[mod_deflate] skip deflate if loadavg too high (fixes #1505)
[lighttpd.git] / src / mod_compress.c
blobbbd3e71a3f6cf7ec82ae6518488401a249f79524
1 #include "first.h"
3 #include "base.h"
4 #include "log.h"
5 #include "buffer.h"
6 #include "response.h"
7 #include "stat_cache.h"
9 #include "plugin.h"
11 #include "crc32.h"
12 #include "etag.h"
14 #include <sys/types.h>
15 #include <sys/stat.h>
17 #include <assert.h>
18 #include <fcntl.h>
19 #include <unistd.h>
20 #include <ctype.h>
21 #include <stdlib.h>
22 #include <string.h>
23 #include <strings.h>
24 #include <errno.h>
25 #include <time.h>
27 #if defined HAVE_ZLIB_H && defined HAVE_LIBZ
28 # define USE_ZLIB
29 # include <zlib.h>
30 #endif
32 #if defined HAVE_BZLIB_H && defined HAVE_LIBBZ2
33 # define USE_BZ2LIB
34 /* we don't need stdio interface */
35 # define BZ_NO_STDIO
36 # include <bzlib.h>
37 #endif
39 #if defined HAVE_SYS_MMAN_H && defined HAVE_MMAP && defined ENABLE_MMAP
40 #define USE_MMAP
42 #include "sys-mmap.h"
43 #include <setjmp.h>
44 #include <signal.h>
46 static volatile int sigbus_jmp_valid;
47 static sigjmp_buf sigbus_jmp;
49 static void sigbus_handler(int sig) {
50 UNUSED(sig);
51 if (sigbus_jmp_valid) siglongjmp(sigbus_jmp, 1);
52 log_failed_assert(__FILE__, __LINE__, "SIGBUS");
54 #endif
56 /* request: accept-encoding */
57 #define HTTP_ACCEPT_ENCODING_IDENTITY BV(0)
58 #define HTTP_ACCEPT_ENCODING_GZIP BV(1)
59 #define HTTP_ACCEPT_ENCODING_DEFLATE BV(2)
60 #define HTTP_ACCEPT_ENCODING_COMPRESS BV(3)
61 #define HTTP_ACCEPT_ENCODING_BZIP2 BV(4)
62 #define HTTP_ACCEPT_ENCODING_X_GZIP BV(5)
63 #define HTTP_ACCEPT_ENCODING_X_BZIP2 BV(6)
65 #ifdef __WIN32
66 # define mkdir(x,y) mkdir(x)
67 #endif
69 typedef struct {
70 buffer *compress_cache_dir;
71 array *compress;
72 off_t compress_max_filesize; /** max filesize in kb */
73 int allowed_encodings;
74 double max_loadavg;
75 } plugin_config;
77 typedef struct {
78 PLUGIN_DATA;
79 buffer *ofn;
80 buffer *b;
82 plugin_config **config_storage;
83 plugin_config conf;
84 } plugin_data;
86 INIT_FUNC(mod_compress_init) {
87 plugin_data *p;
89 p = calloc(1, sizeof(*p));
91 p->ofn = buffer_init();
92 p->b = buffer_init();
94 return p;
97 FREE_FUNC(mod_compress_free) {
98 plugin_data *p = p_d;
100 UNUSED(srv);
102 if (!p) return HANDLER_GO_ON;
104 buffer_free(p->ofn);
105 buffer_free(p->b);
107 if (p->config_storage) {
108 size_t i;
109 for (i = 0; i < srv->config_context->used; i++) {
110 plugin_config *s = p->config_storage[i];
112 if (NULL == s) continue;
114 array_free(s->compress);
115 buffer_free(s->compress_cache_dir);
117 free(s);
119 free(p->config_storage);
123 free(p);
125 return HANDLER_GO_ON;
128 /* 0 on success, -1 for error */
129 static int mkdir_recursive(char *dir) {
130 char *p = dir;
132 if (!dir || !dir[0])
133 return 0;
135 while ((p = strchr(p + 1, '/')) != NULL) {
137 *p = '\0';
138 if ((mkdir(dir, 0700) != 0) && (errno != EEXIST)) {
139 *p = '/';
140 return -1;
143 *p++ = '/';
144 if (!*p) return 0; /* Ignore trailing slash */
147 return (mkdir(dir, 0700) != 0) && (errno != EEXIST) ? -1 : 0;
150 /* 0 on success, -1 for error */
151 static int mkdir_for_file(char *filename) {
152 char *p = filename;
154 if (!filename || !filename[0])
155 return -1;
157 while ((p = strchr(p + 1, '/')) != NULL) {
159 *p = '\0';
160 if ((mkdir(filename, 0700) != 0) && (errno != EEXIST)) {
161 *p = '/';
162 return -1;
165 *p++ = '/';
166 if (!*p) return -1; /* Unexpected trailing slash in filename */
169 return 0;
172 SETDEFAULTS_FUNC(mod_compress_setdefaults) {
173 plugin_data *p = p_d;
174 size_t i = 0;
176 config_values_t cv[] = {
177 { "compress.cache-dir", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION },
178 { "compress.filetype", NULL, T_CONFIG_ARRAY, T_CONFIG_SCOPE_CONNECTION },
179 { "compress.max-filesize", NULL, T_CONFIG_SHORT, T_CONFIG_SCOPE_CONNECTION },
180 { "compress.allowed-encodings", NULL, T_CONFIG_ARRAY, T_CONFIG_SCOPE_CONNECTION },
181 { "compress.max-loadavg", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION },
182 { NULL, NULL, T_CONFIG_UNSET, T_CONFIG_SCOPE_UNSET }
185 p->config_storage = calloc(1, srv->config_context->used * sizeof(plugin_config *));
187 for (i = 0; i < srv->config_context->used; i++) {
188 data_config const* config = (data_config const*)srv->config_context->data[i];
189 plugin_config *s;
190 array *encodings_arr = array_init();
192 s = calloc(1, sizeof(plugin_config));
193 s->compress_cache_dir = buffer_init();
194 s->compress = array_init();
195 s->compress_max_filesize = 0;
196 s->allowed_encodings = 0;
197 s->max_loadavg = 0.0;
199 cv[0].destination = s->compress_cache_dir;
200 cv[1].destination = s->compress;
201 cv[2].destination = &(s->compress_max_filesize);
202 cv[3].destination = encodings_arr; /* temp array for allowed encodings list */
203 cv[4].destination = srv->tmp_buf;
204 buffer_string_set_length(srv->tmp_buf, 0);
206 p->config_storage[i] = s;
208 if (0 != config_insert_values_global(srv, config->value, cv, i == 0 ? T_CONFIG_SCOPE_SERVER : T_CONFIG_SCOPE_CONNECTION)) {
209 return HANDLER_ERROR;
212 if (!buffer_string_is_empty(srv->tmp_buf)) {
213 s->max_loadavg = strtod(srv->tmp_buf->ptr, NULL);
216 if (encodings_arr->used) {
217 size_t j = 0;
218 for (j = 0; j < encodings_arr->used; j++) {
219 #if defined(USE_ZLIB) || defined(USE_BZ2LIB)
220 data_string *ds = (data_string *)encodings_arr->data[j];
221 #endif
222 #ifdef USE_ZLIB
223 if (NULL != strstr(ds->value->ptr, "gzip"))
224 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_GZIP | HTTP_ACCEPT_ENCODING_X_GZIP;
225 if (NULL != strstr(ds->value->ptr, "x-gzip"))
226 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_X_GZIP;
227 if (NULL != strstr(ds->value->ptr, "deflate"))
228 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_DEFLATE;
230 if (NULL != strstr(ds->value->ptr, "compress"))
231 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_COMPRESS;
233 #endif
234 #ifdef USE_BZ2LIB
235 if (NULL != strstr(ds->value->ptr, "bzip2"))
236 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_BZIP2 | HTTP_ACCEPT_ENCODING_X_BZIP2;
237 if (NULL != strstr(ds->value->ptr, "x-bzip2"))
238 s->allowed_encodings |= HTTP_ACCEPT_ENCODING_X_BZIP2;
239 #endif
241 } else {
242 /* default encodings */
243 s->allowed_encodings = 0
244 #ifdef USE_ZLIB
245 | HTTP_ACCEPT_ENCODING_GZIP | HTTP_ACCEPT_ENCODING_X_GZIP | HTTP_ACCEPT_ENCODING_DEFLATE
246 #endif
247 #ifdef USE_BZ2LIB
248 | HTTP_ACCEPT_ENCODING_BZIP2 | HTTP_ACCEPT_ENCODING_X_BZIP2
249 #endif
253 array_free(encodings_arr);
255 if (!buffer_string_is_empty(s->compress_cache_dir)) {
256 struct stat st;
257 mkdir_recursive(s->compress_cache_dir->ptr);
259 if (0 != stat(s->compress_cache_dir->ptr, &st)) {
260 log_error_write(srv, __FILE__, __LINE__, "sbs", "can't stat compress.cache-dir",
261 s->compress_cache_dir, strerror(errno));
263 return HANDLER_ERROR;
268 return HANDLER_GO_ON;
272 #ifdef USE_ZLIB
273 static int deflate_file_to_buffer_gzip(server *srv, connection *con, plugin_data *p, char *start, off_t st_size, time_t mtime) {
274 unsigned char *c;
275 unsigned long crc;
276 z_stream z;
277 size_t outlen;
279 UNUSED(srv);
280 UNUSED(con);
282 z.zalloc = Z_NULL;
283 z.zfree = Z_NULL;
284 z.opaque = Z_NULL;
286 if (Z_OK != deflateInit2(&z,
287 Z_DEFAULT_COMPRESSION,
288 Z_DEFLATED,
289 -MAX_WBITS, /* supress zlib-header */
291 Z_DEFAULT_STRATEGY)) {
292 return -1;
295 z.next_in = (unsigned char *)start;
296 z.avail_in = st_size;
297 z.total_in = 0;
300 buffer_string_prepare_copy(p->b, (z.avail_in * 1.1) + 12 + 18);
302 /* write gzip header */
304 c = (unsigned char *)p->b->ptr;
305 c[0] = 0x1f;
306 c[1] = 0x8b;
307 c[2] = Z_DEFLATED;
308 c[3] = 0; /* options */
309 c[4] = (mtime >> 0) & 0xff;
310 c[5] = (mtime >> 8) & 0xff;
311 c[6] = (mtime >> 16) & 0xff;
312 c[7] = (mtime >> 24) & 0xff;
313 c[8] = 0x00; /* extra flags */
314 c[9] = 0x03; /* UNIX */
316 outlen = 10;
317 z.next_out = (unsigned char *)p->b->ptr + outlen;
318 z.avail_out = p->b->size - outlen - 9;
319 z.total_out = 0;
321 if (Z_STREAM_END != deflate(&z, Z_FINISH)) {
322 deflateEnd(&z);
323 return -1;
326 /* trailer */
327 outlen += z.total_out;
329 crc = generate_crc32c(start, st_size);
331 c = (unsigned char *)p->b->ptr + outlen;
333 c[0] = (crc >> 0) & 0xff;
334 c[1] = (crc >> 8) & 0xff;
335 c[2] = (crc >> 16) & 0xff;
336 c[3] = (crc >> 24) & 0xff;
337 c[4] = (z.total_in >> 0) & 0xff;
338 c[5] = (z.total_in >> 8) & 0xff;
339 c[6] = (z.total_in >> 16) & 0xff;
340 c[7] = (z.total_in >> 24) & 0xff;
341 outlen += 8;
342 buffer_commit(p->b, outlen);
344 if (Z_OK != deflateEnd(&z)) {
345 return -1;
348 return 0;
351 static int deflate_file_to_buffer_deflate(server *srv, connection *con, plugin_data *p, unsigned char *start, off_t st_size) {
352 z_stream z;
354 UNUSED(srv);
355 UNUSED(con);
357 z.zalloc = Z_NULL;
358 z.zfree = Z_NULL;
359 z.opaque = Z_NULL;
361 if (Z_OK != deflateInit2(&z,
362 Z_DEFAULT_COMPRESSION,
363 Z_DEFLATED,
364 -MAX_WBITS, /* supress zlib-header */
366 Z_DEFAULT_STRATEGY)) {
367 return -1;
370 z.next_in = start;
371 z.avail_in = st_size;
372 z.total_in = 0;
374 buffer_string_prepare_copy(p->b, (z.avail_in * 1.1) + 12);
376 z.next_out = (unsigned char *)p->b->ptr;
377 z.avail_out = p->b->size - 1;
378 z.total_out = 0;
380 if (Z_STREAM_END != deflate(&z, Z_FINISH)) {
381 deflateEnd(&z);
382 return -1;
385 if (Z_OK != deflateEnd(&z)) {
386 return -1;
389 /* trailer */
390 buffer_commit(p->b, z.total_out);
392 return 0;
395 #endif
397 #ifdef USE_BZ2LIB
398 static int deflate_file_to_buffer_bzip2(server *srv, connection *con, plugin_data *p, unsigned char *start, off_t st_size) {
399 bz_stream bz;
401 UNUSED(srv);
402 UNUSED(con);
404 bz.bzalloc = NULL;
405 bz.bzfree = NULL;
406 bz.opaque = NULL;
408 if (BZ_OK != BZ2_bzCompressInit(&bz,
409 9, /* blocksize = 900k */
410 0, /* no output */
411 0)) { /* workFactor: default */
412 return -1;
415 bz.next_in = (char *)start;
416 bz.avail_in = st_size;
417 bz.total_in_lo32 = 0;
418 bz.total_in_hi32 = 0;
420 buffer_string_prepare_copy(p->b, (bz.avail_in * 1.1) + 12);
422 bz.next_out = p->b->ptr;
423 bz.avail_out = p->b->size - 1;
424 bz.total_out_lo32 = 0;
425 bz.total_out_hi32 = 0;
427 if (BZ_STREAM_END != BZ2_bzCompress(&bz, BZ_FINISH)) {
428 BZ2_bzCompressEnd(&bz);
429 return -1;
432 if (BZ_OK != BZ2_bzCompressEnd(&bz)) {
433 return -1;
436 /* file is too large for now */
437 if (bz.total_out_hi32) return -1;
439 /* trailer */
440 buffer_commit(p->b, bz.total_out_lo32);
442 return 0;
444 #endif
446 static void mod_compress_note_ratio(server *srv, connection *con, off_t in, off_t out) {
447 /* store compression ratio in con->environment
448 * for possible logging by mod_accesslog
449 * (late in response handling, so not seen by most other modules) */
450 /*(should be called only at end of successful response compression)*/
451 char ratio[LI_ITOSTRING_LENGTH];
452 if (0 == in) return;
453 li_itostrn(ratio, sizeof(ratio), out * 100 / in);
454 array_set_key_value(con->environment,
455 CONST_STR_LEN("ratio"),
456 ratio, strlen(ratio));
457 UNUSED(srv);
460 static int deflate_file_to_file(server *srv, connection *con, plugin_data *p, buffer *fn, stat_cache_entry *sce, int type) {
461 int ifd, ofd;
462 int ret;
463 #ifdef USE_MMAP
464 volatile int mapped = 0;/* quiet warning: might be clobbered by 'longjmp' */
465 #endif
466 void *start;
467 const char *filename = fn->ptr;
468 stat_cache_entry *sce_ofn;
469 ssize_t r;
471 /* overflow */
472 if ((off_t)(sce->st.st_size * 1.1) < sce->st.st_size) return -1;
474 /* don't mmap files > 128Mb
476 * we could use a sliding window, but currently there is no need for it
479 if (sce->st.st_size > 128 * 1024 * 1024) return -1;
481 buffer_reset(p->ofn);
482 buffer_copy_buffer(p->ofn, p->conf.compress_cache_dir);
483 buffer_append_slash(p->ofn);
485 if (0 == strncmp(con->physical.path->ptr, con->physical.doc_root->ptr, buffer_string_length(con->physical.doc_root))) {
486 buffer_append_string(p->ofn, con->physical.path->ptr + buffer_string_length(con->physical.doc_root));
487 } else {
488 buffer_append_string_buffer(p->ofn, con->uri.path);
491 switch(type) {
492 case HTTP_ACCEPT_ENCODING_GZIP:
493 case HTTP_ACCEPT_ENCODING_X_GZIP:
494 buffer_append_string_len(p->ofn, CONST_STR_LEN("-gzip-"));
495 break;
496 case HTTP_ACCEPT_ENCODING_DEFLATE:
497 buffer_append_string_len(p->ofn, CONST_STR_LEN("-deflate-"));
498 break;
499 case HTTP_ACCEPT_ENCODING_BZIP2:
500 case HTTP_ACCEPT_ENCODING_X_BZIP2:
501 buffer_append_string_len(p->ofn, CONST_STR_LEN("-bzip2-"));
502 break;
503 default:
504 log_error_write(srv, __FILE__, __LINE__, "sd", "unknown compression type", type);
505 return -1;
508 buffer_append_string_buffer(p->ofn, sce->etag);
510 if (HANDLER_ERROR != stat_cache_get_entry(srv, con, p->ofn, &sce_ofn)) {
511 if (0 == sce->st.st_size) return -1; /* cache file being created */
512 /* cache-entry exists */
513 #if 0
514 log_error_write(srv, __FILE__, __LINE__, "bs", p->ofn, "compress-cache hit");
515 #endif
516 mod_compress_note_ratio(srv, con, sce->st.st_size, sce_ofn->st.st_size);
517 buffer_copy_buffer(con->physical.path, p->ofn);
518 return 0;
521 if (0.0 < p->conf.max_loadavg && p->conf.max_loadavg < srv->srvconf.loadavg[0]) {
522 return -1;
525 if (-1 == mkdir_for_file(p->ofn->ptr)) {
526 log_error_write(srv, __FILE__, __LINE__, "sb", "couldn't create directory for file", p->ofn);
527 return -1;
530 if (-1 == (ofd = open(p->ofn->ptr, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0600))) {
531 if (errno == EEXIST) {
532 return -1; /* cache file being created */
535 log_error_write(srv, __FILE__, __LINE__, "sbss", "creating cachefile", p->ofn, "failed", strerror(errno));
537 return -1;
539 #if 0
540 log_error_write(srv, __FILE__, __LINE__, "bs", p->ofn, "compress-cache miss");
541 #endif
542 if (-1 == (ifd = open(filename, O_RDONLY | O_BINARY))) {
543 log_error_write(srv, __FILE__, __LINE__, "sbss", "opening plain-file", fn, "failed", strerror(errno));
545 close(ofd);
547 /* Remove the incomplete cache file, so that later hits aren't served from it */
548 if (-1 == unlink(p->ofn->ptr)) {
549 log_error_write(srv, __FILE__, __LINE__, "sbss", "unlinking incomplete cachefile", p->ofn, "failed:", strerror(errno));
552 return -1;
555 #ifdef USE_MMAP
556 if (MAP_FAILED != (start = mmap(NULL, sce->st.st_size, PROT_READ, MAP_SHARED, ifd, 0))) {
557 mapped = 1;
558 signal(SIGBUS, sigbus_handler);
559 sigbus_jmp_valid = 1;
560 if (0 != sigsetjmp(sigbus_jmp, 1)) {
561 sigbus_jmp_valid = 0;
563 log_error_write(srv, __FILE__, __LINE__, "sbd", "SIGBUS in mmap:",
564 fn, ifd);
566 munmap(start, sce->st.st_size);
567 close(ofd);
568 close(ifd);
570 /* Remove the incomplete cache file, so that later hits aren't served from it */
571 if (-1 == unlink(p->ofn->ptr)) {
572 log_error_write(srv, __FILE__, __LINE__, "sbss", "unlinking incomplete cachefile", p->ofn, "failed:", strerror(errno));
575 return -1;
577 } else
578 #endif /* FIXME: might attempt to read very large file completely into memory; see compress.max-filesize config option */
579 if (NULL == (start = malloc(sce->st.st_size)) || sce->st.st_size != read(ifd, start, sce->st.st_size)) {
580 log_error_write(srv, __FILE__, __LINE__, "sbss", "reading", fn, "failed", strerror(errno));
582 close(ofd);
583 close(ifd);
584 free(start);
586 /* Remove the incomplete cache file, so that later hits aren't served from it */
587 if (-1 == unlink(p->ofn->ptr)) {
588 log_error_write(srv, __FILE__, __LINE__, "sbss", "unlinking incomplete cachefile", p->ofn, "failed:", strerror(errno));
591 return -1;
594 ret = -1;
595 switch(type) {
596 #ifdef USE_ZLIB
597 case HTTP_ACCEPT_ENCODING_GZIP:
598 case HTTP_ACCEPT_ENCODING_X_GZIP:
599 ret = deflate_file_to_buffer_gzip(srv, con, p, start, sce->st.st_size, sce->st.st_mtime);
600 break;
601 case HTTP_ACCEPT_ENCODING_DEFLATE:
602 ret = deflate_file_to_buffer_deflate(srv, con, p, start, sce->st.st_size);
603 break;
604 #endif
605 #ifdef USE_BZ2LIB
606 case HTTP_ACCEPT_ENCODING_BZIP2:
607 case HTTP_ACCEPT_ENCODING_X_BZIP2:
608 ret = deflate_file_to_buffer_bzip2(srv, con, p, start, sce->st.st_size);
609 break;
610 #endif
613 if (ret == 0) {
614 r = write(ofd, CONST_BUF_LEN(p->b));
615 if (-1 == r) {
616 log_error_write(srv, __FILE__, __LINE__, "sbss", "writing cachefile", p->ofn, "failed:", strerror(errno));
617 ret = -1;
618 } else if ((size_t)r != buffer_string_length(p->b)) {
619 log_error_write(srv, __FILE__, __LINE__, "sbs", "writing cachefile", p->ofn, "failed: not enough bytes written");
620 ret = -1;
624 #ifdef USE_MMAP
625 if (mapped) {
626 sigbus_jmp_valid = 0;
627 munmap(start, sce->st.st_size);
628 } else
629 #endif
630 free(start);
632 close(ifd);
634 if (0 != close(ofd) || ret != 0) {
635 if (0 == ret) {
636 log_error_write(srv, __FILE__, __LINE__, "sbss", "writing cachefile", p->ofn, "failed:", strerror(errno));
639 /* Remove the incomplete cache file, so that later hits aren't served from it */
640 if (-1 == unlink(p->ofn->ptr)) {
641 log_error_write(srv, __FILE__, __LINE__, "sbss", "unlinking incomplete cachefile", p->ofn, "failed:", strerror(errno));
644 return -1;
647 buffer_copy_buffer(con->physical.path, p->ofn);
648 mod_compress_note_ratio(srv, con, sce->st.st_size,
649 (off_t)buffer_string_length(p->b));
651 return 0;
654 static int deflate_file_to_buffer(server *srv, connection *con, plugin_data *p, buffer *fn, stat_cache_entry *sce, int type) {
655 int ifd;
656 int ret = -1;
657 #ifdef USE_MMAP
658 volatile int mapped = 0;/* quiet warning: might be clobbered by 'longjmp' */
659 #endif
660 void *start;
662 /* overflow */
663 if ((off_t)(sce->st.st_size * 1.1) < sce->st.st_size) return -1;
665 /* don't mmap files > 128M
667 * we could use a sliding window, but currently there is no need for it
670 if (sce->st.st_size > 128 * 1024 * 1024) return -1;
672 if (0.0 < p->conf.max_loadavg && p->conf.max_loadavg < srv->srvconf.loadavg[0]) {
673 return -1;
676 if (-1 == (ifd = open(fn->ptr, O_RDONLY | O_BINARY))) {
677 log_error_write(srv, __FILE__, __LINE__, "sbss", "opening plain-file", fn, "failed", strerror(errno));
679 return -1;
682 #ifdef USE_MMAP
683 if (MAP_FAILED != (start = mmap(NULL, sce->st.st_size, PROT_READ, MAP_SHARED, ifd, 0))) {
684 mapped = 1;
685 signal(SIGBUS, sigbus_handler);
686 sigbus_jmp_valid = 1;
687 if (0 != sigsetjmp(sigbus_jmp, 1)) {
688 sigbus_jmp_valid = 0;
690 log_error_write(srv, __FILE__, __LINE__, "sbd", "SIGBUS in mmap:",
691 fn, ifd);
693 munmap(start, sce->st.st_size);
694 close(ifd);
695 return -1;
697 } else
698 #endif /* FIXME: might attempt to read very large file completely into memory; see compress.max-filesize config option */
699 if (NULL == (start = malloc(sce->st.st_size)) || sce->st.st_size != read(ifd, start, sce->st.st_size)) {
700 log_error_write(srv, __FILE__, __LINE__, "sbss", "reading", fn, "failed", strerror(errno));
702 close(ifd);
703 free(start);
704 return -1;
707 switch(type) {
708 #ifdef USE_ZLIB
709 case HTTP_ACCEPT_ENCODING_GZIP:
710 case HTTP_ACCEPT_ENCODING_X_GZIP:
711 ret = deflate_file_to_buffer_gzip(srv, con, p, start, sce->st.st_size, sce->st.st_mtime);
712 break;
713 case HTTP_ACCEPT_ENCODING_DEFLATE:
714 ret = deflate_file_to_buffer_deflate(srv, con, p, start, sce->st.st_size);
715 break;
716 #endif
717 #ifdef USE_BZ2LIB
718 case HTTP_ACCEPT_ENCODING_BZIP2:
719 case HTTP_ACCEPT_ENCODING_X_BZIP2:
720 ret = deflate_file_to_buffer_bzip2(srv, con, p, start, sce->st.st_size);
721 break;
722 #endif
723 default:
724 ret = -1;
725 break;
728 #ifdef USE_MMAP
729 if (mapped) {
730 sigbus_jmp_valid = 0;
731 munmap(start, sce->st.st_size);
732 } else
733 #endif
734 free(start);
736 close(ifd);
738 if (ret != 0) return -1;
740 mod_compress_note_ratio(srv, con, sce->st.st_size,
741 (off_t)buffer_string_length(p->b));
742 chunkqueue_reset(con->write_queue);
743 chunkqueue_append_buffer(con->write_queue, p->b);
745 buffer_reset(con->physical.path);
747 con->file_finished = 1;
748 con->file_started = 1;
750 return 0;
754 #define PATCH(x) \
755 p->conf.x = s->x;
756 static int mod_compress_patch_connection(server *srv, connection *con, plugin_data *p) {
757 size_t i, j;
758 plugin_config *s = p->config_storage[0];
760 PATCH(compress_cache_dir);
761 PATCH(compress);
762 PATCH(compress_max_filesize);
763 PATCH(allowed_encodings);
764 PATCH(max_loadavg);
766 /* skip the first, the global context */
767 for (i = 1; i < srv->config_context->used; i++) {
768 data_config *dc = (data_config *)srv->config_context->data[i];
769 s = p->config_storage[i];
771 /* condition didn't match */
772 if (!config_check_cond(srv, con, dc)) continue;
774 /* merge config */
775 for (j = 0; j < dc->value->used; j++) {
776 data_unset *du = dc->value->data[j];
778 if (buffer_is_equal_string(du->key, CONST_STR_LEN("compress.cache-dir"))) {
779 PATCH(compress_cache_dir);
780 } else if (buffer_is_equal_string(du->key, CONST_STR_LEN("compress.filetype"))) {
781 PATCH(compress);
782 } else if (buffer_is_equal_string(du->key, CONST_STR_LEN("compress.max-filesize"))) {
783 PATCH(compress_max_filesize);
784 } else if (buffer_is_equal_string(du->key, CONST_STR_LEN("compress.allowed-encodings"))) {
785 PATCH(allowed_encodings);
786 } else if (buffer_is_equal_string(du->key, CONST_STR_LEN("compress.max-loadavg"))) {
787 PATCH(max_loadavg);
792 return 0;
794 #undef PATCH
796 static int mod_compress_contains_encoding(const char *headervalue, const char *encoding, size_t len) {
797 const char *m = headervalue;
798 do {
799 while (*m == ',' || *m == ' ' || *m == '\t') {
800 ++m;
802 if (0 == strncasecmp(m, encoding, len)) {
803 /*(not a full HTTP field parse: not parsing for q-values and not handling q=0)*/
804 m += len;
805 if (*m == '\0' || *m == ',' || *m == ';' || *m == ' ' || *m == '\t')
806 return 1;
807 } else if (*m != '\0') {
808 ++m;
810 } while ((m = strchr(m, ',')));
811 return 0;
814 PHYSICALPATH_FUNC(mod_compress_physical) {
815 plugin_data *p = p_d;
816 size_t m;
817 off_t max_fsize;
818 stat_cache_entry *sce = NULL;
819 buffer *mtime = NULL;
820 buffer *content_type;
822 if (con->mode != DIRECT || con->http_status) return HANDLER_GO_ON;
824 /* only GET and POST can get compressed */
825 if (con->request.http_method != HTTP_METHOD_GET &&
826 con->request.http_method != HTTP_METHOD_POST) {
827 return HANDLER_GO_ON;
830 if (buffer_string_is_empty(con->physical.path)) {
831 return HANDLER_GO_ON;
834 mod_compress_patch_connection(srv, con, p);
836 max_fsize = p->conf.compress_max_filesize;
838 if (con->conf.log_request_handling) {
839 log_error_write(srv, __FILE__, __LINE__, "s", "-- handling file as static file");
842 if (HANDLER_ERROR == stat_cache_get_entry(srv, con, con->physical.path, &sce)) {
843 con->http_status = 403;
845 log_error_write(srv, __FILE__, __LINE__, "sbsb",
846 "not a regular file:", con->uri.path,
847 "->", con->physical.path);
849 return HANDLER_FINISHED;
852 /* we only handle regular files */
853 #ifdef HAVE_LSTAT
854 if ((sce->is_symlink == 1) && !con->conf.follow_symlink) {
855 return HANDLER_GO_ON;
857 #endif
858 if (!S_ISREG(sce->st.st_mode)) {
859 return HANDLER_GO_ON;
862 /* don't compress files that are too large as we need to much time to handle them */
863 if (max_fsize && (sce->st.st_size >> 10) > max_fsize) return HANDLER_GO_ON;
865 /* don't try to compress files less than 128 bytes
867 * - extra overhead for compression
868 * - mmap() fails for st_size = 0 :)
870 if (sce->st.st_size < 128) return HANDLER_GO_ON;
872 /* check if mimetype is in compress-config */
873 content_type = NULL;
874 if (sce->content_type->ptr) {
875 char *c;
876 if ( (c = strchr(sce->content_type->ptr, ';')) != NULL) {
877 content_type = srv->tmp_buf;
878 buffer_copy_string_len(content_type, sce->content_type->ptr, c - sce->content_type->ptr);
882 for (m = 0; m < p->conf.compress->used; m++) {
883 data_string *compress_ds = (data_string *)p->conf.compress->data[m];
885 if (!compress_ds) {
886 log_error_write(srv, __FILE__, __LINE__, "sbb", "evil", con->physical.path, con->uri.path);
888 return HANDLER_GO_ON;
891 if (buffer_is_equal(compress_ds->value, sce->content_type)
892 || (content_type && buffer_is_equal(compress_ds->value, content_type))) {
893 /* mimetype found */
894 data_string *ds;
896 /* the response might change according to Accept-Encoding */
897 response_header_insert(srv, con, CONST_STR_LEN("Vary"), CONST_STR_LEN("Accept-Encoding"));
899 if (NULL != (ds = (data_string *)array_get_element(con->request.headers, "Accept-Encoding"))) {
900 int accept_encoding = 0;
901 char *value = ds->value->ptr;
902 int matched_encodings = 0;
903 int use_etag = sce->etag != NULL && sce->etag->ptr != NULL;
905 /* get client side support encodings */
906 #ifdef USE_ZLIB
907 if (mod_compress_contains_encoding(value, CONST_STR_LEN("gzip"))) accept_encoding |= HTTP_ACCEPT_ENCODING_GZIP;
908 if (mod_compress_contains_encoding(value, CONST_STR_LEN("x-gzip"))) accept_encoding |= HTTP_ACCEPT_ENCODING_X_GZIP;
909 if (mod_compress_contains_encoding(value, CONST_STR_LEN("deflate"))) accept_encoding |= HTTP_ACCEPT_ENCODING_DEFLATE;
910 if (mod_compress_contains_encoding(value, CONST_STR_LEN("compress"))) accept_encoding |= HTTP_ACCEPT_ENCODING_COMPRESS;
911 #endif
912 #ifdef USE_BZ2LIB
913 if (mod_compress_contains_encoding(value, CONST_STR_LEN("bzip2"))) accept_encoding |= HTTP_ACCEPT_ENCODING_BZIP2;
914 if (mod_compress_contains_encoding(value, CONST_STR_LEN("x-bzip2"))) accept_encoding |= HTTP_ACCEPT_ENCODING_X_BZIP2;
915 #endif
916 if (mod_compress_contains_encoding(value, CONST_STR_LEN("identity"))) accept_encoding |= HTTP_ACCEPT_ENCODING_IDENTITY;
918 /* find matching entries */
919 matched_encodings = accept_encoding & p->conf.allowed_encodings;
921 if (matched_encodings) {
922 static const char dflt_gzip[] = "gzip";
923 static const char dflt_x_gzip[] = "x-gzip";
924 static const char dflt_deflate[] = "deflate";
925 static const char dflt_bzip2[] = "bzip2";
926 static const char dflt_x_bzip2[] = "x-bzip2";
928 const char *compression_name = NULL;
929 int compression_type = 0;
931 mtime = strftime_cache_get(srv, sce->st.st_mtime);
933 /* try matching original etag of uncompressed version */
934 if (use_etag) {
935 etag_mutate(con->physical.etag, sce->etag);
936 if (HANDLER_FINISHED == http_response_handle_cachable(srv, con, mtime)) {
937 response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_BUF_LEN(sce->content_type));
938 response_header_overwrite(srv, con, CONST_STR_LEN("Last-Modified"), CONST_BUF_LEN(mtime));
939 response_header_overwrite(srv, con, CONST_STR_LEN("ETag"), CONST_BUF_LEN(con->physical.etag));
940 return HANDLER_FINISHED;
944 /* select best matching encoding */
945 if (matched_encodings & HTTP_ACCEPT_ENCODING_BZIP2) {
946 compression_type = HTTP_ACCEPT_ENCODING_BZIP2;
947 compression_name = dflt_bzip2;
948 } else if (matched_encodings & HTTP_ACCEPT_ENCODING_X_BZIP2) {
949 compression_type = HTTP_ACCEPT_ENCODING_X_BZIP2;
950 compression_name = dflt_x_bzip2;
951 } else if (matched_encodings & HTTP_ACCEPT_ENCODING_GZIP) {
952 compression_type = HTTP_ACCEPT_ENCODING_GZIP;
953 compression_name = dflt_gzip;
954 } else if (matched_encodings & HTTP_ACCEPT_ENCODING_X_GZIP) {
955 compression_type = HTTP_ACCEPT_ENCODING_X_GZIP;
956 compression_name = dflt_x_gzip;
957 } else {
958 force_assert(matched_encodings & HTTP_ACCEPT_ENCODING_DEFLATE);
959 compression_type = HTTP_ACCEPT_ENCODING_DEFLATE;
960 compression_name = dflt_deflate;
963 if (use_etag) {
964 /* try matching etag of compressed version */
965 buffer_copy_buffer(srv->tmp_buf, sce->etag);
966 buffer_append_string_len(srv->tmp_buf, CONST_STR_LEN("-"));
967 buffer_append_string(srv->tmp_buf, compression_name);
968 etag_mutate(con->physical.etag, srv->tmp_buf);
971 if (HANDLER_FINISHED == http_response_handle_cachable(srv, con, mtime)) {
972 response_header_overwrite(srv, con, CONST_STR_LEN("Content-Encoding"), compression_name, strlen(compression_name));
973 response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_BUF_LEN(sce->content_type));
974 response_header_overwrite(srv, con, CONST_STR_LEN("Last-Modified"), CONST_BUF_LEN(mtime));
975 if (use_etag) {
976 response_header_overwrite(srv, con, CONST_STR_LEN("ETag"), CONST_BUF_LEN(con->physical.etag));
978 return HANDLER_FINISHED;
981 /* deflate it */
982 if (use_etag && !buffer_string_is_empty(p->conf.compress_cache_dir)) {
983 if (0 != deflate_file_to_file(srv, con, p, con->physical.path, sce, compression_type))
984 return HANDLER_GO_ON;
985 } else {
986 if (0 != deflate_file_to_buffer(srv, con, p, con->physical.path, sce, compression_type))
987 return HANDLER_GO_ON;
989 response_header_overwrite(srv, con, CONST_STR_LEN("Content-Encoding"), compression_name, strlen(compression_name));
990 response_header_overwrite(srv, con, CONST_STR_LEN("Last-Modified"), CONST_BUF_LEN(mtime));
991 if (use_etag) {
992 response_header_overwrite(srv, con, CONST_STR_LEN("ETag"), CONST_BUF_LEN(con->physical.etag));
994 response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_BUF_LEN(sce->content_type));
995 /* let mod_staticfile handle the cached compressed files, physical path was modified */
996 return (use_etag && !buffer_string_is_empty(p->conf.compress_cache_dir)) ? HANDLER_GO_ON : HANDLER_FINISHED;
1002 return HANDLER_GO_ON;
1005 int mod_compress_plugin_init(plugin *p);
1006 int mod_compress_plugin_init(plugin *p) {
1007 p->version = LIGHTTPD_VERSION_ID;
1008 p->name = buffer_init_string("compress");
1010 p->init = mod_compress_init;
1011 p->set_defaults = mod_compress_setdefaults;
1012 p->handle_subrequest_start = mod_compress_physical;
1013 p->cleanup = mod_compress_free;
1015 p->data = NULL;
1017 return 0;