--- parser3/src/main/pa_common.C 2002/11/26 09:02:14 1.130 +++ 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: 2002/11/26 09:02:14 $"; +static const char* IDENT_COMMON_C="$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 @@ -76,20 +77,6 @@ static int unlock(int fd) { FLOCK(F_TLOC #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; const char* const eob=buf+size; @@ -97,7 +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); @@ -252,8 +239,10 @@ static void http_pass_header(const Hash: Http_pass_header_info& i=*static_cast(info); Pool& pool=i.request->pool(); - *(i.request)<(value))->get_string()<<"\n"; - + *(i.request)<(value), String::UL_HTTP_HEADER, false) + <<"\n"; + if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT") i.user_agent_specified=true; } @@ -267,16 +256,20 @@ static void file_read_http(Pool& pool, c int timeout=2; Value *vheaders=0; - char* connect_string=file_spec.cstr(String::UL_UNSPECIFIED); - if(strncmp(connect_string, "http://", 7)!=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_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; @@ -298,20 +291,20 @@ static void file_read_http(Pool& pool, c throw Exception("parser.runtime", 0, "invalid option passed"); - } + } //making request String request(pool); request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; bool user_agent_specified=false; if(vheaders && !vheaders->is_string()) { // allow empty - if(Hash *headers=vheaders->get_hash(&file_spec)) { + 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", - &file_spec, + &connect_string, "headers param must be hash"); }; if(!user_agent_specified) // defaulting @@ -321,13 +314,13 @@ static void file_read_http(Pool& pool, c //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); @@ -342,7 +335,7 @@ 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 @@ -351,7 +344,7 @@ static void file_read_http(Pool& pool, c pos=line->pos(": ", 2); if(pos<1) throw Exception("http.response", - &file_spec, + &connect_string, "bad response from host - bad header \"%s\"", line->cstr()); headers.put( @@ -359,7 +352,7 @@ static void file_read_http(Pool& pool, c 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()); } @@ -416,6 +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; @@ -456,7 +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'", @@ -926,3 +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; +}