--- parser3/src/classes/curl.C 2010/05/16 00:23:00 1.4 +++ parser3/src/classes/curl.C 2012/04/20 20:02:03 1.16 @@ -1,15 +1,11 @@ /** @file Parser: @b curl parser class. - Copyright(c) 2001-2009 ArtLebedev Group(http://www.artlebedev.com) + Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) */ #include "pa_config_includes.h" -#ifdef HAVE_CURL - -static const char * const IDENT_INET_C="$Date: 2010/05/16 00:23:00 $"; - #include "pa_vmethod_frame.h" #include "pa_request.h" #include "pa_vfile.h" @@ -17,23 +13,20 @@ static const char * const IDENT_INET_C=" #include "pa_vstring.h" #include "pa_vtable.h" #include "pa_common.h" -#include "ltdl.h" +#include "pa_http.h" +#include "ltdl.h" + +volatile const char * IDENT_CURL_C="$Id: curl.C,v 1.16 2012/04/20 20:02:03 moko Exp $"; class MCurl: public Methoded { public: MCurl(); - -public: // Methoded - bool used_directly() { return true; } }; // global variables DECLARE_CLASS_VAR(curl, new MCurl, 0); -// from file.C -extern bool is_text_mode(const String& mode); - #include "curl.h" typedef CURL *(*t_curl_easy_init)(); t_curl_easy_init f_curl_easy_init; @@ -50,7 +43,7 @@ typedef void (*t_curl_formfree)(struct c #define GLINK(name) f_##name=(t_##name)lt_dlsym(handle, #name); #define DLINK(name) GLINK(name) if(!f_##name) return "function " #name " was not found"; -const char *dlink(const char *dlopen_file_spec) { +static const char *dlink(const char *dlopen_file_spec) { if(lt_dlinit()) return lt_dlerror(); @@ -78,26 +71,25 @@ const char *dlink(const char *dlopen_fil class ParserOptions { public: - const char *filename; + const String *filename; const String *content_type; bool is_text; Charset *charset, *response_charset; struct curl_httppost *f_post; + FILE *stderr; - ParserOptions() : filename(0), content_type(0), is_text(true), charset(0), response_charset(0), f_post(0){} + ParserOptions() : filename(0), content_type(0), is_text(true), charset(0), response_charset(0), f_post(0), stderr(0){} ~ParserOptions() { f_curl_formfree(f_post); + if(stderr) + fclose(stderr); } }; -// using thread local variables instead of keeping them in request -// not necessary for cgi version -#ifdef WIN32 -#define __thread __declspec(thread) -#endif -__thread CURL *fcurl = 0; -__thread ParserOptions *foptions = 0; +// using TLS instead of keeping variables in request +THREAD_LOCAL CURL *fcurl = 0; +THREAD_LOCAL ParserOptions *foptions = 0; static CURL *curl(){ if(!fcurl) @@ -117,8 +109,8 @@ class Temp_curl { ParserOptions *saved_options; public: Temp_curl() : saved_curl(fcurl), saved_options(foptions){ - fcurl = f_curl_easy_init(); - foptions = new ParserOptions(); + fcurl = f_curl_easy_init(); + foptions = new ParserOptions(); f_curl_easy_setopt(fcurl, CURLOPT_POSTFIELDSIZE, 0); // fix libcurl bug } ~Temp_curl() { @@ -161,7 +153,7 @@ static void _curl_session(Request& r, Me temp_curl(_curl_session_action, r, params); } -static void _curl_version_action(Request& r, MethodParams& params){ +static void _curl_version_action(Request& r, MethodParams& ){ r.write_no_lang(*new VString(*new String(f_curl_version(), String::L_TAINTED))); } @@ -195,6 +187,7 @@ public: CURL_FORM, CURL_HEADERS, CURL_FILE, + CURL_STDERR, PARSER_LIBRARY, PARSER_NAME, PARSER_CONTENT_TYPE, @@ -218,13 +211,15 @@ public: CURL_OPT(CURL_INT, LOCALPORT); CURL_OPT(CURL_INT, PORT); + CURL_OPT(CURL_INT, VERBOSE); + CURL_OPT(CURL_STDERR, STDERR); + CURL_OPT(CURL_INT, MAXFILESIZE); + CURL_OPT(CURL_INT, HTTPAUTH); CURL_OPT(CURL_STRING, USERPWD); -#ifdef CURLOPT_USERNAME CURL_OPT(CURL_STRING, USERNAME); CURL_OPT(CURL_STRING, PASSWORD); -#endif CURL_OPT(CURL_URLENCODE, USERAGENT); CURL_OPT(CURL_URLENCODE, REFERER); @@ -283,13 +278,8 @@ public: CURL_OPT(CURL_STRING, SSLENGINE); CURL_OPT(CURL_STRING, SSLENGINE_DEFAULT); -#ifdef CURLOPT_ISSUERCERT CURL_OPT(CURL_FILE, ISSUERCERT); -#endif - -#ifdef CURLOPT_CRLFILE CURL_OPT(CURL_FILE, CRLFILE); -#endif CURL_OPT(CURL_STRING, CAINFO); CURL_OPT(CURL_STRING, CAPATH); @@ -321,9 +311,9 @@ static struct curl_slist *curl_headers(H for(HashStringValue::Iterator i(*value_hash); i; i.next() ){ String header = - String(capitalize(i.key().cstr()), String::L_URI) + String(pa_http_safe_header_name(capitalize(i.key().cstr())), String::L_AS_IS) << ": " - << String(i.value()->as_string(), String::L_URI); + << String(i.value()->as_string(), String::L_HTTP_HEADER); slist=f_curl_slist_append(slist, curl_urlencode(header, r)); } @@ -391,9 +381,9 @@ static void curl_setopt(HashStringValue: } case CurlOption::CURL_URL:{ // url-encoded string curl_url option - const String url = v.as_string(); - if(!url.starts_with("http://") && !url.starts_with("https://")) - throw Exception("curl", 0, "failed to set option '%s': invalid url scheme '%s'", key.cstr(), url.cstr()); + const String url = v.as_string(); + if(!url.starts_with("http://") && !url.starts_with("https://")) + throw Exception("curl", 0, "failed to set option '%s': invalid url scheme '%s'", key.cstr(), url.cstr()); const char *value_str=curl_urlencode(url, r); res=f_curl_easy_setopt(curl(), opt->id, value_str); break; @@ -420,12 +410,12 @@ static void curl_setopt(HashStringValue: HashStringValue *value_hash = v.get_hash(); if(value_hash){ curl_form(value_hash, r); - } else if(v.get_string()->is_empty()){ + } else if(v.get_string()->is_empty()){ f_curl_formfree(options().f_post); options().f_post = 0; } else { - throw Exception("curl", 0, "%s must be a hash", key.cstr()); - } + throw Exception("curl", 0, "failed to set option '%s': value must be a hash", key.cstr()); + } res=f_curl_easy_setopt(curl(), CURLOPT_HTTPPOST, foptions->f_post); break; } @@ -441,17 +431,28 @@ static void curl_setopt(HashStringValue: res=f_curl_easy_setopt(curl(), opt->id, value_str); break; } + case CurlOption::CURL_STDERR:{ + // verbose output redirection from stderr to file curl option + const char *value_str=r.absolute(v.as_string()).taint_cstr(String::L_FILE_SPEC); + FILE *stderr=options().stderr=fopen(value_str, "at"); + if (stderr){ + res=f_curl_easy_setopt(curl(), opt->id, stderr); + } else { + throw Exception("curl", 0, "failed to set option '%s': unable to open file %s", key.cstr(), value_str); + } + break; + } case CurlOption::PARSER_LIBRARY:{ // 'library' parser option if(fcurl==0){ curl_library=v.as_string().taint_cstr(String::L_FILE_SPEC); } else - throw Exception("curl", 0, "failed to set option '%s': %s", key.cstr(), "already loaded"); + throw Exception("curl", 0, "failed to set option '%s': already loaded", key.cstr()); break; } case CurlOption::PARSER_NAME:{ // 'name' parser option - options().filename=v.as_string().taint_cstr(String::L_FILE_SPEC); + options().filename=&v.as_string(); break; } case CurlOption::PARSER_CONTENT_TYPE:{ @@ -461,7 +462,7 @@ static void curl_setopt(HashStringValue: } case CurlOption::PARSER_MODE:{ // 'mode' parser option - options().is_text=is_text_mode(v.as_string()); + options().is_text=VFile::is_text_mode(v.as_string()); break; } case CurlOption::PARSER_CHARSET:{ @@ -470,7 +471,7 @@ static void curl_setopt(HashStringValue: break; } case CurlOption::PARSER_RESPONSE_CHARSET:{ - // 'charset' parser option + // 'response-charset' parser option options().response_charset=&::charsets.get(v.as_string().change_case(r.charsets.source(), String::CC_UPPER)); break; } @@ -487,7 +488,7 @@ static void _curl_options(Request& r, Me if(HashStringValue* options=params.as_no_junction(0, OPTIONS_MUST_NOT_BE_CODE).get_hash()){ options->for_each(curl_setopt, r); } else - throw Exception("curl", 0, "options must be hash"); + throw Exception("curl", 0, OPTIONS_MUST_BE_HASH); } @@ -553,7 +554,7 @@ static void _curl_load_action(Request& r CURL_SETOPT(CURLOPT_WRITEHEADER, &headers, "curl header buffer"); if((res=f_curl_easy_perform(curl())) != CURLE_OK){ - char *ex_type = 0; + const char *ex_type = 0; switch(res){ case CURLE_OPERATION_TIMEDOUT: ex_type = "curl.timeout"; break; @@ -576,15 +577,14 @@ static void _curl_load_action(Request& r // assure trailing zero body.buf[body.length]=0; - Value* vcontent_type= - options().content_type ? new VString(*options().content_type) : - options().filename ? new VString(r.mime_type_of(options().filename)) : 0; - VFile& result=*new VFile; String::Body ct_header = headers.get(HTTP_CONTENT_TYPE_UPPER); - Charset *remote_charset = ct_header.is_empty() ? 0 : detect_charset(ct_header.trim(String::TRIM_BOTH, " \t\n\r").cstr()); - Charset *asked_charset = options().response_charset ? options().response_charset : (remote_charset ? remote_charset : options().charset); + Charset *asked_charset = options().response_charset; + if (asked_charset == 0){ + Charset *remote_charset = ct_header.is_empty() ? 0 : detect_charset(ct_header.trim(String::TRIM_BOTH, " \t\n\r").cstr()); + asked_charset = remote_charset ? remote_charset : options().charset; + } if(options().is_text && asked_charset != 0){ String::C c=Charset::transcode(String::C(body.buf, body.length), *asked_charset, r.charsets.source()); @@ -592,7 +592,9 @@ static void _curl_load_action(Request& r body.length=c.length; } - result.set(true /*tainted*/, body.buf, body.length, options().filename, vcontent_type); + result.set(true /*tainted*/, body.buf, body.length, options().filename + , options().content_type ? new VString(*options().content_type) : 0 + , &r); result.set_mode(options().is_text); long http_status = 0; @@ -624,11 +626,3 @@ MCurl::MCurl(): Methoded("curl") { add_native_method("options", Method::CT_STATIC, _curl_options, 1, 1); add_native_method("load", Method::CT_STATIC, _curl_load, 0, 1); } - -#else // HAVE_CURL - -#include "classes.h" -// global variable -DECLARE_CLASS_VAR(curl, 0, 0); // fictive - -#endif // HAVE_CURL