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