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