1libcurl-tutorial(3) libcurl programming libcurl-tutorial(3)
2
3
4
6 libcurl-tutorial - libcurl programming tutorial
7
9 This document attempts to describe the general principles and some ba‐
10 sic approaches to consider when programming with libcurl. The text will
11 focus mainly on the C interface but might apply fairly well on other
12 interfaces as well as they usually follow the C one pretty closely.
13
14 This document will refer to 'the user' as the person writing the source
15 code that uses libcurl. That would probably be you or someone in your
16 position. What will be generally referred to as 'the program' will be
17 the collected source code that you write that is using libcurl for
18 transfers. The program is outside libcurl and libcurl is outside of the
19 program.
20
21 To get more details on all options and functions described herein,
22 please refer to their respective man pages.
23
24
26 There are many different ways to build C programs. This chapter will
27 assume a Unix style build process. If you use a different build system,
28 you can still read this to get general information that may apply to
29 your environment as well.
30
31 Compiling the Program
32 Your compiler needs to know where the libcurl headers are lo‐
33 cated. Therefore you must set your compiler's include path to
34 point to the directory where you installed them. The 'curl-con‐
35 fig'[3] tool can be used to get this information:
36
37 $ curl-config --cflags
38
39
40 Linking the Program with libcurl
41 When having compiled the program, you need to link your object
42 files to create a single executable. For that to succeed, you
43 need to link with libcurl and possibly also with other libraries
44 that libcurl itself depends on. Like the OpenSSL libraries, but
45 even some standard OS libraries may be needed on the command
46 line. To figure out which flags to use, once again the 'curl-
47 config' tool comes to the rescue:
48
49 $ curl-config --libs
50
51
52 SSL or Not
53 libcurl can be built and customized in many ways. One of the
54 things that varies from different libraries and builds is the
55 support for SSL-based transfers, like HTTPS and FTPS. If a sup‐
56 ported SSL library was detected properly at build-time, libcurl
57 will be built with SSL support. To figure out if an installed
58 libcurl has been built with SSL support enabled, use 'curl-con‐
59 fig' like this:
60
61 $ curl-config --feature
62
63 And if SSL is supported, the keyword 'SSL' will be written to
64 stdout, possibly together with a few other features that could
65 be either on or off on for different libcurls.
66
67 See also the "Features libcurl Provides" further down.
68
69 autoconf macro
70 When you write your configure script to detect libcurl and setup
71 variables accordingly, we offer a prewritten macro that probably
72 does everything you need in this area. See
73 docs/libcurl/libcurl.m4 file - it includes docs on how to use
74 it.
75
76
78 The people behind libcurl have put a considerable effort to make
79 libcurl work on a large amount of different operating systems and envi‐
80 ronments.
81
82 You program libcurl the same way on all platforms that libcurl runs on.
83 There are only very few minor considerations that differ. If you just
84 make sure to write your code portable enough, you may very well create
85 yourself a very portable program. libcurl shouldn't stop you from that.
86
87
89 The program must initialize some of the libcurl functionality globally.
90 That means it should be done exactly once, no matter how many times you
91 intend to use the library. Once for your program's entire life time.
92 This is done using
93
94 curl_global_init()
95
96 and it takes one parameter which is a bit pattern that tells libcurl
97 what to initialize. Using CURL_GLOBAL_ALL will make it initialize all
98 known internal sub modules, and might be a good default option. The
99 current two bits that are specified are:
100
101 CURL_GLOBAL_WIN32
102 which only does anything on Windows machines. When used
103 on a Windows machine, it'll make libcurl initialize the
104 win32 socket stuff. Without having that initialized prop‐
105 erly, your program cannot use sockets properly. You
106 should only do this once for each application, so if your
107 program already does this or of another library in use
108 does it, you should not tell libcurl to do this as well.
109
110 CURL_GLOBAL_SSL
111 which only does anything on libcurls compiled and built
112 SSL-enabled. On these systems, this will make libcurl
113 initialize the SSL library properly for this application.
114 This only needs to be done once for each application so
115 if your program or another library already does this,
116 this bit should not be needed.
117
118 libcurl has a default protection mechanism that detects if
119 curl_global_init(3) hasn't been called by the time curl_easy_perform(3)
120 is called and if that is the case, libcurl runs the function itself
121 with a guessed bit pattern. Please note that depending solely on this
122 is not considered nice nor very good.
123
124 When the program no longer uses libcurl, it should call
125 curl_global_cleanup(3), which is the opposite of the init call. It will
126 then do the reversed operations to cleanup the resources the
127 curl_global_init(3) call initialized.
128
129 Repeated calls to curl_global_init(3) and curl_global_cleanup(3) should
130 be avoided. They should only be called once each.
131
132
134 It is considered best-practice to determine libcurl features at run-
135 time rather than at build-time (if possible of course). By calling
136 curl_version_info(3) and checking out the details of the returned
137 struct, your program can figure out exactly what the currently running
138 libcurl supports.
139
140
142 libcurl first introduced the so called easy interface. All operations
143 in the easy interface are prefixed with 'curl_easy'. The easy interface
144 lets you do single transfers with a synchronous and blocking function
145 call.
146
147 libcurl also offers another interface that allows multiple simultaneous
148 transfers in a single thread, the so called multi interface. More about
149 that interface is detailed in a separate chapter further down. You
150 still need to understand the easy interface first, so please continue
151 reading for better understanding.
152
154 To use the easy interface, you must first create yourself an easy han‐
155 dle. You need one handle for each easy session you want to perform. Ba‐
156 sically, you should use one handle for every thread you plan to use for
157 transferring. You must never share the same handle in multiple threads.
158
159 Get an easy handle with
160
161 easyhandle = curl_easy_init();
162
163 It returns an easy handle. Using that you proceed to the next step:
164 setting up your preferred actions. A handle is just a logic entity for
165 the upcoming transfer or series of transfers.
166
167 You set properties and options for this handle using curl_easy_se‐
168 topt(3). They control how the subsequent transfer or transfers will be
169 made. Options remain set in the handle until set again to something
170 different. They are sticky. Multiple requests using the same handle
171 will use the same options.
172
173 If you at any point would like to blank all previously set options for
174 a single easy handle, you can call curl_easy_reset(3) and you can also
175 make a clone of an easy handle (with all its set options) using
176 curl_easy_duphandle(3).
177
178 Many of the options you set in libcurl are "strings", pointers to data
179 terminated with a zero byte. When you set strings with curl_easy_se‐
180 topt(3), libcurl makes its own copy so that they don't need to be kept
181 around in your application after being set[4].
182
183 One of the most basic properties to set in the handle is the URL. You
184 set your preferred URL to transfer with CURLOPT_URL(3) in a manner sim‐
185 ilar to:
186
187 curl_easy_setopt(handle, CURLOPT_URL, "http://domain.com/");
188
189 Let's assume for a while that you want to receive data as the URL iden‐
190 tifies a remote resource you want to get here. Since you write a sort
191 of application that needs this transfer, I assume that you would like
192 to get the data passed to you directly instead of simply getting it
193 passed to stdout. So, you write your own function that matches this
194 prototype:
195
196 size_t write_data(void *buffer, size_t size, size_t nmemb, void
197 *userp);
198
199 You tell libcurl to pass all data to this function by issuing a func‐
200 tion similar to this:
201
202 curl_easy_setopt(easyhandle, CURLOPT_WRITEFUNCTION, write_data);
203
204 You can control what data your callback function gets in the fourth ar‐
205 gument by setting another property:
206
207 curl_easy_setopt(easyhandle, CURLOPT_WRITEDATA, &internal_struct);
208
209 Using that property, you can easily pass local data between your appli‐
210 cation and the function that gets invoked by libcurl. libcurl itself
211 won't touch the data you pass with CURLOPT_WRITEDATA(3).
212
213 libcurl offers its own default internal callback that will take care of
214 the data if you don't set the callback with CURLOPT_WRITEFUNCTION(3).
215 It will then simply output the received data to stdout. You can have
216 the default callback write the data to a different file handle by pass‐
217 ing a 'FILE *' to a file opened for writing with the CURLOPT_WRITE‐
218 DATA(3) option.
219
220 Now, we need to take a step back and have a deep breath. Here's one of
221 those rare platform-dependent nitpicks. Did you spot it? On some plat‐
222 forms[2], libcurl won't be able to operate on files opened by the pro‐
223 gram. Thus, if you use the default callback and pass in an open file
224 with CURLOPT_WRITEDATA(3), it will crash. You should therefore avoid
225 this to make your program run fine virtually everywhere.
226
227 (CURLOPT_WRITEDATA(3) was formerly known as CURLOPT_FILE. Both names
228 still work and do the same thing).
229
230 If you're using libcurl as a win32 DLL, you MUST use the CURLOPT_WRITE‐
231 FUNCTION(3) if you set CURLOPT_WRITEDATA(3) - or you will experience
232 crashes.
233
234 There are of course many more options you can set, and we'll get back
235 to a few of them later. Let's instead continue to the actual transfer:
236
237 success = curl_easy_perform(easyhandle);
238
239 curl_easy_perform(3) will connect to the remote site, do the necessary
240 commands and receive the transfer. Whenever it receives data, it calls
241 the callback function we previously set. The function may get one byte
242 at a time, or it may get many kilobytes at once. libcurl delivers as
243 much as possible as often as possible. Your callback function should
244 return the number of bytes it "took care of". If that is not the exact
245 same amount of bytes that was passed to it, libcurl will abort the op‐
246 eration and return with an error code.
247
248 When the transfer is complete, the function returns a return code that
249 informs you if it succeeded in its mission or not. If a return code
250 isn't enough for you, you can use the CURLOPT_ERRORBUFFER(3) to point
251 libcurl to a buffer of yours where it'll store a human readable error
252 message as well.
253
254 If you then want to transfer another file, the handle is ready to be
255 used again. Mind you, it is even preferred that you re-use an existing
256 handle if you intend to make another transfer. libcurl will then at‐
257 tempt to re-use the previous connection.
258
259 For some protocols, downloading a file can involve a complicated
260 process of logging in, setting the transfer mode, changing the current
261 directory and finally transferring the file data. libcurl takes care of
262 all that complication for you. Given simply the URL to a file, libcurl
263 will take care of all the details needed to get the file moved from one
264 machine to another.
265
266
268 libcurl is thread safe but there are a few exceptions. Refer to
269 libcurl-thread(3) for more information.
270
271
273 There will always be times when the transfer fails for some reason. You
274 might have set the wrong libcurl option or misunderstood what the
275 libcurl option actually does, or the remote server might return non-
276 standard replies that confuse the library which then confuses your pro‐
277 gram.
278
279 There's one golden rule when these things occur: set the CURLOPT_VER‐
280 BOSE(3) option to 1. It'll cause the library to spew out the entire
281 protocol details it sends, some internal info and some received proto‐
282 col data as well (especially when using FTP). If you're using HTTP,
283 adding the headers in the received output to study is also a clever way
284 to get a better understanding why the server behaves the way it does.
285 Include headers in the normal body output with CURLOPT_HEADER(3) set 1.
286
287 Of course, there are bugs left. We need to know about them to be able
288 to fix them, so we're quite dependent on your bug reports! When you do
289 report suspected bugs in libcurl, please include as many details as you
290 possibly can: a protocol dump that CURLOPT_VERBOSE(3) produces, library
291 version, as much as possible of your code that uses libcurl, operating
292 system name and version, compiler name and version etc.
293
294 If CURLOPT_VERBOSE(3) is not enough, you increase the level of debug
295 data your application receive by using the CURLOPT_DEBUGFUNCTION(3).
296
297 Getting some in-depth knowledge about the protocols involved is never
298 wrong, and if you're trying to do funny things, you might very well un‐
299 derstand libcurl and how to use it better if you study the appropriate
300 RFC documents at least briefly.
301
302
304 libcurl tries to keep a protocol independent approach to most trans‐
305 fers, thus uploading to a remote FTP site is very similar to uploading
306 data to an HTTP server with a PUT request.
307
308 Of course, first you either create an easy handle or you re-use one ex‐
309 isting one. Then you set the URL to operate on just like before. This
310 is the remote URL, that we now will upload.
311
312 Since we write an application, we most likely want libcurl to get the
313 upload data by asking us for it. To make it do that, we set the read
314 callback and the custom pointer libcurl will pass to our read callback.
315 The read callback should have a prototype similar to:
316
317 size_t function(char *bufptr, size_t size, size_t nitems, void
318 *userp);
319
320 Where bufptr is the pointer to a buffer we fill in with data to upload
321 and size*nitems is the size of the buffer and therefore also the maxi‐
322 mum amount of data we can return to libcurl in this call. The 'userp'
323 pointer is the custom pointer we set to point to a struct of ours to
324 pass private data between the application and the callback.
325
326 curl_easy_setopt(easyhandle, CURLOPT_READFUNCTION, read_function);
327
328 curl_easy_setopt(easyhandle, CURLOPT_READDATA, &filedata);
329
330 Tell libcurl that we want to upload:
331
332 curl_easy_setopt(easyhandle, CURLOPT_UPLOAD, 1L);
333
334 A few protocols won't behave properly when uploads are done without any
335 prior knowledge of the expected file size. So, set the upload file size
336 using the CURLOPT_INFILESIZE_LARGE(3) for all known file sizes like
337 this[1]:
338
339 /* in this example, file_size must be an curl_off_t variable */
340 curl_easy_setopt(easyhandle, CURLOPT_INFILESIZE_LARGE, file_size);
341
342 When you call curl_easy_perform(3) this time, it'll perform all the
343 necessary operations and when it has invoked the upload it'll call your
344 supplied callback to get the data to upload. The program should return
345 as much data as possible in every invoke, as that is likely to make the
346 upload perform as fast as possible. The callback should return the num‐
347 ber of bytes it wrote in the buffer. Returning 0 will signal the end of
348 the upload.
349
350
352 Many protocols use or even require that user name and password are pro‐
353 vided to be able to download or upload the data of your choice. libcurl
354 offers several ways to specify them.
355
356 Most protocols support that you specify the name and password in the
357 URL itself. libcurl will detect this and use them accordingly. This is
358 written like this:
359
360 protocol://user:password@example.com/path/
361
362 If you need any odd letters in your user name or password, you should
363 enter them URL encoded, as %XX where XX is a two-digit hexadecimal num‐
364 ber.
365
366 libcurl also provides options to set various passwords. The user name
367 and password as shown embedded in the URL can instead get set with the
368 CURLOPT_USERPWD(3) option. The argument passed to libcurl should be a
369 char * to a string in the format "user:password". In a manner like
370 this:
371
372 curl_easy_setopt(easyhandle, CURLOPT_USERPWD, "myname:thesecret");
373
374 Another case where name and password might be needed at times, is for
375 those users who need to authenticate themselves to a proxy they use.
376 libcurl offers another option for this, the CURLOPT_PROXYUSERPWD(3). It
377 is used quite similar to the CURLOPT_USERPWD(3) option like this:
378
379 curl_easy_setopt(easyhandle, CURLOPT_PROXYUSERPWD, "myname:these‐
380 cret");
381
382 There's a long time Unix "standard" way of storing FTP user names and
383 passwords, namely in the $HOME/.netrc file. The file should be made
384 private so that only the user may read it (see also the "Security Con‐
385 siderations" chapter), as it might contain the password in plain text.
386 libcurl has the ability to use this file to figure out what set of user
387 name and password to use for a particular host. As an extension to the
388 normal functionality, libcurl also supports this file for non-FTP pro‐
389 tocols such as HTTP. To make curl use this file, use the CUR‐
390 LOPT_NETRC(3) option:
391
392 curl_easy_setopt(easyhandle, CURLOPT_NETRC, 1L);
393
394 And a very basic example of how such a .netrc file may look like:
395
396 machine myhost.mydomain.com
397 login userlogin
398 password secretword
399
400 All these examples have been cases where the password has been op‐
401 tional, or at least you could leave it out and have libcurl attempt to
402 do its job without it. There are times when the password isn't op‐
403 tional, like when you're using an SSL private key for secure transfers.
404
405 To pass the known private key password to libcurl:
406
407 curl_easy_setopt(easyhandle, CURLOPT_KEYPASSWD, "keypassword");
408
409
411 The previous chapter showed how to set user name and password for get‐
412 ting URLs that require authentication. When using the HTTP protocol,
413 there are many different ways a client can provide those credentials to
414 the server and you can control which way libcurl will (attempt to) use
415 them. The default HTTP authentication method is called 'Basic', which
416 is sending the name and password in clear-text in the HTTP request,
417 base64-encoded. This is insecure.
418
419 At the time of this writing, libcurl can be built to use: Basic, Di‐
420 gest, NTLM, Negotiate (SPNEGO). You can tell libcurl which one to use
421 with CURLOPT_HTTPAUTH(3) as in:
422
423 curl_easy_setopt(easyhandle, CURLOPT_HTTPAUTH, CURLAUTH_DIGEST);
424
425 And when you send authentication to a proxy, you can also set authenti‐
426 cation type the same way but instead with CURLOPT_PROXYAUTH(3):
427
428 curl_easy_setopt(easyhandle, CURLOPT_PROXYAUTH, CURLAUTH_NTLM);
429
430 Both these options allow you to set multiple types (by ORing them to‐
431 gether), to make libcurl pick the most secure one out of the types the
432 server/proxy claims to support. This method does however add a round-
433 trip since libcurl must first ask the server what it supports:
434
435 curl_easy_setopt(easyhandle, CURLOPT_HTTPAUTH,
436 CURLAUTH_DIGEST|CURLAUTH_BASIC);
437
438 For convenience, you can use the 'CURLAUTH_ANY' define (instead of a
439 list with specific types) which allows libcurl to use whatever method
440 it wants.
441
442 When asking for multiple types, libcurl will pick the available one it
443 considers "best" in its own internal order of preference.
444
445
447 We get many questions regarding how to issue HTTP POSTs with libcurl
448 the proper way. This chapter will thus include examples using both dif‐
449 ferent versions of HTTP POST that libcurl supports.
450
451 The first version is the simple POST, the most common version, that
452 most HTML pages using the <form> tag uses. We provide a pointer to the
453 data and tell libcurl to post it all to the remote site:
454
455 char *data="name=daniel&project=curl";
456 curl_easy_setopt(easyhandle, CURLOPT_POSTFIELDS, data);
457 curl_easy_setopt(easyhandle, CURLOPT_URL, "http://posthere.com/");
458
459 curl_easy_perform(easyhandle); /* post away! */
460
461 Simple enough, huh? Since you set the POST options with the CUR‐
462 LOPT_POSTFIELDS(3), this automatically switches the handle to use POST
463 in the upcoming request.
464
465 Ok, so what if you want to post binary data that also requires you to
466 set the Content-Type: header of the post? Well, binary posts prevent
467 libcurl from being able to do strlen() on the data to figure out the
468 size, so therefore we must tell libcurl the size of the post data. Set‐
469 ting headers in libcurl requests are done in a generic way, by building
470 a list of our own headers and then passing that list to libcurl.
471
472 struct curl_slist *headers=NULL;
473 headers = curl_slist_append(headers, "Content-Type: text/xml");
474
475 /* post binary data */
476 curl_easy_setopt(easyhandle, CURLOPT_POSTFIELDS, binaryptr);
477
478 /* set the size of the postfields data */
479 curl_easy_setopt(easyhandle, CURLOPT_POSTFIELDSIZE, 23L);
480
481 /* pass our list of custom made headers */
482 curl_easy_setopt(easyhandle, CURLOPT_HTTPHEADER, headers);
483
484 curl_easy_perform(easyhandle); /* post away! */
485
486 curl_slist_free_all(headers); /* free the header list */
487
488 While the simple examples above cover the majority of all cases where
489 HTTP POST operations are required, they don't do multi-part formposts.
490 Multi-part formposts were introduced as a better way to post (possibly
491 large) binary data and were first documented in the RFC1867 (updated in
492 RFC2388). They're called multi-part because they're built by a chain of
493 parts, each part being a single unit of data. Each part has its own
494 name and contents. You can in fact create and post a multi-part form‐
495 post with the regular libcurl POST support described above, but that
496 would require that you build a formpost yourself and provide to
497 libcurl. To make that easier, libcurl provides a MIME API consisting in
498 several functions: using those, you can create and fill a multi-part
499 form. Function curl_mime_init(3) creates a multi-part body; you can
500 then append new parts to a multi-part body using curl_mime_addpart(3).
501 There are three possible data sources for a part: memory using
502 curl_mime_data(3), file using curl_mime_filedata(3) and user-defined
503 data read callback using curl_mime_data_cb(3). curl_mime_name(3) sets
504 a part's (i.e.: form field) name, while curl_mime_filename(3) fills in
505 the remote file name. With curl_mime_type(3), you can tell the MIME
506 type of a part, curl_mime_headers(3) allows defining the part's head‐
507 ers. When a multi-part body is no longer needed, you can destroy it us‐
508 ing curl_mime_free(3).
509
510 The following example sets two simple text parts with plain textual
511 contents, and then a file with binary contents and uploads the whole
512 thing.
513
514 curl_mime *multipart = curl_mime_init(easyhandle);
515 curl_mimepart *part = curl_mime_addpart(multipart);
516 curl_mime_name(part, "name");
517 curl_mime_data(part, "daniel", CURL_ZERO_TERMINATED);
518 part = curl_mime_addpart(multipart);
519 curl_mime_name(part, "project");
520 curl_mime_data(part, "curl", CURL_ZERO_TERMINATED);
521 part = curl_mime_addpart(multipart);
522 curl_mime_name(part, "logotype-image");
523 curl_mime_filedata(part, "curl.png");
524
525 /* Set the form info */
526 curl_easy_setopt(easyhandle, CURLOPT_MIMEPOST, multipart);
527
528 curl_easy_perform(easyhandle); /* post away! */
529
530 /* free the post data again */
531 curl_mime_free(multipart);
532
533 To post multiple files for a single form field, you must supply each
534 file in a separate part, all with the same field name. Although func‐
535 tion curl_mime_subparts(3) implements nested multi-parts, this way of
536 multiple files posting is deprecated by RFC 7578, chapter 4.3.
537
538 To set the data source from an already opened FILE pointer, use:
539
540 curl_mime_data_cb(part, filesize, (curl_read_callback) fread,
541 (curl_seek_callback) fseek, NULL, filepointer);
542
543 A deprecated curl_formadd(3) function is still supported in libcurl.
544 It should however not be used anymore for new designs and programs us‐
545 ing it ought to be converted to the MIME API. It is however described
546 here as an aid to conversion.
547
548 Using curl_formadd, you add parts to the form. When you're done adding
549 parts, you post the whole form.
550
551 The MIME API example above is expressed as follows using this function:
552
553 struct curl_httppost *post=NULL;
554 struct curl_httppost *last=NULL;
555 curl_formadd(&post, &last,
556 CURLFORM_COPYNAME, "name",
557 CURLFORM_COPYCONTENTS, "daniel", CURLFORM_END);
558 curl_formadd(&post, &last,
559 CURLFORM_COPYNAME, "project",
560 CURLFORM_COPYCONTENTS, "curl", CURLFORM_END);
561 curl_formadd(&post, &last,
562 CURLFORM_COPYNAME, "logotype-image",
563 CURLFORM_FILECONTENT, "curl.png", CURLFORM_END);
564
565 /* Set the form info */
566 curl_easy_setopt(easyhandle, CURLOPT_HTTPPOST, post);
567
568 curl_easy_perform(easyhandle); /* post away! */
569
570 /* free the post data again */
571 curl_formfree(post);
572
573 Multipart formposts are chains of parts using MIME-style separators and
574 headers. It means that each one of these separate parts get a few head‐
575 ers set that describe the individual content-type, size etc. To enable
576 your application to handicraft this formpost even more, libcurl allows
577 you to supply your own set of custom headers to such an individual form
578 part. You can of course supply headers to as many parts as you like,
579 but this little example will show how you set headers to one specific
580 part when you add that to the post handle:
581
582 struct curl_slist *headers=NULL;
583 headers = curl_slist_append(headers, "Content-Type: text/xml");
584
585 curl_formadd(&post, &last,
586 CURLFORM_COPYNAME, "logotype-image",
587 CURLFORM_FILECONTENT, "curl.xml",
588 CURLFORM_CONTENTHEADER, headers,
589 CURLFORM_END);
590
591 curl_easy_perform(easyhandle); /* post away! */
592
593 curl_formfree(post); /* free post */
594 curl_slist_free_all(headers); /* free custom header list */
595
596 Since all options on an easyhandle are "sticky", they remain the same
597 until changed even if you do call curl_easy_perform(3), you may need to
598 tell curl to go back to a plain GET request if you intend to do one as
599 your next request. You force an easyhandle to go back to GET by using
600 the CURLOPT_HTTPGET(3) option:
601
602 curl_easy_setopt(easyhandle, CURLOPT_HTTPGET, 1L);
603
604 Just setting CURLOPT_POSTFIELDS(3) to "" or NULL will *not* stop
605 libcurl from doing a POST. It will just make it POST without any data
606 to send!
607
608
610 Four rules have to be respected in building the multi-part:
611 - The easy handle must be created before building the multi-part.
612 - The multi-part is always created by a call to curl_mime_init(easyhan‐
613 dle).
614 - Each part is created by a call to curl_mime_addpart(multipart).
615 - When complete, the multi-part must be bound to the easy handle using
616 CURLOPT_MIMEPOST(3) instead of CURLOPT_HTTPPOST(3).
617
618 Here are some example of curl_formadd calls to MIME API sequences:
619
620 curl_formadd(&post, &last,
621 CURLFORM_COPYNAME, "id",
622 CURLFORM_COPYCONTENTS, "daniel", CURLFORM_END);
623 CURLFORM_CONTENTHEADER, headers,
624 CURLFORM_END);
625 becomes:
626 part = curl_mime_addpart(multipart);
627 curl_mime_name(part, "id");
628 curl_mime_data(part, "daniel", CURL_ZERO_TERMINATED);
629 curl_mime_headers(part, headers, FALSE);
630
631 Setting the last curl_mime_headers argument to TRUE would have caused
632 the headers to be automatically released upon destroyed the multi-part,
633 thus saving a clean-up call to curl_slist_free_all(3).
634
635 curl_formadd(&post, &last,
636 CURLFORM_PTRNAME, "logotype-image",
637 CURLFORM_FILECONTENT, "-",
638 CURLFORM_END);
639 becomes:
640 part = curl_mime_addpart(multipart);
641 curl_mime_name(part, "logotype-image");
642 curl_mime_data_cb(part, (curl_off_t) -1, fread, fseek, NULL, stdin);
643
644 curl_mime_name always copies the field name. The special file name "-"
645 is not supported by curl_mime_file: to read an open file, use a call‐
646 back source using fread(). The transfer will be chunked since the data
647 size is unknown.
648
649 curl_formadd(&post, &last,
650 CURLFORM_COPYNAME, "datafile[]",
651 CURLFORM_FILE, "file1",
652 CURLFORM_FILE, "file2",
653 CURLFORM_END);
654 becomes:
655 part = curl_mime_addpart(multipart);
656 curl_mime_name(part, "datafile[]");
657 curl_mime_filedata(part, "file1");
658 part = curl_mime_addpart(multipart);
659 curl_mime_name(part, "datafile[]");
660 curl_mime_filedata(part, "file2");
661
662 The deprecated multipart/mixed implementation of multiple files field
663 is translated to two distinct parts with the same name.
664
665 curl_easy_setopt(easyhandle, CURLOPT_READFUNCTION, myreadfunc);
666 curl_formadd(&post, &last,
667 CURLFORM_COPYNAME, "stream",
668 CURLFORM_STREAM, arg,
669 CURLFORM_CONTENTLEN, (curl_off_t) datasize,
670 CURLFORM_FILENAME, "archive.zip",
671 CURLFORM_CONTENTTYPE, "application/zip",
672 CURLFORM_END);
673 becomes:
674 part = curl_mime_addpart(multipart);
675 curl_mime_name(part, "stream");
676 curl_mime_data_cb(part, (curl_off_t) datasize,
677 myreadfunc, NULL, NULL, arg);
678 curl_mime_filename(part, "archive.zip");
679 curl_mime_type(part, "application/zip");
680
681 CURLOPT_READFUNCTION callback is not used: it is replace by directly
682 setting the part source data from the callback read function.
683
684 curl_formadd(&post, &last,
685 CURLFORM_COPYNAME, "memfile",
686 CURLFORM_BUFFER, "memfile.bin",
687 CURLFORM_BUFFERPTR, databuffer,
688 CURLFORM_BUFFERLENGTH, (long) sizeof databuffer,
689 CURLFORM_END);
690 becomes:
691 part = curl_mime_addpart(multipart);
692 curl_mime_name(part, "memfile");
693 curl_mime_data(part, databuffer, (curl_off_t) sizeof databuffer);
694 curl_mime_filename(part, "memfile.bin");
695
696 curl_mime_data always copies the initial data: data buffer is thus free
697 for immediate reuse.
698
699 curl_formadd(&post, &last,
700 CURLFORM_COPYNAME, "message",
701 CURLFORM_FILECONTENT, "msg.txt",
702 CURLFORM_END);
703 becomes:
704 part = curl_mime_addpart(multipart);
705 curl_mime_name(part, "message");
706 curl_mime_filedata(part, "msg.txt");
707 curl_mime_filename(part, NULL);
708
709 Use of curl_mime_filedata sets the remote file name as a side effect:
710 it is therefore necessary to clear it for CURLFORM_FILECONTENT emula‐
711 tion.
712
713
715 For historical and traditional reasons, libcurl has a built-in progress
716 meter that can be switched on and then makes it present a progress me‐
717 ter in your terminal.
718
719 Switch on the progress meter by, oddly enough, setting CURLOPT_NO‐
720 PROGRESS(3) to zero. This option is set to 1 by default.
721
722 For most applications however, the built-in progress meter is useless
723 and what instead is interesting is the ability to specify a progress
724 callback. The function pointer you pass to libcurl will then be called
725 on irregular intervals with information about the current transfer.
726
727 Set the progress callback by using CURLOPT_PROGRESSFUNCTION(3). And
728 pass a pointer to a function that matches this prototype:
729
730 int progress_callback(void *clientp,
731 double dltotal,
732 double dlnow,
733 double ultotal,
734 double ulnow);
735
736 If any of the input arguments is unknown, a 0 will be passed. The first
737 argument, the 'clientp' is the pointer you pass to libcurl with CUR‐
738 LOPT_PROGRESSDATA(3). libcurl won't touch it.
739
740
742 There's basically only one thing to keep in mind when using C++ instead
743 of C when interfacing libcurl:
744
745 The callbacks CANNOT be non-static class member functions
746
747 Example C++ code:
748
749 class AClass {
750 static size_t write_data(void *ptr, size_t size, size_t nmemb,
751 void *ourpointer)
752 {
753 /* do what you want with the data */
754 }
755 }
756
757
759 What "proxy" means according to Merriam-Webster: "a person authorized
760 to act for another" but also "the agency, function, or office of a
761 deputy who acts as a substitute for another".
762
763 Proxies are exceedingly common these days. Companies often only offer
764 Internet access to employees through their proxies. Network clients or
765 user-agents ask the proxy for documents, the proxy does the actual re‐
766 quest and then it returns them.
767
768 libcurl supports SOCKS and HTTP proxies. When a given URL is wanted,
769 libcurl will ask the proxy for it instead of trying to connect to the
770 actual host identified in the URL.
771
772 If you're using a SOCKS proxy, you may find that libcurl doesn't quite
773 support all operations through it.
774
775 For HTTP proxies: the fact that the proxy is an HTTP proxy puts certain
776 restrictions on what can actually happen. A requested URL that might
777 not be a HTTP URL will be still be passed to the HTTP proxy to deliver
778 back to libcurl. This happens transparently, and an application may not
779 need to know. I say "may", because at times it is very important to un‐
780 derstand that all operations over an HTTP proxy use the HTTP protocol.
781 For example, you can't invoke your own custom FTP commands or even
782 proper FTP directory listings.
783
784
785 Proxy Options
786
787 To tell libcurl to use a proxy at a given port number:
788
789 curl_easy_setopt(easyhandle, CURLOPT_PROXY, "proxy-
790 host.com:8080");
791
792 Some proxies require user authentication before allowing a re‐
793 quest, and you pass that information similar to this:
794
795 curl_easy_setopt(easyhandle, CURLOPT_PROXYUSERPWD, "user:pass‐
796 word");
797
798 If you want to, you can specify the host name only in the CUR‐
799 LOPT_PROXY(3) option, and set the port number separately with
800 CURLOPT_PROXYPORT(3).
801
802 Tell libcurl what kind of proxy it is with CURLOPT_PROXYTYPE(3)
803 (if not, it will default to assume an HTTP proxy):
804
805 curl_easy_setopt(easyhandle, CURLOPT_PROXYTYPE, CURL‐
806 PROXY_SOCKS4);
807
808
809 Environment Variables
810
811 libcurl automatically checks and uses a set of environment vari‐
812 ables to know what proxies to use for certain protocols. The
813 names of the variables are following an ancient de facto stan‐
814 dard and are built up as "[protocol]_proxy" (note the lower cas‐
815 ing). Which makes the variable 'http_proxy' checked for a name
816 of a proxy to use when the input URL is HTTP. Following the same
817 rule, the variable named 'ftp_proxy' is checked for FTP URLs.
818 Again, the proxies are always HTTP proxies, the different names
819 of the variables simply allows different HTTP proxies to be
820 used.
821
822 The proxy environment variable contents should be in the format
823 "[protocol://][user:password@]machine[:port]". Where the proto‐
824 col:// part is simply ignored if present (so http://proxy and
825 bluerk://proxy will do the same) and the optional port number
826 specifies on which port the proxy operates on the host. If not
827 specified, the internal default port number will be used and
828 that is most likely *not* the one you would like it to be.
829
830 There are two special environment variables. 'all_proxy' is what
831 sets proxy for any URL in case the protocol specific variable
832 wasn't set, and 'no_proxy' defines a list of hosts that should
833 not use a proxy even though a variable may say so. If 'no_proxy'
834 is a plain asterisk ("*") it matches all hosts.
835
836 To explicitly disable libcurl's checking for and using the proxy
837 environment variables, set the proxy name to "" - an empty
838 string - with CURLOPT_PROXY(3).
839
840 SSL and Proxies
841
842 SSL is for secure point-to-point connections. This involves
843 strong encryption and similar things, which effectively makes it
844 impossible for a proxy to operate as a "man in between" which
845 the proxy's task is, as previously discussed. Instead, the only
846 way to have SSL work over an HTTP proxy is to ask the proxy to
847 tunnel trough everything without being able to check or fiddle
848 with the traffic.
849
850 Opening an SSL connection over an HTTP proxy is therefore a mat‐
851 ter of asking the proxy for a straight connection to the target
852 host on a specified port. This is made with the HTTP request
853 CONNECT. ("please mr proxy, connect me to that remote host").
854
855 Because of the nature of this operation, where the proxy has no
856 idea what kind of data that is passed in and out through this
857 tunnel, this breaks some of the very few advantages that come
858 from using a proxy, such as caching. Many organizations prevent
859 this kind of tunneling to other destination port numbers than
860 443 (which is the default HTTPS port number).
861
862
863 Tunneling Through Proxy
864 As explained above, tunneling is required for SSL to work and
865 often even restricted to the operation intended for SSL; HTTPS.
866
867 This is however not the only time proxy-tunneling might offer
868 benefits to you or your application.
869
870 As tunneling opens a direct connection from your application to
871 the remote machine, it suddenly also re-introduces the ability
872 to do non-HTTP operations over an HTTP proxy. You can in fact
873 use things such as FTP upload or FTP custom commands this way.
874
875 Again, this is often prevented by the administrators of proxies
876 and is rarely allowed.
877
878 Tell libcurl to use proxy tunneling like this:
879
880 curl_easy_setopt(easyhandle, CURLOPT_HTTPPROXYTUNNEL, 1L);
881
882 In fact, there might even be times when you want to do plain
883 HTTP operations using a tunnel like this, as it then enables you
884 to operate on the remote server instead of asking the proxy to
885 do so. libcurl will not stand in the way for such innovative ac‐
886 tions either!
887
888
889 Proxy Auto-Config
890
891 Netscape first came up with this. It is basically a web page
892 (usually using a .pac extension) with a Javascript that when ex‐
893 ecuted by the browser with the requested URL as input, returns
894 information to the browser on how to connect to the URL. The re‐
895 turned information might be "DIRECT" (which means no proxy
896 should be used), "PROXY host:port" (to tell the browser where
897 the proxy for this particular URL is) or "SOCKS host:port" (to
898 direct the browser to a SOCKS proxy).
899
900 libcurl has no means to interpret or evaluate Javascript and
901 thus it doesn't support this. If you get yourself in a position
902 where you face this nasty invention, the following advice have
903 been mentioned and used in the past:
904
905 - Depending on the Javascript complexity, write up a script that
906 translates it to another language and execute that.
907
908 - Read the Javascript code and rewrite the same logic in another
909 language.
910
911 - Implement a Javascript interpreter; people have successfully
912 used the Mozilla Javascript engine in the past.
913
914 - Ask your admins to stop this, for a static proxy setup or sim‐
915 ilar.
916
917
919 Re-cycling the same easy handle several times when doing multiple re‐
920 quests is the way to go.
921
922 After each single curl_easy_perform(3) operation, libcurl will keep the
923 connection alive and open. A subsequent request using the same easy
924 handle to the same host might just be able to use the already open con‐
925 nection! This reduces network impact a lot.
926
927 Even if the connection is dropped, all connections involving SSL to the
928 same host again, will benefit from libcurl's session ID cache that
929 drastically reduces re-connection time.
930
931 FTP connections that are kept alive save a lot of time, as the command-
932 response round-trips are skipped, and also you don't risk getting
933 blocked without permission to login again like on many FTP servers only
934 allowing N persons to be logged in at the same time.
935
936 libcurl caches DNS name resolving results, to make lookups of a previ‐
937 ously looked up name a lot faster.
938
939 Other interesting details that improve performance for subsequent re‐
940 quests may also be added in the future.
941
942 Each easy handle will attempt to keep the last few connections alive
943 for a while in case they are to be used again. You can set the size of
944 this "cache" with the CURLOPT_MAXCONNECTS(3) option. Default is 5.
945 There is very seldom any point in changing this value, and if you think
946 of changing this it is often just a matter of thinking again.
947
948 To force your upcoming request to not use an already existing connec‐
949 tion (it will even close one first if there happens to be one alive to
950 the same host you're about to operate on), you can do that by setting
951 CURLOPT_FRESH_CONNECT(3) to 1. In a similar spirit, you can also forbid
952 the upcoming request to be "lying" around and possibly get re-used af‐
953 ter the request by setting CURLOPT_FORBID_REUSE(3) to 1.
954
955
957 When you use libcurl to do HTTP requests, it'll pass along a series of
958 headers automatically. It might be good for you to know and understand
959 these. You can replace or remove them by using the CURLOPT_HTTP‐
960 HEADER(3) option.
961
962
963 Host This header is required by HTTP 1.1 and even many 1.0 servers
964 and should be the name of the server we want to talk to. This
965 includes the port number if anything but default.
966
967
968 Accept "*/*".
969
970
971 Expect When doing POST requests, libcurl sets this header to "100-con‐
972 tinue" to ask the server for an "OK" message before it proceeds
973 with sending the data part of the post. If the POSTed data
974 amount is deemed "small", libcurl will not use this header.
975
976
978 There is an ongoing development today where more and more protocols are
979 built upon HTTP for transport. This has obvious benefits as HTTP is a
980 tested and reliable protocol that is widely deployed and has excellent
981 proxy-support.
982
983 When you use one of these protocols, and even when doing other kinds of
984 programming you may need to change the traditional HTTP (or FTP or...)
985 manners. You may need to change words, headers or various data.
986
987 libcurl is your friend here too.
988
989
990 CUSTOMREQUEST
991 If just changing the actual HTTP request keyword is what you
992 want, like when GET, HEAD or POST is not good enough for you,
993 CURLOPT_CUSTOMREQUEST(3) is there for you. It is very simple to
994 use:
995
996 curl_easy_setopt(easyhandle, CURLOPT_CUSTOMREQUEST, "MYOWNRE‐
997 QUEST");
998
999 When using the custom request, you change the request keyword of
1000 the actual request you are performing. Thus, by default you make
1001 a GET request but you can also make a POST operation (as de‐
1002 scribed before) and then replace the POST keyword if you want
1003 to. You're the boss.
1004
1005
1006 Modify Headers
1007 HTTP-like protocols pass a series of headers to the server when
1008 doing the request, and you're free to pass any amount of extra
1009 headers that you think fit. Adding headers is this easy:
1010
1011 struct curl_slist *headers=NULL; /* init to NULL is important */
1012
1013 headers = curl_slist_append(headers, "Hey-server-hey: how are you?");
1014 headers = curl_slist_append(headers, "X-silly-content: yes");
1015
1016 /* pass our list of custom made headers */
1017 curl_easy_setopt(easyhandle, CURLOPT_HTTPHEADER, headers);
1018
1019 curl_easy_perform(easyhandle); /* transfer http */
1020
1021 curl_slist_free_all(headers); /* free the header list */
1022
1023 ... and if you think some of the internally generated headers,
1024 such as Accept: or Host: don't contain the data you want them to
1025 contain, you can replace them by simply setting them too:
1026
1027 headers = curl_slist_append(headers, "Accept: Agent-007");
1028 headers = curl_slist_append(headers, "Host: munged.host.line");
1029
1030
1031 Delete Headers
1032 If you replace an existing header with one with no contents, you
1033 will prevent the header from being sent. For instance, if you
1034 want to completely prevent the "Accept:" header from being sent,
1035 you can disable it with code similar to this:
1036
1037 headers = curl_slist_append(headers, "Accept:");
1038
1039 Both replacing and canceling internal headers should be done
1040 with careful consideration and you should be aware that you may
1041 violate the HTTP protocol when doing so.
1042
1043
1044 Enforcing chunked transfer-encoding
1045
1046 By making sure a request uses the custom header "Transfer-Encod‐
1047 ing: chunked" when doing a non-GET HTTP operation, libcurl will
1048 switch over to "chunked" upload, even though the size of the
1049 data to upload might be known. By default, libcurl usually
1050 switches over to chunked upload automatically if the upload data
1051 size is unknown.
1052
1053
1054 HTTP Version
1055
1056 All HTTP requests includes the version number to tell the server
1057 which version we support. libcurl speaks HTTP 1.1 by default.
1058 Some very old servers don't like getting 1.1-requests and when
1059 dealing with stubborn old things like that, you can tell libcurl
1060 to use 1.0 instead by doing something like this:
1061
1062 curl_easy_setopt(easyhandle, CURLOPT_HTTP_VERSION,
1063 CURL_HTTP_VERSION_1_0);
1064
1065
1066 FTP Custom Commands
1067
1068 Not all protocols are HTTP-like, and thus the above may not help
1069 you when you want to make, for example, your FTP transfers to
1070 behave differently.
1071
1072 Sending custom commands to an FTP server means that you need to
1073 send the commands exactly as the FTP server expects them (RFC959
1074 is a good guide here), and you can only use commands that work
1075 on the control-connection alone. All kinds of commands that re‐
1076 quire data interchange and thus need a data-connection must be
1077 left to libcurl's own judgement. Also be aware that libcurl will
1078 do its very best to change directory to the target directory be‐
1079 fore doing any transfer, so if you change directory (with CWD or
1080 similar) you might confuse libcurl and then it might not attempt
1081 to transfer the file in the correct remote directory.
1082
1083 A little example that deletes a given file before an operation:
1084
1085 headers = curl_slist_append(headers, "DELE file-to-remove");
1086
1087 /* pass the list of custom commands to the handle */
1088 curl_easy_setopt(easyhandle, CURLOPT_QUOTE, headers);
1089
1090 curl_easy_perform(easyhandle); /* transfer ftp data! */
1091
1092 curl_slist_free_all(headers); /* free the header list */
1093
1094 If you would instead want this operation (or chain of opera‐
1095 tions) to happen _after_ the data transfer took place the option
1096 to curl_easy_setopt(3) would instead be called CUR‐
1097 LOPT_POSTQUOTE(3) and used the exact same way.
1098
1099 The custom FTP command will be issued to the server in the same
1100 order they are added to the list, and if a command gets an error
1101 code returned back from the server, no more commands will be is‐
1102 sued and libcurl will bail out with an error code
1103 (CURLE_QUOTE_ERROR). Note that if you use CURLOPT_QUOTE(3) to
1104 send commands before a transfer, no transfer will actually take
1105 place when a quote command has failed.
1106
1107 If you set the CURLOPT_HEADER(3) to 1, you will tell libcurl to
1108 get information about the target file and output "headers" about
1109 it. The headers will be in "HTTP-style", looking like they do in
1110 HTTP.
1111
1112 The option to enable headers or to run custom FTP commands may
1113 be useful to combine with CURLOPT_NOBODY(3). If this option is
1114 set, no actual file content transfer will be performed.
1115
1116
1117 FTP Custom CUSTOMREQUEST
1118 If you do want to list the contents of an FTP directory using
1119 your own defined FTP command, CURLOPT_CUSTOMREQUEST(3) will do
1120 just that. "NLST" is the default one for listing directories but
1121 you're free to pass in your idea of a good alternative.
1122
1123
1125 In the HTTP sense, a cookie is a name with an associated value. A
1126 server sends the name and value to the client, and expects it to get
1127 sent back on every subsequent request to the server that matches the
1128 particular conditions set. The conditions include that the domain name
1129 and path match and that the cookie hasn't become too old.
1130
1131 In real-world cases, servers send new cookies to replace existing ones
1132 to update them. Server use cookies to "track" users and to keep "ses‐
1133 sions".
1134
1135 Cookies are sent from server to clients with the header Set-Cookie: and
1136 they're sent from clients to servers with the Cookie: header.
1137
1138 To just send whatever cookie you want to a server, you can use CUR‐
1139 LOPT_COOKIE(3) to set a cookie string like this:
1140
1141 curl_easy_setopt(easyhandle, CURLOPT_COOKIE, "name1=var1;
1142 name2=var2;");
1143
1144 In many cases, that is not enough. You might want to dynamically save
1145 whatever cookies the remote server passes to you, and make sure those
1146 cookies are then used accordingly on later requests.
1147
1148 One way to do this, is to save all headers you receive in a plain file
1149 and when you make a request, you tell libcurl to read the previous
1150 headers to figure out which cookies to use. Set the header file to read
1151 cookies from with CURLOPT_COOKIEFILE(3).
1152
1153 The CURLOPT_COOKIEFILE(3) option also automatically enables the cookie
1154 parser in libcurl. Until the cookie parser is enabled, libcurl will not
1155 parse or understand incoming cookies and they will just be ignored.
1156 However, when the parser is enabled the cookies will be understood and
1157 the cookies will be kept in memory and used properly in subsequent re‐
1158 quests when the same handle is used. Many times this is enough, and you
1159 may not have to save the cookies to disk at all. Note that the file you
1160 specify to CURLOPT_COOKIEFILE(3) doesn't have to exist to enable the
1161 parser, so a common way to just enable the parser and not read any
1162 cookies is to use the name of a file you know doesn't exist.
1163
1164 If you would rather use existing cookies that you've previously re‐
1165 ceived with your Netscape or Mozilla browsers, you can make libcurl use
1166 that cookie file as input. The CURLOPT_COOKIEFILE(3) is used for that
1167 too, as libcurl will automatically find out what kind of file it is and
1168 act accordingly.
1169
1170 Perhaps the most advanced cookie operation libcurl offers, is saving
1171 the entire internal cookie state back into a Netscape/Mozilla formatted
1172 cookie file. We call that the cookie-jar. When you set a file name with
1173 CURLOPT_COOKIEJAR(3), that file name will be created and all received
1174 cookies will be stored in it when curl_easy_cleanup(3) is called. This
1175 enables cookies to get passed on properly between multiple handles
1176 without any information getting lost.
1177
1178
1180 FTP transfers use a second TCP/IP connection for the data transfer.
1181 This is usually a fact you can forget and ignore but at times this fact
1182 will come back to haunt you. libcurl offers several different ways to
1183 customize how the second connection is being made.
1184
1185 libcurl can either connect to the server a second time or tell the
1186 server to connect back to it. The first option is the default and it is
1187 also what works best for all the people behind firewalls, NATs or IP-
1188 masquerading setups. libcurl then tells the server to open up a new
1189 port and wait for a second connection. This is by default attempted
1190 with EPSV first, and if that doesn't work it tries PASV instead. (EPSV
1191 is an extension to the original FTP spec and does not exist nor work on
1192 all FTP servers.)
1193
1194 You can prevent libcurl from first trying the EPSV command by setting
1195 CURLOPT_FTP_USE_EPSV(3) to zero.
1196
1197 In some cases, you will prefer to have the server connect back to you
1198 for the second connection. This might be when the server is perhaps be‐
1199 hind a firewall or something and only allows connections on a single
1200 port. libcurl then informs the remote server which IP address and port
1201 number to connect to. This is made with the CURLOPT_FTPPORT(3) option.
1202 If you set it to "-", libcurl will use your system's "default IP ad‐
1203 dress". If you want to use a particular IP, you can set the full IP ad‐
1204 dress, a host name to resolve to an IP address or even a local network
1205 interface name that libcurl will get the IP address from.
1206
1207 When doing the "PORT" approach, libcurl will attempt to use the EPRT
1208 and the LPRT before trying PORT, as they work with more protocols. You
1209 can disable this behavior by setting CURLOPT_FTP_USE_EPRT(3) to zero.
1210
1211
1213 In addition to support HTTP multi-part form fields, the MIME API can be
1214 used to build structured e-mail messages and send them via SMTP or ap‐
1215 pend such messages to IMAP directories.
1216
1217 A structured e-mail message may contain several parts: some are dis‐
1218 played inline by the MUA, some are attachments. Parts can also be
1219 structured as multi-part, for example to include another e-mail message
1220 or to offer several text formats alternatives. This can be nested to
1221 any level.
1222
1223 To build such a message, you prepare the nth-level multi-part and then
1224 include it as a source to the parent multi-part using function
1225 curl_mime_subparts(3). Once it has been bound to its parent multi-part,
1226 a nth-level multi-part belongs to it and should not be freed explic‐
1227 itly.
1228
1229 E-mail messages data is not supposed to be non-ascii and line length is
1230 limited: fortunately, some transfer encodings are defined by the stan‐
1231 dards to support the transmission of such incompatible data. Function
1232 curl_mime_encoder(3) tells a part that its source data must be encoded
1233 before being sent. It also generates the corresponding header for that
1234 part. If the part data you want to send is already encoded in such a
1235 scheme, do not use this function (this would over-encode it), but ex‐
1236 plicitly set the corresponding part header.
1237
1238 Upon sending such a message, libcurl prepends it with the header list
1239 set with CURLOPT_HTTPHEADER(3), as 0th-level mime part headers.
1240
1241 Here is an example building an e-mail message with an inline plain/html
1242 text alternative and a file attachment encoded in base64:
1243
1244 curl_mime *message = curl_mime_init(easyhandle);
1245
1246 /* The inline part is an alternative proposing the html and the text
1247 versions of the e-mail. */
1248 curl_mime *alt = curl_mime_init(easyhandle);
1249
1250 /* HTML message. */
1251 curl_mimepart *part = curl_mime_addpart(alt);
1252 curl_mime_data(part, "<html><body><p>This is HTML</p></body></html>",
1253 CURL_ZERO_TERMINATED);
1254 curl_mime_type(part, "text/html");
1255
1256 /* Text message. */
1257 part = curl_mime_addpart(alt);
1258 curl_mime_data(part, "This is plain text message",
1259 CURL_ZERO_TERMINATED);
1260
1261 /* Create the inline part. */
1262 part = curl_mime_addpart(message);
1263 curl_mime_subparts(part, alt);
1264 curl_mime_type(part, "multipart/alternative");
1265 struct curl_slist *headers = curl_slist_append(NULL,
1266 "Content-Disposition: inline");
1267 curl_mime_headers(part, headers, TRUE);
1268
1269 /* Add the attachment. */
1270 part = curl_mime_addpart(message);
1271 curl_mime_filedata(part, "manual.pdf");
1272 curl_mime_encoder(part, "base64");
1273
1274 /* Build the mail headers. */
1275 headers = curl_slist_append(NULL, "From: me@example.com");
1276 headers = curl_slist_append(headers, "To: you@example.com");
1277
1278 /* Set these into the easy handle. */
1279 curl_easy_setopt(easyhandle, CURLOPT_HTTPHEADER, headers);
1280 curl_easy_setopt(easyhandle, CURLOPT_MIMEPOST, mime);
1281
1282 It should be noted that appending a message to an IMAP directory re‐
1283 quires the message size to be known prior upload. It is therefore not
1284 possible to include parts with unknown data size in this context.
1285
1286
1288 Some protocols provide "headers", meta-data separated from the normal
1289 data. These headers are by default not included in the normal data
1290 stream, but you can make them appear in the data stream by setting CUR‐
1291 LOPT_HEADER(3) to 1.
1292
1293 What might be even more useful, is libcurl's ability to separate the
1294 headers from the data and thus make the callbacks differ. You can for
1295 example set a different pointer to pass to the ordinary write callback
1296 by setting CURLOPT_HEADERDATA(3).
1297
1298 Or, you can set an entirely separate function to receive the headers,
1299 by using CURLOPT_HEADERFUNCTION(3).
1300
1301 The headers are passed to the callback function one by one, and you can
1302 depend on that fact. It makes it easier for you to add custom header
1303 parsers etc.
1304
1305 "Headers" for FTP transfers equal all the FTP server responses. They
1306 aren't actually true headers, but in this case we pretend they are! ;-)
1307
1308
1310 See curl_easy_getinfo(3).
1311
1313 The easy interface as described in detail in this document is a syn‐
1314 chronous interface that transfers one file at a time and doesn't return
1315 until it is done.
1316
1317 The multi interface, on the other hand, allows your program to transfer
1318 multiple files in both directions at the same time, without forcing you
1319 to use multiple threads. The name might make it seem that the multi
1320 interface is for multi-threaded programs, but the truth is almost the
1321 reverse. The multi interface allows a single-threaded application to
1322 perform the same kinds of multiple, simultaneous transfers that multi-
1323 threaded programs can perform. It allows many of the benefits of
1324 multi-threaded transfers without the complexity of managing and syn‐
1325 chronizing many threads.
1326
1327 To complicate matters somewhat more, there are even two versions of the
1328 multi interface. The event based one, also called multi_socket and the
1329 "normal one" designed for using with select(). See the libcurl-multi.3
1330 man page for details on the multi_socket event based API, this descrip‐
1331 tion here is for the select() oriented one.
1332
1333 To use this interface, you are better off if you first understand the
1334 basics of how to use the easy interface. The multi interface is simply
1335 a way to make multiple transfers at the same time by adding up multiple
1336 easy handles into a "multi stack".
1337
1338 You create the easy handles you want, one for each concurrent transfer,
1339 and you set all the options just like you learned above, and then you
1340 create a multi handle with curl_multi_init(3) and add all those easy
1341 handles to that multi handle with curl_multi_add_handle(3).
1342
1343 When you've added the handles you have for the moment (you can still
1344 add new ones at any time), you start the transfers by calling
1345 curl_multi_perform(3).
1346
1347 curl_multi_perform(3) is asynchronous. It will only perform what can be
1348 done now and then return back control to your program. It is designed
1349 to never block. You need to keep calling the function until all trans‐
1350 fers are completed.
1351
1352 The best usage of this interface is when you do a select() on all pos‐
1353 sible file descriptors or sockets to know when to call libcurl again.
1354 This also makes it easy for you to wait and respond to actions on your
1355 own application's sockets/handles. You figure out what to select() for
1356 by using curl_multi_fdset(3), that fills in a set of fd_set variables
1357 for you with the particular file descriptors libcurl uses for the mo‐
1358 ment.
1359
1360 When you then call select(), it'll return when one of the file handles
1361 signal action and you then call curl_multi_perform(3) to allow libcurl
1362 to do what it wants to do. Take note that libcurl does also feature
1363 some time-out code so we advise you to never use very long timeouts on
1364 select() before you call curl_multi_perform(3) again. curl_multi_time‐
1365 out(3) is provided to help you get a suitable timeout period.
1366
1367 Another precaution you should use: always call curl_multi_fdset(3) im‐
1368 mediately before the select() call since the current set of file de‐
1369 scriptors may change in any curl function invoke.
1370
1371 If you want to stop the transfer of one of the easy handles in the
1372 stack, you can use curl_multi_remove_handle(3) to remove individual
1373 easy handles. Remember that easy handles should be
1374 curl_easy_cleanup(3)ed.
1375
1376 When a transfer within the multi stack has finished, the counter of
1377 running transfers (as filled in by curl_multi_perform(3)) will de‐
1378 crease. When the number reaches zero, all transfers are done.
1379
1380 curl_multi_info_read(3) can be used to get information about completed
1381 transfers. It then returns the CURLcode for each easy transfer, to al‐
1382 low you to figure out success on each individual transfer.
1383
1384
1386 [ seeding, passwords, keys, certificates, ENGINE, ca certs ]
1387
1388
1390 You can share some data between easy handles when the easy interface is
1391 used, and some data is share automatically when you use the multi in‐
1392 terface.
1393
1394 When you add easy handles to a multi handle, these easy handles will
1395 automatically share a lot of the data that otherwise would be kept on a
1396 per-easy handle basis when the easy interface is used.
1397
1398 The DNS cache is shared between handles within a multi handle, making
1399 subsequent name resolving faster, and the connection pool that is kept
1400 to better allow persistent connections and connection re-use is also
1401 shared. If you're using the easy interface, you can still share these
1402 between specific easy handles by using the share interface, see
1403 libcurl-share(3).
1404
1405 Some things are never shared automatically, not within multi handles,
1406 like for example cookies so the only way to share that is with the
1407 share interface.
1408
1410 [1] libcurl 7.10.3 and later have the ability to switch over to
1411 chunked Transfer-Encoding in cases where HTTP uploads are done
1412 with data of an unknown size.
1413
1414 [2] This happens on Windows machines when libcurl is built and used
1415 as a DLL. However, you can still do this on Windows if you link
1416 with a static library.
1417
1418 [3] The curl-config tool is generated at build-time (on Unix-like
1419 systems) and should be installed with the 'make install' or sim‐
1420 ilar instruction that installs the library, header files, man
1421 pages etc.
1422
1423 [4] This behavior was different in versions before 7.17.0, where
1424 strings had to remain valid past the end of the curl_easy_se‐
1425 topt(3) call.
1426
1428 libcurl-errors(3), libcurl-multi(3), libcurl-easy(3)
1429
1430
1431
1432libcurl 7.76.1 November 05, 2020 libcurl-tutorial(3)