--- parser3/src/main/pa_common.C 2002/11/25 14:10:52 1.126 +++ parser3/src/main/pa_common.C 2002/12/05 12:53:48 1.137 @@ -5,13 +5,14 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_COMMON_C="$Date: 2002/11/25 14:10:52 $"; +static const char* IDENT_COMMON_C="$Date: 2002/12/05 12:53:48 $"; #include "pa_common.h" #include "pa_exception.h" #include "pa_globals.h" #include "pa_hash.h" #include "pa_vstring.h" +#include "pa_vdate.h" #ifdef WIN32 # include @@ -73,19 +74,8 @@ static int unlock(int fd) { FLOCK(F_TLOC #endif #endif +#define DEFAULT_USER_AGENT "parser3" -static char* strnchr(char* buf, size_t size, char c) { - // sanity check - if(!buf) - return 0; - - for(; size-->0; buf++) { - if(*buf==c) - return buf; - } - - return 0; -} void fix_line_breaks(char* buf, size_t& size) { //_asm int 3; @@ -94,7 +84,7 @@ void fix_line_breaks(char* buf, size_t& // fix DOS: \r\n -> \n // fix Macintosh: \r -> \n char* bol=buf; - while(char* eol=strnchr(bol, eob -bol, '\r')) { + while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { size_t len=eol-bol; if(dest!=bol) memcpy(dest, bol, len); @@ -137,13 +127,32 @@ static bool set_addr(struct sockaddr_in } static void http_read_response(String& response, int sock){ + bool check_status=true; + ssize_t EOLat=0; while(true) { char *buf=(char *)response.pool().malloc(MAX_STRING); ssize_t size=recv(sock, buf, MAX_STRING, 0); if(size<=0) break; - response.APPEND_TAINTED(buf, size, "http response", 0); + 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; + + 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") + throw Exception("http.status", + status_code, + "invalid HTTP response status"); + } } + if(check_status) + throw Exception("http.response", + 0, + "bad response from host - no status found (size=%lu)", response.size()); } /* ********************** request *************************** */ @@ -187,21 +196,21 @@ static void http_request(String& respons if(!set_addr(&dest, host, port)) throw Exception("http.host", origin_string, - "can't resolve hostname \"%s\"", host); + "can not resolve hostname \"%s\"", host); if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) throw Exception("http.connect", origin_string, - "can't make socket : %s (%d)", strerror(errno), errno); + "can not make socket: %s (%d)", strerror(errno), errno); if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) throw Exception("http.connect", origin_string, - "can't connect to host \"%s\": %s (%d)", host, strerror(errno), errno); + "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); + "error sending request: %s (%d)", strerror(errno), errno); http_read_response(response, sock); if(sock>=0) @@ -219,63 +228,99 @@ static void http_request(String& respons } } -static void headers_foreacher(const Hash::Key& key, Hash::Val *value, void *info) +#ifndef DOXYGEN +struct Http_pass_header_info { + String* request; + bool user_agent_specified; +}; +#endif +static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info) { - String * request=static_cast(info); - *(request)<(value))->get_string()<<"\n"; + Http_pass_header_info& i=*static_cast(info); + Pool& pool=i.request->pool(); + + *(i.request)<(value), String::UL_HTTP_HEADER, false) + <<"\n"; + + if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT") + i.user_agent_specified=true; } - static void file_read_http(Pool& pool, const String& file_spec, void*& data, size_t& data_size, - Hash *params=0, Hash** out_fields=0) { + Hash *options=0, Hash** out_fields=0) { char host[MAX_STRING]; - char* uri; + const char* uri; int port; const char* method="GET"; int timeout=2; + Value *vheaders=0; + + String connect_string(pool); + // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one + connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces - char* connect_string=file_spec.cstr(String::UL_UNSPECIFIED); - if(strncmp(connect_string, "http://", 7)!=0) + char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED); + if(strncmp(connect_string_cstr, "http://", 7)!=0) throw Exception(0, - &file_spec, + &connect_string, "does not start with http://"); //never - connect_string+=7; + connect_string_cstr+=7; - strncpy(host, connect_string, sizeof(host)-1); host[sizeof(host)-1]=0; + strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0; char* host_uri=lsplit(host, '/'); - uri=host_uri?connect_string+(host_uri-1-host):"/"; + uri=host_uri?connect_string_cstr+(host_uri-1-host):"/"; char* port_cstr=lsplit(host, ':'); char* error_pos=0; port=port_cstr?strtol(port_cstr, &error_pos, 0):80; - if(params) { - if(Value *method_element=static_cast(params->get(*http_method_name))) - method=method_element->as_string().cstr(); - if(Value *timeout_element=static_cast(params->get(*http_timeout_name))) - timeout=timeout_element->as_int(); - } + if(options) { + int valid_options=0; + if(Value *vmethod=static_cast(options->get(*http_method_name))) { + valid_options++; + method=vmethod->as_string().cstr(); + } + if(Value *vtimeout=static_cast(options->get(*http_timeout_name))) { + valid_options++; + timeout=vtimeout->as_int(); + } + if(vheaders=static_cast(options->get(*http_headers_name))) { + valid_options++; + } + if(valid_options!=options->size()) + throw Exception("parser.runtime", + 0, + "invalid option passed"); + } //making request - String request(pool); + String request(pool); request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; - if(params) - if(Value *headers_element=static_cast(params->get(*http_headers_name))) - { - Hash *headers=headers_element->get_hash(&file_spec); - headers->for_each(headers_foreacher, &request); - } + bool user_agent_specified=false; + if(vheaders && !vheaders->is_string()) { // allow empty + if(Hash *headers=vheaders->get_hash(&connect_string)) { + Http_pass_header_info info={&request}; + headers->for_each(http_pass_header, &info); + user_agent_specified=info.user_agent_specified; + } else + throw Exception("parser.runtime", + &connect_string, + "headers param must be hash"); + }; + if(!user_agent_specified) // defaulting + request << "user-agent: " DEFAULT_USER_AGENT "\n"; request<<"\n"; //sending request String response(pool); http_request(response, - &file_spec, host, port, request.cstr(String::UL_UNSPECIFIED), timeout); + &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout); //processing results int pos=response.pos("\r\n\r\n", 4); if(pos<1){ throw Exception("http.response", - &file_spec, + &connect_string, "bad response from host - no headers found"); } String header_block=response.mid(0, pos); @@ -290,36 +335,31 @@ static void file_read_http(Pool& pool, c const String *status_line=aheaders.get_string(0); if(!status_line){ throw Exception("http.response", - &file_spec, + &connect_string, "bad response from host - no status line "); } //processing headers - Array astatus(pool); - pos_after_ref=0; - status_line->split(astatus, &pos_after_ref, " ", 1); - const String * status_code=astatus.get_string(1); - for(int i=1;ipos(": ", 2); if(pos<1) throw Exception("http.response", - &file_spec, + &connect_string, "bad response from host - bad header \"%s\"", line->cstr()); - headers.put(line->mid(0, pos), new(pool) VString(line->mid(pos+2, line->size()))); + headers.put( + line->mid(0, pos).change_case(pool, String::CC_UPPER), + new(pool) VString(line->mid(pos+2, line->size()))); } else throw Exception("http.response", - &file_spec, + &connect_string, "bad response from host - bad headers \"%s\"", header_block.cstr()); } // output response data=body.cstr(); data_size=body.size(); - if(out_fields) { - headers.put(*file_status_name, new(pool) VString(*status_code)); + if(out_fields) *out_fields=&headers; - } } #ifndef DOXYGEN @@ -369,6 +409,15 @@ bool file_read(Pool& pool, const String& } if(result && as_text) { + // UTF-8 signature: EF BB BF + if(data_size>=3) { + char *in=(char *)data; + if((in[0] == '\xEF') && (in[1] == '\xBB') && + (in[2] == '\xBF')) { + data=in+3; data_size-=3;// skip prefix + } + } + fix_line_breaks((char *)(data), data_size); // note: after fixing ((char*&)(data))[data_size]=0; @@ -409,7 +458,7 @@ bool file_read_action_under_lock(Pool& p strerror(errno), errno, fname); #ifdef NO_FOREIGN_GROUP_FILES - if(finfo.st_gid/*foreign?*/!=getegid()) { + if(finfo.st_gid/*foreign?*/!=getegid()) throw Exception("parser.runtime", &file_spec, "parser reading files of foreign group disabled [recompile parser without --disable-foreign-group-files configure option], actual filename '%s'", @@ -879,3 +928,76 @@ int pa_sleep(unsigned long secs, unsigne return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0); #endif } + + +// 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; +}