|
|
| version 1.142, 2003/01/16 09:07:20 | version 1.205, 2005/08/09 08:14:51 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: commonly functions. | Parser: commonly functions. |
| Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) | Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| */ | */ |
| static const char* IDENT_COMMON_C="$Date$"; | static const char * const IDENT_COMMON_C="$Date$"; |
| #include "pa_common.h" | #include "pa_common.h" |
| #include "pa_exception.h" | #include "pa_exception.h" |
| #include "pa_globals.h" | |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_globals.h" | |
| #include "pa_request_charsets.h" | |
| #include "pa_charsets.h" | |
| #define PA_HTTP | |
| #ifdef PA_HTTP | |
| #include "pa_vstring.h" | #include "pa_vstring.h" |
| #include "pa_vdate.h" | #include "pa_vint.h" |
| #include "pa_vhash.h" | |
| #include "pa_vtable.h" | |
| #include "pa_socks.h" | |
| #ifdef CYGWIN | |
| #define _GNU_H_WINDOWS32_SOCKETS | |
| // for PASCAL | |
| #include <windows.h> | |
| // SOCKET | |
| typedef u_int SOCKET; | |
| int PASCAL closesocket(SOCKET); | |
| #else | |
| # if defined(WIN32) | |
| # include <windows.h> | |
| # else | |
| # define closesocket close | |
| # endif | |
| #endif | |
| #ifdef WIN32 | |
| # include <windows.h> | |
| #else | #else |
| # define closesocket close | |
| # if defined(WIN32) | |
| # include <windows.h> | |
| # endif | |
| #endif | #endif |
| // some maybe-undefined constants | // some maybe-undefined constants |
| Line 39 static const char* IDENT_COMMON_C="$Date | Line 65 static const char* IDENT_COMMON_C="$Date |
| # endif | # endif |
| #endif | #endif |
| // locking constants | # ifndef INADDR_NONE |
| # define INADDR_NONE ((ulong) -1) | |
| #ifdef HAVE_FLOCK | # endif |
| 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 | // defines for globals |
| #ifdef HAVE__LOCKING | |
| #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1) | #define FILE_STATUS_NAME "status" |
| 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 | // globals |
| #ifdef HAVE_FCNTL | |
| #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls) | const String file_status_name(FILE_STATUS_NAME); |
| 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 | // defines |
| #ifdef HAVE_LOCKF | |
| #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1) | #define HTTP_METHOD_NAME "method" |
| static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking | #define HTTP_FORM_NAME "form" |
| static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); } | #define HTTP_BODY_NAME "body" |
| static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); } | #define HTTP_TIMEOUT_NAME "timeout" |
| static int unlock(int fd) { FLOCK(F_TLOCK); } | #define HTTP_HEADERS_NAME "headers" |
| #define HTTP_ANY_STATUS_NAME "any-status" | |
| #define HTTP_CHARSET_NAME "charset" | |
| #define HTTP_TABLES_NAME "tables" | |
| #define HTTP_USER "user" | |
| #define HTTP_PASSWORD "password" | |
| #else | // defines |
| #error unable to find file locking func | |
| #endif | |
| #endif | |
| #endif | |
| #endif | |
| #define DEFAULT_USER_AGENT "parser3" | #define DEFAULT_USER_AGENT "parser3" |
| // functions | |
| void fix_line_breaks(char* buf, size_t& size) { | void fix_line_breaks(char *str, size_t& length) { |
| if(size==0) | |
| return; | |
| //_asm int 3; | //_asm int 3; |
| const char* const eob=buf+size; | const char* const eob=str+length; |
| char* dest=buf; | char* dest=str; |
| // fix DOS: \r\n -> \n | // fix DOS: \r\n -> \n |
| // fix Macintosh: \r -> \n | // fix Macintosh: \r -> \n |
| char* bol=buf; | char* bol=str; |
| while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { | while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { |
| size_t len=eol-bol; | size_t len=eol-bol; |
| if(dest!=bol) | if(dest!=bol) |
| Line 106 void fix_line_breaks(char* buf, size_t& | Line 112 void fix_line_breaks(char* buf, size_t& |
| if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS | if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS |
| bol=eol+2; | bol=eol+2; |
| size--; | length--; |
| } else // \r, not \n = Macintosh | } else // \r, not \n = Macintosh |
| bol=eol+1; | bol=eol+1; |
| } | } |
| // last piece without \r, including terminating 0 | // last piece without \r |
| if(dest!=bol) | if(dest!=bol) |
| memcpy(dest, bol, eob-bol); | memcpy(dest, bol, eob-bol); |
| str[length]=0; // terminating | |
| } | } |
| char* file_read_text(Pool& pool, const String& file_spec, | char* file_read_text(Request_charsets& charsets, |
| bool fail_on_read_problem, | const String& file_spec, |
| Hash *params, Hash** out_fields) { | bool fail_on_read_problem, |
| void *result; size_t size; | HashStringValue* params/*, HashStringValue* * out_fields*/) { |
| return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0; | File_read_result file= |
| file_read(charsets, file_spec, true, params, fail_on_read_problem); | |
| return file.success?file.str:0; | |
| } | } |
| //http request stuff | //http request stuff |
| /* ************************ http stuff *********************** */ | #ifdef PA_HTTP |
| #undef CRLF | |
| #define CRLF "\r\n" | |
| static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){ | static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){ |
| memset(addr, 0, sizeof(*addr)); | memset(addr, 0, sizeof(*addr)); |
| addr->sin_family=AF_INET; | addr->sin_family=AF_INET; |
| addr->sin_port=htons(port); | addr->sin_port=htons(port); |
| if(host) { | if(host) { |
| if(struct hostent *hostIP=gethostbyname(host)) | ulong packed_ip=inet_addr(host); |
| memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); | if(packed_ip!=INADDR_NONE) |
| else | memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip)); |
| return false; | else { |
| struct hostent *hostIP=gethostbyname(host); | |
| if(hostIP) | |
| memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); | |
| else | |
| return false; | |
| } | |
| } else | } else |
| addr->sin_addr.s_addr=INADDR_ANY; | addr->sin_addr.s_addr=INADDR_ANY; |
| return true; | 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){ |
| const String* status_code=0; | response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type |
| ssize_t EOLat=0; | response[(response_size=0)]=0; |
| int result=0; | |
| char* EOLat=0; | |
| while(true) { | while(true) { |
| char *buf=(char *)response.pool().malloc(MAX_STRING); | char buf[MAX_STRING*10]; |
| ssize_t size=recv(sock, buf, MAX_STRING, 0); | ssize_t received_size=recv(sock, buf, sizeof(buf), 0); |
| if(size<=0) | if(received_size==0) |
| break; | break; |
| response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0); | if(received_size<0) { |
| if(!status_code && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response | if(int no=pa_socks_errno()) |
| const String& status_line=response.mid(0, (size_t)EOLat); | throw Exception("http.timeout", |
| Array astatus(response.pool()); | 0, |
| size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1); | "error receiving response: %s (%d)", pa_socks_strerr(no), no); |
| status_code=astatus.get_string(1); | 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, "\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(astatus.count()>1?1:0); | |
| result=status_code.as_int(); | |
| if(fail_on_status_ne_200 && *status_code!="200") | if(fail_on_status_ne_200 && result!=200) |
| throw Exception("http.status", | throw Exception("http.status", |
| status_code, | &status_code, |
| "invalid HTTP response status"); | "invalid HTTP response status"); |
| } | } |
| } | } |
| if(status_code) | if(result) |
| return status_code->as_int(); | return result; |
| else | else |
| throw Exception("http.response", | throw Exception("http.response", |
| 0, | 0, |
| "bad response from host - no status found (size=%lu)", response.size()); | "bad response from host - no status found (size=%u)", response_size); |
| } | } |
| /* ********************** request *************************** */ | /* ********************** request *************************** */ |
| #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) | #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) |
| # define WE_CAN_USE_ALARM | # define PA_USE_ALARM |
| #endif | #endif |
| #ifdef WE_CAN_USE_ALARM | #ifdef PA_USE_ALARM |
| static sigjmp_buf timeout_env; | static sigjmp_buf timeout_env; |
| static void timeout_handler(int sig){ | static void timeout_handler(int /*sig*/){ |
| siglongjmp(timeout_env, 1); | siglongjmp(timeout_env, 1); |
| } | } |
| #endif | #endif |
| static int http_request(String& response, | static int http_request(char*& response, size_t& response_size, |
| const String *origin_string, | const char* host, short port, |
| const char* host, int port, | const char* request, |
| const char* request, | int timeout_secs, |
| int timeout, | bool fail_on_status_ne_200) { |
| bool fail_on_status_ne_200){ | |
| if(!host) | if(!host) |
| throw Exception("http.host", | throw Exception("http.host", |
| origin_string, | 0, |
| "zero hostname"); //never | "zero hostname"); //never |
| #ifdef WE_CAN_USE_ALARM | volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning] |
| signal(SIGALRM, timeout_handler); | int sock=-1; |
| #endif | #ifdef PA_USE_ALARM |
| int sock=-1; | signal(SIGALRM, timeout_handler); |
| try { | #endif |
| int result; | #ifdef PA_USE_ALARM |
| #ifdef WE_CAN_USE_ALARM | if(sigsetjmp(timeout_env, 1)) { |
| if(sigsetjmp(timeout_env, 1)) | // stupid gcc [2.95.4] generated bad code |
| throw Exception("http.timeout", | // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code. |
| origin_string, | // rewritten simplier [athough duplicating closesocket code] |
| "timeout occured while retrieving document"); | if(sock>=0) |
| else { | closesocket(sock); |
| alarm(timeout); | throw Exception("http.timeout", |
| 0, | |
| "timeout occured while retrieving document"); | |
| return 0; // never | |
| } else { | |
| alarm(timeout_secs); | |
| #endif | #endif |
| try { | |
| int result; | |
| struct sockaddr_in dest; | struct sockaddr_in dest; |
| if(!set_addr(&dest, host, port)) | if(!set_addr(&dest, host, port)) |
| throw Exception("http.host", | throw Exception("http.host", |
| origin_string, | 0, |
| "can not resolve hostname \"%s\"", host); | "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", | throw Exception("http.connect", |
| origin_string, | 0, |
| "can not make socket: %s (%d)", strerror(errno), errno); | "can not make socket: %s (%d)", pa_socks_strerr(no), no); |
| if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) | } |
| // 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", | throw Exception("http.connect", |
| origin_string, | 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); | size_t request_size=strlen(request); |
| if(send(sock, request, request_size, 0)!=(ssize_t)request_size) | if(send(sock, request, request_size, 0)!=(ssize_t)request_size) { |
| throw Exception("http.connect", | int no=pa_socks_errno(); |
| origin_string, | throw Exception("http.timeout", |
| "error sending request: %s (%d)", strerror(errno), errno); | 0, |
| "error sending request: %s (%d)", pa_socks_strerr(no), no); | |
| } | |
| 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); | closesocket(sock); |
| #ifdef WE_CAN_USE_ALARM | #ifdef PA_USE_ALARM |
| alarm(0); | alarm(0); |
| } | |
| #endif | #endif |
| return result; | return result; |
| } catch(...) { | } catch(...) { |
| if(sock>=0) | #ifdef PA_USE_ALARM |
| closesocket(sock); | alarm(0); |
| #ifdef WE_CAN_USE_ALARM | |
| alarm(0); | |
| #endif | #endif |
| /*re*/throw; | if(sock>=0) |
| closesocket(sock); | |
| rethrow; | |
| } | |
| #ifdef PA_USE_ALARM | |
| } | } |
| #endif | |
| } | } |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct Http_pass_header_info { | struct Http_pass_header_info { |
| Request_charsets* charsets; | |
| String* request; | String* request; |
| bool user_agent_specified; | bool user_agent_specified; |
| }; | }; |
| #endif | #endif |
| static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info) | static void http_pass_header(HashStringValue::key_type key, |
| HashStringValue::value_type value, | |
| Http_pass_header_info *info) { | |
| *info->request <<key<<": " | |
| << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false) | |
| << CRLF; | |
| if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT") | |
| info->user_agent_specified=true; | |
| } | |
| static Charset* detect_charset(Charset& source_charset, const String& 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" | |
| // content-type: xxx/xxx; source_charset="WE-NEED-THIS"; | |
| size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET="); | |
| if(before_charseteq_pos!=STRING_NOT_FOUND) { | |
| size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/; | |
| size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin); | |
| bool quoted=open_quote_pos==charset_begin; | |
| if(quoted) | |
| charset_begin++; // skip opening '"' | |
| size_t charset_end=CONTENT_TYPE_VALUE.length(); | |
| if(quoted) { | |
| size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin); | |
| if(close_quote_pos!=STRING_NOT_FOUND) | |
| charset_end=close_quote_pos; | |
| } else { | |
| size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin); | |
| if(delim_pos!=STRING_NOT_FOUND) | |
| charset_end=delim_pos; | |
| } | |
| const String::Body CHARSET_NAME_BODY= | |
| CONTENT_TYPE_VALUE.mid(charset_begin, charset_end); | |
| return &charsets.get(CHARSET_NAME_BODY); | |
| } | |
| return 0; | |
| } | |
| static const String* basic_authorization_field(const char* user, const char* pass) { | |
| if(!user&& !pass) | |
| return 0; | |
| String combined; | |
| if(user) | |
| combined<<user; | |
| combined<<":"; | |
| if(pass) | |
| combined<<pass; | |
| String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length()); | |
| return result; | |
| } | |
| static void form_string_value2string( | |
| HashStringValue::key_type key, | |
| const String& value, | |
| String& result) | |
| { | { |
| Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info); | result << String(key, String::L_TAINTED) << "="; |
| Pool& pool=i.request->pool(); | result.append(value, String::L_URI, true); |
| result<< "&"; | |
| *(i.request)<<key<<": " | } |
| << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false) | #ifndef DOXYGEN |
| <<"\n"; | struct Form_table_value2string_info { |
| HashStringValue::key_type key; | |
| if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT") | String& result; |
| i.user_agent_specified=true; | |
| } | Form_table_value2string_info(HashStringValue::key_type akey, String& aresult): |
| static void file_read_http(Pool& pool, const String& file_spec, | key(akey), result(aresult) {} |
| void*& data, size_t& data_size, | }; |
| Hash *options=0, Hash** out_fields=0) { | #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) | |
| { | |
| 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 string.cstr(String::L_UNSPECIFIED); | |
| } | |
| #ifndef DOXYGEN | |
| struct File_read_http_result { | |
| char *str; size_t length; | |
| 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, | |
| bool as_text, | |
| HashStringValue *options=0) { | |
| File_read_http_result result; | |
| char host[MAX_STRING]; | char host[MAX_STRING]; |
| const char* uri; | const char* uri; |
| int port; | short port; |
| const char* method="GET"; | const char* method="GET"; bool method_is_get; |
| int timeout=2; | HashStringValue* form=0; |
| const char* body_cstr=0; | |
| int timeout_secs=2; | |
| bool fail_on_status_ne_200=true; | bool fail_on_status_ne_200=true; |
| Value *vheaders=0; | Value* vheaders=0; |
| Charset *asked_remote_charset=0; | |
| String connect_string(pool); | const char* user_cstr=0; |
| // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one | const char* password_cstr=0; |
| 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, | |
| &connect_string, | |
| "does not start with http://"); //never | |
| connect_string_cstr+=7; | |
| strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0; | |
| char* host_uri=lsplit(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(options) { | if(options) { |
| int valid_options=0; | int valid_options=0; |
| if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) { | if(Value* vmethod=options->get(HTTP_METHOD_NAME)) { |
| valid_options++; | valid_options++; |
| method=vmethod->as_string().cstr(); | method=vmethod->as_string().cstr(); |
| } | } |
| if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) { | if(Value* vform=options->get(HTTP_FORM_NAME)) { |
| valid_options++; | valid_options++; |
| timeout=vtimeout->as_int(); | form=vform->get_hash(); |
| } | } |
| if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) { | if(Value* vbody=options->get(HTTP_BODY_NAME)) { |
| valid_options++; | valid_options++; |
| } | body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED); |
| if(Value *vany_status=static_cast<Value *>(options->get(*http_any_status_name))) { | } |
| if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { | |
| valid_options++; | |
| timeout_secs=vtimeout->as_int(); | |
| } | |
| if((vheaders=options->get(HTTP_HEADERS_NAME))) { | |
| valid_options++; | |
| } | |
| if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) { | |
| valid_options++; | valid_options++; |
| fail_on_status_ne_200=!vany_status->as_bool(); | fail_on_status_ne_200=!vany_status->as_bool(); |
| } | |
| 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)); | |
| } | |
| if(Value* vuser=options->get(HTTP_USER)) { | |
| valid_options++; | |
| user_cstr=vuser->as_string().cstr(); | |
| } | |
| if(Value* vpassword=options->get(HTTP_PASSWORD)) { | |
| valid_options++; | |
| password_cstr=vpassword->as_string().cstr(); | |
| } | } |
| if(valid_options!=options->size()) | if(valid_options!=options->count()) |
| throw Exception("parser.runtime", | throw Exception("parser.runtime", |
| 0, | 0, |
| "invalid option passed"); | "invalid option passed"); |
| } | } |
| if(!asked_remote_charset) // defaulting to $request:charset | |
| asked_remote_charset=&charsets.source(); | |
| method_is_get=strcmp(method, "GET")==0; | |
| if(method_is_get && body_cstr) | |
| throw Exception("parser.runtime", | |
| 0, | |
| "you can not use $."HTTP_BODY_NAME" option with method GET"); | |
| //preparing request | |
| String& connect_string=*new String; | |
| // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one | |
| connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces | |
| String request_head_and_body; | |
| { | |
| // influence URLencoding of tainted pieces to String::L_URI lang | |
| Temp_client_charset temp(charsets, *asked_remote_charset); | |
| const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED); | |
| const char* current=connect_string_cstr; | |
| if(strncmp(current, "http://", 7)!=0) | |
| throw Exception(0, | |
| &connect_string, | |
| "does not start with http://"); //never | |
| current+=7; | |
| strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0; | |
| 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; | |
| //making request | if(strchr(uri, '?') && form) |
| String request(pool); | throw Exception("parser.runtime", |
| request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; | 0, |
| bool user_agent_specified=false; | "use either uri with ?params or $."HTTP_FORM_NAME" option"); |
| if(vheaders && !vheaders->is_string()) { // allow empty | |
| if(Hash *headers=vheaders->get_hash(&connect_string)) { | //making request head |
| Http_pass_header_info info={&request}; | String head; |
| headers->for_each(http_pass_header, &info); | head << method; |
| user_agent_specified=info.user_agent_specified; | head << " " << uri; |
| } else | if(form) |
| throw Exception("parser.runtime", | if(method_is_get) |
| &connect_string, | head << "?" << form2string(*form); |
| "headers param must be hash"); | head <<" HTTP/1.0" CRLF |
| }; | "host: "<< host << CRLF; |
| if(!user_agent_specified) // defaulting | if(form && !method_is_get) { |
| request << "user-agent: " DEFAULT_USER_AGENT "\n"; | head << "content-type: application/x-www-form-urlencoded" CRLF; |
| request<<"\n"; | body_cstr = form2string(*form); |
| } | |
| // http://www.ietf.org/rfc/rfc2617.txt | |
| if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr)) | |
| head<<"authorization: "<<*authorization_field_value<<CRLF; | |
| bool user_agent_specified=false; | |
| if(vheaders && !vheaders->is_string()) { // allow empty | |
| if(HashStringValue *headers=vheaders->get_hash()) { | |
| Http_pass_header_info info={&charsets, &head, false}; | |
| 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 | |
| head << "user-agent: " DEFAULT_USER_AGENT CRLF; | |
| if(body_cstr) { | |
| // recode those pieces which are not in String::L_URI lang | |
| // [those violating HTTP standard, but widly used] | |
| body_cstr=Charset::transcode( | |
| String::C(body_cstr, strlen(body_cstr)), | |
| charsets.source(), | |
| *asked_remote_charset); | |
| 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); | |
| // head + end of header | |
| request_head_and_body << head_cstr << CRLF; | |
| // body | |
| if(body_cstr) | |
| request_head_and_body << body_cstr; | |
| } | |
| //sending request | //sending request |
| String response(pool); | char* response; |
| int status_code=http_request(response, | size_t response_size; |
| &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), | int status_code=http_request(response, response_size, |
| timeout, fail_on_status_ne_200); | host, port, request_head_and_body.cstr(), |
| timeout_secs, fail_on_status_ne_200); | |
| //processing results | //processing results |
| int pos=response.pos("\r\n\r\n", 4); | char* raw_body; size_t raw_body_size; |
| if(pos<1){ | char* headers_end_at; |
| throw Exception("http.response", | find_headers_end(response, |
| &connect_string, | headers_end_at, |
| "bad response from host - no headers found"); | raw_body); |
| } | raw_body_size=response_size-(raw_body-response); |
| String header_block=response.mid(0, pos); | |
| String body=response.mid(pos+4, response.size()); | result.headers=new HashStringValue; |
| VHash* vtables=new VHash; | |
| Array aheaders(pool); | result.headers->put(HTTP_TABLES_NAME, vtables); |
| Hash& headers=*new(pool) Hash(pool); | Charset* real_remote_charset=0; // undetected, yet |
| size_t pos_after_ref=0; | |
| header_block.split(aheaders, &pos_after_ref, "\r\n", 2); | if(headers_end_at) { |
| *headers_end_at=0; | |
| //processing headers | const String header_block(String::C(response, headers_end_at-response), true); |
| for(int i=1;i<aheaders.size();i++) { | |
| if(const String *line=aheaders.get_string(i)) { | ArrayString aheaders; |
| pos=line->pos(": ", 2); | HashStringValue& tables=vtables->hash(); |
| if(pos<1) | |
| 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; i<aheaders_count; i++) { | |
| const String& line=*aheaders.get(i); | |
| size_t pos=line.pos(':'); | |
| if(pos==STRING_NOT_FOUND || pos<1) | |
| throw Exception("http.response", | throw Exception("http.response", |
| &connect_string, | &connect_string, |
| "bad response from host - bad header \"%s\"", line->cstr()); | "bad response from host - bad header \"%s\"", line.cstr()); |
| const String::Body HEADER_NAME= | |
| headers.put( | line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER); |
| line->mid(0, pos).change_case(pool, String::CC_UPPER), | const String& header_value=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r"); |
| new(pool) VString(line->mid(pos+2, line->size()))); | if(as_text && HEADER_NAME=="CONTENT-TYPE") |
| } else | real_remote_charset=detect_charset(charsets.source(), header_value); |
| throw Exception("http.response", | |
| &connect_string, | // tables |
| "bad response from host - bad headers \"%s\"", header_block.cstr()); | { |
| Value *valready=(Value *)tables.get(HEADER_NAME); | |
| bool existed=valready!=0; | |
| Table *table; | |
| if(existed) { | |
| // second+ appearence | |
| table=valready->get_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)); | |
| } | |
| } | } |
| // output response | // output response |
| data=body.cstr(); data_size=body.size(); | String::C real_body=String::C(raw_body, raw_body_size); |
| if(out_fields) { | if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below |
| headers.put(*file_status_name, new(pool) VInt(pool, status_code)); | // defaulting to used-asked charset [it's never empty!] |
| *out_fields=&headers; | 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<char *>(real_body.str); // hacking a little | |
| result.length=real_body.length; | |
| result.headers->put(file_status_name, new VInt(status_code)); | |
| return result; | |
| } | } |
| #endif | |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct File_read_action_info { | struct File_read_action_info { |
| void **data; size_t *data_size; | char **data; size_t *data_size; |
| char* buf; size_t offset; size_t count; | |
| }; | }; |
| #endif | #endif |
| static void file_read_action(Pool& pool, | static void file_read_action( |
| struct stat& finfo, | struct stat& finfo, |
| int f, | int f, |
| const String& file_spec, const char* fname, bool as_text, | const String& file_spec, const char* /*fname*/, bool as_text, |
| void *context) { | void *context) { |
| File_read_action_info& info=*static_cast<File_read_action_info *>(context); | File_read_action_info& info=*static_cast<File_read_action_info *>(context); |
| if(size_t to_read_size=(size_t)finfo.st_size) { | size_t to_read_size=info.count; |
| *info.data=pool.malloc(to_read_size+(as_text?1:0), 3); | 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); | *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) | if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) |
| throw Exception(0, | throw Exception(0, |
| &file_spec, | &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); | *info.data_size, to_read_size); |
| } else { // empty file | } else { // empty file |
| if(as_text) { | if(as_text) { |
| *info.data=pool.malloc(1); | *info.data=new(PointerFreeGC) char[1]; |
| *(char*)(*info.data)=0; | *(char*)(*info.data)=0; |
| } else | } else |
| *info.data=0; | *info.data=0; |
| Line 410 static void file_read_action(Pool& pool, | Line 728 static void file_read_action(Pool& pool, |
| return; | return; |
| } | } |
| } | } |
| bool file_read(Pool& pool, const String& file_spec, | File_read_result file_read(Request_charsets& charsets, const String& file_spec, |
| void*& data, size_t& data_size, | bool as_text, HashStringValue *params, |
| bool as_text, Hash *params, Hash** out_fields, | bool fail_on_read_problem, |
| bool fail_on_read_problem) { | char* buf, size_t offset, size_t count) { |
| bool result; | File_read_result result={false, 0, 0, 0}; |
| if(file_spec.starts_with("http://", 7)) { | #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 | // fail on read problem |
| file_read_http(pool, file_spec, data, data_size, params, out_fields); | File_read_http_result http=file_read_http(charsets, file_spec, as_text, params); |
| result=true; | result.success=true; |
| result.str=http.str; | |
| result.length=http.length; | |
| result.headers=http.headers; | |
| } else { | } else { |
| File_read_action_info info={&data, &data_size}; | #endif |
| result=file_read_action_under_lock(pool, file_spec, | if(params && params->count()) |
| throw Exception("parser.runtime", | |
| 0, | |
| "invalid option passed"); | |
| 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, | "read", file_read_action, &info, |
| as_text, fail_on_read_problem); | as_text, fail_on_read_problem); |
| #ifdef PA_HTTP | |
| } | } |
| #endif | |
| if(result && as_text) { | if(result.success && as_text) { |
| // UTF-8 signature: EF BB BF | // UTF-8 signature: EF BB BF |
| if(data_size>=3) { | if(result.length>=3) { |
| char *in=(char *)data; | char *in=(char *)result.str; |
| if((in[0] == '\xEF') && (in[1] == '\xBB') && | if(strncmp(in, "\xEF\xBB\xBF", 3)==0) { |
| (in[2] == '\xBF')) { | result.str=in+3; result.length-=3;// skip prefix |
| data=in+3; data_size-=3;// skip prefix | |
| } | } |
| } | } |
| fix_line_breaks((char *)(data), data_size); | fix_line_breaks((char *)(result.str), result.length); |
| // note: after fixing | |
| ((char*&)(data))[data_size]=0; | |
| } | } |
| return result; | return result; |
| } | } |
| bool file_read_action_under_lock(Pool& pool, const String& file_spec, | #ifdef PA_SAFE_MODE |
| void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { | |
| if(finfo.st_uid/*foreign?*/!=geteuid() | |
| && finfo.st_gid/*foreign?*/!=getegid()) | |
| throw Exception("parser.runtime", | |
| &file_spec, | |
| "parser is in safe mode: " | |
| "reading files of foreign group and user disabled " | |
| "[recompile parser with --disable-safe-mode configure option], " | |
| "actual filename '%s', " | |
| "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)", | |
| fname, | |
| finfo.st_uid, geteuid(), | |
| finfo.st_gid, getegid()); | |
| } | |
| #endif | |
| bool file_read_action_under_lock(const String& file_spec, | |
| const char* action_name, File_read_action action, void *context, | const char* action_name, File_read_action action, void *context, |
| bool as_text, | bool as_text, |
| bool fail_on_read_problem) { | bool fail_on_read_problem) { |
| const char* fname=file_spec.cstr(String::UL_FILE_SPEC); | const char* fname=file_spec.cstr(String::L_FILE_SPEC); |
| int f; | int f; |
| // first open, next stat: | // first open, next stat: |
| Line 462 bool file_read_action_under_lock(Pool& p | Line 812 bool file_read_action_under_lock(Pool& p |
| // later: it seems, even this does not help sometimes | // later: it seems, even this does not help sometimes |
| if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { | if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { |
| try { | try { |
| if(lock_shared_blocking(f)!=0) | if(pa_lock_shared_blocking(f)!=0) |
| throw Exception("file.lock", | throw Exception("file.lock", |
| &file_spec, | &file_spec, |
| "shared lock failed: %s (%d), actual filename '%s'", | "shared lock failed: %s (%d), actual filename '%s'", |
| strerror(errno), errno, fname); | strerror(errno), errno, fname); |
| struct stat finfo; | struct stat finfo; |
| if(stat(fname, &finfo)!=0) | if(stat(fname, &finfo)!=0) |
| throw Exception("file.missing", // hardly possible: we just opened it OK | throw Exception("file.missing", // hardly possible: we just opened it OK |
| &file_spec, | &file_spec, |
| "stat failed: %s (%d), actual filename '%s'", | "stat failed: %s (%d), actual filename '%s'", |
| strerror(errno), errno, fname); | strerror(errno), errno, fname); |
| #ifdef PA_SAFE_MODE | #ifdef PA_SAFE_MODE |
| if(finfo.st_uid/*foreign?*/!=geteuid() | check_safe_mode(finfo, file_spec, fname); |
| && finfo.st_gid/*foreign?*/!=getegid()) | |
| throw Exception("parser.runtime", | |
| &file_spec, | |
| "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'", | |
| fname); | |
| #endif | #endif |
| action(pool, finfo, f, file_spec, fname, as_text, context); | action(finfo, f, file_spec, fname, as_text, context); |
| } catch(...) { | } catch(...) { |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| /*re*/throw; | rethrow; |
| return false; | return false; |
| } | } |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| return true; | return true; |
| } else { | } else { |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, |
| &file_spec, | &file_spec, |
| "%s failed: %s (%d), actual filename '%s'", | "%s failed: %s (%d), actual filename '%s'", |
| action_name, strerror(errno), errno, fname); | action_name, strerror(errno), errno, fname); |
| return false; | return false; |
| } | } |
| } | } |
| 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_after=1; |
| int pos_before; | size_t pos_before; |
| while((pos_before=file_spec.pos("/", 1, pos_after))>=0) { | while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { |
| mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775); | mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775); |
| pos_after=pos_before+1; | pos_after=pos_before+1; |
| } | } |
| } | } |
| Line 520 bool file_write_action_under_lock( | Line 865 bool file_write_action_under_lock( |
| bool do_append, | bool do_append, |
| bool do_block, | bool do_block, |
| bool fail_on_lock_problem) { | bool fail_on_lock_problem) { |
| const char* fname=file_spec.cstr(String::UL_FILE_SPEC); | const char* fname=file_spec.cstr(String::L_FILE_SPEC); |
| int f; | int f; |
| if(access(fname, W_OK)!=0) // no | if(access(fname, W_OK)!=0) // no |
| create_dir_for_file(file_spec); | create_dir_for_file(file_spec); |
| Line 529 bool file_write_action_under_lock( | Line 874 bool file_write_action_under_lock( |
| O_CREAT|O_RDWR | O_CREAT|O_RDWR |
| |(as_text?_O_TEXT:_O_BINARY) | |(as_text?_O_TEXT:_O_BINARY) |
| |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { | |(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", | Exception e("file.lock", |
| &file_spec, | &file_spec, |
| "shared lock failed: %s (%d), actual filename '%s'", | "shared lock failed: %s (%d), actual filename '%s'", |
| strerror(errno), errno, fname); | strerror(errno), errno, fname); |
| close(f); | close(f); |
| if(fail_on_lock_problem) | if(fail_on_lock_problem) |
| throw e; | throw e; |
| Line 547 bool file_write_action_under_lock( | Line 892 bool file_write_action_under_lock( |
| if(!do_append) | if(!do_append) |
| ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower | ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower |
| #endif | #endif |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| /*re*/throw; | rethrow; |
| } | } |
| #ifdef HAVE_FTRUNCATE | #ifdef HAVE_FTRUNCATE |
| if(!do_append) | if(!do_append) |
| ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering] | ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering] |
| #endif | #endif |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| return true; | return true; |
| } else | } else |
| throw Exception(errno==EACCES?"file.access":0, | throw Exception(errno==EACCES?"file.access":0, |
| &file_spec, | &file_spec, |
| "%s failed: %s (%d), actual filename '%s'", | "%s failed: %s (%d), actual filename '%s'", |
| action_name, strerror(errno), errno, fname); | action_name, strerror(errno), errno, fname); |
| // here should be nothing, see rethrow above | // here should be nothing, see rethrow above |
| } | } |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct File_write_action_info { | struct File_write_action_info { |
| const void *data; size_t size; | const char* str; size_t length; |
| }; | }; |
| #endif | #endif |
| static void file_write_action(int f, void *context) { | static void file_write_action(int f, void *context) { |
| File_write_action_info& info=*static_cast<File_write_action_info *>(context); | File_write_action_info& info=*static_cast<File_write_action_info *>(context); |
| if(info.size) { | if(info.length) { |
| int written=write(f, info.data, info.size); | int written=write(f, info.str, info.length); |
| if(written<0) | if(written<0) |
| throw Exception(0, | throw Exception(0, |
| 0, | 0, |
| Line 582 static void file_write_action(int f, voi | Line 927 static void file_write_action(int f, voi |
| } | } |
| void file_write( | void file_write( |
| const String& file_spec, | const String& file_spec, |
| const void *data, size_t size, | const char* data, size_t size, |
| bool as_text, | bool as_text, |
| bool do_append) { | bool do_append) { |
| File_write_action_info info={data, size}; | File_write_action_info info={data, size}; |
| file_write_action_under_lock( | file_write_action_under_lock( |
| file_spec, | file_spec, |
| "write", file_write_action, &info, | "write", file_write_action, &info, |
| as_text, | as_text, |
| do_append); | do_append); |
| } | } |
| // throws nothing! [this is required in file_move & file_delete] | // throws nothing! [this is required in file_move & file_delete] |
| static void rmdir(const String& file_spec, size_t pos_after) { | static void rmdir(const String& file_spec, size_t pos_after) { |
| int pos_before; | size_t pos_before; |
| if((pos_before=file_spec.pos("/", 1, pos_after))>=0) | if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) |
| rmdir(file_spec, pos_before+1); | rmdir(file_spec, pos_before+1); |
| rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC)); | 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::UL_FILE_SPEC); | const char* fname=file_spec.cstr(String::L_FILE_SPEC); |
| if(unlink(fname)!=0) | if(unlink(fname)!=0) |
| if(fail_on_read_problem) | if(fail_on_problem) |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, |
| &file_spec, | &file_spec, |
| "unlink failed: %s (%d), actual filename '%s'", | "unlink failed: %s (%d), actual filename '%s'", |
| strerror(errno), errno, fname); | strerror(errno), errno, fname); |
| else | else |
| return false; | return false; |
| Line 616 bool file_delete(const String& file_spec | Line 961 bool file_delete(const String& file_spec |
| return true; | return true; |
| } | } |
| void file_move(const String& old_spec, const String& new_spec) { | void file_move(const String& old_spec, const String& new_spec) { |
| const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC); | const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC); |
| const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC); | const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC); |
| create_dir_for_file(new_spec); | create_dir_for_file(new_spec); |
| Line 625 void file_move(const String& old_spec, c | Line 970 void file_move(const String& old_spec, c |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, |
| &old_spec, | &old_spec, |
| "rename failed: %s (%d), actual filename '%s' to '%s'", | "rename failed: %s (%d), actual filename '%s' to '%s'", |
| strerror(errno), errno, old_spec_cstr, new_spec_cstr); | strerror(errno), errno, old_spec_cstr, new_spec_cstr); |
| rmdir(old_spec, 1); | rmdir(old_spec, 1); |
| } | } |
| Line 640 bool entry_exists(const char* fname, str | Line 985 bool entry_exists(const char* fname, str |
| } | } |
| bool entry_exists(const String& file_spec) { | bool entry_exists(const String& file_spec) { |
| const char* fname=file_spec.cstr(String::UL_FILE_SPEC); | const char* fname=file_spec.cstr(String::L_FILE_SPEC); |
| return entry_exists(fname, 0); | return entry_exists(fname, 0); |
| } | } |
| static bool entry_readable(const String& file_spec, bool need_dir) { | static bool entry_readable(const String& file_spec, bool need_dir) { |
| char* fname=file_spec.cstr(String::UL_FILE_SPEC); | char* fname=file_spec.cstrm(String::L_FILE_SPEC); |
| if(need_dir) { | if(need_dir) { |
| size_t size=strlen(fname); | size_t size=strlen(fname); |
| while(size) { | while(size) { |
| Line 669 bool file_readable(const String& file_sp | Line 1014 bool file_readable(const String& file_sp |
| bool dir_readable(const String& file_spec) { | bool dir_readable(const String& file_spec) { |
| return entry_readable(file_spec, true); | return entry_readable(file_spec, true); |
| } | } |
| String *file_readable(const String& path, const String& name) { | const String* file_readable(const String& path, const String& name) { |
| String *result=new(path.pool()) String(path); | String& result=*new String(path); |
| *result << "/"; | result << "/"; |
| *result << name; | result << name; |
| return file_readable(*result)?result:0; | return file_readable(result)?&result:0; |
| } | } |
| bool file_executable(const String& file_spec) { | bool file_executable(const String& file_spec) { |
| return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0; | return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0; |
| } | } |
| bool file_stat(const String& file_spec, | bool file_stat(const String& file_spec, |
| Line 685 bool file_stat(const String& file_spec, | Line 1030 bool file_stat(const String& file_spec, |
| time_t& rmtime, | time_t& rmtime, |
| time_t& rctime, | time_t& rctime, |
| bool fail_on_read_problem) { | bool fail_on_read_problem) { |
| Pool& pool=file_spec.pool(); | const char* fname=file_spec.cstr(String::L_FILE_SPEC); |
| const char* fname=file_spec.cstr(String::UL_FILE_SPEC); | struct stat finfo; |
| struct stat finfo; | |
| if(stat(fname, &finfo)!=0) | if(stat(fname, &finfo)!=0) |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception("file.missing", | throw Exception("file.missing", |
| &file_spec, | &file_spec, |
| "getting file size failed: %s (%d), real filename '%s'", | "getting file size failed: %s (%d), real filename '%s'", |
| strerror(errno), errno, fname); | strerror(errno), errno, fname); |
| else | else |
| return false; | return false; |
| rsize=finfo.st_size; | rsize=finfo.st_size; |
| Line 745 char* rsplit(char* string, char delim) { | Line 1089 char* rsplit(char* string, char delim) { |
| } | } |
| /// @todo less stupid type detection | /// @todo less stupid type detection |
| char* format(Pool& pool, double value, char* fmt) { | const char* format(double value, char* fmt) { |
| char local_buf[MAX_NUMBER]; | char local_buf[MAX_NUMBER]; |
| size_t size; | size_t size; |
| Line 760 char* format(Pool& pool, double value, c | Line 1104 char* format(Pool& pool, double value, c |
| else | else |
| size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); | size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); |
| char* pool_buf=(char *)pool.malloc(size+1, 4); | return pa_strdup(local_buf, size); |
| memcpy(pool_buf, local_buf, size+1); | |
| return pool_buf; | |
| } | } |
| size_t stdout_write(const void *buf, size_t size) { | size_t stdout_write(const void *buf, size_t size) { |
| #ifdef WIN32 | #ifdef WIN32 |
| size_t to_write = size; | |
| do{ | do{ |
| int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout); | int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); |
| if(chunk_written<=0) | if(chunk_written<=0) |
| break; | break; |
| size-=chunk_written; | size-=chunk_written; |
| buf=((const char*)buf)+chunk_written; | buf=((const char*)buf)+chunk_written; |
| } while(size>0); | } while(size>0); |
| return size; | return to_write-size; |
| #else | #else |
| return fwrite(buf, 1, size, stdout); | return fwrite(buf, 1, size, stdout); |
| #endif | #endif |
| } | } |
| char* unescape_chars(Pool& pool, const char* cp, int len) { | char* unescape_chars(const char* cp, int len) { |
| char* s=(char *)pool.malloc(len + 1, 5); | char* s=new(PointerFreeGC) char[len + 1]; |
| enum EscapeState { | enum EscapeState { |
| EscapeRest, | EscapeRest, |
| EscapeFirst, | EscapeFirst, |
| EscapeSecond | EscapeSecond |
| } escapeState=EscapeRest; | } escapeState=EscapeRest; |
| int escapedValue=0; | uchar escapedValue=0; |
| int srcPos=0; | int srcPos=0; |
| int dstPos=0; | int dstPos=0; |
| while(srcPos < len) { | while(srcPos < len) { |
| int ch=cp[srcPos]; | uchar ch=(uchar)cp[srcPos]; |
| switch(escapeState) { | switch(escapeState) { |
| case EscapeRest: | case EscapeRest: |
| if(ch=='%') { | if(ch=='%') { |
| Line 804 char* unescape_chars(Pool& pool, const c | Line 1147 char* unescape_chars(Pool& pool, const c |
| } | } |
| break; | break; |
| case EscapeFirst: | case EscapeFirst: |
| escapedValue=hex_value[ch] << 4; | escapedValue=(uchar)(hex_value[ch] << 4); |
| escapeState=EscapeSecond; | escapeState=EscapeSecond; |
| break; | break; |
| case EscapeSecond: | case EscapeSecond: |
| Line 845 bool StrEqNc(const char* s1, const char* | Line 1188 bool StrEqNc(const char* s1, const char* |
| return !strict; | return !strict; |
| } else if(!(*s2)) | } else if(!(*s2)) |
| return !strict; | return !strict; |
| if(isalpha(*s1)) { | if(isalpha((unsigned char)*s1)) { |
| if(tolower(*s1) !=tolower(*s2)) | if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2)) |
| return false; | return false; |
| } else if((*s1) !=(*s2)) | } else if((*s1) !=(*s2)) |
| return false; | return false; |
| Line 898 int __vsnprintf(char* b, size_t s, const | Line 1241 int __vsnprintf(char* b, size_t s, const |
| // note: on win32& maybe somewhere else | // note: on win32& maybe somewhere else |
| // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing | // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing |
| --s; | --s; |
| // clients do not check for negative 's', feature: ignore such prints | |
| if((ssize_t)s<0) | |
| return 0; | |
| #if _MSC_VER | #if _MSC_VER |
| /* | /* |
| win32: | win32: |
| mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm | mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm |
| if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned | if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned |
| */ | */ |
| r=_vsnprintf(b, s, f, l); | r=_vsnprintf(b, s, f, l); |
| if(r<0) | if(r<0) |
| Line 923 int __vsnprintf(char* b, size_t s, const | Line 1271 int __vsnprintf(char* b, size_t s, const |
| if(r<0) | if(r<0) |
| r=0; | r=0; |
| else if(r>s) | else if((size_t)r>s) |
| r=s; | r=s; |
| #endif | #endif |
| b[r]=0; | b[r]=0; |
| Line 938 int __snprintf(char* b, size_t s, const | Line 1286 int __snprintf(char* b, size_t s, const |
| return r; | return r; |
| } | } |
| int pa_sleep(unsigned long secs, unsigned long usecs) { | /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */ |
| for (; usecs >= 1000000; ++secs, usecs -= 1000000); | /* |
| * Authors: Michael Zucchi <notzed@helixcode.com> | |
| #ifdef WIN32 | * Jeffrey Stedfast <fejj@helixcode.com> |
| Sleep(secs * 1000 + usecs / 1000); | * |
| return 0; | * Copyright 2000 Helix Code, Inc. (www.helixcode.com) |
| #else | * |
| struct timeval t; | * This program is free software; you can redistribute it and/or modify |
| t.tv_sec = secs; | * it under the terms of the GNU General Public License as published by |
| t.tv_usec = usecs; | * the Free Software Foundation; either version 2 of the License, or |
| return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0); | * (at your option) any later version. |
| #endif | * |
| * This program is distributed in the hope that it will be useful, | |
| * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
| * GNU General Public License for more details. | |
| * | |
| * You should have received a copy of the GNU General Public License | |
| * along with this program; if not, write to the Free Software | |
| * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA. | |
| * | |
| */ | |
| static char *base64_alphabet = | |
| "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; | |
| /** | |
| * g_mime_utils_base64_encode_step: | |
| * @in: input stream | |
| * @inlen: length of the input | |
| * @out: output string | |
| * @state: holds the number of bits that are stored in @save | |
| * @save: leftover bits that have not yet been encoded | |
| * | |
| * Base64 encodes a chunk of data. Performs an 'encode step', only | |
| * encodes blocks of 3 characters to the output at a time, saves | |
| * left-over state in state and save (initialise to 0 on first | |
| * invocation). | |
| * | |
| * Returns the number of bytes encoded. | |
| **/ | |
| static size_t | |
| g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) | |
| { | |
| register const unsigned char *inptr; | |
| register unsigned char *outptr; | |
| if (inlen <= 0) | |
| return 0; | |
| inptr = in; | |
| outptr = out; | |
| if (inlen + ((unsigned char *)save)[0] > 2) { | |
| const unsigned char *inend = in + inlen - 2; | |
| register int c1 = 0, c2 = 0, c3 = 0; | |
| register int already; | |
| already = *state; | |
| switch (((char *)save)[0]) { | |
| case 1: c1 = ((unsigned char *)save)[1]; goto skip1; | |
| case 2: c1 = ((unsigned char *)save)[1]; | |
| c2 = ((unsigned char *)save)[2]; goto skip2; | |
| } | |
| /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */ | |
| while (inptr < inend) { | |
| c1 = *inptr++; | |
| skip1: | |
| c2 = *inptr++; | |
| skip2: | |
| c3 = *inptr++; | |
| *outptr++ = base64_alphabet [c1 >> 2]; | |
| *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)]; | |
| *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)]; | |
| *outptr++ = base64_alphabet [c3 & 0x3f]; | |
| /* this is a bit ugly ... */ | |
| if ((++already) >= 19) { | |
| *outptr++ = '\n'; | |
| already = 0; | |
| } | |
| } | |
| ((unsigned char *)save)[0] = 0; | |
| inlen = 2 - (inptr - inend); | |
| *state = already; | |
| } | |
| //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen)); | |
| if (inlen > 0) { | |
| register char *saveout; | |
| /* points to the slot for the next char to save */ | |
| saveout = & (((char *)save)[1]) + ((char *)save)[0]; | |
| /* inlen can only be 0 1 or 2 */ | |
| switch (inlen) { | |
| case 2: *saveout++ = *inptr++; | |
| case 1: *saveout++ = *inptr++; | |
| } | |
| ((char *)save)[0] += inlen; | |
| } | |
| /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n", | |
| (int)((char *)save)[0], | |
| (int)((char *)save)[1], | |
| (int)((char *)save)[2]));*/ | |
| return (outptr - out); | |
| } | } |
| /** | |
| // attributed meaning | * g_mime_utils_base64_encode_close: |
| * @in: input stream | |
| static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) { | * @inlen: length of the input |
| const char month_names[12][4]={ | * @out: output string |
| "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"}; | * @state: holds the number of bits that are stored in @save |
| const char days[7][4]={ | * @save: leftover bits that have not yet been encoded |
| "Sun","Mon","Tue","Wed","Thu","Fri","Sat"}; | * |
| * Base64 encodes the input stream to the output stream. Call this | |
| time_t when=vdate.get_time(); | * when finished encoding data with g_mime_utils_base64_encode_step to |
| struct tm *tms=gmtime(&when); | * flush off the last little bit. |
| if(!tms) | * |
| throw Exception(0, | * Returns the number of bytes encoded. |
| 0, | **/ |
| "bad time in attribute value (seconds from epoch=%ld)", when); | static size_t |
| return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT", | g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) |
| days[tms->tm_wday], | { |
| tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900, | unsigned char *outptr = out; |
| tms->tm_hour,tms->tm_min,tms->tm_sec); | int c1, c2; |
| } | |
| static void append_attribute_meaning(String& result, | if (inlen > 0) |
| Value& value, String::Untaint_lang lang, bool forced) { | outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save); |
| if(const String *string=value.get_string()) | |
| result.append(string->join_chains(result.pool(), 0), lang, forced); | c1 = ((unsigned char *)save)[1]; |
| else | c2 = ((unsigned char *)save)[2]; |
| if(Value *vdate=value.as(VDATE_TYPE, false)) { | |
| char *buf=(char *)result.malloc(MAX_STRING); | switch (((unsigned char *)save)[0]) { |
| size_t size=date_attribute(*static_cast<VDate *>(vdate), | case 2: |
| buf, MAX_STRING); | outptr[2] = base64_alphabet [(c2 & 0x0f) << 2]; |
| goto skip; | |
| result.APPEND_CLEAN(buf, size, "converted from date", 0); | case 1: |
| } else | outptr[2] = '='; |
| throw Exception("parser.runtime", | skip: |
| &result, | outptr[0] = base64_alphabet [c1 >> 2]; |
| "trying to append here neither string nor date (%s)", | outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)]; |
| value.type()); | outptr[3] = '='; |
| outptr += 4; | |
| break; | |
| } | |
| *outptr++ = 0; | |
| *save = 0; | |
| *state = 0; | |
| return (outptr - out); | |
| } | } |
| #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<Attributed_meaning_info *>(info); | |
| // ...; charset=windows1251 | char* pa_base64(const char *in, size_t len) |
| *ami.header << "; "; | { |
| ami.header->append(akey, ami.lang, ami.forced); | /* wont go to more than 2x size (overly conservative) */ |
| *ami.header << "="; | char* result=new(PointerFreeGC) char[len * 2 + 6]; |
| append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced); | int state=0; |
| } | int save=0; |
| const String& attributed_meaning_to_string(Value& meaning, | #ifndef NDEBUG |
| String::Untaint_lang lang, bool forced) { | size_t filled= |
| String &result=*new(meaning.pool()) String(meaning.pool()); | #endif |
| if(Hash *hash=meaning.get_hash(0)) { | g_mime_utils_base64_encode_close ((const unsigned char*)in, len, |
| // $value(value) $subattribute(subattribute value) | (unsigned char*)result, &state, &save); |
| if(Value *value=static_cast<Value *>(hash->get(*value_name))) | assert(filled <= len * 2 + 6); |
| 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; | return result; |
| } | } |