--- parser3/src/main/pa_http.C 2009/09/10 09:44:07 1.41 +++ parser3/src/main/pa_http.C 2010/12/29 12:46:44 1.51 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_HTTP_C="$Date: 2009/09/10 09:44:07 $"; +static const char * const IDENT_HTTP_C="$Date: 2010/12/29 12:46:44 $"; #include "pa_http.h" #include "pa_common.h" @@ -194,29 +194,6 @@ static void timeout_handler(int /*sig*/) } #endif -static size_t file_untaint(const char* str, size_t len) { - // untaint file from L_FILE_POST encoding - char* j=(char *)str; - const char* end=str+len-1; - for(const char* i=str; i<=end; i++, j++){ - if(*i=='\\' && i!=end){ - switch(*(i+1)){ - case '0': - *j='\0'; - i++; - continue; - case '\\': - *j='\\'; - i++; - continue; - } - } - if(i!=j) - *j=*i; - } - return j-str; // new length -} - static int http_request(char*& response, size_t& response_size, const char* host, short port, const char* request, size_t request_size, @@ -318,18 +295,30 @@ struct Http_pass_header_info { bool* content_type_url_encoded; }; #endif + +char *pa_http_safe_header_name(const char *name) { + char *result=pa_strdup(name); + char *n=result; + if(!isalpha(*n)) + *n++ = '_'; + for(; *n; ++n) { + if (!isalnum(*n) && *n != '-' && *n != '_') + *n = '_'; + } + return result; +} + static void http_pass_header(HashStringValue::key_type aname, HashStringValue::value_type avalue, Http_pass_header_info *info) { const char* name_cstr=aname.cstr(); - String name=String(capitalize(name_cstr), String::L_URI); - if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0) return; - String value=attributed_meaning_to_string(*avalue, String::L_URI, false); + String name=String(pa_http_safe_header_name(capitalize(name_cstr)), String::L_AS_IS); + String value=attributed_meaning_to_string(*avalue, String::L_HTTP_HEADER, true); *info->request << name << ": " << value << CRLF; @@ -406,27 +395,63 @@ static void form_value2string( const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) { String string; form.for_each(form_value2string, &string); - return string.transcode_and_untaint_cstr(String::L_URI, &charsets); + return string.untaint_and_transcode_cstr(String::L_URI, &charsets); } struct FormPart { Request* r; const char* boundary; - String string; + String* string; Form_table_value2string_info* info; + + struct BinaryBlock{ + const char* ptr; + size_t length; + + BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){} + BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){} + }; + + Array blocks; + + FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){} + + const char *post(size_t &length){ + if(blocks.count()){ + blocks+=BinaryBlock(string, r); + + length=0; + for(size_t i=0; icharsets.source(), part.r->charsets.client()) + << name << "\""; if(file_name){ if(strcmp(file_name, NONAME_DAT)!=0) - part.string << "; filename=\"" << file_name << "\""; - part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name); + *part.string << "; filename=\"" << file_name << "\""; + *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name); } - part.string << CRLF CRLF; + *part.string << CRLF CRLF; } static void form_string_value2part( @@ -435,7 +460,7 @@ static void form_string_value2part( FormPart& part) { form_part_boundary_header(part, key); - part.string << Charset::transcode(value, part.r->charsets.source(), part.r->charsets.client()) << CRLF; + *part.string << value << CRLF; } static void form_file_value2part( @@ -444,8 +469,10 @@ static void form_file_value2part( FormPart& part) { form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr()); - part.string.append_know_length(vfile.value_ptr(), vfile.value_size(), String::L_FILE_POST); - part.string << CRLF; + part.blocks+=FormPart::BinaryBlock(part.string, part.r); + part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size()); + part.string=new String(); + *part.string << CRLF; } static void form_table_value2part(Table::element_type row, FormPart* part) { @@ -460,7 +487,7 @@ static void form_value2part( if(const String* svalue=value->get_string()) form_string_value2part(key, *svalue, part); else if(Table* tvalue=value->get_table()) { - Form_table_value2string_info info(key, part.string); + Form_table_value2string_info info(key, *part.string); part.info = &info; tvalue->for_each(form_table_value2part, &part); } else if(VFile* vfile=static_cast(value->as("file"))){ @@ -472,14 +499,11 @@ static void form_value2part( } const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){ - FormPart formpart; - formpart.r=&r; - formpart.boundary=boundary; - formpart.info=NULL; + FormPart formpart(&r, boundary); form.for_each(form_value2part, formpart); - formpart.string << "--" << boundary << "--"; - post_size=formpart.string.length(); // very surprizing, but it calculates correct post_size even with binary files! - return formpart.string.untaint_cstr(String::L_AS_IS); // without transcoding + *formpart.string << "--" << boundary << "--"; + // @todo: return binary blocks here to save memory in pa_internal_file_read_http + return formpart.post(post_size); } static void find_headers_end(char* p, @@ -510,11 +534,10 @@ File_read_http_result pa_internal_file_r File_read_http_result result; char host[MAX_STRING]; const char* uri; - short port; + short port=80; const char* method="GET"; bool method_is_get=true; HashStringValue* form=0; - const char* body_cstr=0; int timeout_secs=2; bool fail_on_status_ne_200=true; bool omit_post_charset=false; @@ -578,9 +601,7 @@ File_read_http_result pa_internal_file_r } if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - INVALID_OPTION_PASSED); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } if(!asked_remote_charset) // defaulting to $request:charset asked_remote_charset=&(r.charsets).source(); @@ -614,12 +635,13 @@ File_read_http_result pa_internal_file_r //preparing request String& connect_string=*new String(file_spec); - String request_head_and_body; + const char* request; + size_t request_size; { // influence URLencoding of tainted pieces to String::L_URI lang Temp_client_charset temp(r.charsets, *asked_remote_charset); - const char* connect_string_cstr=connect_string.transcode_and_untaint_cstr(String::L_URI, &(r.charsets)); + const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets)); const char* current=connect_string_cstr; if(strncmp(current, "http://", 7)!=0) @@ -632,8 +654,13 @@ File_read_http_result pa_internal_file_r char* host_uri=lsplit(host, '/'); uri=host_uri?current+(host_uri-1-host):"/"; char* port_cstr=lsplit(host, ':'); - char* error_pos=0; - port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80; + + if (port_cstr){ + char* error_pos=0; + port=(short)strtol(port_cstr, &error_pos, 10); + if(port==0 || *error_pos) + throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr); + } // making request head String head; @@ -641,7 +668,10 @@ File_read_http_result pa_internal_file_r if(method_is_get && form) head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets); - head <<" HTTP/1.0" CRLF "Host: "<< host << CRLF; + head <<" HTTP/1.0" CRLF "Host: "<< host; + if (port != 80) + head << ":" << port_cstr; + head << CRLF; char* boundary=0; @@ -676,6 +706,7 @@ File_read_http_result pa_internal_file_r "headers param must be hash"); }; + const char* request_body=0; size_t post_size=0; if(form && !method_is_get) { head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED); @@ -685,26 +716,26 @@ File_read_http_result pa_internal_file_r if(multipart) { head << "; boundary=" << boundary; - body_cstr=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/); + request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/); } else { - body_cstr=pa_form2string(*form, r.charsets); - post_size=strlen(body_cstr); + request_body=pa_form2string(*form, r.charsets); + post_size=strlen(request_body); } head << CRLF; } else if(vbody) { // $.body was specified if(content_type_url_encoded){ // transcode + url-encode - body_cstr=vbody->as_string().transcode_and_untaint_cstr(String::L_URI, &(r.charsets)); + request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets)); } else { // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode! - body_cstr=Charset::transcode( + request_body=Charset::transcode( String::C(vbody->as_string().cstr(), vbody->as_string().length()), r.charsets.source(), *asked_remote_charset ); } - post_size=strlen(body_cstr); + post_size=strlen(request_body); } // http://www.ietf.org/rfc/rfc2617.txt @@ -730,32 +761,35 @@ File_read_http_result pa_internal_file_r } else throw Exception(PARSER_RUNTIME, 0, - "cookies param must be hash"); + "cookies param must be hash"); } - if(body_cstr) + if(request_body) head << "Content-Length: " << format(post_size, "%u") << CRLF; + + head << CRLF; + + const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets)); - // head + end of header - request_head_and_body << head.transcode_and_untaint_cstr(String::L_URI, &(r.charsets)) << CRLF; - - // body - if(body_cstr) - request_head_and_body << body_cstr; + if(request_body){ + size_t head_size = strlen(request_head); + request_size=post_size + head_size; + char *ptr=(char *)pa_malloc_atomic(request_size); + memcpy(ptr, request_head, head_size); + memcpy(ptr+head_size, request_body, post_size); + request=ptr; + } else { + request_size=strlen(request_head); + request=request_head; + } } - const char* request_cstr=request_head_and_body.cstr(); - size_t request_size=strlen(request_cstr); - - if(multipart) - request_size=file_untaint(request_cstr, request_size); - char* response; size_t response_size; // sending request int status_code=http_request(response, response_size, - host, port, request_cstr, request_size, + host, port, request, request_size, timeout_secs, fail_on_status_ne_200); // processing results @@ -826,6 +860,8 @@ File_read_http_result pa_internal_file_r // skip UTF-8 signature (BOM code) raw_body+=3; raw_body_size-=3; + if(!real_remote_charset) + real_remote_charset=&UTF8_charset; } // output response