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