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