--- parser3/src/main/pa_common.C 2003/04/15 06:44:34 1.140.2.3 +++ parser3/src/main/pa_common.C 2003/01/24 14:36:10 1.143.2.3 @@ -1,19 +1,17 @@ /** @file Parser: commonly functions. - Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_COMMON_C="$Date: 2003/04/15 06:44:34 $"; +static const char* IDENT_COMMON_C="$Date: 2003/01/24 14:36:10 $"; #include "pa_common.h" #include "pa_exception.h" #include "pa_globals.h" #include "pa_hash.h" #include "pa_vstring.h" -#include "pa_vdate.h" -#include "pa_request.h" #ifdef WIN32 # include @@ -116,12 +114,13 @@ void fix_line_breaks(char* buf, size_t& memcpy(dest, bol, eob-bol); } -char* file_read_text(Pool& pool, const String& file_spec, +char* file_read_text(Pool& pool, Charset& charset, + ConstStringPtr file_spec, bool fail_on_read_problem, - Hash* options, Request* r, - Hash** out_fields) { + HashStringValue *params, HashStringValue** out_fields) { void *result; size_t size; - return file_read(pool, file_spec, result, size, true, options, r, out_fields, fail_on_read_problem)?(char *)result:0; + return file_read(pool, charset, file_spec, result, size, true, params, out_fields, fail_on_read_problem)? + (char *)result:0; } //http request stuff @@ -141,106 +140,107 @@ static bool set_addr(struct sockaddr_in return true; } -static void http_read_response(String& response, int sock){ - bool check_status=true; +static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){ + int result=0; ssize_t EOLat=0; while(true) { - char *buf=(char *)response.pool().malloc(MAX_STRING); + char *buf=(char *)pool.malloc(MAX_STRING); ssize_t size=recv(sock, buf, MAX_STRING, 0); if(size<=0) break; response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0); - if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response - check_status=false; + if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response + ConstStringPtr status_line=response.mid(0, (size_t)EOLat); + Array astatus; + status_line->split(astatus, 0, " ", 1); + ConstStringPtr status_code=astatus.get(1); + result=status_code->as_int(); - const String& status_line=response.mid(0, (size_t)EOLat); - Array astatus(response.pool()); - size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1); - const String* status_code=astatus.get_string(1); - - if(!status_code || *status_code!="200") + if(fail_on_status_ne_200 && result!=200) throw Exception("http.status", status_code, "invalid HTTP response status"); } - } - if(check_status) + } + if(result) + return result; + else throw Exception("http.response", - 0, + Exception::undefined_source, "bad response from host - no status found (size=%lu)", response.size()); } /* ********************** request *************************** */ #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) -# define PA_USE_ALARM +# define WE_CAN_USE_ALARM #endif -#ifdef PA_USE_ALARM +#ifdef WE_CAN_USE_ALARM static sigjmp_buf timeout_env; static void timeout_handler(int sig){ siglongjmp(timeout_env, 1); } #endif -static void http_request(String& response, - const String *origin_string, - const char* host, int port, - const char* request, - int timeout) { +static int http_request(Pool& pool, + String& response, + ConstStringPtr origin_string, + const char* host, int port, + const char* request, + int timeout, + bool fail_on_status_ne_200) { if(!host) throw Exception("http.host", origin_string, "zero hostname"); //never -#ifdef PA_USE_ALARM +#ifdef WE_CAN_USE_ALARM signal(SIGALRM, timeout_handler); #endif int sock=-1; -#ifdef PA_USE_ALARM - if(sigsetjmp(timeout_env, 1)) { - // stupid gcc [2.95.4] generated bad code - // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code. - // rewritten simplier [though duplicating closesocket code] - if(sock>=0) - closesocket(sock); - throw Exception("http.timeout", - origin_string, - "timeout occured while retrieving document"); - } else - alarm(timeout); -#endif try { - struct sockaddr_in dest; - - if(!set_addr(&dest, host, port)) - throw Exception("http.host", + int result; +#ifdef WE_CAN_USE_ALARM + if(sigsetjmp(timeout_env, 1)) + throw Exception("http.timeout", origin_string, - "can not resolve hostname \"%s\"", host); + "timeout occured while retrieving document"); + else { + alarm(timeout); +#endif + struct sockaddr_in dest; - if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) - throw Exception("http.connect", - origin_string, - "can not make socket: %s (%d)", strerror(errno), errno); - if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) - throw Exception("http.connect", - origin_string, - "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno); - size_t request_size=strlen(request); - if(send(sock, request, request_size, 0)!=(ssize_t)request_size) - throw Exception("http.connect", - origin_string, - "error sending request: %s (%d)", strerror(errno), errno); + if(!set_addr(&dest, host, port)) + throw Exception("http.host", + origin_string, + "can not resolve hostname \"%s\"", host); + + if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) + throw Exception("http.connect", + origin_string, + "can not make socket: %s (%d)", strerror(errno), errno); + if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) + throw Exception("http.connect", + origin_string, + "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno); + size_t request_size=strlen(request); + if(send(sock, request, request_size, 0)!=(ssize_t)request_size) + throw Exception("http.connect", + origin_string, + "error sending request: %s (%d)", strerror(errno), errno); - http_read_response(response, sock); - closesocket(sock); -#ifdef PA_USE_ALARM - alarm(0); + result=http_read_response(pool, response, sock, fail_on_status_ne_200); + closesocket(sock); +#ifdef WE_CAN_USE_ALARM + alarm(0); + } #endif + return result; } catch(...) { if(sock>=0) closesocket(sock); -#ifdef PA_USE_ALARM +#ifdef WE_CAN_USE_ALARM alarm(0); #endif /*re*/throw; @@ -249,81 +249,92 @@ static void http_request(String& respons #ifndef DOXYGEN struct Http_pass_header_info { + Pool* pool; + Charset* charset; String* request; bool user_agent_specified; }; #endif -static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info) -{ - Http_pass_header_info& i=*static_cast(info); - Pool& pool=i.request->pool(); - - *(i.request)<(value), String::UL_HTTP_HEADER, false) +static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value, + Http_pass_header_info *info) { + *(info->request)<<*key<<": " + << *attributed_meaning_to_string(value, String::UL_HTTP_HEADER, false) <<"\n"; - if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT") - i.user_agent_specified=true; + if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT") + info->user_agent_specified=true; } -static void file_read_http(Pool& pool, const String& file_spec, - void*& data, size_t& data_size, - Hash *options=0, Hash** out_fields=0) { +static HashStringValuePtr file_read_http(Pool& pool, Charset& charset, const String& file_spec, + char*& data, size_t& data_size, + HashStringValue *options=0, bool return_outfields=false) { + HashStringValuePtr result(0); char host[MAX_STRING]; const char* uri; int port; - const char* method="GET"; + ConstCharPtr method(0); int timeout=2; - Value *vheaders=0; + bool fail_on_status_ne_200=true; + ValuePtr vheaders(0); - String connect_string(pool); + StringPtr connect_string(new String()); // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one - connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces + connect_string->append(file_spec, String::UL_URI); // tainted pieces -> URI pieces - char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED); - if(strncmp(connect_string_cstr, "http://", 7)!=0) + CharPtr connect_string_cstr=connect_string->cstr(String::UL_UNSPECIFIED); + const char *current=connect_string_cstr.get(); + if(strncmp(current, "http://", 7)!=0) throw Exception(0, - &connect_string, + connect_string, "does not start with http://"); //never - connect_string_cstr+=7; + current+=7; - strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0; + strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0; char* host_uri=lsplit(host, '/'); - uri=host_uri?connect_string_cstr+(host_uri-1-host):"/"; + uri=host_uri?current+(host_uri-1-host):"/"; char* port_cstr=lsplit(host, ':'); char* error_pos=0; port=port_cstr?strtol(port_cstr, &error_pos, 0):80; if(options) { int valid_options=0; - if(Value *vmethod=static_cast(options->get(*http_method_name))) { + if(ValuePtr vmethod=options->get(http_method_name)) { valid_options++; - method=vmethod->as_string().cstr(); + method=vmethod->as_string(pool)->cstr(); } - if(Value *vtimeout=static_cast(options->get(*http_timeout_name))) { + if(ValuePtr vtimeout=options->get(http_timeout_name)) { valid_options++; timeout=vtimeout->as_int(); } - if(vheaders=static_cast(options->get(*http_headers_name))) { + if(vheaders=options->get(http_headers_name)) { + valid_options++; + } + if(ValuePtr vany_status=options->get(http_any_status_name)) { valid_options++; + fail_on_status_ne_200=!vany_status->as_bool(); } - if(valid_options!=options->size()) + + if(valid_options!=options->count()) throw Exception("parser.runtime", - 0, + Exception::undefined_source, "invalid option passed"); } //making request - String request(pool); - request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; + String request; + if(method) + request<is_string()) { // allow empty - if(Hash *headers=vheaders->get_hash(&connect_string)) { - Http_pass_header_info info={&request}; + if(HashStringValuePtr headers=vheaders->get_hash(connect_string)) { + Http_pass_header_info info={&pool, &charset, &request}; headers->for_each(http_pass_header, &info); user_agent_specified=info.user_agent_specified; } else throw Exception("parser.runtime", - &connect_string, + connect_string, "headers param must be hash"); }; if(!user_agent_specified) // defaulting @@ -331,89 +342,72 @@ static void file_read_http(Pool& pool, c request<<"\n"; //sending request - String response(pool); - http_request(response, - &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout); + String response; + ConstCharPtr request_cstr=request.cstr(String::UL_UNSPECIFIED); + int status_code=http_request(pool, response, + connect_string, host, port, request_cstr.get(), + timeout, fail_on_status_ne_200); //processing results int pos=response.pos("\r\n\r\n", 4); if(pos<1){ throw Exception("http.response", - &connect_string, + connect_string, "bad response from host - no headers found"); } - String header_block=response.mid(0, pos); - String body=response.mid(pos+4, response.size()); + StringPtr header_block=response.mid(0, pos); + StringPtr body=response.mid(pos+4, response.size()); - Array aheaders(pool); - Hash& headers=*new(pool) Hash(pool); - size_t pos_after_ref=0; - header_block.split(aheaders, &pos_after_ref, "\r\n", 2); + Array aheaders; + HashStringValuePtr headers(new HashStringValue); + header_block->split(aheaders, 0, "\r\n", 2); - //processing status code - const String *status_line=aheaders.get_string(0); - if(!status_line){ - throw Exception("http.response", - &connect_string, - "bad response from host - no status line "); - } //processing headers - for(int i=1;ipos(": ", 2); - if(pos<1) - throw Exception("http.response", - &connect_string, - "bad response from host - bad header \"%s\"", line->cstr()); - - headers.put( - line->mid(0, pos).change_case(pool, String::CC_UPPER), - new(pool) VString(line->mid(pos+2, line->size()))); - } else + for(int i=1; ipos(": ", 2); + if(pos<1) throw Exception("http.response", - &connect_string, - "bad response from host - bad headers \"%s\"", header_block.cstr()); + connect_string, + "bad response from host - bad header \"%s\"", line->cstr()); + + headers->put( + line->mid(0, pos)->change_case(pool, charset, String::CC_UPPER), + ValuePtr(new VString(line->mid(pos+2, line->size())))); } // output response - data=body.cstr(); data_size=body.size(); - if(out_fields) + data=new(pool) char[data_size=body->size()]; + body->store_to(data, String::UL_AS_IS); + if(return_outfields) { + headers.put(file_status_name, VIntPtr); *out_fields=&headers; + } + + return result; } #ifndef DOXYGEN struct File_read_action_info { void **data; size_t *data_size; - Hash* options; Request* r; }; #endif -static void file_read_action(Pool& pool, +static void file_read_action( struct stat& finfo, int f, const String& file_spec, const char* fname, bool as_text, void *context) { File_read_action_info& info=*static_cast(context); + if(size_t to_read_size=(size_t)finfo.st_size) { + *info.data=pool.malloc(to_read_size+(as_text?1:0), 3); + *info.data_size=(size_t)read(f, *info.data, to_read_size); - int offset=0; - int limit=-1; - if(info.options) { - int valid_options=0; - if(Value *voffset=(Value *)info.options->get(*sql_offset_name)) { - valid_options++; - offset=info.r->process_to_value(*voffset).as_int(); - } - if(Value *vlimit=(Value *)info.options->get(*sql_limit_name)) { - valid_options++; - limit=info.r->process_to_value(*vlimit).as_int(); - } - if(valid_options!=info.options->size()) - throw Exception("parser.runtime", - &file_spec, - "load called with invalid option"); - } - - int m=finfo.st_size; - if(offset<0 || offset>=m || limit==0) { // bad options/empty file: returning empty file + if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) + throw Exception(0, + &file_spec, + "read failed: actually read %lu bytes count not in [0..%lu] valid range", + *info.data_size, to_read_size); + } else { // empty file if(as_text) { *info.data=pool.malloc(1); *(char*)(*info.data)=0; @@ -422,38 +416,18 @@ static void file_read_action(Pool& pool, *info.data_size=0; return; } - if(limit<0 || limit>m) - limit=m; - - if(offset) - if(lseek(f, offset, SEEK_SET)<0) - throw Exception("file.read", - &file_spec, - "seek to %d position failed: %s (%d), actual filename '%s'", - offset, strerror(errno), errno, fname); - - *info.data=pool.malloc(limit+(as_text?1:0), 3); - *info.data_size=(size_t)read(f, *info.data, limit); - - if((ssize_t)*info.data_size<0) - throw Exception(0, - &file_spec, - "failed to read %d bytes from %d offset of %d size (stat)", // may be in text mode... - limit, offset, m); } -bool file_read(Pool& pool, const String& file_spec, +bool file_read(ConstStringPtr file_spec, void*& data, size_t& data_size, - bool as_text, - Hash* options, Request* r, - Hash** out_fields, + bool as_text, HashStringValue *params, HashStringValue** out_fields, bool fail_on_read_problem) { bool result; if(file_spec.starts_with("http://", 7)) { // fail on read problem - file_read_http(pool, file_spec, data, data_size, options, out_fields); + file_read_http(pool, file_spec, data, data_size, params, out_fields); result=true; } else { - File_read_action_info info={&data, &data_size, options, r}; + File_read_action_info info={&data, &data_size}; result=file_read_action_under_lock(pool, file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); @@ -477,9 +451,10 @@ bool file_read(Pool& pool, const String& return result; } -bool file_read_action_under_lock(Pool& pool, const String& file_spec, +bool file_read_action_under_lock(const String& file_spec, const char* action_name, File_read_action action, void *context, - bool as_text, bool fail_on_read_problem) { + bool as_text, + bool fail_on_read_problem) { const char* fname=file_spec.cstr(String::UL_FILE_SPEC); int f; @@ -777,7 +752,7 @@ char* rsplit(char* string, char delim) { } /// @todo less stupid type detection -char* format(Pool& pool, double value, char* fmt) { +char* format(double value, char* fmt) { char local_buf[MAX_NUMBER]; size_t size; @@ -813,7 +788,7 @@ size_t stdout_write(const void *buf, siz #endif } -char* unescape_chars(Pool& pool, const char* cp, int len) { +char* unescape_chars(const char* cp, int len) { char* s=(char *)pool.malloc(len + 1, 5); enum EscapeState { EscapeRest, @@ -985,74 +960,3 @@ int pa_sleep(unsigned long secs, unsigne } -// attributed meaning - -static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) { - const char month_names[12][4]={ - "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"}; - const char days[7][4]={ - "Sun","Mon","Tue","Wed","Thu","Fri","Sat"}; - - time_t when=vdate.get_time(); - struct tm *tms=gmtime(&when); - if(!tms) - throw Exception(0, - 0, - "bad time in attribute value (seconds from epoch=%ld)", when); - return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT", - days[tms->tm_wday], - tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900, - tms->tm_hour,tms->tm_min,tms->tm_sec); -} -static void append_attribute_meaning(String& result, - Value& value, String::Untaint_lang lang, bool forced) { - if(const String *string=value.get_string()) - result.append(string->join_chains(result.pool(), 0), lang, forced); - else - if(Value *vdate=value.as(VDATE_TYPE, false)) { - char *buf=(char *)result.malloc(MAX_STRING); - size_t size=date_attribute(*static_cast(vdate), - buf, MAX_STRING); - - result.APPEND_CLEAN(buf, size, "converted from date", 0); - } else - throw Exception("parser.runtime", - &result, - "trying to append here neither string nor date (%s)", - value.type()); -} -#ifndef DOXYGEN -struct Attributed_meaning_info { - String *header; // header line being constructed - String::Untaint_lang lang; // language in which to append to that line - bool forced; // do they force that lang? -}; -#endif -static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue, - void *info) { - if(akey==VALUE_NAME) - return; - - Attributed_meaning_info& ami=*static_cast(info); - - // ...; charset=windows1251 - *ami.header << "; "; - ami.header->append(akey, ami.lang, ami.forced); - *ami.header << "="; - append_attribute_meaning(*ami.header, *static_cast(avalue), ami.lang, ami.forced); -} -const String& attributed_meaning_to_string(Value& meaning, - String::Untaint_lang lang, bool forced) { - String &result=*new(meaning.pool()) String(meaning.pool()); - if(Hash *hash=meaning.get_hash(0)) { - // $value(value) $subattribute(subattribute value) - if(Value *value=static_cast(hash->get(*value_name))) - append_attribute_meaning(result, *value, lang, forced); - - Attributed_meaning_info attributed_meaning_info={&result, lang}; - hash->for_each(append_attribute_subattribute, &attributed_meaning_info); - } else // result value - append_attribute_meaning(result, meaning, lang, forced); - - return result; -}