--- parser3/src/main/pa_common.C 2004/03/05 15:14:47 1.180 +++ parser3/src/main/pa_common.C 2005/11/16 14:18:06 1.204.6.2 @@ -1,11 +1,11 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_COMMON_C="$Date: 2004/03/05 15:14:47 $"; +static const char * const IDENT_COMMON_C="$Date: 2005/11/16 14:18:06 $"; #include "pa_common.h" #include "pa_exception.h" @@ -21,6 +21,7 @@ static const char * const IDENT_COMMON_C #include "pa_vint.h" #include "pa_vhash.h" #include "pa_vtable.h" +#include "pa_socks.h" #ifdef CYGWIN #define _GNU_H_WINDOWS32_SOCKETS @@ -130,6 +131,21 @@ char* file_read_text(Request_charsets& c return file.success?file.str:0; } +/// these options were handled but not checked elsewhere, now check them +static int get_valid_file_options_count(HashStringValue& options) +{ + int result=0; + if(options.get(PA_SQL_LIMIT_NAME)) + result++; + if(options.get(PA_SQL_OFFSET_NAME)) + result++; + if(options.get(PA_COLUMN_SEPARATOR_NAME)) + result++; + if(options.get(PA_COLUMN_ENCLOSER_NAME)) + result++; + return result; +} + //http request stuff #ifdef PA_HTTP @@ -142,13 +158,15 @@ static bool set_addr(struct sockaddr_in addr->sin_port=htons(port); if(host) { ulong packed_ip=inet_addr(host); - struct hostent *hostIP = packed_ip==INADDR_NONE? - gethostbyname(host) - : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET); - if(hostIP) - memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); - else - return false; + if(packed_ip!=INADDR_NONE) + memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip)); + else { + struct hostent *hostIP=gethostbyname(host); + if(hostIP) + memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); + else + return false; + } } else addr->sin_addr.s_addr=INADDR_ANY; return true; @@ -162,19 +180,26 @@ static int http_read_response(char*& res while(true) { char buf[MAX_STRING*10]; ssize_t received_size=recv(sock, buf, sizeof(buf), 0); - if(received_size<=0) + if(received_size==0) + break; + if(received_size<0) { + if(int no=pa_socks_errno()) + throw Exception("http.timeout", + 0, + "error receiving response: %s (%d)", pa_socks_strerr(no), no); break; + } response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/); memcpy(response+response_size, buf, received_size); response_size+=received_size; response[response_size]=0; - if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response + if(!result && (EOLat=strstr(response, "\n"))) { // checking status in first response const String status_line(pa_strdup(response, EOLat-response)); ArrayString astatus; size_t pos_after=0; status_line.split(astatus, pos_after, " "); - const String& status_code=*astatus.get(1); + const String& status_code=*astatus.get(astatus.count()>1?1:0); result=status_code.as_int(); if(fail_on_status_ne_200 && result!=200) @@ -199,42 +224,39 @@ static int http_read_response(char*& res #ifdef PA_USE_ALARM static sigjmp_buf timeout_env; -static void timeout_handler(int sig){ +static void timeout_handler(int /*sig*/){ siglongjmp(timeout_env, 1); } #endif static int http_request(char*& response, size_t& response_size, - const char* host, int port, + const char* host, short port, const char* request, - int -#ifdef PA_USE_ALARM - timeout -#endif - , + int timeout_secs, bool fail_on_status_ne_200) { if(!host) throw Exception("http.host", 0, "zero hostname"); //never + volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning] + int sock=-1; #ifdef PA_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] + // rewritten simplier [athough duplicating closesocket code] if(sock>=0) closesocket(sock); throw Exception("http.timeout", - origin_string, + 0, "timeout occured while retrieving document"); return 0; // never } else { - alarm(timeout); + alarm(timeout_secs); #endif try { int result; @@ -245,19 +267,40 @@ static int http_request(char*& response, 0, "can not resolve hostname \"%s\"", host); - if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) + if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) { + int no=pa_socks_errno(); throw Exception("http.connect", 0, - "can not make socket: %s (%d)", strerror(errno), errno); - if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) + "can not make socket: %s (%d)", pa_socks_strerr(no), no); + } + + // To enable SO_DONTLINGER (that is, disable SO_LINGER) + // l_onoff should be set to zero and setsockopt should be called + linger dont_linger={0,0}; + setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger)); + +#ifdef WIN32 +// SO_*TIMEO can be defined in .h but not implemlemented in protocol, +// failing subsequently with Option not supported by protocol (99) message +// could not suppress that, so leaving this only for win32 + int timeout_ms=timeout_secs*1000; + setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms)); + setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms)); +#endif + + if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) { + int no=pa_socks_errno(); throw Exception("http.connect", 0, - "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno); + "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no); + } size_t request_size=strlen(request); - if(send(sock, request, request_size, 0)!=(ssize_t)request_size) - throw Exception("http.connect", + if(send(sock, request, request_size, 0)!=(ssize_t)request_size) { + int no=pa_socks_errno(); + throw Exception("http.timeout", 0, - "error sending request: %s (%d)", strerror(errno), errno); + "error sending request: %s (%d)", pa_socks_strerr(no), no); + } result=http_read_response(response, response_size, sock, fail_on_status_ne_200); closesocket(sock); @@ -344,19 +387,46 @@ static const String* basic_authorization return result; } -/// @todo table value +static void form_string_value2string( + HashStringValue::key_type key, + const String& value, + String& result) +{ + result << String(key, String::L_TAINTED) << "="; + result.append(value, String::L_URI, true); + result<< "&"; +} +#ifndef DOXYGEN +struct Form_table_value2string_info { + HashStringValue::key_type key; + String& result; + + Form_table_value2string_info(HashStringValue::key_type akey, String& aresult): + key(akey), result(aresult) {} +}; +#endif +static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) { + form_string_value2string(info->key, *row->get(0), info->result); +} static void form_value2string( HashStringValue::key_type key, HashStringValue::value_type value, String* result) { - *result << String(key, String::L_TAINTED) << "="; - result->append(value->as_string(), String::L_URI, true); + if(const String* svalue=value->get_string()) + form_string_value2string(key, *svalue, *result); + else if(Table* tvalue=value->get_table()) { + Form_table_value2string_info info(key, *result); + tvalue->for_each(form_table_value2string, &info); + } else + throw Exception(0, + new String(key, String::L_TAINTED), + "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type()); } static const char* form2string(HashStringValue& form) { String string; form.for_each(form_value2string, &string); - return 0; // todo + return string.cstr(String::L_UNSPECIFIED); } #ifndef DOXYGEN struct File_read_http_result { @@ -364,6 +434,25 @@ struct File_read_http_result { HashStringValue* headers; }; #endif +static void find_headers_end(char* p, + char*& headers_end_at, + char*& raw_body) +{ + raw_body=p; + // \n\n + // \r\n\r\n + while((p=strchr(p, '\n'))) { + headers_end_at=++p; // \n>.< + if(*p=='\r') // \r\n>\r?<\n + p++; + if(*p=='\n') { // \r\n\r>\n?< + raw_body=p+1; + return; + } + } + headers_end_at=0; +} + /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now static File_read_http_result file_read_http(Request_charsets& charsets, const String& file_spec, @@ -372,11 +461,11 @@ static File_read_http_result file_read_h File_read_http_result result; char host[MAX_STRING]; const char* uri; - int port; + short port; const char* method="GET"; bool method_is_get; HashStringValue* form=0; const char* body_cstr=0; - int timeout=2; + int timeout_secs=2; bool fail_on_status_ne_200=true; Value* vheaders=0; Charset *asked_remote_charset=0; @@ -384,32 +473,41 @@ static File_read_http_result file_read_h const char* password_cstr=0; if(options) { - int valid_options=0; + int valid_options=get_valid_file_options_count(*options); + if(Value* vmethod=options->get(HTTP_METHOD_NAME)) { valid_options++; method=vmethod->as_string().cstr(); - } else if(Value* vform=options->get(HTTP_FORM_NAME)) { + } + if(Value* vform=options->get(HTTP_FORM_NAME)) { valid_options++; form=vform->get_hash(); - } else if(Value* vbody=options->get(HTTP_BODY_NAME)) { + } + if(Value* vbody=options->get(HTTP_BODY_NAME)) { valid_options++; body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED); - } else if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { + } + if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { valid_options++; - timeout=vtimeout->as_int(); - } else if((vheaders=options->get(HTTP_HEADERS_NAME))) { + timeout_secs=vtimeout->as_int(); + } + if((vheaders=options->get(HTTP_HEADERS_NAME))) { valid_options++; - } else if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) { + } + if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) { valid_options++; fail_on_status_ne_200=!vany_status->as_bool(); - } else if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) { + } + if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) { valid_options++; asked_remote_charset=&::charsets.get(vcharset_name->as_string(). change_case(charsets.source(), String::CC_UPPER)); - } else if(Value* vuser=options->get(HTTP_USER)) { + } + if(Value* vuser=options->get(HTTP_USER)) { valid_options++; user_cstr=vuser->as_string().cstr(); - } else if(Value* vpassword=options->get(HTTP_PASSWORD)) { + } + if(Value* vpassword=options->get(HTTP_PASSWORD)) { valid_options++; password_cstr=vpassword->as_string().cstr(); } @@ -452,7 +550,7 @@ static File_read_http_result file_read_h 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; + port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80; if(strchr(uri, '?') && form) throw Exception("parser.runtime", @@ -465,11 +563,13 @@ static File_read_http_result file_read_h head << " " << uri; if(form) if(method_is_get) - head << form2string(*form); - else - body_cstr = form2string(*form); + head << "?" << form2string(*form); head <<" HTTP/1.0" CRLF "host: "<< host << CRLF; + if(form && !method_is_get) { + head << "content-type: application/x-www-form-urlencoded" CRLF; + body_cstr = form2string(*form); + } // http://www.ietf.org/rfc/rfc2617.txt if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr)) @@ -489,15 +589,6 @@ static File_read_http_result file_read_h if(!user_agent_specified) // defaulting head << "user-agent: " DEFAULT_USER_AGENT CRLF; - const char* head_cstr=head.cstr(String::L_UNSPECIFIED); - - // recode those pieces which are not in String::L_URI lang - // [those violating HTTP standard, but widly used] - head_cstr=Charset::transcode( - String::C(head_cstr, strlen(head_cstr)), - charsets.source(), - *asked_remote_charset); - if(body_cstr) { // recode those pieces which are not in String::L_URI lang // [those violating HTTP standard, but widly used] @@ -505,14 +596,21 @@ static File_read_http_result file_read_h String::C(body_cstr, strlen(body_cstr)), charsets.source(), *asked_remote_charset); - } - request_head_and_body << head_cstr; - if(body_cstr) head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF; + } + + const char* head_cstr=head.cstr(String::L_UNSPECIFIED); + + // recode those pieces which are not in String::L_URI lang + // [those violating HTTP standard, but widly used] + head_cstr=Charset::transcode( + String::C(head_cstr, strlen(head_cstr)), + charsets.source(), + *asked_remote_charset); - // end of header - request_head_and_body << CRLF; + // head + end of header + request_head_and_body << head_cstr << CRLF; // body if(body_cstr) request_head_and_body << body_cstr; @@ -523,80 +621,81 @@ static File_read_http_result file_read_h size_t response_size; int status_code=http_request(response, response_size, host, port, request_head_and_body.cstr(), - timeout, fail_on_status_ne_200); + timeout_secs, fail_on_status_ne_200); //processing results char* raw_body; size_t raw_body_size; - char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/); - if(headers_end_at) { - raw_body=headers_end_at+4; - raw_body_size=response_size-(raw_body-response); - } else - throw Exception("http.response", - &connect_string, - "bad response from host - no headers found"); - - *headers_end_at=0; - const String header_block(response, headers_end_at-response, true); + char* headers_end_at; + find_headers_end(response, + headers_end_at, + raw_body); + raw_body_size=response_size-(raw_body-response); - ArrayString aheaders; result.headers=new HashStringValue; VHash* vtables=new VHash; result.headers->put(HTTP_TABLES_NAME, vtables); - HashStringValue& tables=vtables->hash(); - - size_t pos_after=0; - header_block.split(aheaders, pos_after, CRLF); - - //processing headers - size_t aheaders_count=aheaders.count(); Charset* real_remote_charset=0; // undetected, yet - for(size_t i=1; iget_table(); - } else { - // first appearence - Table::columns_type columns =new ArrayString(1); - *columns+=new String("value"); - table=new Table(columns); + + if(headers_end_at) { + *headers_end_at=0; + const String header_block(String::C(response, headers_end_at-response), true); + + ArrayString aheaders; + HashStringValue& tables=vtables->hash(); + + size_t pos_after=0; + header_block.split(aheaders, pos_after, "\n"); + + //processing headers + size_t aheaders_count=aheaders.count(); + for(size_t i=1; iget_table(); + } else { + // first appearence + Table::columns_type columns =new ArrayString(1); + *columns+=new String("value"); + table=new Table(columns); + } + // this string becomes next row + ArrayString& row=*new ArrayString(1); + row+=&header_value; + *table+=&row; + // not existed before? add it + if(!existed) + tables.put(HEADER_NAME, new VTable(table)); } - // this string becomes next row - ArrayString& row=*new ArrayString(1); - row+=&header_value; - *table+=&row; - // not existed before? add it - if(!existed) - tables.put(HEADER_NAME, new VTable(table)); - } - result.headers->put(HEADER_NAME, new VString(header_value)); - } - // defaulting to used-asked charset [it's never empty!] - if(!real_remote_charset) - real_remote_charset=asked_remote_charset; + result.headers->put(HEADER_NAME, new VString(header_value)); + } + } // output response String::C real_body=String::C(raw_body, raw_body_size); - if(as_text && raw_body_size) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below + if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below + // defaulting to used-asked charset [it's never empty!] + if(!real_remote_charset) + real_remote_charset=asked_remote_charset; real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source()); + } result.str=const_cast(real_body.str); // hacking a little result.length=real_body.length; @@ -609,6 +708,7 @@ static File_read_http_result file_read_h #ifndef DOXYGEN struct File_read_action_info { char **data; size_t *data_size; + char* buf; size_t offset; size_t count; }; #endif static void file_read_action( @@ -617,8 +717,16 @@ static void file_read_action( 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=new(PointerFreeGC) char[to_read_size+(as_text?1:0)]; + size_t to_read_size=info.count; + if(!to_read_size) + to_read_size=(size_t)finfo.st_size; + assert( !(info.buf && as_text) ); + if(to_read_size) { + if(info.offset) + lseek(f, info.offset, SEEK_SET); + *info.data=info.buf + ? info.buf + : new(PointerFreeGC) char[to_read_size+(as_text?1:0)]; *info.data_size=(size_t)read(f, *info.data, to_read_size); if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) @@ -638,10 +746,16 @@ static void file_read_action( } File_read_result file_read(Request_charsets& charsets, const String& file_spec, bool as_text, HashStringValue *params, - bool fail_on_read_problem) { + bool fail_on_read_problem, + char* buf, size_t offset, size_t count) { File_read_result result={false, 0, 0, 0}; #ifdef PA_HTTP if(file_spec.starts_with("http://")) { + if(offset || count) + throw Exception("parser.runtime", + 0, + "offset and load options are not supported for HTTP:// file load"); + // fail on read problem File_read_http_result http=file_read_http(charsets, file_spec, as_text, params); result.success=true; @@ -650,12 +764,16 @@ File_read_result file_read(Request_chars result.headers=http.headers; } else { #endif - if(params && params->count()) - throw Exception("parser.runtime", - 0, - "invalid option passed"); + if(params) { + int valid_options=get_valid_file_options_count(*params); + if(valid_options!=params->count()) + throw Exception("parser.runtime", + 0, + "invalid option passed"); + } - File_read_action_info info={&result.str, &result.length}; + File_read_action_info info={&result.str, &result.length, + buf, offset, count}; result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); @@ -750,7 +868,7 @@ bool file_read_action_under_lock(const S } } -static void create_dir_for_file(const String& file_spec) { +void create_dir_for_file(const String& file_spec) { size_t pos_after=1; size_t pos_before; while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { @@ -1010,6 +1128,7 @@ const char* format(double value, char* f size_t stdout_write(const void *buf, size_t size) { #ifdef WIN32 + size_t to_write = size; do{ int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); if(chunk_written<=0) @@ -1018,7 +1137,7 @@ size_t stdout_write(const void *buf, siz buf=((const char*)buf)+chunk_written; } while(size>0); - return size; + return to_write-size; #else return fwrite(buf, 1, size, stdout); #endif @@ -1031,11 +1150,11 @@ char* unescape_chars(const char* cp, int EscapeFirst, EscapeSecond } escapeState=EscapeRest; - int escapedValue=0; + uchar escapedValue=0; int srcPos=0; int dstPos=0; while(srcPos < len) { - int ch=cp[srcPos]; + uchar ch=(uchar)cp[srcPos]; switch(escapeState) { case EscapeRest: if(ch=='%') { @@ -1047,7 +1166,7 @@ char* unescape_chars(const char* cp, int } break; case EscapeFirst: - escapedValue=hex_value[ch] << 4; + escapedValue=(uchar)(hex_value[ch] << 4); escapeState=EscapeSecond; break; case EscapeSecond: @@ -1088,8 +1207,8 @@ bool StrEqNc(const char* s1, const char* return !strict; } else if(!(*s2)) return !strict; - if(isalpha(*s1)) { - if(tolower(*s1) !=tolower(*s2)) + if(isalpha((unsigned char)*s1)) { + if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2)) return false; } else if((*s1) !=(*s2)) return false; @@ -1229,7 +1348,7 @@ static char *base64_alphabet = static size_t g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) { - const register unsigned char *inptr; + register const unsigned char *inptr; register unsigned char *outptr; if (inlen <= 0) @@ -1352,7 +1471,10 @@ char* pa_base64(const char *in, size_t l char* result=new(PointerFreeGC) char[len * 2 + 6]; int state=0; int save=0; - size_t filled=g_mime_utils_base64_encode_close ((const unsigned char*)in, len, +#ifndef NDEBUG + size_t filled= +#endif + g_mime_utils_base64_encode_close ((const unsigned char*)in, len, (unsigned char*)result, &state, &save); assert(filled <= len * 2 + 6);