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