--- parser3/src/classes/curl.C 2020/10/10 09:05:42 1.62 +++ parser3/src/classes/curl.C 2026/07/19 19:06:14 1.80 @@ -1,7 +1,8 @@ /** @file Parser: @b curl parser class. - Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) + Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com) + Authors: Konstantin Morshnev */ #include "pa_config_includes.h" @@ -14,10 +15,10 @@ #include "pa_vdate.h" #include "pa_vtable.h" #include "pa_common.h" -#include "pa_http.h" +#include "pa_http.h" #include "ltdl.h" -volatile const char * IDENT_CURL_C="$Id: curl.C,v 1.62 2020/10/10 09:05:42 moko Exp $"; +volatile const char * IDENT_CURL_C="$Id: curl.C,v 1.80 2026/07/19 19:06:14 moko Exp $"; class MCurl: public Methoded { public: @@ -43,16 +44,21 @@ 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"; - -static const char *dlink(const char *dlopen_file_spec) { + +static const char *dlink(char *dlopen_file_spec) { pa_dlinit(); - lt_dlhandle handle=lt_dlopen(dlopen_file_spec); + lt_dlhandle handle; + do { + char *next=lsplit(dlopen_file_spec, ','); + handle=lt_dlopen(dlopen_file_spec); + dlopen_file_spec=next; + } while (!handle && dlopen_file_spec); if(!handle){ if(const char* result=lt_dlerror()) return result; - return "can not open the dynamic link module"; + return "cannot open the dynamic link module"; } DLINK(curl_easy_init); @@ -68,6 +74,16 @@ static const char *dlink(const char *dlo return 0; } +#ifdef WIN32 +#define CURL_LIBRARY "libcurl" LT_MODULE_EXT +#else +#define CURL_LIBRARY "libcurl" LT_MODULE_EXT ",libcurl" LT_MODULE_EXT ".4" +#endif + +// 'parser' options can be set outside of 'session' operator +const char *curl_library=CURL_LIBRARY; +bool curl_parser_resolve = false; + struct ParserOptions : public PA_Allocated { // real options @@ -81,13 +97,20 @@ struct ParserOptions : public PA_Allocat struct curl_httppost *f_post; FILE *f_stderr; + // resolving + struct curl_slist *resolve; + bool parser_resolve; + long resolve_family; + long port; + // if response content-length check required bool no_body; // stuff to walkaround curl request content-length bugs bool is_post; bool has_content_length; - ParserOptions() : filename(0), content_type(0), is_text(true), charset(0), response_charset(0), url(0), f_post(0), f_stderr(0), no_body(false), is_post(false), has_content_length(false){} + ParserOptions() : filename(0), content_type(0), is_text(true), charset(0), response_charset(0), url(0), f_post(0), f_stderr(0), resolve(0), + parser_resolve(curl_parser_resolve), resolve_family(CURL_IPRESOLVE_V4), port(0), no_body(false), is_post(false), has_content_length(false){} ~ParserOptions() { f_curl_formfree(f_post); if(f_stderr) @@ -116,10 +139,14 @@ static ParserOptions &options(){ class Temp_curl { CURL *saved_curl; ParserOptions *saved_options; + + // every TLS should be referenced elsewhere, or GC will collect it + CURL *thread_curl; + ParserOptions *thread_options; public: Temp_curl() : saved_curl(fcurl), saved_options(foptions){ - fcurl = f_curl_easy_init(); - foptions = new ParserOptions(); + thread_curl = fcurl = f_curl_easy_init(); + thread_options = foptions = new ParserOptions(); f_curl_easy_setopt(fcurl, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4); // avoid ipv6 by default } @@ -133,18 +160,18 @@ public: bool curl_linked = false; const char *curl_status = 0; -const char *curl_library="libcurl" LT_MODULE_EXT; static void temp_curl(void (*action)(Request&, MethodParams&), Request& r, MethodParams& params){ if(!curl_linked) - curl_status=dlink(curl_library); + curl_status=dlink(pa_strdup(curl_library)); if(curl_status == 0){ curl_linked=true; Temp_curl temp_curl; action(r,params); } else { - throw Exception("curl", 0, "failed to load curl library %s: %s", curl_library, curl_status); + const char *hint=strcmp(curl_library, CURL_LIBRARY) ? "" : " (before use, call ^curl:options[ $.library[correct.libcurl" LT_MODULE_EXT ".name] ])"; + throw Exception("curl", 0, "failed to load curl library %s%s", curl_status, hint); } } @@ -158,7 +185,7 @@ static void _curl_session(Request& r, Me } static void _curl_version_action(Request& r, MethodParams& ){ - r.write(*new VString(*new String(f_curl_version(), String::L_TAINTED))); + r.write(*new VString(f_curl_version())); } static void _curl_version(Request& r, MethodParams& params){ @@ -177,6 +204,7 @@ struct CurlOption : public PA_Allocated{ CURL_POSTFIELDS, CURL_FORM, CURL_HEADERS, + CURL_RESOLVE, CURL_FILE, CURL_STDERR, CURL_HTTP_VERSION, @@ -223,6 +251,7 @@ public: CURL_OPT(CURL_INT, FOLLOWLOCATION); CURL_OPT(CURL_INT, UNRESTRICTED_AUTH); CURL_OPT(CURL_INT, IPRESOLVE); + CURL_OPT(CURL_RESOLVE, RESOLVE); CURL_OPT(CURL_POST, POST); CURL_OPT(CURL_INT, HTTPGET); @@ -371,6 +400,24 @@ static struct curl_slist *curl_headers(H return slist; } +static struct curl_slist *curl_resolve(HashStringValue *value_hash) { + struct curl_slist *slist=NULL; + + for(HashStringValue::Iterator i(*value_hash); i; i.next() ){ + const char *key=i.key().cstr(); + const String &address=i.value()->as_string(); + + if(address.is_empty() || strchr(key, ':')){ + slist=f_curl_slist_append(slist, address.is_empty() ? key : pa_strcat(key, ":", address.cstr()) ); + } else { + // port-less specification covers both http and https ports by default + slist=f_curl_slist_append(slist, pa_strcat(key, ":80:", address.cstr()) ); + slist=f_curl_slist_append(slist, pa_strcat(key, ":443:", address.cstr()) ); + } + } + return slist; +} + static const char* curl_transcode(const String &s, Request& r){ return options().charset ? Charset::transcode(s.cstr(), r.charsets.source(), *options().charset).cstr() : s.cstr(); } @@ -393,7 +440,7 @@ static void curl_form(HashStringValue *v CURLFORM_PTRCONTENTS, curl_transcode(String(tvalue->get(t)->get(0)->cstr()), r), CURLFORM_END); } - } else if(VFile* fvalue=static_cast(i.value()->as("file"))){ + } else if(VFile* fvalue=dynamic_cast(i.value())){ // file f_curl_formadd(&options().f_post, &f_last, CURLFORM_PTRNAME, key, @@ -473,6 +520,10 @@ static void curl_setopt(HashStringValue: // integer curl option long value_int=(long)v.as_double(); res=f_curl_easy_setopt(curl(), opt->id, value_int); + if(opt->id==CURLOPT_PORT) + options().port=value_int; // can be used instead of :port in url, saving for 'parser' resolve mode + if(opt->id==CURLOPT_IPRESOLVE) + options().resolve_family=value_int; break; } case CurlOption::CURL_NO_BODY:{ @@ -495,7 +546,7 @@ static void curl_setopt(HashStringValue: if( (res=f_curl_easy_setopt(curl(), CURLOPT_POSTFIELDSIZE, -1L)) == CURLE_OK ) res=f_curl_easy_setopt(curl(), opt->id, curl_urlencode(v.as_string(), r)); } else { - VFile *file=v.as_vfile(String::L_AS_IS); + VFile *file=v.as_vfile(); if( (res=f_curl_easy_setopt(curl(), CURLOPT_POSTFIELDSIZE, (long)file->value_size())) == CURLE_OK ) res=f_curl_easy_setopt(curl(), opt->id, file->value_ptr()); } @@ -521,15 +572,38 @@ static void curl_setopt(HashStringValue: res=f_curl_easy_setopt(curl(), opt->id, value_hash ? curl_headers(value_hash, r) : 0); break; } + case CurlOption::CURL_RESOLVE:{ + // 'curl|parser' resolver mode + if(const String* mode=v.get_string()){ + bool parser_resolve; + if(*mode == "parser") + parser_resolve=true; + else if(*mode == "curl" || mode->is_empty()) + parser_resolve=false; + else + throw Exception("curl", mode, "invalid resolve option value, must be 'parser' or 'curl'"); + + if(foptions) + foptions->parser_resolve=parser_resolve; + else + curl_parser_resolve=parser_resolve; + break; + } + + // hash of $.[host:port][address[,address]] entries to populate DNS cache with, bypassing the resolver + HashStringValue *value_hash=v.as_hash("failed to set option 'resolve': value"); + res=f_curl_easy_setopt(curl(), opt->id, options().resolve=curl_resolve(value_hash)); + break; + } case CurlOption::CURL_FILE:{ // file-spec curl option - const char *file_spec_cstr=curl_check_file(r.absolute(v.as_string())); + const char *file_spec_cstr=curl_check_file(r.full_disk_path(v.as_string())); res=f_curl_easy_setopt(curl(), opt->id, file_spec_cstr); break; } case CurlOption::CURL_STDERR:{ // verbose output redirection from stderr to file curl option - const char *file_spec_cstr=curl_check_file(r.absolute(v.as_string())); + const char *file_spec_cstr=curl_check_file(r.full_disk_path(v.as_string())); FILE *f_stderr=options().f_stderr=pa_fopen(file_spec_cstr, "wt"); if (f_stderr){ res=f_curl_easy_setopt(curl(), opt->id, f_stderr); @@ -548,6 +622,8 @@ static void curl_setopt(HashStringValue: // 'library' parser option if(!curl_linked){ curl_library=v.as_string().taint_cstr(String::L_FILE_SPEC); + if(!curl_library[0]) + curl_library=CURL_LIBRARY; } else throw Exception("curl", 0, "failed to set option '%s': already loaded", key.cstr()); break; @@ -624,7 +700,7 @@ static Value *curl_getinfo(const String: case CurlInfo::CURL_HTTP_VERSION:{ long l=0; CURL_GETINFO(l); - return new VString(*new String(curl_http_version_name(l), String::L_TAINTED)); + return new VString(curl_http_version_name(l)); } } return VVoid::get(); @@ -704,6 +780,61 @@ static int curl_header(char *data, size_ throw Exception("curl", 0, "failed to set " message ": %s", f_curl_easy_strerror(res)); \ } +// defined in inet.C +const char *pa_name2ip(const char *name, int family, Table *table, const char *exception_type); + +static void pa_curl_parser_resolve(){ + const char *url=options().url; + if(!url) + return; + + // url scheme was validated to be http:// or https:// before + bool is_http=!pa_strncasecmp(url, "http://"); + + char *host=pa_strdup(url+(is_http ? 7 : 8)); + if(char *uri=strpbrk(host, "/?#")) + *uri=0; + + char *name=host; + if(char *userinfo_host=lsplit(name, '@')) + name=userinfo_host; + + if(*name=='[') // no resolving for IPv6 literal + return; + + long port=options().port; + if(char *port_cstr=lsplit(name, ':')) + if(!port) + port=atol(port_cstr); + if(!port) + port=is_http ? 80 : 443; + + if(!*name) + return; + + const char *prefix=pa_strcat(name, ":", pa_itoa(port)); + size_t prefix_length=strlen(prefix); + + // explicit entries have priority over auto-resolving + last duplicate wins in curl, so we should check + for(struct curl_slist *e=options().resolve; e; e=e->next){ + const char *data=e->data; + // support optional '+host:port:address' curl 7.75+ syntax, '-host:port' should not match + if(*data=='+') + data++; + if(!pa_strncasecmp(data, prefix, prefix_length) && data[prefix_length]==':') + return; + } + + int family=options().resolve_family==CURL_IPRESOLVE_V4 ? AF_INET : options().resolve_family==CURL_IPRESOLVE_V6 ? AF_INET6 : AF_UNSPEC; + const char *address=pa_name2ip(name, family, 0, "curl.host"); + if(!address) + return; + + options().resolve=f_curl_slist_append(options().resolve, pa_strcat(prefix, ":", address)); + CURLcode res; + CURL_SETOPT(CURLOPT_RESOLVE, options().resolve, "curl resolve list"); +} + static void _curl_load_action(Request& r, MethodParams& params){ if(params.count()==1) _curl_options(r, params); @@ -719,13 +850,20 @@ static void _curl_load_action(Request& r CURL_SETOPT(CURLOPT_WRITEFUNCTION, curl_writer, "curl writer function"); CURL_SETOPT(CURLOPT_WRITEDATA, &body, "curl write buffer"); + char error[CURL_ERROR_SIZE+1]=""; + CURL_SETOPT(CURLOPT_ERRORBUFFER, error, "curl error buffer"); + if(options().is_post && !options().has_content_length){ // libcurl bug walkaround. Prior to 7.38 (Debian Jessie) curl passed Content-length: -1 // after that no Content-length header is passed, that hangs request to nginx. CURL_SETOPT(CURLOPT_POSTFIELDSIZE, 0, "post content-length"); } - if((res=f_curl_easy_perform(curl())) != CURLE_OK){ + if(options().parser_resolve) + pa_curl_parser_resolve(); + + ALTER_EXCEPTION_SOURCE(res=f_curl_easy_perform(curl()), new String(options().url)); + if(res != CURLE_OK){ const char *ex_type = 0; switch(res){ case CURLE_OPERATION_TIMEDOUT: @@ -743,10 +881,10 @@ static void _curl_load_action(Request& r case CURLE_SSL_ENGINE_INITFAILED: ex_type = "curl.ssl"; break; case CURLE_WRITE_ERROR: - check_file_size(response.content_length, *new String(options().url)); break; + check_file_size(response.content_length, new String(options().url)); break; default: break; } - throw Exception( PA_DEFAULT(ex_type, "curl.fail"), 0, "%s", f_curl_easy_strerror(res)); + throw Exception( PA_DEFAULT(ex_type, "curl.fail"), new String(options().url), "%s", error[0] ? error : f_curl_easy_strerror(res)); } // assure trailing zero @@ -777,13 +915,13 @@ static void _curl_load_action(Request& r long http_status = 0; if(f_curl_easy_getinfo(curl(), CURLINFO_RESPONSE_CODE, &http_status) == CURLE_OK){ - result.fields().put("status", new VInt(http_status)); + HASH_PUT_CSTR(result.fields(), "status", new VInt(http_status)); } VHash* vtables=new VHash; - result.fields().put("tables", vtables); + HASH_PUT_CSTR(result.fields(), "tables", vtables); - for(Array_iterator i(response.headers); i.has_next(); ){ + for(Array_iterator i(response.headers); i; ){ HTTP_Headers::Header header=i.next(); if(asked_charset) @@ -797,7 +935,7 @@ static void _curl_load_action(Request& r // filling $.cookies if(Value *vcookies=vtables->hash().get("SET-COOKIE")) - result.fields().put(HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table()))); + HASH_PUT_CSTR(result.fields(), HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table()))); r.write(result); }