL) {
|
|
1110 return NULL;
|
|
1111 }
|
|
1112
|
|
1113 cl->next = NULL;
|
|
1114
|
|
1115 cl->buf->tag = (ngx_buf_tag_t) &ngx_http_charset_filter_module;
|
|
1116
|
|
1117 return cl;
|
|
1118 }
|
|
1119
|
|
1120
|
|
1121 static ngx_chain_t *
|
|
1122 ngx_http_charset_get_buffer(ngx_pool_t *pool, ngx_http_charset_ctx_t *ctx,
|
|
1123 size_t size)
|
|
1124 {
|
|
1125 ngx_buf_t *b;
|
|
1126 ngx_chain_t *cl, **ll;
|
|
1127
|
|
1128 for (ll = &ctx->free_buffers, cl = ctx->free_buffers;
|
|
1129 cl;
|
|
1130 ll = &cl->next, cl = cl->next)
|
|
1131 {
|
|
1132 b = cl->buf;
|
|
1133
|
|
1134 if ((size_t) (b->end - b->start) >= size) {
|
|
1135 *ll = cl->next;
|
|
1136 cl->next = NULL;
|
|
1137
|
|
1138 b->pos = b->start;
|
|
1139 b->temporary = 1;
|
|
1140 b->shadow = NULL;
|
|
1141
|
|
1142 return cl;
|
|
1143 }
|
|
1144 }
|
|
1145
|
|
1146 cl = ngx_alloc_chain_link(pool);
|
|
1147 if (cl == NULL) {
|
|
1148 return NULL;
|
|
1149 }
|
|
1150
|
|
1151 cl->buf = ngx_create_temp_buf(pool, size);
|
|
1152 if (cl->buf == NULL) {
|
|
1153 return NULL;
|
|
1154 }
|
|
1155
|
|
1156 cl->next = NULL;
|
|
1157
|
|
1158 cl->buf->temporary = 1;
|
|
1159 cl->buf->tag = (ngx_buf_tag_t) &ngx_http_charset_filter_module;
|
|
1160
|
|
1161 return cl;
|
|
1162 }
|
|
1163
|
|
1164
|
50
|
1165 static char *
|
206
|
1166 ngx_http_charset_map_block(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
|
50
|
1167 {
|
|
1168 ngx_http_charset_main_conf_t *mcf = conf;
|
|
1169
|
206
|
1170 char *rv;
|
|
1171 u_char *p, *dst2src, **pp;
|
|
1172 ngx_int_t src, dst;
|
|
1173 ngx_uint_t i, n;
|
|
1174 ngx_str_t *value;
|
|
1175 ngx_conf_t pvcf;
|
|
1176 ngx_http_charset_t *charset;
|
|
1177 ngx_http_charset_tables_t *table;
|
|
1178 ngx_http_charset_conf_ctx_t ctx;
|
50
|
1179
|
|
1180 value = cf->args->elts;
|
|
1181
|
|
1182 src = ngx_http_add_charset(&mcf->charsets, &value[1]);
|
|
1183 if (src == NGX_ERROR) {
|
|
1184 return NGX_CONF_ERROR;
|
|
1185 }
|
|
1186
|
|
1187 dst = ngx_http_add_charset(&mcf->charsets, &value[2]);
|
|
1188 if (dst == NGX_ERROR) {
|
|
1189 return NGX_CONF_ERROR;
|
|
1190 }
|
|
1191
|
|
1192 if (src == dst) {
|
|
1193 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1194 "\"charset_map\" between the same charsets "
|
|
1195 "\"%V\" and \"%V\"", &value[1], &value[2]);
|
|
1196 return NGX_CONF_ERROR;
|
|
1197 }
|
|
1198
|
|
1199 table = mcf->tables.elts;
|
|
1200 for (i = 0; i < mcf->tables.nelts; i++) {
|
|
1201 if ((src == table->src && dst == table->dst)
|
|
1202 || (src == table->dst && dst == table->src))
|
|
1203 {
|
|
1204 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1205 "duplicate \"charset_map\" between "
|
|
1206 "\"%V\" and \"%V\"", &value[1], &value[2]);
|
|
1207 return NGX_CONF_ERROR;
|
|
1208 }
|
|
1209 }
|
|
1210
|
|
1211 table = ngx_array_push(&mcf->tables);
|
|
1212 if (table == NULL) {
|
|
1213 return NGX_CONF_ERROR;
|
|
1214 }
|
|
1215
|
|
1216 table->src = src;
|
|
1217 table->dst = dst;
|
|
1218
|
286
|
1219 if (ngx_strcasecmp(value[2].data, (u_char *) "utf-8") == 0) {
|
206
|
1220 table->src2dst = ngx_pcalloc(cf->pool, 256 * NGX_UTF_LEN);
|
|
1221 if (table->src2dst == NULL) {
|
|
1222 return NGX_CONF_ERROR;
|
|
1223 }
|
|
1224
|
|
1225 table->dst2src = ngx_pcalloc(cf->pool, 256 * sizeof(void *));
|
|
1226 if (table->dst2src == NULL) {
|
|
1227 return NGX_CONF_ERROR;
|
|
1228 }
|
|
1229
|
|
1230 dst2src = ngx_pcalloc(cf->pool, 256);
|
|
1231 if (dst2src == NULL) {
|
|
1232 return NGX_CONF_ERROR;
|
|
1233 }
|
|
1234
|
|
1235 pp = (u_char **) &table->dst2src[0];
|
|
1236 pp[0] = dst2src;
|
|
1237
|
|
1238 for (i = 0; i < 128; i++) {
|
|
1239 p = &table->src2dst[i * NGX_UTF_LEN];
|
|
1240 p[0] = '\1';
|
|
1241 p[1] = (u_char) i;
|
|
1242 dst2src[i] = (u_char) i;
|
|
1243 }
|
50
|
1244
|
206
|
1245 for (/* void */; i < 256; i++) {
|
|
1246 p = &table->src2dst[i * NGX_UTF_LEN];
|
|
1247 p[0] = '\1';
|
|
1248 p[1] = '?';
|
|
1249 }
|
|
1250
|
|
1251 } else {
|
|
1252 table->src2dst = ngx_palloc(cf->pool, 256);
|
|
1253 if (table->src2dst == NULL) {
|
|
1254 return NGX_CONF_ERROR;
|
|
1255 }
|
|
1256
|
|
1257 table->dst2src = ngx_palloc(cf->pool, 256);
|
|
1258 if (table->dst2src == NULL) {
|
|
1259 return NGX_CONF_ERROR;
|
|
1260 }
|
|
1261
|
|
1262 for (i = 0; i < 128; i++) {
|
|
1263 table->src2dst[i] = (u_char) i;
|
|
1264 table->dst2src[i] = (u_char) i;
|
|
1265 }
|
|
1266
|
|
1267 for (/* void */; i < 256; i++) {
|
|
1268 table->src2dst[i] = '?';
|
|
1269 table->dst2src[i] = '?';
|
|
1270 }
|
50
|
1271 }
|
|
1272
|
206
|
1273 charset = mcf->charsets.elts;
|
50
|
1274
|
206
|
1275 ctx.table = table;
|
|
1276 ctx.charset = &charset[dst];
|
|
1277 ctx.characters = 0;
|
50
|
1278
|
|
1279 pvcf = *cf;
|
206
|
1280 cf->ctx = &ctx;
|
|
1281 cf->handler = ngx_http_charset_map;
|
50
|
1282 cf->handler_conf = conf;
|
|
1283
|
|
1284 rv = ngx_conf_parse(cf, NULL);
|
|
1285
|
|
1286 *cf = pvcf;
|
|
1287
|
206
|
1288 if (ctx.characters) {
|
|
1289 n = ctx.charset->length;
|
|
1290 ctx.charset->length /= ctx.characters;
|
|
1291
|
|
1292 if (((n * 10) / ctx.characters) % 10 > 4) {
|
|
1293 ctx.charset->length++;
|
|
1294 }
|
|
1295 }
|
|
1296
|
50
|
1297 return rv;
|
|
1298 }
|
|
1299
|
|
1300
|
|
1301 static char *
|
206
|
1302 ngx_http_charset_map(ngx_conf_t *cf, ngx_command_t *dummy, void *conf)
|
50
|
1303 {
|
206
|
1304 u_char *p, *dst2src, **pp;
|
|
1305 uint32_t n;
|
|
1306 ngx_int_t src, dst;
|
|
1307 ngx_str_t *value;
|
|
1308 ngx_uint_t i;
|
|
1309 ngx_http_charset_tables_t *table;
|
|
1310 ngx_http_charset_conf_ctx_t *ctx;
|
50
|
1311
|
|
1312 if (cf->args->nelts != 2) {
|
|
1313 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, "invalid parameters number");
|
|
1314 return NGX_CONF_ERROR;
|
|
1315 }
|
|
1316
|
|
1317 value = cf->args->elts;
|
|
1318
|
|
1319 src = ngx_hextoi(value[0].data, value[0].len);
|
|
1320 if (src == NGX_ERROR || src > 255) {
|
|
1321 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1322 "invalid value \"%V\"", &value[0]);
|
|
1323 return NGX_CONF_ERROR;
|
|
1324 }
|
|
1325
|
206
|
1326 ctx = cf->ctx;
|
|
1327 table = ctx->table;
|
|
1328
|
|
1329 if (ctx->charset->utf8) {
|
|
1330 p = &table->src2dst[src * NGX_UTF_LEN];
|
|
1331
|
|
1332 *p++ = (u_char) (value[1].len / 2);
|
|
1333
|
|
1334 for (i = 0; i < value[1].len; i += 2) {
|
|
1335 dst = ngx_hextoi(&value[1].data[i], 2);
|
|
1336 if (dst == NGX_ERROR || dst > 255) {
|
|
1337 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1338 "invalid value \"%V\"", &value[1]);
|
|
1339 return NGX_CONF_ERROR;
|
|
1340 }
|
|
1341
|
|
1342 *p++ = (u_char) dst;
|
|
1343 }
|
|
1344
|
|
1345 i /= 2;
|
|
1346
|
|
1347 ctx->charset->length += i;
|
|
1348 ctx->characters++;
|
|
1349
|
|
1350 p = &table->src2dst[src * NGX_UTF_LEN] + 1;
|
|
1351
|
392
|
1352 n = ngx_utf8_decode(&p, i);
|
206
|
1353
|
|
1354 if (n > 0xffff) {
|
|
1355 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1356 "invalid value \"%V\"", &value[1]);
|
|
1357 return NGX_CONF_ERROR;
|
|
1358 }
|
|
1359
|
|
1360 pp = (u_char **) &table->dst2src[0];
|
|
1361
|
|
1362 dst2src = pp[n >> 8];
|
|
1363
|
|
1364 if (dst2src == NULL) {
|
|
1365 dst2src = ngx_pcalloc(cf->pool, 256);
|
|
1366 if (dst2src == NULL) {
|
|
1367 return NGX_CONF_ERROR;
|
|
1368 }
|
|
1369
|
|
1370 pp[n >> 8] = dst2src;
|
|
1371 }
|
|
1372
|
|
1373 dst2src[n & 0xff] = (u_char) src;
|
|
1374
|
|
1375 } else {
|
|
1376 dst = ngx_hextoi(value[1].data, value[1].len);
|
|
1377 if (dst == NGX_ERROR || dst > 255) {
|
|
1378 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
1379 "invalid value \"%V\"", &value[1]);
|
|
1380 return NGX_CONF_ERROR;
|
|
1381 }
|
|
1382
|
|
1383 table->src2dst[src] = (u_char) dst;
|
|
1384 table->dst2src[dst] = (u_char) src;
|
50
|
1385 }
|
|
1386
|
|
1387 return NGX_CONF_OK;
|
|
1388 }
|
|
1389
|
|
1390
|
|
1391 static char *
|
|
1392 ngx_http_set_charset_slot(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
|
|
1393 {
|
|
1394 char *p = conf;
|
|
1395
|
|
1396 ngx_int_t *cp;
|
244
|
1397 ngx_str_t *value, var;
|
50
|
1398 ngx_http_charset_main_conf_t *mcf;
|
|
1399
|
|
1400 cp = (ngx_int_t *) (p + cmd->offset);
|
|
1401
|
|
1402 if (*cp != NGX_CONF_UNSET) {
|
|
1403 return "is duplicate";
|
|
1404 }
|
|
1405
|
78
|
1406 value = cf->args->elts;
|
|
1407
|
|
1408 if (cmd->offset == offsetof(ngx_http_charset_loc_conf_t, charset)
|
|
1409 && ngx_strcmp(value[1].data, "off") == 0)
|
|
1410 {
|
505
|
1411 *cp = NGX_HTTP_CHARSET_OFF;
|
78
|
1412 return NGX_CONF_OK;
|
|
1413 }
|
|
1414
|
244
|
1415
|
|
1416 if (value[1].data[0] == '$') {
|
|
1417 var.len = value[1].len - 1;
|
|
1418 var.data = value[1].data + 1;
|
|
1419
|
|
1420 *cp = ngx_http_get_variable_index(cf, &var);
|
|
1421
|
|
1422 if (*cp == NGX_ERROR) {
|
|
1423 return NGX_CONF_ERROR;
|
|
1424 }
|
|
1425
|
|
1426 *cp += NGX_HTTP_CHARSET_VAR;
|
|
1427
|
|
1428 return NGX_CONF_OK;
|
|
1429 }
|
|
1430
|
50
|
1431 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
1432 ngx_http_charset_filter_module);
|
|
1433
|
|
1434 *cp = ngx_http_add_charset(&mcf->charsets, &value[1]);
|
|
1435 if (*cp == NGX_ERROR) {
|
|
1436 return NGX_CONF_ERROR;
|
|
1437 }
|
|
1438
|
|
1439 return NGX_CONF_OK;
|
|
1440 }
|
|
1441
|
|
1442
|
|
1443 static ngx_int_t
|
|
1444 ngx_http_add_charset(ngx_array_t *charsets, ngx_str_t *name)
|
|
1445 {
|
|
1446 ngx_uint_t i;
|
|
1447 ngx_http_charset_t *c;
|
|
1448
|
|
1449 c = charsets->elts;
|
|
1450 for (i = 0; i < charsets->nelts; i++) {
|
|
1451 if (name->len != c[i].name.len) {
|
|
1452 continue;
|
|
1453 }
|
|
1454
|
|
1455 if (ngx_strcasecmp(name->data, c[i].name.data) == 0) {
|
|
1456 break;
|
|
1457 }
|
|
1458 }
|
|
1459
|
|
1460 if (i < charsets->nelts) {
|
|
1461 return i;
|
|
1462 }
|
|
1463
|
|
1464 c = ngx_array_push(charsets);
|
|
1465 if (c == NULL) {
|
|
1466 return NGX_ERROR;
|
|
1467 }
|
|
1468
|
|
1469 c->tables = NULL;
|
|
1470 c->name = *name;
|
206
|
1471 c->length = 0;
|
50
|
1472
|
286
|
1473 if (ngx_strcasecmp(name->data, (u_char *) "utf-8") == 0) {
|
72
|
1474 c->utf8 = 1;
|
216
|
1475
|
|
1476 } else {
|
|
1477 c->utf8 = 0;
|
72
|
1478 }
|
|
1479
|
50
|
1480 return i;
|
|
1481 }
|
|
1482
|
|
1483
|
|
1484 static void *
|
|
1485 ngx_http_charset_create_main_conf(ngx_conf_t *cf)
|
|
1486 {
|
|
1487 ngx_http_charset_main_conf_t *mcf;
|
|
1488
|
|
1489 mcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_main_conf_t));
|
|
1490 if (mcf == NULL) {
|
505
|
1491 return NULL;
|
50
|
1492 }
|
|
1493
|
|
1494 if (ngx_array_init(&mcf->charsets, cf->pool, 2, sizeof(ngx_http_charset_t))
|
463
|
1495 != NGX_OK)
|
50
|
1496 {
|
505
|
1497 return NULL;
|
50
|
1498 }
|
|
1499
|
78
|
1500 if (ngx_array_init(&mcf->tables, cf->pool, 1,
|
463
|
1501 sizeof(ngx_http_charset_tables_t))
|
|
1502 != NGX_OK)
|
50
|
1503 {
|
505
|
1504 return NULL;
|
50
|
1505 }
|
|
1506
|
78
|
1507 if (ngx_array_init(&mcf->recodes, cf->pool, 2,
|
463
|
1508 sizeof(ngx_http_charset_recode_t))
|
|
1509 != NGX_OK)
|
78
|
1510 {
|
505
|
1511 return NULL;
|
78
|
1512 }
|
|
1513
|
50
|
1514 return mcf;
|
|
1515 }
|
|
1516
|
|
1517
|
|
1518 static void *
|
|
1519 ngx_http_charset_create_loc_conf(ngx_conf_t *cf)
|
|
1520 {
|
|
1521 ngx_http_charset_loc_conf_t *lcf;
|
|
1522
|
|
1523 lcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_loc_conf_t));
|
|
1524 if (lcf == NULL) {
|
505
|
1525 return NULL;
|
50
|
1526 }
|
|
1527
|
397
|
1528 /*
|
|
1529 * set by ngx_pcalloc():
|
|
1530 *
|
|
1531 * lcf->types = { NULL };
|
|
1532 * lcf->types_keys = NULL;
|
|
1533 */
|
|
1534
|
78
|
1535 lcf->charset = NGX_CONF_UNSET;
|
50
|
1536 lcf->source_charset = NGX_CONF_UNSET;
|
184
|
1537 lcf->override_charset = NGX_CONF_UNSET;
|
50
|
1538
|
|
1539 return lcf;
|
|
1540 }
|
|
1541
|
|
1542
|
|
1543 static char *
|
|
1544 ngx_http_charset_merge_loc_conf(ngx_conf_t *cf, void *parent, void *child)
|
|
1545 {
|
|
1546 ngx_http_charset_loc_conf_t *prev = parent;
|
|
1547 ngx_http_charset_loc_conf_t *conf = child;
|
|
1548
|
78
|
1549 ngx_uint_t i;
|
|
1550 ngx_http_charset_recode_t *recode;
|
|
1551 ngx_http_charset_main_conf_t *mcf;
|
50
|
1552
|
400
|
1553 if (ngx_http_merge_types(cf, conf->types_keys, &conf->types,
|
|
1554 prev->types_keys, &prev->types,
|
|
1555 ngx_http_charset_default_types)
|
|
1556 != NGX_OK)
|
|
1557 {
|
|
1558 return NGX_CONF_ERROR;
|
|
1559 }
|
|
1560
|
184
|
1561 ngx_conf_merge_value(conf->override_charset, prev->override_charset, 0);
|
505
|
1562 ngx_conf_merge_value(conf->charset, prev->charset, NGX_HTTP_CHARSET_OFF);
|
|
1563 ngx_conf_merge_value(conf->source_charset, prev->source_charset,
|
|
1564 NGX_HTTP_CHARSET_OFF);
|
50
|
1565
|
505
|
1566 if (conf->charset == NGX_HTTP_CHARSET_OFF
|
|
1567 || conf->source_charset == NGX_HTTP_CHARSET_OFF
|
78
|
1568 || conf->charset == conf->source_charset)
|
50
|
1569 {
|
78
|
1570 return NGX_CONF_OK;
|
50
|
1571 }
|
|
1572
|
342
|
1573 if (conf->source_charset >= NGX_HTTP_CHARSET_VAR
|
|
1574 || conf->charset >= NGX_HTTP_CHARSET_VAR)
|
|
1575 {
|
|
1576 return NGX_CONF_OK;
|
|
1577 }
|
|
1578
|
78
|
1579 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
1580 ngx_http_charset_filter_module);
|
|
1581 recode = mcf->recodes.elts;
|
|
1582 for (i = 0; i < mcf->recodes.nelts; i++) {
|
|
1583 if (conf->source_charset == recode[i].src
|
|
1584 && conf->charset == recode[i].dst)
|
|
1585 {
|
|
1586 return NGX_CONF_OK;
|
|
1587 }
|
50
|
1588 }
|
|
1589
|
78
|
1590 recode = ngx_array_push(&mcf->recodes);
|
|
1591 if (recode == NULL) {
|
50
|
1592 return NGX_CONF_ERROR;
|
|
1593 }
|
|
1594
|
78
|
1595 recode->src = conf->source_charset;
|
|
1596 recode->dst = conf->charset;
|
|
1597
|
50
|
1598 return NGX_CONF_OK;
|
|
1599 }
|
78
|
1600
|
|
1601
|
|
1602 static ngx_int_t
|
|
1603 ngx_http_charset_postconfiguration(ngx_conf_t *cf)
|
|
1604 {
|
184
|
1605 u_char **src, **dst;
|
78
|
1606 ngx_int_t c;
|
|
1607 ngx_uint_t i, t;
|
|
1608 ngx_http_charset_t *charset;
|
|
1609 ngx_http_charset_recode_t *recode;
|
|
1610 ngx_http_charset_tables_t *tables;
|
|
1611 ngx_http_charset_main_conf_t *mcf;
|
|
1612
|
|
1613 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
1614 ngx_http_charset_filter_module);
|
|
1615
|
|
1616 recode = mcf->recodes.elts;
|
|
1617 tables = mcf->tables.elts;
|
|
1618 charset = mcf->charsets.elts;
|
|
1619
|
|
1620 for (i = 0; i < mcf->recodes.nelts; i++) {
|
|
1621
|
|
1622 c = recode[i].src;
|
|
1623
|
|
1624 for (t = 0; t < mcf->tables.nelts; t++) {
|
|
1625
|
|
1626 if (c == tables[t].src && recode[i].dst == tables[t].dst) {
|
|
1627 goto next;
|
|
1628 }
|
|
1629
|
|
1630 if (c == tables[t].dst && recode[i].dst == tables[t].src) {
|
|
1631 goto next;
|
|
1632 }
|
|
1633 }
|
|
1634
|
|
1635 ngx_log_error(NGX_LOG_EMERG, cf->log, 0,
|
342
|
1636 "no \"charset_map\" between the charsets \"%V\" and \"%V\"",
|
|
1637 &charset[c].name, &charset[recode[i].dst].name);
|
78
|
1638 return NGX_ERROR;
|
|
1639
|
|
1640 next:
|
|
1641 continue;
|
|
1642 }
|
|
1643
|
184
|
1644
|
|
1645 for (t = 0; t < mcf->tables.nelts; t++) {
|
|
1646
|
|
1647 src = charset[tables[t].src].tables;
|
|
1648
|
|
1649 if (src == NULL) {
|
|
1650 src = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
|
|
1651 if (src == NULL) {
|
|
1652 return NGX_ERROR;
|
|
1653 }
|
|
1654
|
|
1655 charset[tables[t].src].tables = src;
|
|
1656 }
|
|
1657
|
|
1658 dst = charset[tables[t].dst].tables;
|
|
1659
|
|
1660 if (dst == NULL) {
|
|
1661 dst = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
|
|
1662 if (dst == NULL) {
|
|
1663 return NGX_ERROR;
|
|
1664 }
|
|
1665
|
|
1666 charset[tables[t].dst].tables = dst;
|
|
1667 }
|
|
1668
|
|
1669 src[tables[t].dst] = tables[t].src2dst;
|
|
1670 dst[tables[t].src] = tables[t].dst2src;
|
|
1671 }
|
|
1672
|
230
|
1673 ngx_http_next_header_filter = ngx_http_top_header_filter;
|
|
1674 ngx_http_top_header_filter = ngx_http_charset_header_filter;
|
|
1675
|
|
1676 ngx_http_next_body_filter = ngx_http_top_body_filter;
|
|
1677 ngx_http_top_body_filter = ngx_http_charset_body_filter;
|
|
1678
|
78
|
1679 return NGX_OK;
|
|
1680 }
|