1 | /*************************************************************************** |
2 | * _ _ ____ _ |
3 | * Project ___| | | | _ \| | |
4 | * / __| | | | |_) | | |
5 | * | (__| |_| | _ <| |___ |
6 | * \___|\___/|_| \_\_____| |
7 | * |
8 | * Copyright (C) 1998 - 2021, Daniel Stenberg, <daniel@haxx.se>, et al. |
9 | * |
10 | * This software is licensed as described in the file COPYING, which |
11 | * you should have received as part of this distribution. The terms |
12 | * are also available at https://curl.se/docs/copyright.html. |
13 | * |
14 | * You may opt to use, copy, modify, merge, publish, distribute and/or sell |
15 | * copies of the Software, and permit persons to whom the Software is |
16 | * furnished to do so, under the terms of the COPYING file. |
17 | * |
18 | * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY |
19 | * KIND, either express or implied. |
20 | * |
21 | ***************************************************************************/ |
22 | |
23 | #include "curl_setup.h" |
24 | |
25 | #ifndef CURL_DISABLE_FILE |
26 | |
27 | #ifdef HAVE_NETINET_IN_H |
28 | #include <netinet/in.h> |
29 | #endif |
30 | #ifdef HAVE_NETDB_H |
31 | #include <netdb.h> |
32 | #endif |
33 | #ifdef HAVE_ARPA_INET_H |
34 | #include <arpa/inet.h> |
35 | #endif |
36 | #ifdef HAVE_NET_IF_H |
37 | #include <net/if.h> |
38 | #endif |
39 | #ifdef HAVE_SYS_IOCTL_H |
40 | #include <sys/ioctl.h> |
41 | #endif |
42 | |
43 | #ifdef HAVE_SYS_PARAM_H |
44 | #include <sys/param.h> |
45 | #endif |
46 | |
47 | #ifdef HAVE_FCNTL_H |
48 | #include <fcntl.h> |
49 | #endif |
50 | |
51 | #include "strtoofft.h" |
52 | #include "urldata.h" |
53 | #include <curl/curl.h> |
54 | #include "progress.h" |
55 | #include "sendf.h" |
56 | #include "escape.h" |
57 | #include "file.h" |
58 | #include "speedcheck.h" |
59 | #include "getinfo.h" |
60 | #include "transfer.h" |
61 | #include "url.h" |
62 | #include "parsedate.h" /* for the week day and month names */ |
63 | #include "warnless.h" |
64 | #include "curl_range.h" |
65 | /* The last 3 #include files should be in this order */ |
66 | #include "curl_printf.h" |
67 | #include "curl_memory.h" |
68 | #include "memdebug.h" |
69 | |
70 | #if defined(WIN32) || defined(MSDOS) || defined(__EMX__) |
71 | #define DOS_FILESYSTEM 1 |
72 | #endif |
73 | |
74 | #ifdef OPEN_NEEDS_ARG3 |
75 | # define open_readonly(p,f) open((p),(f),(0)) |
76 | #else |
77 | # define open_readonly(p,f) open((p),(f)) |
78 | #endif |
79 | |
80 | /* |
81 | * Forward declarations. |
82 | */ |
83 | |
84 | static CURLcode file_do(struct Curl_easy *data, bool *done); |
85 | static CURLcode file_done(struct Curl_easy *data, |
86 | CURLcode status, bool premature); |
87 | static CURLcode file_connect(struct Curl_easy *data, bool *done); |
88 | static CURLcode file_disconnect(struct Curl_easy *data, |
89 | struct connectdata *conn, |
90 | bool dead_connection); |
91 | static CURLcode file_setup_connection(struct Curl_easy *data, |
92 | struct connectdata *conn); |
93 | |
94 | /* |
95 | * FILE scheme handler. |
96 | */ |
97 | |
98 | const struct Curl_handler Curl_handler_file = { |
99 | "FILE" , /* scheme */ |
100 | file_setup_connection, /* setup_connection */ |
101 | file_do, /* do_it */ |
102 | file_done, /* done */ |
103 | ZERO_NULL, /* do_more */ |
104 | file_connect, /* connect_it */ |
105 | ZERO_NULL, /* connecting */ |
106 | ZERO_NULL, /* doing */ |
107 | ZERO_NULL, /* proto_getsock */ |
108 | ZERO_NULL, /* doing_getsock */ |
109 | ZERO_NULL, /* domore_getsock */ |
110 | ZERO_NULL, /* perform_getsock */ |
111 | file_disconnect, /* disconnect */ |
112 | ZERO_NULL, /* readwrite */ |
113 | ZERO_NULL, /* connection_check */ |
114 | ZERO_NULL, /* attach connection */ |
115 | 0, /* defport */ |
116 | CURLPROTO_FILE, /* protocol */ |
117 | CURLPROTO_FILE, /* family */ |
118 | PROTOPT_NONETWORK | PROTOPT_NOURLQUERY /* flags */ |
119 | }; |
120 | |
121 | |
122 | static CURLcode file_setup_connection(struct Curl_easy *data, |
123 | struct connectdata *conn) |
124 | { |
125 | (void)conn; |
126 | /* allocate the FILE specific struct */ |
127 | data->req.p.file = calloc(1, sizeof(struct FILEPROTO)); |
128 | if(!data->req.p.file) |
129 | return CURLE_OUT_OF_MEMORY; |
130 | |
131 | return CURLE_OK; |
132 | } |
133 | |
134 | /* |
135 | * file_connect() gets called from Curl_protocol_connect() to allow us to |
136 | * do protocol-specific actions at connect-time. We emulate a |
137 | * connect-then-transfer protocol and "connect" to the file here |
138 | */ |
139 | static CURLcode file_connect(struct Curl_easy *data, bool *done) |
140 | { |
141 | char *real_path; |
142 | struct FILEPROTO *file = data->req.p.file; |
143 | int fd; |
144 | #ifdef DOS_FILESYSTEM |
145 | size_t i; |
146 | char *actual_path; |
147 | #endif |
148 | size_t real_path_len; |
149 | |
150 | CURLcode result = Curl_urldecode(data, data->state.up.path, 0, &real_path, |
151 | &real_path_len, REJECT_ZERO); |
152 | if(result) |
153 | return result; |
154 | |
155 | #ifdef DOS_FILESYSTEM |
156 | /* If the first character is a slash, and there's |
157 | something that looks like a drive at the beginning of |
158 | the path, skip the slash. If we remove the initial |
159 | slash in all cases, paths without drive letters end up |
160 | relative to the current directory which isn't how |
161 | browsers work. |
162 | |
163 | Some browsers accept | instead of : as the drive letter |
164 | separator, so we do too. |
165 | |
166 | On other platforms, we need the slash to indicate an |
167 | absolute pathname. On Windows, absolute paths start |
168 | with a drive letter. |
169 | */ |
170 | actual_path = real_path; |
171 | if((actual_path[0] == '/') && |
172 | actual_path[1] && |
173 | (actual_path[2] == ':' || actual_path[2] == '|')) { |
174 | actual_path[2] = ':'; |
175 | actual_path++; |
176 | real_path_len--; |
177 | } |
178 | |
179 | /* change path separators from '/' to '\\' for DOS, Windows and OS/2 */ |
180 | for(i = 0; i < real_path_len; ++i) |
181 | if(actual_path[i] == '/') |
182 | actual_path[i] = '\\'; |
183 | else if(!actual_path[i]) { /* binary zero */ |
184 | Curl_safefree(real_path); |
185 | return CURLE_URL_MALFORMAT; |
186 | } |
187 | |
188 | fd = open_readonly(actual_path, O_RDONLY|O_BINARY); |
189 | file->path = actual_path; |
190 | #else |
191 | if(memchr(real_path, 0, real_path_len)) { |
192 | /* binary zeroes indicate foul play */ |
193 | Curl_safefree(real_path); |
194 | return CURLE_URL_MALFORMAT; |
195 | } |
196 | |
197 | fd = open_readonly(real_path, O_RDONLY); |
198 | file->path = real_path; |
199 | #endif |
200 | file->freepath = real_path; /* free this when done */ |
201 | |
202 | file->fd = fd; |
203 | if(!data->set.upload && (fd == -1)) { |
204 | failf(data, "Couldn't open file %s" , data->state.up.path); |
205 | file_done(data, CURLE_FILE_COULDNT_READ_FILE, FALSE); |
206 | return CURLE_FILE_COULDNT_READ_FILE; |
207 | } |
208 | *done = TRUE; |
209 | |
210 | return CURLE_OK; |
211 | } |
212 | |
213 | static CURLcode file_done(struct Curl_easy *data, |
214 | CURLcode status, bool premature) |
215 | { |
216 | struct FILEPROTO *file = data->req.p.file; |
217 | (void)status; /* not used */ |
218 | (void)premature; /* not used */ |
219 | |
220 | if(file) { |
221 | Curl_safefree(file->freepath); |
222 | file->path = NULL; |
223 | if(file->fd != -1) |
224 | close(file->fd); |
225 | file->fd = -1; |
226 | } |
227 | |
228 | return CURLE_OK; |
229 | } |
230 | |
231 | static CURLcode file_disconnect(struct Curl_easy *data, |
232 | struct connectdata *conn, |
233 | bool dead_connection) |
234 | { |
235 | (void)dead_connection; /* not used */ |
236 | (void)conn; |
237 | return file_done(data, 0, 0); |
238 | } |
239 | |
240 | #ifdef DOS_FILESYSTEM |
241 | #define DIRSEP '\\' |
242 | #else |
243 | #define DIRSEP '/' |
244 | #endif |
245 | |
246 | static CURLcode file_upload(struct Curl_easy *data) |
247 | { |
248 | struct FILEPROTO *file = data->req.p.file; |
249 | const char *dir = strchr(file->path, DIRSEP); |
250 | int fd; |
251 | int mode; |
252 | CURLcode result = CURLE_OK; |
253 | char *buf = data->state.buffer; |
254 | curl_off_t bytecount = 0; |
255 | struct_stat file_stat; |
256 | const char *buf2; |
257 | |
258 | /* |
259 | * Since FILE: doesn't do the full init, we need to provide some extra |
260 | * assignments here. |
261 | */ |
262 | data->req.upload_fromhere = buf; |
263 | |
264 | if(!dir) |
265 | return CURLE_FILE_COULDNT_READ_FILE; /* fix: better error code */ |
266 | |
267 | if(!dir[1]) |
268 | return CURLE_FILE_COULDNT_READ_FILE; /* fix: better error code */ |
269 | |
270 | #ifdef O_BINARY |
271 | #define MODE_DEFAULT O_WRONLY|O_CREAT|O_BINARY |
272 | #else |
273 | #define MODE_DEFAULT O_WRONLY|O_CREAT |
274 | #endif |
275 | |
276 | if(data->state.resume_from) |
277 | mode = MODE_DEFAULT|O_APPEND; |
278 | else |
279 | mode = MODE_DEFAULT|O_TRUNC; |
280 | |
281 | fd = open(file->path, mode, data->set.new_file_perms); |
282 | if(fd < 0) { |
283 | failf(data, "Can't open %s for writing" , file->path); |
284 | return CURLE_WRITE_ERROR; |
285 | } |
286 | |
287 | if(-1 != data->state.infilesize) |
288 | /* known size of data to "upload" */ |
289 | Curl_pgrsSetUploadSize(data, data->state.infilesize); |
290 | |
291 | /* treat the negative resume offset value as the case of "-" */ |
292 | if(data->state.resume_from < 0) { |
293 | if(fstat(fd, &file_stat)) { |
294 | close(fd); |
295 | failf(data, "Can't get the size of %s" , file->path); |
296 | return CURLE_WRITE_ERROR; |
297 | } |
298 | data->state.resume_from = (curl_off_t)file_stat.st_size; |
299 | } |
300 | |
301 | while(!result) { |
302 | size_t nread; |
303 | size_t nwrite; |
304 | size_t readcount; |
305 | result = Curl_fillreadbuffer(data, data->set.buffer_size, &readcount); |
306 | if(result) |
307 | break; |
308 | |
309 | if(!readcount) |
310 | break; |
311 | |
312 | nread = readcount; |
313 | |
314 | /*skip bytes before resume point*/ |
315 | if(data->state.resume_from) { |
316 | if((curl_off_t)nread <= data->state.resume_from) { |
317 | data->state.resume_from -= nread; |
318 | nread = 0; |
319 | buf2 = buf; |
320 | } |
321 | else { |
322 | buf2 = buf + data->state.resume_from; |
323 | nread -= (size_t)data->state.resume_from; |
324 | data->state.resume_from = 0; |
325 | } |
326 | } |
327 | else |
328 | buf2 = buf; |
329 | |
330 | /* write the data to the target */ |
331 | nwrite = write(fd, buf2, nread); |
332 | if(nwrite != nread) { |
333 | result = CURLE_SEND_ERROR; |
334 | break; |
335 | } |
336 | |
337 | bytecount += nread; |
338 | |
339 | Curl_pgrsSetUploadCounter(data, bytecount); |
340 | |
341 | if(Curl_pgrsUpdate(data)) |
342 | result = CURLE_ABORTED_BY_CALLBACK; |
343 | else |
344 | result = Curl_speedcheck(data, Curl_now()); |
345 | } |
346 | if(!result && Curl_pgrsUpdate(data)) |
347 | result = CURLE_ABORTED_BY_CALLBACK; |
348 | |
349 | close(fd); |
350 | |
351 | return result; |
352 | } |
353 | |
354 | /* |
355 | * file_do() is the protocol-specific function for the do-phase, separated |
356 | * from the connect-phase above. Other protocols merely setup the transfer in |
357 | * the do-phase, to have it done in the main transfer loop but since some |
358 | * platforms we support don't allow select()ing etc on file handles (as |
359 | * opposed to sockets) we instead perform the whole do-operation in this |
360 | * function. |
361 | */ |
362 | static CURLcode file_do(struct Curl_easy *data, bool *done) |
363 | { |
364 | /* This implementation ignores the host name in conformance with |
365 | RFC 1738. Only local files (reachable via the standard file system) |
366 | are supported. This means that files on remotely mounted directories |
367 | (via NFS, Samba, NT sharing) can be accessed through a file:// URL |
368 | */ |
369 | CURLcode result = CURLE_OK; |
370 | struct_stat statbuf; /* struct_stat instead of struct stat just to allow the |
371 | Windows version to have a different struct without |
372 | having to redefine the simple word 'stat' */ |
373 | curl_off_t expected_size = -1; |
374 | bool size_known; |
375 | bool fstated = FALSE; |
376 | char *buf = data->state.buffer; |
377 | curl_off_t bytecount = 0; |
378 | int fd; |
379 | struct FILEPROTO *file; |
380 | |
381 | *done = TRUE; /* unconditionally */ |
382 | |
383 | Curl_pgrsStartNow(data); |
384 | |
385 | if(data->set.upload) |
386 | return file_upload(data); |
387 | |
388 | file = data->req.p.file; |
389 | |
390 | /* get the fd from the connection phase */ |
391 | fd = file->fd; |
392 | |
393 | /* VMS: This only works reliable for STREAMLF files */ |
394 | if(-1 != fstat(fd, &statbuf)) { |
395 | if(!S_ISDIR(statbuf.st_mode)) |
396 | expected_size = statbuf.st_size; |
397 | /* and store the modification time */ |
398 | data->info.filetime = statbuf.st_mtime; |
399 | fstated = TRUE; |
400 | } |
401 | |
402 | if(fstated && !data->state.range && data->set.timecondition) { |
403 | if(!Curl_meets_timecondition(data, data->info.filetime)) { |
404 | *done = TRUE; |
405 | return CURLE_OK; |
406 | } |
407 | } |
408 | |
409 | if(fstated) { |
410 | time_t filetime; |
411 | struct tm buffer; |
412 | const struct tm *tm = &buffer; |
413 | char [80]; |
414 | int ; |
415 | char accept_ranges[24]= { "Accept-ranges: bytes\r\n" }; |
416 | if(expected_size >= 0) { |
417 | headerlen = msnprintf(header, sizeof(header), |
418 | "Content-Length: %" CURL_FORMAT_CURL_OFF_T "\r\n" , |
419 | expected_size); |
420 | result = Curl_client_write(data, CLIENTWRITE_HEADER, header, headerlen); |
421 | if(result) |
422 | return result; |
423 | |
424 | result = Curl_client_write(data, CLIENTWRITE_HEADER, |
425 | accept_ranges, strlen(accept_ranges)); |
426 | if(result != CURLE_OK) |
427 | return result; |
428 | } |
429 | |
430 | filetime = (time_t)statbuf.st_mtime; |
431 | result = Curl_gmtime(filetime, &buffer); |
432 | if(result) |
433 | return result; |
434 | |
435 | /* format: "Tue, 15 Nov 1994 12:45:26 GMT" */ |
436 | headerlen = msnprintf(header, sizeof(header), |
437 | "Last-Modified: %s, %02d %s %4d %02d:%02d:%02d GMT\r\n%s" , |
438 | Curl_wkday[tm->tm_wday?tm->tm_wday-1:6], |
439 | tm->tm_mday, |
440 | Curl_month[tm->tm_mon], |
441 | tm->tm_year + 1900, |
442 | tm->tm_hour, |
443 | tm->tm_min, |
444 | tm->tm_sec, |
445 | data->set.opt_no_body ? "" : "\r\n" ); |
446 | result = Curl_client_write(data, CLIENTWRITE_HEADER, header, headerlen); |
447 | if(result) |
448 | return result; |
449 | /* set the file size to make it available post transfer */ |
450 | Curl_pgrsSetDownloadSize(data, expected_size); |
451 | if(data->set.opt_no_body) |
452 | return result; |
453 | } |
454 | |
455 | /* Check whether file range has been specified */ |
456 | result = Curl_range(data); |
457 | if(result) |
458 | return result; |
459 | |
460 | /* Adjust the start offset in case we want to get the N last bytes |
461 | * of the stream if the filesize could be determined */ |
462 | if(data->state.resume_from < 0) { |
463 | if(!fstated) { |
464 | failf(data, "Can't get the size of file." ); |
465 | return CURLE_READ_ERROR; |
466 | } |
467 | data->state.resume_from += (curl_off_t)statbuf.st_size; |
468 | } |
469 | |
470 | if(data->state.resume_from > 0) { |
471 | /* We check explicitly if we have a start offset, because |
472 | * expected_size may be -1 if we don't know how large the file is, |
473 | * in which case we should not adjust it. */ |
474 | if(data->state.resume_from <= expected_size) |
475 | expected_size -= data->state.resume_from; |
476 | else { |
477 | failf(data, "failed to resume file:// transfer" ); |
478 | return CURLE_BAD_DOWNLOAD_RESUME; |
479 | } |
480 | } |
481 | |
482 | /* A high water mark has been specified so we obey... */ |
483 | if(data->req.maxdownload > 0) |
484 | expected_size = data->req.maxdownload; |
485 | |
486 | if(!fstated || (expected_size <= 0)) |
487 | size_known = FALSE; |
488 | else |
489 | size_known = TRUE; |
490 | |
491 | /* The following is a shortcut implementation of file reading |
492 | this is both more efficient than the former call to download() and |
493 | it avoids problems with select() and recv() on file descriptors |
494 | in Winsock */ |
495 | if(size_known) |
496 | Curl_pgrsSetDownloadSize(data, expected_size); |
497 | |
498 | if(data->state.resume_from) { |
499 | if(data->state.resume_from != |
500 | lseek(fd, data->state.resume_from, SEEK_SET)) |
501 | return CURLE_BAD_DOWNLOAD_RESUME; |
502 | } |
503 | |
504 | Curl_pgrsTime(data, TIMER_STARTTRANSFER); |
505 | |
506 | while(!result) { |
507 | ssize_t nread; |
508 | /* Don't fill a whole buffer if we want less than all data */ |
509 | size_t bytestoread; |
510 | |
511 | if(size_known) { |
512 | bytestoread = (expected_size < data->set.buffer_size) ? |
513 | curlx_sotouz(expected_size) : (size_t)data->set.buffer_size; |
514 | } |
515 | else |
516 | bytestoread = data->set.buffer_size-1; |
517 | |
518 | nread = read(fd, buf, bytestoread); |
519 | |
520 | if(nread > 0) |
521 | buf[nread] = 0; |
522 | |
523 | if(nread <= 0 || (size_known && (expected_size == 0))) |
524 | break; |
525 | |
526 | bytecount += nread; |
527 | if(size_known) |
528 | expected_size -= nread; |
529 | |
530 | result = Curl_client_write(data, CLIENTWRITE_BODY, buf, nread); |
531 | if(result) |
532 | return result; |
533 | |
534 | Curl_pgrsSetDownloadCounter(data, bytecount); |
535 | |
536 | if(Curl_pgrsUpdate(data)) |
537 | result = CURLE_ABORTED_BY_CALLBACK; |
538 | else |
539 | result = Curl_speedcheck(data, Curl_now()); |
540 | } |
541 | if(Curl_pgrsUpdate(data)) |
542 | result = CURLE_ABORTED_BY_CALLBACK; |
543 | |
544 | return result; |
545 | } |
546 | |
547 | #endif |
548 | |