--- parser3/src/main/pa_common.C 2003/07/24 11:31:23 1.154 +++ parser3/src/main/pa_common.C 2004/03/01 12:16:00 1.176 @@ -1,11 +1,11 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_COMMON_C="$Date: 2003/07/24 11:31:23 $"; +static const char * const IDENT_COMMON_C="$Date: 2004/03/01 12:16:00 $"; #include "pa_common.h" #include "pa_exception.h" @@ -19,6 +19,8 @@ static const char* IDENT_COMMON_C="$Date #ifdef PA_HTTP #include "pa_vstring.h" #include "pa_vint.h" +#include "pa_vhash.h" +#include "pa_vtable.h" #ifdef CYGWIN #define _GNU_H_WINDOWS32_SOCKETS @@ -62,61 +64,9 @@ int PASCAL closesocket(SOCKET); # endif #endif -// locking constants -//#define PA_DEBUG_NO_LOCKING - -#ifdef PA_DEBUG_NO_LOCKING - -#ifdef HAVE_FLOCK - -static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); } -static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); } -static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); } -static int unlock(int fd) { return flock(fd, LOCK_UN); } - -#else -#ifdef HAVE__LOCKING - -#define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1) -static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); } -static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); } -static int unlock(int fd) { FLOCK(_LK_UNLCK); } - -#else -#ifdef HAVE_FCNTL - -#define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls) -static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); } -static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); } -static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); } - -#else -#ifdef HAVE_LOCKF - -#define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1) -static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking -static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); } -static int unlock(int fd) { FLOCK(F_TLOCK); } - -#else - -#error unable to find file locking func - -#endif -#endif -#endif -#endif - -#else -static int lock_shared_blocking(int fd) { return 0; } -static int lock_exclusive_blocking(int fd) { return 0; } -static int lock_exclusive_nonblocking(int fd) { return 0; } -static int unlock(int fd) { return 0; } - -#endif +# ifndef INADDR_NONE +# define INADDR_NONE ((ulong) -1) +# endif // defines for globals @@ -133,6 +83,7 @@ const String file_status_name(FILE_STATU #define HTTP_HEADERS_NAME "headers" #define HTTP_ANY_STATUS_NAME "any-status" #define HTTP_CHARSET_NAME "charset" +#define HTTP_TABLES_NAME "tables" // statics @@ -141,6 +92,7 @@ static const String http_timeout_name(HT static const String http_headers_name(HTTP_HEADERS_NAME); static const String http_any_status_name(HTTP_ANY_STATUS_NAME); static const String http_charset_name(HTTP_CHARSET_NAME); +static const String http_tables_name(HTTP_TABLES_NAME); // defines @@ -194,7 +146,11 @@ static bool set_addr(struct sockaddr_in addr->sin_family=AF_INET; addr->sin_port=htons(port); if(host) { - if(struct hostent *hostIP=gethostbyname(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; @@ -203,17 +159,23 @@ static bool set_addr(struct sockaddr_in return true; } -static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){ +static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){ + response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type + response[(response_size=0)]=0; int result=0; - size_t EOLat=0; + char* EOLat=0; while(true) { - char *buf=new(PointerFreeGC) char[MAX_STRING]; - ssize_t size=recv(sock, buf, MAX_STRING, 0); - if(size<=0) + char buf[MAX_STRING*10]; + ssize_t received_size=recv(sock, buf, sizeof(buf), 0); + if(received_size<=0) break; - response.append_strdup(buf, size, String::L_TAINTED); - if(!result && (EOLat=response.pos(CRLF, 2))!=STRING_NOT_FOUND) { // checking status in first response - const String& status_line=response.mid(0, (size_t)EOLat); + 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 + const String status_line(pa_strdup(response, EOLat-response)); ArrayString astatus; size_t pos_after=0; status_line.split(astatus, pos_after, " "); @@ -231,7 +193,7 @@ static int http_read_response(String& re else throw Exception("http.response", 0, - "bad response from host - no status found (size=%lu)", response.length()); + "bad response from host - no status found (size=%u)", response_size); } /* ********************** request *************************** */ @@ -247,10 +209,14 @@ static void timeout_handler(int sig){ } #endif -static int http_request(String& response, +static int http_request(char*& response, size_t& response_size, const char* host, int port, const char* request, - int timeout, + int +#ifdef PA_USE_ALARM + timeout +#endif + , bool fail_on_status_ne_200) { if(!host) throw Exception("http.host", @@ -298,7 +264,7 @@ static int http_request(String& response 0, "error sending request: %s (%d)", strerror(errno), errno); - result=http_read_response(response, sock, fail_on_status_ne_200); + result=http_read_response(response, response_size, sock, fail_on_status_ne_200); closesocket(sock); #ifdef PA_USE_ALARM alarm(0); @@ -337,7 +303,7 @@ static void http_pass_header(HashStringV static Charset* detect_charset(Charset& source_charset, const String& content_type_value) { - const StringBody CONTENT_TYPE_VALUE= + const String::Body CONTENT_TYPE_VALUE= content_type_value.change_case(source_charset, String::CC_UPPER); // content-type: xxx/xxx; source_charset=WE-NEED-THIS // content-type: xxx/xxx; source_charset="WE-NEED-THIS" @@ -359,7 +325,7 @@ static Charset* detect_charset(Charset& if(delim_pos!=STRING_NOT_FOUND) charset_end=delim_pos; } - const StringBody CHARSET_NAME_BODY= + const String::Body CHARSET_NAME_BODY= CONTENT_TYPE_VALUE.mid(charset_begin, charset_end); return &charsets.get(CHARSET_NAME_BODY); @@ -375,9 +341,11 @@ struct File_read_http_result { HashStringValue* headers; }; #endif +/// @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, - HashStringValue *options=0) { + bool as_text, + HashStringValue *options=0) { File_read_http_result result; char host[MAX_STRING]; const char* uri; @@ -398,7 +366,7 @@ static File_read_http_result file_read_h valid_options++; timeout=vtimeout->as_int(); } - if(vheaders=options->get(http_headers_name)) { + if((vheaders=options->get(http_headers_name))) { valid_options++; } if(Value* vany_status=options->get(http_any_status_name)) { @@ -457,7 +425,7 @@ static File_read_http_result file_read_h bool user_agent_specified=false; if(vheaders && !vheaders->is_string()) { // allow empty if(HashStringValue *headers=vheaders->get_hash()) { - Http_pass_header_info info={&charsets, &request}; + Http_pass_header_info info={&charsets, &request, false}; headers->for_each(http_pass_header, &info); user_agent_specified=info.user_agent_specified; } else @@ -479,23 +447,32 @@ static File_read_http_result file_read_h *asked_remote_charset); //sending request - String response; - int status_code=http_request(response, + char* response; + size_t response_size; + int status_code=http_request(response, response_size, host, port, request_cstr, timeout, fail_on_status_ne_200); //processing results - size_t pos=response.pos(CRLF CRLF, 4); - if(pos==STRING_NOT_FOUND || pos<1) { + 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"); - } - const String& header_block=response.mid(0, pos); - const String& raw_body=response.mid(pos+4, response.length()); + + *headers_end_at=0; + const String header_block(response, headers_end_at-response, true); 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); @@ -504,16 +481,40 @@ static File_read_http_result file_read_h 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); + } + // 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!] @@ -521,10 +522,9 @@ static File_read_http_result file_read_h real_remote_charset=asked_remote_charset; // output response - String::C real_body=Charset::transcode( - String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()), - *real_remote_charset, - charsets.source()); + 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 + 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; @@ -542,7 +542,7 @@ struct File_read_action_info { static void file_read_action( struct stat& finfo, int f, - const String& file_spec, const char* fname, bool as_text, + 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) { @@ -552,7 +552,7 @@ static void file_read_action( 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", + "read failed: actually read %u bytes count not in [0..%u] valid range", *info.data_size, to_read_size); } else { // empty file if(as_text) { @@ -567,17 +567,22 @@ 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) { - File_read_result result={false}; + File_read_result result={false, 0, 0, 0}; #ifdef PA_HTTP if(file_spec.starts_with("http://")) { // fail on read problem - File_read_http_result http=file_read_http(charsets, file_spec, params); + File_read_http_result http=file_read_http(charsets, file_spec, as_text, params); result.success=true; result.str=http.str; result.length=http.length; result.headers=http.headers; } else { #endif + if(params && params->count()) + throw Exception("parser.runtime", + 0, + "invalid option passed"); + File_read_action_info info={&result.str, &result.length}; result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, @@ -590,14 +595,12 @@ File_read_result file_read(Request_chars // UTF-8 signature: EF BB BF if(result.length>=3) { char *in=(char *)result.str; - if(strncmp(in, "\xEB\xBB\xBF", 3)==0) { + if(strncmp(in, "\xEF\xBB\xBF", 3)==0) { result.str=in+3; result.length-=3;// skip prefix } } fix_line_breaks((char *)(result.str), result.length); - // note: after fixing - ((char*&)(result.str))[result.length]=0; } return result; @@ -638,7 +641,7 @@ bool file_read_action_under_lock(const S // later: it seems, even this does not help sometimes if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { try { - if(lock_shared_blocking(f)!=0) + if(pa_lock_shared_blocking(f)!=0) throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", @@ -657,13 +660,13 @@ bool file_read_action_under_lock(const S action(finfo, f, file_spec, fname, as_text, context); } catch(...) { - unlock(f);close(f); + pa_unlock(f);close(f); if(fail_on_read_problem) rethrow; return false; } - unlock(f);close(f); + pa_unlock(f);close(f); return true; } else { if(fail_on_read_problem) @@ -700,7 +703,7 @@ bool file_write_action_under_lock( O_CREAT|O_RDWR |(as_text?_O_TEXT:_O_BINARY) |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { - if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) { + if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) { Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", @@ -718,7 +721,7 @@ bool file_write_action_under_lock( if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower #endif - unlock(f);close(f); + pa_unlock(f);close(f); rethrow; } @@ -726,7 +729,7 @@ bool file_write_action_under_lock( if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov for discovering] #endif - unlock(f);close(f); + pa_unlock(f);close(f); return true; } else throw Exception(errno==EACCES?"file.access":0, @@ -772,10 +775,10 @@ static void rmdir(const String& file_spe rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC)); } -bool file_delete(const String& file_spec, bool fail_on_read_problem) { +bool file_delete(const String& file_spec, bool fail_on_problem) { const char* fname=file_spec.cstr(String::L_FILE_SPEC); if(unlink(fname)!=0) - if(fail_on_read_problem) + if(fail_on_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, &file_spec, "unlink failed: %s (%d), actual filename '%s'", @@ -1066,6 +1069,11 @@ int __vsnprintf(char* b, size_t s, const // note: on win32& maybe somewhere else // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing --s; + + // clients do not check for negative 's', feature: ignore such prints + if((ssize_t)s<0) + return 0; + #if _MSC_VER /* win32: @@ -1091,7 +1099,7 @@ int __vsnprintf(char* b, size_t s, const if(r<0) r=0; - else if(r>s) + else if((size_t)r>s) r=s; #endif b[r]=0; @@ -1105,19 +1113,3 @@ int __snprintf(char* b, size_t s, const va_end(l); return r; } - -int pa_sleep(unsigned long secs, unsigned long usecs) { - for (; usecs >= 1000000; ++secs, usecs -= 1000000); - -#ifdef WIN32 - Sleep(secs * 1000 + usecs / 1000); - return 0; -#else - struct timeval t; - t.tv_sec = secs; - t.tv_usec = usecs; - return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0); -#endif -} - -