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