Annotation of parser3/src/classes/curl.C, revision 1.82
1.1 misha 1: /** @file
2: Parser: @b curl parser class.
3:
1.78 moko 4: Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com)
1.70 moko 5: Authors: Konstantin Morshnev <moko@design.ru>
1.1 misha 6: */
7:
1.3 misha 8: #include "pa_config_includes.h"
9:
1.1 misha 10: #include "pa_vmethod_frame.h"
11: #include "pa_request.h"
12: #include "pa_vfile.h"
13: #include "pa_charsets.h"
1.4 misha 14: #include "pa_vstring.h"
1.32 moko 15: #include "pa_vdate.h"
1.4 misha 16: #include "pa_vtable.h"
17: #include "pa_common.h"
1.80 moko 18: #include "pa_http.h"
1.5 misha 19: #include "ltdl.h"
1.1 misha 20:
1.82 ! moko 21: #ifdef _MSC_VER
! 22: #include "winsock2.h" // AF_INET & co
! 23: #endif
! 24:
! 25: volatile const char * IDENT_CURL_C="$Id: curl.C,v 1.81 2026/07/19 19:42:00 moko Exp $";
1.13 moko 26:
1.1 misha 27: class MCurl: public Methoded {
28: public:
29: MCurl();
30: };
31:
32: // global variables
33:
1.36 moko 34: DECLARE_CLASS_VAR(curl, new MCurl);
1.1 misha 35:
36: #include "curl.h"
37:
38: typedef CURL *(*t_curl_easy_init)(); t_curl_easy_init f_curl_easy_init;
39: typedef CURLcode (*t_curl_easy_setopt)(CURL *, CURLoption option, ...); t_curl_easy_setopt f_curl_easy_setopt;
40: typedef CURLcode (*t_curl_easy_perform)(CURL *); t_curl_easy_perform f_curl_easy_perform;
41: typedef void (*t_curl_easy_cleanup)(CURL *); t_curl_easy_cleanup f_curl_easy_cleanup;
42: typedef const char *(*t_curl_easy_strerror)(CURLcode); t_curl_easy_strerror f_curl_easy_strerror;
1.4 misha 43: typedef CURLcode (*t_curl_easy_getinfo)(CURL *curl, CURLINFO info, ...); t_curl_easy_getinfo f_curl_easy_getinfo;
1.1 misha 44: typedef struct curl_slist *(*t_curl_slist_append)(struct curl_slist *,const char *); t_curl_slist_append f_curl_slist_append;
1.4 misha 45: typedef const char *(*t_curl_version)(); t_curl_version f_curl_version;
46: typedef CURLFORMcode (*t_curl_formadd)(struct curl_httppost **httppost, struct curl_httppost **last_post, ...); t_curl_formadd f_curl_formadd;
47: typedef void (*t_curl_formfree)(struct curl_httppost *form); t_curl_formfree f_curl_formfree;
1.1 misha 48:
49: #define GLINK(name) f_##name=(t_##name)lt_dlsym(handle, #name);
50: #define DLINK(name) GLINK(name) if(!f_##name) return "function " #name " was not found";
1.80 moko 51:
1.75 moko 52: static const char *dlink(char *dlopen_file_spec) {
1.28 moko 53: pa_dlinit();
1.1 misha 54:
1.75 moko 55: lt_dlhandle handle;
56: do {
57: char *next=lsplit(dlopen_file_spec, ',');
58: handle=lt_dlopen(dlopen_file_spec);
59: dlopen_file_spec=next;
60: } while (!handle && dlopen_file_spec);
1.1 misha 61:
62: if(!handle){
63: if(const char* result=lt_dlerror())
64: return result;
1.73 moko 65: return "cannot open the dynamic link module";
1.1 misha 66: }
67:
68: DLINK(curl_easy_init);
69: DLINK(curl_easy_cleanup);
1.4 misha 70: DLINK(curl_version);
1.1 misha 71: DLINK(curl_easy_setopt);
72: DLINK(curl_easy_perform);
73: DLINK(curl_easy_strerror);
1.4 misha 74: DLINK(curl_easy_getinfo);
1.1 misha 75: DLINK(curl_slist_append);
1.4 misha 76: DLINK(curl_formadd);
77: DLINK(curl_formfree);
1.1 misha 78: return 0;
79: }
80:
1.80 moko 81: #ifdef WIN32
82: #define CURL_LIBRARY "libcurl" LT_MODULE_EXT
83: #else
84: #define CURL_LIBRARY "libcurl" LT_MODULE_EXT ",libcurl" LT_MODULE_EXT ".4"
85: #endif
86:
87: // 'parser' options can be set outside of 'session' operator
88: const char *curl_library=CURL_LIBRARY;
89: bool curl_parser_resolve = false;
90:
1.1 misha 91:
1.55 moko 92: struct ParserOptions : public PA_Allocated {
1.39 moko 93: // real options
1.12 misha 94: const String *filename;
1.1 misha 95: const String *content_type;
96: bool is_text;
1.4 misha 97: Charset *charset, *response_charset;
1.39 moko 98:
99: // stuff for internal use
100: const char *url;
1.4 misha 101: struct curl_httppost *f_post;
1.19 moko 102: FILE *f_stderr;
1.1 misha 103:
1.80 moko 104: // resolving
105: struct curl_slist *resolve;
106: bool parser_resolve;
107: long resolve_family;
108: long port;
109:
1.60 moko 110: // if response content-length check required
111: bool no_body;
112: // stuff to walkaround curl request content-length bugs
1.57 moko 113: bool is_post;
114: bool has_content_length;
115:
1.80 moko 116: ParserOptions() : filename(0), content_type(0), is_text(true), charset(0), response_charset(0), url(0), f_post(0), f_stderr(0), resolve(0),
117: parser_resolve(curl_parser_resolve), resolve_family(CURL_IPRESOLVE_V4), port(0), no_body(false), is_post(false), has_content_length(false){}
1.4 misha 118: ~ParserOptions() {
119: f_curl_formfree(f_post);
1.19 moko 120: if(f_stderr)
121: fclose(f_stderr);
1.4 misha 122: }
1.55 moko 123:
1.1 misha 124: };
125:
1.9 moko 126: // using TLS instead of keeping variables in request
127: THREAD_LOCAL CURL *fcurl = 0;
128: THREAD_LOCAL ParserOptions *foptions = 0;
1.1 misha 129:
130: static CURL *curl(){
131: if(!fcurl)
132: throw Exception("curl", 0, "outside of 'session' operator");
133: return fcurl;
134: }
135:
136: static ParserOptions &options(){
137: if(!foptions)
138: throw Exception("curl", 0, "outside of 'session' operator");
139: return *foptions;
140: }
141:
142: // using temporal object scheme to garanty cleanup call
143: class Temp_curl {
1.4 misha 144: CURL *saved_curl;
145: ParserOptions *saved_options;
1.66 moko 146:
147: // every TLS should be referenced elsewhere, or GC will collect it
148: CURL *thread_curl;
149: ParserOptions *thread_options;
1.1 misha 150: public:
1.4 misha 151: Temp_curl() : saved_curl(fcurl), saved_options(foptions){
1.66 moko 152: thread_curl = fcurl = f_curl_easy_init();
153: thread_options = foptions = new ParserOptions();
1.27 moko 154: f_curl_easy_setopt(fcurl, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4); // avoid ipv6 by default
1.1 misha 155: }
1.56 moko 156:
1.1 misha 157: ~Temp_curl() {
158: f_curl_easy_cleanup(fcurl);
159: fcurl = saved_curl;
160: delete foptions;
161: foptions = saved_options;
162: }
163: };
164:
165: bool curl_linked = false;
1.41 moko 166: const char *curl_status = 0;
1.75 moko 167:
1.1 misha 168: static void temp_curl(void (*action)(Request&, MethodParams&), Request& r, MethodParams& params){
1.41 moko 169: if(!curl_linked)
1.75 moko 170: curl_status=dlink(pa_strdup(curl_library));
1.1 misha 171:
172: if(curl_status == 0){
1.41 moko 173: curl_linked=true;
1.1 misha 174: Temp_curl temp_curl;
175: action(r,params);
176: } else {
1.75 moko 177: const char *hint=strcmp(curl_library, CURL_LIBRARY) ? "" : " (before use, call ^curl:options[ $.library[correct.libcurl" LT_MODULE_EXT ".name] ])";
1.69 moko 178: throw Exception("curl", 0, "failed to load curl library %s%s", curl_status, hint);
1.1 misha 179: }
180: }
181:
182: static void _curl_session_action(Request& r, MethodParams& params){
183: Value& body_code=params.as_junction(0, "body must be code");
184: r.process_write(body_code);
185: }
186:
187: static void _curl_session(Request& r, MethodParams& params){
188: temp_curl(_curl_session_action, r, params);
189: }
190:
1.5 misha 191: static void _curl_version_action(Request& r, MethodParams& ){
1.77 moko 192: r.write(*new VString(f_curl_version()));
1.4 misha 193: }
194:
195: static void _curl_version(Request& r, MethodParams& params){
196: fcurl ? _curl_version_action(r, params) : temp_curl(_curl_version_action, r, params);
197: }
1.1 misha 198:
1.55 moko 199: struct CurlOption : public PA_Allocated{
1.1 misha 200:
201: enum OptionType {
202: CURL_STRING,
203: CURL_URLENCODE, // url-encoded string
1.4 misha 204: CURL_URL,
1.1 misha 205: CURL_INT,
1.60 moko 206: CURL_NO_BODY,
1.1 misha 207: CURL_POST,
1.57 moko 208: CURL_POSTFIELDS,
1.4 misha 209: CURL_FORM,
1.1 misha 210: CURL_HEADERS,
1.80 moko 211: CURL_RESOLVE,
1.1 misha 212: CURL_FILE,
1.15 moko 213: CURL_STDERR,
1.56 moko 214: CURL_HTTP_VERSION,
1.1 misha 215: PARSER_LIBRARY,
216: PARSER_NAME,
217: PARSER_CONTENT_TYPE,
218: PARSER_MODE,
1.4 misha 219: PARSER_CHARSET,
220: PARSER_RESPONSE_CHARSET
1.1 misha 221: };
222:
223: CURLoption id;
224: OptionType type;
225: CurlOption(CURLoption aid, OptionType atype): id(aid), type(atype) {}
226: };
227:
228: class CurlOptionHash: public HashString<CurlOption*> {
229: public:
230: CurlOptionHash() {
231: #define CURL_OPT(type, name) put(str_lower(#name),new CurlOption(CURLOPT_##name, CurlOption::type));
232: #define PARSER_OPT(type, name) put(name,new CurlOption((CURLoption)0, CurlOption::type));
1.4 misha 233: CURL_OPT(CURL_URL, URL);
1.1 misha 234: CURL_OPT(CURL_STRING, INTERFACE);
235: CURL_OPT(CURL_INT, LOCALPORT);
236: CURL_OPT(CURL_INT, PORT);
1.4 misha 237:
1.15 moko 238: CURL_OPT(CURL_INT, VERBOSE);
239: CURL_OPT(CURL_STDERR, STDERR);
240: CURL_OPT(CURL_INT, MAXFILESIZE);
241:
1.1 misha 242: CURL_OPT(CURL_INT, HTTPAUTH);
243: CURL_OPT(CURL_STRING, USERPWD);
244:
245: CURL_OPT(CURL_STRING, USERNAME);
246: CURL_OPT(CURL_STRING, PASSWORD);
247:
1.4 misha 248: CURL_OPT(CURL_URLENCODE, USERAGENT);
249: CURL_OPT(CURL_URLENCODE, REFERER);
1.1 misha 250: CURL_OPT(CURL_INT, AUTOREFERER);
1.25 misha 251:
1.1 misha 252: CURL_OPT(CURL_STRING, ENCODING); // gzip or deflate
1.25 misha 253: CURL_OPT(CURL_STRING, ACCEPT_ENCODING); // gzip or deflate
254:
1.1 misha 255: CURL_OPT(CURL_INT, FOLLOWLOCATION);
256: CURL_OPT(CURL_INT, UNRESTRICTED_AUTH);
1.27 moko 257: CURL_OPT(CURL_INT, IPRESOLVE);
1.80 moko 258: CURL_OPT(CURL_RESOLVE, RESOLVE);
1.1 misha 259:
1.57 moko 260: CURL_OPT(CURL_POST, POST);
1.1 misha 261: CURL_OPT(CURL_INT, HTTPGET);
1.60 moko 262: CURL_OPT(CURL_NO_BODY, NOBODY);
1.4 misha 263: CURL_OPT(CURL_STRING, CUSTOMREQUEST);
1.1 misha 264:
1.57 moko 265: CURL_OPT(CURL_POSTFIELDS, POSTFIELDS); // hopefully is safe too
266: CURL_OPT(CURL_POSTFIELDS, COPYPOSTFIELDS);
1.4 misha 267: CURL_OPT(CURL_FORM, HTTPPOST);
1.1 misha 268:
269: CURL_OPT(CURL_HEADERS, HTTPHEADER);
270: CURL_OPT(CURL_URLENCODE, COOKIE);
271: CURL_OPT(CURL_URLENCODE, COOKIELIST);
272: CURL_OPT(CURL_INT, COOKIESESSION);
273:
274: CURL_OPT(CURL_INT, IGNORE_CONTENT_LENGTH);
275: CURL_OPT(CURL_INT, HTTP_CONTENT_DECODING);
276: CURL_OPT(CURL_INT, HTTP_TRANSFER_DECODING);
277:
1.4 misha 278: CURL_OPT(CURL_INT, MAXREDIRS);
279: CURL_OPT(CURL_INT, POSTREDIR);
280:
281: CURL_OPT(CURL_STRING, RANGE);
282:
1.1 misha 283: CURL_OPT(CURL_INT, TIMEOUT);
284: CURL_OPT(CURL_INT, TIMEOUT_MS);
285: CURL_OPT(CURL_INT, LOW_SPEED_LIMIT);
286: CURL_OPT(CURL_INT, LOW_SPEED_TIME);
287: CURL_OPT(CURL_INT, MAXCONNECTS);
288:
1.4 misha 289: CURL_OPT(CURL_STRING, PROXY);
290: CURL_OPT(CURL_INT, PROXYPORT);
291: CURL_OPT(CURL_INT, PROXYTYPE);
292: CURL_OPT(CURL_INT, HTTPPROXYTUNNEL);
293: CURL_OPT(CURL_STRING, PROXYUSERPWD);
294: CURL_OPT(CURL_INT, PROXYAUTH);
295:
1.1 misha 296: CURL_OPT(CURL_INT, FRESH_CONNECT);
297: CURL_OPT(CURL_INT, FORBID_REUSE);
298: CURL_OPT(CURL_INT, CONNECTTIMEOUT);
299: CURL_OPT(CURL_INT, CONNECTTIMEOUT_MS);
1.4 misha 300: CURL_OPT(CURL_INT, FAILONERROR);
1.1 misha 301:
302: CURL_OPT(CURL_FILE, SSLCERT);
303: CURL_OPT(CURL_STRING, SSLCERTTYPE);
304: CURL_OPT(CURL_FILE, SSLKEY);
305: CURL_OPT(CURL_STRING, SSLKEYTYPE);
306: CURL_OPT(CURL_STRING, KEYPASSWD);
307: CURL_OPT(CURL_STRING, SSLENGINE);
308: CURL_OPT(CURL_STRING, SSLENGINE_DEFAULT);
309:
310: CURL_OPT(CURL_FILE, ISSUERCERT);
311: CURL_OPT(CURL_FILE, CRLFILE);
312:
313: CURL_OPT(CURL_STRING, CAINFO);
1.18 moko 314: CURL_OPT(CURL_FILE, CAPATH);
1.1 misha 315: CURL_OPT(CURL_INT, SSL_VERIFYPEER);
316: CURL_OPT(CURL_INT, SSL_VERIFYHOST);
317: CURL_OPT(CURL_STRING, SSL_CIPHER_LIST);
318: CURL_OPT(CURL_INT, SSL_SESSIONID_CACHE);
1.43 moko 319: CURL_OPT(CURL_INT, SSLVERSION);
1.56 moko 320: CURL_OPT(CURL_HTTP_VERSION, HTTP_VERSION);
1.1 misha 321:
322: PARSER_OPT(PARSER_LIBRARY, "library");
323: PARSER_OPT(PARSER_NAME, "name");
324: PARSER_OPT(PARSER_CONTENT_TYPE, "content-type");
325: PARSER_OPT(PARSER_MODE, "mode");
326: PARSER_OPT(PARSER_CHARSET, "charset");
1.4 misha 327: PARSER_OPT(PARSER_RESPONSE_CHARSET, "response-charset");
1.1 misha 328: }
329:
330: } *curl_options=0;
331:
1.55 moko 332: struct CurlInfo : public PA_Allocated{
1.32 moko 333:
334: enum OptionType {
335: CURL_STRING,
336: CURL_INT,
1.56 moko 337: CURL_DOUBLE,
338: CURL_HTTP_VERSION
1.32 moko 339: };
340:
341: CURLINFO id;
342: OptionType type;
343: CurlInfo(CURLINFO aid, OptionType atype): id(aid), type(atype) {}
344: };
345:
1.61 moko 346: class CurlInfoHash: public OrderedHashString<CurlInfo*> {
1.32 moko 347: public:
348: CurlInfoHash() {
349: #define CURL_INF(type, name) put(str_lower(#name),new CurlInfo(CURLINFO_##name, CurlInfo::type));
1.61 moko 350: CURL_INF(CURL_STRING, SCHEME);
351: CURL_INF(CURL_HTTP_VERSION, HTTP_VERSION);
352: CURL_INF(CURL_STRING, EFFECTIVE_URL);
353: CURL_INF(CURL_STRING, CONTENT_TYPE);
354: CURL_INF(CURL_INT, RESPONSE_CODE);
355: CURL_INF(CURL_INT, OS_ERRNO);
356:
357: CURL_INF(CURL_DOUBLE, NAMELOOKUP_TIME);
1.32 moko 358: CURL_INF(CURL_DOUBLE, APPCONNECT_TIME);
1.61 moko 359: CURL_INF(CURL_DOUBLE, PRETRANSFER_TIME);
360: CURL_INF(CURL_DOUBLE, STARTTRANSFER_TIME);
1.32 moko 361: CURL_INF(CURL_DOUBLE, CONNECT_TIME);
1.61 moko 362: CURL_INF(CURL_DOUBLE, TOTAL_TIME);
363:
1.32 moko 364: CURL_INF(CURL_DOUBLE, CONTENT_LENGTH_DOWNLOAD);
365: CURL_INF(CURL_DOUBLE, CONTENT_LENGTH_UPLOAD);
366: CURL_INF(CURL_INT, HEADER_SIZE);
1.61 moko 367: CURL_INF(CURL_INT, REQUEST_SIZE);
368: CURL_INF(CURL_DOUBLE, SIZE_DOWNLOAD);
369: CURL_INF(CURL_DOUBLE, SIZE_UPLOAD);
370: CURL_INF(CURL_DOUBLE, SPEED_DOWNLOAD);
371: CURL_INF(CURL_DOUBLE, SPEED_UPLOAD);
372:
1.32 moko 373: CURL_INF(CURL_INT, NUM_CONNECTS);
374: CURL_INF(CURL_STRING, PRIMARY_IP);
1.61 moko 375: CURL_INF(CURL_INT, HTTPAUTH_AVAIL);
1.32 moko 376: CURL_INF(CURL_INT, PROXYAUTH_AVAIL);
377: CURL_INF(CURL_INT, REDIRECT_COUNT);
378: CURL_INF(CURL_DOUBLE, REDIRECT_TIME);
379: CURL_INF(CURL_STRING, REDIRECT_URL);
380: CURL_INF(CURL_INT, SSL_VERIFYRESULT);
381: }
382:
383: } *curl_infos=0;
384:
1.1 misha 385: static const char *curl_urlencode(const String &s, Request& r){
386: if(options().charset){
387: Temp_client_charset temp(r.charsets, *options().charset);
388: return s.untaint_and_transcode_cstr(String::L_URI, &r.charsets);
389: } else
390: return s.untaint_cstr(String::L_URI);
391: }
392:
393: static struct curl_slist *curl_headers(HashStringValue *value_hash, Request& r) {
394: struct curl_slist *slist=NULL;
395:
396: for(HashStringValue::Iterator i(*value_hash); i; i.next() ){
397: String header =
1.10 moko 398: String(pa_http_safe_header_name(capitalize(i.key().cstr())), String::L_AS_IS)
1.1 misha 399: << ": "
1.10 moko 400: << String(i.value()->as_string(), String::L_HTTP_HEADER);
1.1 misha 401:
402: slist=f_curl_slist_append(slist, curl_urlencode(header, r));
403: }
404: return slist;
405: }
406:
1.80 moko 407: static struct curl_slist *curl_resolve(HashStringValue *value_hash) {
408: struct curl_slist *slist=NULL;
409:
410: for(HashStringValue::Iterator i(*value_hash); i; i.next() ){
411: const char *key=i.key().cstr();
412: const String &address=i.value()->as_string();
413:
414: if(address.is_empty() || strchr(key, ':')){
415: slist=f_curl_slist_append(slist, address.is_empty() ? key : pa_strcat(key, ":", address.cstr()) );
416: } else {
417: // port-less specification covers both http and https ports by default
418: slist=f_curl_slist_append(slist, pa_strcat(key, ":80:", address.cstr()) );
419: slist=f_curl_slist_append(slist, pa_strcat(key, ":443:", address.cstr()) );
420: }
421: }
422: return slist;
423: }
424:
1.4 misha 425: static const char* curl_transcode(const String &s, Request& r){
426: return options().charset ? Charset::transcode(s.cstr(), r.charsets.source(), *options().charset).cstr() : s.cstr();
427: }
428:
429: static void curl_form(HashStringValue *value_hash, Request& r){
430: struct curl_httppost *f_last=0;
431: for(HashStringValue::Iterator i(*value_hash); i; i.next() ){
432: const char *key = curl_transcode(String(i.key().cstr()), r);
433: if(const String* svalue = i.value()->get_string()){
434: // string
435: f_curl_formadd(&options().f_post, &f_last,
436: CURLFORM_PTRNAME, key,
437: CURLFORM_PTRCONTENTS, curl_transcode(String(svalue->cstr()), r),
438: CURLFORM_END);
439: } else if(Table* tvalue = i.value()->get_table()){
440: // table
441: for(size_t t = 0; t < tvalue->count(); t++) {
442: f_curl_formadd(&options().f_post, &f_last,
443: CURLFORM_PTRNAME, key,
444: CURLFORM_PTRCONTENTS, curl_transcode(String(tvalue->get(t)->get(0)->cstr()), r),
445: CURLFORM_END);
446: }
1.72 moko 447: } else if(VFile* fvalue=dynamic_cast<VFile *>(i.value())){
1.4 misha 448: // file
449: f_curl_formadd(&options().f_post, &f_last,
450: CURLFORM_PTRNAME, key,
451: CURLFORM_BUFFER, curl_transcode(String(fvalue->fields().get("name")->as_string(), String::L_FILE_SPEC), r),
452: CURLFORM_BUFFERLENGTH, (long)fvalue->value_size(),
453: CURLFORM_BUFFERPTR, fvalue->value_ptr(),
454: CURLFORM_CONTENTTYPE, fvalue->fields().get("content-type")->as_string().taint_cstr(String::L_URI),
455: CURLFORM_END);
456: } else {
1.56 moko 457: throw Exception("curl", new String(i.key(), String::L_TAINTED), "is %s, form option value can be string, table or file only", i.value()->type());
1.4 misha 458: }
459: }
460: }
461:
1.18 moko 462: static const char *curl_check_file(const String &file_spec){
463: const char *file_spec_cstr=file_spec.taint_cstr(String::L_FILE_SPEC);
464: struct stat finfo;
1.49 moko 465: if(pa_stat(file_spec_cstr, &finfo)==0)
1.18 moko 466: check_safe_mode(finfo, file_spec, file_spec_cstr);
467: return file_spec_cstr;
468: }
469:
1.56 moko 470: static long curl_http_version(const String &name){
471: if(name.is_empty()) return CURL_HTTP_VERSION_NONE;
472:
473: if(name == "1.0") return CURL_HTTP_VERSION_1_0;
474: if(name == "1.1") return CURL_HTTP_VERSION_1_1;
475: if(name == "2") return CURL_HTTP_VERSION_2;
476: if(name == "2.0") return CURL_HTTP_VERSION_2_0;
477:
478: const char *sname = str_upper(name.cstr());
479: if(!strcmp(sname,"2TLS")) return CURL_HTTP_VERSION_2TLS;
480: if(!strcmp(sname,"2ONLY")) return CURL_HTTP_VERSION_2_PRIOR_KNOWLEDGE;
481: throw Exception("curl", &name, "invalid http_version option value");
482: }
483:
484: static const char *curl_http_version_name(long value){
485: if(value == CURL_HTTP_VERSION_NONE) return "none";
486: if(value == CURL_HTTP_VERSION_1_0) return "1.0";
487: if(value == CURL_HTTP_VERSION_1_1) return "1.1";
488: if(value == CURL_HTTP_VERSION_2) return "2";
489: throw Exception("curl", 0, "invalid http version '%d' in info", value);
490: }
491:
1.1 misha 492: static void curl_setopt(HashStringValue::key_type key, HashStringValue::value_type value, Request& r) {
493: CurlOption *opt=curl_options->get(key);
494:
495: if(opt==0)
496: throw Exception("curl", 0, "called with invalid option '%s'", key.cstr());
497:
498: CURLcode res = CURLE_OK;
1.44 moko 499: Value &v=r.process(*value);
1.1 misha 500:
501: switch (opt->type){
502: case CurlOption::CURL_STRING:{
503: // string curl option
504: const char *value_str=v.as_string().cstr();
505: res=f_curl_easy_setopt(curl(), opt->id, value_str);
506: break;
507: }
508: case CurlOption::CURL_URLENCODE:{
509: // url-encoded string curl option
510: const char *value_str=curl_urlencode(v.as_string(), r);
511: res=f_curl_easy_setopt(curl(), opt->id, value_str);
512: break;
513: }
1.4 misha 514: case CurlOption::CURL_URL:{
515: // url-encoded string curl_url option
1.5 misha 516: const String url = v.as_string();
517: if(!url.starts_with("http://") && !url.starts_with("https://"))
518: throw Exception("curl", 0, "failed to set option '%s': invalid url scheme '%s'", key.cstr(), url.cstr());
1.39 moko 519: options().url=curl_urlencode(url, r);
520: res=f_curl_easy_setopt(curl(), opt->id, options().url);
1.4 misha 521: break;
522: }
1.1 misha 523: case CurlOption::CURL_INT:{
1.2 misha 524: // integer curl option
525: long value_int=(long)v.as_double();
1.1 misha 526: res=f_curl_easy_setopt(curl(), opt->id, value_int);
1.80 moko 527: if(opt->id==CURLOPT_PORT)
528: options().port=value_int; // can be used instead of :port in url, saving for 'parser' resolve mode
529: if(opt->id==CURLOPT_IPRESOLVE)
530: options().resolve_family=value_int;
1.1 misha 531: break;
532: }
1.60 moko 533: case CurlOption::CURL_NO_BODY:{
534: // integer curl option
535: long value_int=(long)v.as_double();
536: res=f_curl_easy_setopt(curl(), opt->id, value_int);
537: options().no_body=value_int != 0;
538: break;
539: }
1.1 misha 540: case CurlOption::CURL_POST:{
1.57 moko 541: // integer curl option
542: long value_int=(long)v.as_double();
543: res=f_curl_easy_setopt(curl(), opt->id, value_int);
544: options().is_post=value_int != 0;
545: break;
546: }
547: case CurlOption::CURL_POSTFIELDS:{
1.1 misha 548: // http post curl option
549: if(v.get_string()){
1.2 misha 550: if( (res=f_curl_easy_setopt(curl(), CURLOPT_POSTFIELDSIZE, -1L)) == CURLE_OK )
1.1 misha 551: res=f_curl_easy_setopt(curl(), opt->id, curl_urlencode(v.as_string(), r));
552: } else {
1.68 moko 553: VFile *file=v.as_vfile();
1.2 misha 554: if( (res=f_curl_easy_setopt(curl(), CURLOPT_POSTFIELDSIZE, (long)file->value_size())) == CURLE_OK )
1.1 misha 555: res=f_curl_easy_setopt(curl(), opt->id, file->value_ptr());
556: }
1.57 moko 557: options().has_content_length=true;
1.1 misha 558: break;
559: }
1.4 misha 560: case CurlOption::CURL_FORM:{
1.48 moko 561: HashStringValue *value_hash = v.as_hash("failed to set option 'httppost': value");
1.4 misha 562: if(value_hash){
563: curl_form(value_hash, r);
1.48 moko 564: } else {
1.57 moko 565: if(options().f_post)
566: f_curl_formfree(options().f_post);
1.4 misha 567: options().f_post = 0;
1.5 misha 568: }
1.4 misha 569: res=f_curl_easy_setopt(curl(), CURLOPT_HTTPPOST, foptions->f_post);
1.57 moko 570: options().has_content_length=true;
1.4 misha 571: break;
572: }
1.1 misha 573: case CurlOption::CURL_HEADERS:{
574: // http headers curl option
1.48 moko 575: HashStringValue *value_hash=v.as_hash("failed to set option 'httpheader': value");
1.1 misha 576: res=f_curl_easy_setopt(curl(), opt->id, value_hash ? curl_headers(value_hash, r) : 0);
577: break;
578: }
1.80 moko 579: case CurlOption::CURL_RESOLVE:{
580: // 'curl|parser' resolver mode
581: if(const String* mode=v.get_string()){
582: bool parser_resolve;
583: if(*mode == "parser")
584: parser_resolve=true;
585: else if(*mode == "curl" || mode->is_empty())
586: parser_resolve=false;
587: else
588: throw Exception("curl", mode, "invalid resolve option value, must be 'parser' or 'curl'");
589:
590: if(foptions)
591: foptions->parser_resolve=parser_resolve;
592: else
593: curl_parser_resolve=parser_resolve;
594: break;
595: }
596:
597: // hash of $.[host:port][address[,address]] entries to populate DNS cache with, bypassing the resolver
598: HashStringValue *value_hash=v.as_hash("failed to set option 'resolve': value");
1.81 moko 599: res=f_curl_easy_setopt(curl(), opt->id, options().resolve=value_hash ? curl_resolve(value_hash) : 0);
1.80 moko 600: break;
601: }
1.1 misha 602: case CurlOption::CURL_FILE:{
603: // file-spec curl option
1.65 moko 604: const char *file_spec_cstr=curl_check_file(r.full_disk_path(v.as_string()));
1.18 moko 605: res=f_curl_easy_setopt(curl(), opt->id, file_spec_cstr);
1.1 misha 606: break;
607: }
1.16 moko 608: case CurlOption::CURL_STDERR:{
609: // verbose output redirection from stderr to file curl option
1.65 moko 610: const char *file_spec_cstr=curl_check_file(r.full_disk_path(v.as_string()));
1.53 moko 611: FILE *f_stderr=options().f_stderr=pa_fopen(file_spec_cstr, "wt");
1.19 moko 612: if (f_stderr){
613: res=f_curl_easy_setopt(curl(), opt->id, f_stderr);
1.16 moko 614: } else {
1.18 moko 615: throw Exception("curl", 0, "failed to set option '%s': unable to open file '%s'", key.cstr(), file_spec_cstr);
1.16 moko 616: }
617: break;
618: }
1.56 moko 619: case CurlOption::CURL_HTTP_VERSION:{
620: // http protocol version name curl option
621: long value_int=curl_http_version(v.as_string());
622: res=f_curl_easy_setopt(curl(), opt->id, value_int);
623: break;
624: }
1.1 misha 625: case CurlOption::PARSER_LIBRARY:{
626: // 'library' parser option
1.41 moko 627: if(!curl_linked){
1.1 misha 628: curl_library=v.as_string().taint_cstr(String::L_FILE_SPEC);
1.75 moko 629: if(!curl_library[0])
630: curl_library=CURL_LIBRARY;
1.1 misha 631: } else
1.16 moko 632: throw Exception("curl", 0, "failed to set option '%s': already loaded", key.cstr());
1.1 misha 633: break;
634: }
635: case CurlOption::PARSER_NAME:{
636: // 'name' parser option
1.12 misha 637: options().filename=&v.as_string();
1.1 misha 638: break;
639: }
640: case CurlOption::PARSER_CONTENT_TYPE:{
641: // 'content-type' parser option
642: options().content_type=&v.as_string();
643: break;
644: }
645: case CurlOption::PARSER_MODE:{
646: // 'mode' parser option
1.12 misha 647: options().is_text=VFile::is_text_mode(v.as_string());
1.1 misha 648: break;
649: }
650: case CurlOption::PARSER_CHARSET:{
1.30 moko 651: // 'charset' parser option should be processed before other options
1.1 misha 652: break;
653: }
1.4 misha 654: case CurlOption::PARSER_RESPONSE_CHARSET:{
1.12 misha 655: // 'response-charset' parser option
1.42 moko 656: options().response_charset=&pa_charsets.get(v.as_string());
1.4 misha 657: break;
658: }
1.1 misha 659: }
660:
661: if(res != CURLE_OK)
662: throw Exception("curl", 0, "failed to set option '%s': %s", key.cstr(), f_curl_easy_strerror(res));
663: }
664:
1.3 misha 665: static void _curl_options(Request& r, MethodParams& params){
1.1 misha 666: if(curl_options==0)
667: curl_options=new CurlOptionHash();
668:
1.30 moko 669: if(HashStringValue* options_hash=params.as_hash(0)){
670: if(Value* value=options_hash->get("charset")){
671: // charset should be handled first as params may require transcode
1.44 moko 672: Value &v=r.process(*value);
1.42 moko 673: options().charset=&pa_charsets.get(v.as_string());
1.30 moko 674: }
675: options_hash->for_each<Request&>(curl_setopt, r);
676: }
1.1 misha 677: }
678:
1.32 moko 679: #define CURL_GETINFO(arg) \
680: if((res=f_curl_easy_getinfo(curl(), info->id, &arg)) != CURLE_OK){ \
1.58 moko 681: if (fail_on_error) \
682: throw Exception("curl", 0, "failed to get %s info: %s", key.cstr(), f_curl_easy_strerror(res)); \
683: return 0; \
1.32 moko 684: }
685:
1.58 moko 686: static Value *curl_getinfo(const String::Body &key, CurlInfo *info, bool fail_on_error=false) {
1.32 moko 687: CURLcode res;
688: switch (info->type){
689: case CurlInfo::CURL_STRING:{
690: char *str=0;
691: CURL_GETINFO(str);
1.33 moko 692: return new VString(str ? *new String(pa_strdup(str), String::L_TAINTED) : String::Empty);
1.32 moko 693: }
694: case CurlInfo::CURL_INT:{
695: long l=0;
696: CURL_GETINFO(l);
697: return new VInt(l);
698: }
699: case CurlInfo::CURL_DOUBLE:{
700: double d=0;
701: CURL_GETINFO(d);
702: return new VDouble(d);
703: }
1.56 moko 704: case CurlInfo::CURL_HTTP_VERSION:{
705: long l=0;
706: CURL_GETINFO(l);
1.77 moko 707: return new VString(curl_http_version_name(l));
1.56 moko 708: }
1.32 moko 709: }
710: return VVoid::get();
711: }
712:
713: static void _curl_info(Request& r, MethodParams& params){
714: if(curl_infos==0)
715: curl_infos=new CurlInfoHash();
716: if(params.count()==1){
717: const String &name=params.as_string(0, "name must be string");
1.58 moko 718: CurlInfo *info=curl_infos->get(name);
719: if(info==0)
720: throw Exception("curl", 0, "called with invalid parameter '%s'", name.cstr());
1.59 moko 721: r.write(*curl_getinfo(name, info, true));
1.32 moko 722: } else {
723: VHash& result=*new VHash;
724: for(CurlInfoHash::Iterator i(*curl_infos); i; i.next() ){
1.59 moko 725: Value *value=curl_getinfo(i.key(), i.value());
1.58 moko 726: if(value)
727: result.get_hash()->put(i.key(), value);
1.32 moko 728: }
1.46 moko 729: r.write(result);
1.32 moko 730: }
731: }
732:
1.1 misha 733: class Curl_buffer{
734: public:
735: char *buf;
736: size_t length;
737: size_t buf_size;
1.62 moko 738: HTTP_Headers& headers;
1.1 misha 739:
1.62 moko 740: Curl_buffer(HTTP_Headers& aheaders) : buf((char *)pa_malloc_atomic(MAX_STRING)), length(0), buf_size(MAX_STRING-1), headers(aheaders){}
1.50 moko 741:
742: void resize(size_t size){
743: buf_size=size;
744: buf=(char *)pa_realloc(buf, size+1);
745: }
1.1 misha 746: };
747:
748: static int curl_writer(char *data, size_t size, size_t nmemb, Curl_buffer *result){
749: if(result == 0)
750: return 0;
751:
752: size=size*nmemb;
753: if(size>0){
1.50 moko 754: size_t buf_required = result->length + size;
755: if(buf_required > result->buf_size)
1.51 moko 756: result->resize(buf_required <= result->headers.content_length ? (size_t)result->headers.content_length : result->buf_size*2 + size);
1.1 misha 757: memcpy(result->buf+result->length, data, size);
758: result->length += size;
759: }
760: return size;
761: }
762:
1.62 moko 763: static int curl_header(char *data, size_t size, size_t nmemb, HTTP_Headers *result){
1.1 misha 764: if(result == 0)
765: return 0;
766:
767: size=size*nmemb;
768: if(size>0){
1.47 moko 769: char *header=pa_strdup(data, size);
770: if(!pa_strncasecmp(header, "HTTP/") && !strchr(header, ':')){
771: // response code, clearing possible headers from previous requests
772: result->clear();
773: } else {
774: result->add_header(header);
1.60 moko 775: if(result->content_length>pa_file_size_limit && !options().no_body)
1.50 moko 776: return 0;
1.47 moko 777: }
1.1 misha 778: }
779: return size;
780: }
781:
782: #define CURL_SETOPT(option, arg, message) \
783: if( (res=f_curl_easy_setopt(curl(), option, arg)) != CURLE_OK){ \
784: throw Exception("curl", 0, "failed to set " message ": %s", f_curl_easy_strerror(res)); \
785: }
786:
1.80 moko 787: // defined in inet.C
788: const char *pa_name2ip(const char *name, int family, Table *table, const char *exception_type);
789:
790: static void pa_curl_parser_resolve(){
791: const char *url=options().url;
792: if(!url)
793: return;
794:
795: // url scheme was validated to be http:// or https:// before
796: bool is_http=!pa_strncasecmp(url, "http://");
797:
798: char *host=pa_strdup(url+(is_http ? 7 : 8));
799: if(char *uri=strpbrk(host, "/?#"))
800: *uri=0;
801:
802: char *name=host;
803: if(char *userinfo_host=lsplit(name, '@'))
804: name=userinfo_host;
805:
806: if(*name=='[') // no resolving for IPv6 literal
807: return;
808:
809: long port=options().port;
810: if(char *port_cstr=lsplit(name, ':'))
811: if(!port)
812: port=atol(port_cstr);
813: if(!port)
814: port=is_http ? 80 : 443;
815:
816: if(!*name)
817: return;
818:
819: const char *prefix=pa_strcat(name, ":", pa_itoa(port));
820: size_t prefix_length=strlen(prefix);
821:
822: // explicit entries have priority over auto-resolving + last duplicate wins in curl, so we should check
823: for(struct curl_slist *e=options().resolve; e; e=e->next){
824: const char *data=e->data;
825: // support optional '+host:port:address' curl 7.75+ syntax, '-host:port' should not match
826: if(*data=='+')
827: data++;
828: if(!pa_strncasecmp(data, prefix, prefix_length) && data[prefix_length]==':')
829: return;
830: }
831:
832: int family=options().resolve_family==CURL_IPRESOLVE_V4 ? AF_INET : options().resolve_family==CURL_IPRESOLVE_V6 ? AF_INET6 : AF_UNSPEC;
833: const char *address=pa_name2ip(name, family, 0, "curl.host");
834: if(!address)
835: return;
836:
837: options().resolve=f_curl_slist_append(options().resolve, pa_strcat(prefix, ":", address));
838: CURLcode res;
839: CURL_SETOPT(CURLOPT_RESOLVE, options().resolve, "curl resolve list");
840: }
841:
1.1 misha 842: static void _curl_load_action(Request& r, MethodParams& params){
843: if(params.count()==1)
1.3 misha 844: _curl_options(r, params);
1.1 misha 845:
846: CURLcode res;
847:
848: // we need a container for headers as VFile fields can be put only after VFile.set
1.62 moko 849: HTTP_Headers response;
1.1 misha 850: CURL_SETOPT(CURLOPT_HEADERFUNCTION, curl_header, "curl header function");
1.24 misha 851: CURL_SETOPT(CURLOPT_WRITEHEADER, &response, "curl header buffer");
1.1 misha 852:
1.50 moko 853: Curl_buffer body(response);
854: CURL_SETOPT(CURLOPT_WRITEFUNCTION, curl_writer, "curl writer function");
855: CURL_SETOPT(CURLOPT_WRITEDATA, &body, "curl write buffer");
856:
1.79 moko 857: char error[CURL_ERROR_SIZE+1]="";
858: CURL_SETOPT(CURLOPT_ERRORBUFFER, error, "curl error buffer");
859:
1.57 moko 860: if(options().is_post && !options().has_content_length){
861: // libcurl bug walkaround. Prior to 7.38 (Debian Jessie) curl passed Content-length: -1
862: // after that no Content-length header is passed, that hangs request to nginx.
863: CURL_SETOPT(CURLOPT_POSTFIELDSIZE, 0, "post content-length");
864: }
865:
1.80 moko 866: if(options().parser_resolve)
867: pa_curl_parser_resolve();
868:
1.63 moko 869: ALTER_EXCEPTION_SOURCE(res=f_curl_easy_perform(curl()), new String(options().url));
870: if(res != CURLE_OK){
1.15 moko 871: const char *ex_type = 0;
1.4 misha 872: switch(res){
873: case CURLE_OPERATION_TIMEDOUT:
874: ex_type = "curl.timeout"; break;
875: case CURLE_COULDNT_RESOLVE_HOST:
876: ex_type = "curl.host"; break;
877: case CURLE_COULDNT_CONNECT:
878: ex_type = "curl.connect"; break;
879: case CURLE_HTTP_RETURNED_ERROR:
880: ex_type = "curl.status"; break;
881: case CURLE_SSL_CONNECT_ERROR:
882: case CURLE_SSL_CERTPROBLEM:
883: case CURLE_SSL_CIPHER:
884: case CURLE_SSL_CACERT:
885: case CURLE_SSL_ENGINE_INITFAILED:
886: ex_type = "curl.ssl"; break;
1.50 moko 887: case CURLE_WRITE_ERROR:
1.64 moko 888: check_file_size(response.content_length, new String(options().url)); break;
1.31 moko 889: default: break;
1.4 misha 890: }
1.79 moko 891: throw Exception( PA_DEFAULT(ex_type, "curl.fail"), new String(options().url), "%s", error[0] ? error : f_curl_easy_strerror(res));
1.1 misha 892: }
893:
894: // assure trailing zero
895: body.buf[body.length]=0;
896:
1.4 misha 897: VFile& result=*new VFile;
898:
1.8 moko 899: Charset *asked_charset = options().response_charset;
1.38 moko 900: if (!asked_charset && !response.content_type.is_empty())
901: asked_charset=detect_charset(response.content_type.cstr());
902:
903: if(options().is_text)
1.42 moko 904: asked_charset=pa_charsets.checkBOM(body.buf, body.length, asked_charset);
1.38 moko 905:
906: if (!asked_charset)
907: asked_charset = options().charset;
1.1 misha 908:
909: if(options().is_text && asked_charset != 0){
910: String::C c=Charset::transcode(String::C(body.buf, body.length), *asked_charset, r.charsets.source());
911: body.buf=(char *)c.str;
912: body.length=c.length;
913: }
914:
1.38 moko 915: const String *content_type = PA_DEFAULT(options().content_type, response.content_type.is_empty() ? 0 : new String(response.content_type, String::L_TAINTED));
1.39 moko 916: const String *filename = PA_DEFAULT(options().filename, new String(options().url));
1.38 moko 917:
1.39 moko 918: result.set(true/*tainted*/, options().is_text, body.buf, body.length, filename, content_type ? new VString(*content_type) : 0, &r);
1.38 moko 919:
1.4 misha 920: long http_status = 0;
921: if(f_curl_easy_getinfo(curl(), CURLINFO_RESPONSE_CODE, &http_status) == CURLE_OK){
1.76 moko 922: HASH_PUT_CSTR(result.fields(), "status", new VInt(http_status));
1.4 misha 923: }
924:
1.38 moko 925: VHash* vtables=new VHash;
1.76 moko 926: HASH_PUT_CSTR(result.fields(), "tables", vtables);
1.23 misha 927:
1.71 moko 928: for(Array_iterator<HTTP_Headers::Header> i(response.headers); i; ){
1.62 moko 929: HTTP_Headers::Header header=i.next();
1.24 misha 930:
931: if(asked_charset)
1.38 moko 932: header.transcode(*asked_charset, r.charsets.source());
933:
934: String &header_value=*new String(header.value, String::L_TAINTED);
935:
936: tables_update(vtables->hash(), header.name, header_value);
937: result.fields().put(header.name, new VString(header_value));
1.1 misha 938: }
939:
1.38 moko 940: // filling $.cookies
941: if(Value *vcookies=vtables->hash().get("SET-COOKIE"))
1.76 moko 942: HASH_PUT_CSTR(result.fields(), HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table())));
1.23 misha 943:
1.46 moko 944: r.write(result);
1.1 misha 945: }
946:
947: static void _curl_load(Request& r, MethodParams& params){
948: fcurl ? _curl_load_action(r, params) : temp_curl(_curl_load_action, r, params);
949: }
950:
951: // constructor
952: MCurl::MCurl(): Methoded("curl") {
953: add_native_method("session", Method::CT_STATIC, _curl_session, 1, 1);
1.4 misha 954: add_native_method("version", Method::CT_STATIC, _curl_version, 0, 0);
1.3 misha 955: add_native_method("options", Method::CT_STATIC, _curl_options, 1, 1);
1.32 moko 956: add_native_method("info", Method::CT_STATIC, _curl_info, 0, 1);
1.1 misha 957: add_native_method("load", Method::CT_STATIC, _curl_load, 0, 1);
1.3 misha 958: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to curl.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / classes