--- parser3/src/main/pa_common.C 2016/12/21 21:14:42 1.296 +++ parser3/src/main/pa_common.C 2025/07/02 23:42:10 1.337 @@ -1,30 +1,9 @@ /** @file Parser: commonly functions. - Copyright (c) 2000-2015 Art. Lebedev Studio (http://www.artlebedev.com) - Author: Alexandr Petrosian (http://paf.design.ru) - - * BASE64 part - * Authors: Michael Zucchi - * Jeffrey Stedfast - * - * Copyright 2000-2004 Ximian, Inc. (www.ximian.com) - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * 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. - * - */ + Copyright (c) 2000-2024 Art. Lebedev Studio (http://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian +*/ #include "pa_common.h" #include "pa_exception.h" @@ -33,7 +12,6 @@ #include "pa_charsets.h" #include "pa_http.h" #include "pa_request_charsets.h" -#include "pcre.h" #include "pa_request.h" #include "pa_idna.h" @@ -50,7 +28,7 @@ #define pa_mkdir(path, mode) mkdir(path, mode) #endif -volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.296 2016/12/21 21:14:42 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; +volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.337 2025/07/02 23:42:10 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; // some maybe-undefined constants @@ -79,18 +57,48 @@ volatile const char * IDENT_PA_COMMON_C= const String file_status_name(FILE_STATUS_NAME); +String sql_bind_name(SQL_BIND_NAME); +String sql_limit_name(PA_SQL_LIMIT_NAME); +String sql_offset_name(PA_SQL_OFFSET_NAME); +String sql_default_name(SQL_DEFAULT_NAME); +String sql_distinct_name(SQL_DISTINCT_NAME); +String sql_value_type_name(SQL_VALUE_TYPE_NAME); + +// forwards + +const UTF16* pa_utf16_encode(const char* in, Charset& source_charset); + // functions -char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) { - File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result); - return file.success?file.str:0; +#ifdef _MSC_VER + +int pa_stat(const char *pathname, struct stat *buffer){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + return _wstat64((const wchar_t *)utf16name, buffer); +} + +int pa_open(const char *pathname, int flags, int mode){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + return _wopen((const wchar_t *)utf16name, flags, mode); +} + +FILE *pa_fopen(const char *pathname, const char *mode){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + const UTF16* utf16mode=pa_utf16_encode(mode, pa_thread_request().charsets.source()); + return _wfopen((const wchar_t *)utf16name, (const wchar_t *)utf16mode); } -char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) { - File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result); - return file.success?file.str:0; +int pa_unlink(const char *pathname){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + return _wunlink((const wchar_t *)utf16name); } +#else + +#define pa_unlink unlink + +#endif + /// these options were handled but not checked elsewhere, now check them int pa_get_valid_file_options_count(HashStringValue& options) { int result=0; @@ -110,62 +118,68 @@ int pa_get_valid_file_options_count(Hash #ifndef DOXYGEN struct File_read_action_info { char **data; size_t *data_size; - char* buf; size_t offset; size_t count; + char* buf; uint64_t offset; size_t limit; }; #endif static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) { - File_read_action_info& info=*static_cast(context); - size_t to_read_size=info.count; - if(!to_read_size) - to_read_size=(size_t)finfo.st_size; + File_read_action_info& info = *static_cast(context); + size_t to_read_size = check_file_size(info.limit && info.limit < (size_t)finfo.st_size ? info.limit : (size_t)finfo.st_size, &file_spec); if(to_read_size) { if(info.offset) - lseek(f, info.offset, SEEK_SET); - *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1); - ssize_t result=read(f, *info.data, to_read_size); + pa_lseek(f, info.offset, SEEK_SET); // seek never fails as POSIX allows the file offset to be set beyond the EOF + *info.data = info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1); + ssize_t result = read(f, *info.data, to_read_size); if(result<0) throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno); - *info.data_size=result; + *info.data_size = result; } else { // empty file // for both, text and binary: for text we need that terminator, for binary we need nonzero pointer to be able to save such files - *info.data=(char *)pa_malloc_atomic(1); - *(char*)(*info.data)=0; - *info.data_size=0; + *info.data = (char *)pa_malloc_atomic(1); + *(char*)(*info.data) = 0; + *info.data_size = 0; return; } } +File_read_result file_read_binary(const String& file_spec, bool fail_on_read_problem, char* buf, uint64_t offset, size_t limit) { + File_read_result result = {false, 0, 0, 0}; + File_read_action_info info = {&result.str, &result.length, buf, offset, limit}; + + result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, 0, fail_on_read_problem); + return result; +} + File_read_result file_read(Request_charsets& charsets, const String& file_spec, - bool as_text, HashStringValue *params, + bool as_text, HashStringValue *options, bool fail_on_read_problem, - char* buf, size_t offset, size_t count, bool transcode_text_result) { - File_read_result result={false, 0, 0, 0}; - if(params){ - int valid_options=pa_get_valid_file_options_count(*params); - if(valid_options!=params->count()) + size_t offset = 0, size_t limit = 0, bool transcode_text_result = true) { + File_read_result result = {false, 0, 0, 0}; + if(options){ + int valid_options = pa_get_valid_file_options_count(*options); + if(valid_options != options->count()) throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } - File_read_action_info info={&result.str, &result.length, buf, offset, count}; + File_read_action_info info = {&result.str, &result.length, 0, offset, limit}; - result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); + result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); if(as_text){ if(result.success){ - Charset* asked_charset=0; - if(params) - if(Value* vcharset_name=params->get(PA_CHARSET_NAME)) - asked_charset=&pa_charsets.get(vcharset_name->as_string()); + Charset* asked_charset = 0; + if(options) + if(Value* vcharset_name = options->get(PA_CHARSET_NAME)) + asked_charset = &pa_charsets.get(vcharset_name->as_string()); - asked_charset=pa_charsets.checkBOM(result.str, result.length, asked_charset); + asked_charset = pa_charsets.checkBOM(result.str, result.length, asked_charset); if(result.length && transcode_text_result && asked_charset){ // length must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below - String::C body=String::C(result.str, result.length); + String::C body = String::C(result.str, result.length); body=Charset::transcode(body, *asked_charset, charsets.source()); - result.str=const_cast(body.str); // hacking a little - result.length=body.length; + result.str = const_cast(body.str); // hacking a little + result.length = body.length; } } if(result.length) @@ -176,28 +190,43 @@ File_read_result file_read(Request_chars } File_read_result file_load(Request& r, const String& file_spec, - bool as_text, HashStringValue *params, + bool as_text, HashStringValue *options, bool fail_on_read_problem, - char* buf, size_t offset, size_t count, bool transcode_text_result) { + bool transcode_text_result) { + + size_t offset = 0; + size_t limit = 0; + + if(options){ + if(Value *voffset = (Value *)options->get(sql_offset_name)) + offset = r.process(*voffset).as_int(); + if(Value *vlimit = (Value *)options->get(sql_limit_name)) + limit = r.process(*vlimit).as_int(); + // no check on options count here + } - File_read_result result={false, 0, 0, 0}; if(file_spec.starts_with("http://")) { - if(offset || count) + if(offset || limit) 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=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result); - result.success=true; - result.str=http.str; - result.length=http.length; - result.headers=http.headers; + File_read_http_result http = pa_internal_file_read_http(r, file_spec, as_text, options, transcode_text_result); + + File_read_result result = {true, http.str, http.length, http.headers}; + return result; } else - result= - file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result); + return file_read(r.charsets, file_spec, as_text, options, fail_on_read_problem, offset, limit, transcode_text_result); +} - return result; +char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem) { + File_read_result file = file_read(charsets, file_spec, true, 0, fail_on_read_problem); + return file.success ? file.str : 0; } +char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* options, bool transcode_result) { + File_read_result file = file_load(r, file_spec, true, options, fail_on_read_problem, transcode_result); + return file.success ? file.str : 0; +} #ifdef PA_SAFE_MODE void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { @@ -217,12 +246,10 @@ void check_safe_mode(struct stat, const #endif +bool file_read_action_under_lock(const String& file_spec, const char* action_name, File_read_action action, void *context, + bool as_text, bool fail_on_read_problem) { -bool file_read_action_under_lock(const String& file_spec, - const char* action_name, File_read_action action, void *context, - bool as_text, - bool fail_on_read_problem) { - const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; // first open, next stat: @@ -234,13 +261,14 @@ bool file_read_action_under_lock(const S // they delay update till open, so we would receive "!^test[" string // if would do stat, next open. // later: it seems, even this does not help sometimes - if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { + if((f=pa_open(fname, O_RDONLY | (as_text ? _O_TEXT : _O_BINARY) ))>=0) { try { - if(pa_lock_shared_blocking(f)!=0) - throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); + int pa_errno=pa_lock_shared_blocking(f); + if(pa_errno!=0) + throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(pa_errno), pa_errno, fname); struct stat finfo; - if(fstat(f, &finfo)!=0) + if(pa_fstat(f, &finfo)!=0) throw Exception("file.missing", // hardly possible: we just opened it OK &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); @@ -265,35 +293,29 @@ bool file_read_action_under_lock(const S } 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) { - pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775); - pos_after=pos_before+1; + const char *str=file_spec.taint_cstr(String::L_FILE_SPEC); + if(str[0]){ + const char *pos=str+1; + while((pos=strchr(pos,'/')) && pos[1]) { // to avoid trailing /, see #1166 + pa_mkdir(pa_strdup(str,pos-str), 0775); + pos++; + } } } -bool file_write_action_under_lock( - const String& file_spec, - const char* action_name, - File_write_action action, - void *context, - bool as_text, - bool do_append, - bool do_block, - bool fail_on_lock_problem) { +bool file_write_action_under_lock(const String& file_spec, const char* action_name, File_write_action action, void *context, + bool as_text, bool do_append, bool do_block, bool fail_on_lock_problem) { + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; if(access(fname, W_OK)!=0) // no create_dir_for_file(file_spec); - if((f=open(fname, - O_CREAT|O_RDWR - |(as_text?_O_TEXT:_O_BINARY) - |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=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'", strerror(errno), errno, fname); - close(f); + if((f=pa_open(fname, O_CREAT | O_RDWR | (as_text ? _O_TEXT : _O_BINARY) | (do_append ? O_APPEND : PA_O_TRUNC), 0664))>=0) { + int pa_errno=do_block ? pa_lock_exclusive_blocking(f) : pa_lock_exclusive_nonblocking(f); + if(pa_errno!=0) { + Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(pa_errno), pa_errno, fname); + close(f); if(fail_on_lock_problem) throw e; return false; @@ -302,24 +324,24 @@ bool file_write_action_under_lock( try { #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE)) struct stat finfo; - if(fstat(f, &finfo)==0 && finfo.st_mode & 0111) + if(pa_fstat(f, &finfo)==0 && finfo.st_mode & 0111) fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any #endif action(f, context); } catch(...) { #ifdef HAVE_FTRUNCATE if(!do_append) - ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower + PA_UNUSED int ignore_result=ftruncate(f, lseek(f, 0, SEEK_CUR)); // one cannot use O_TRUNC, read lower #endif - pa_unlock(f);close(f); + pa_unlock(f);close(f); rethrow; } #ifdef HAVE_FTRUNCATE if(!do_append) - ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov for discovering] + PA_UNUSED int ignore_result=ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov for discovering] #endif - pa_unlock(f);close(f); + pa_unlock(f);close(f); return true; } else throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); @@ -384,7 +406,19 @@ static size_t get_dir(char* fname, size_ return pos; } -static bool entry_readable(char* fname, bool need_dir) { +bool entry_exists(const char* fname, struct stat *afinfo) { + struct stat lfinfo; + bool result=pa_stat(fname, &lfinfo)==0; + if(afinfo) + *afinfo=lfinfo; + return result; +} + +bool entry_exists(const String& file_spec) { + return entry_exists(file_spec.taint_cstr(String::L_FILE_SPEC), 0); +} + +static bool entry_ifdir(char *fname, bool need_dir) { if(need_dir){ size_t size=strlen(fname); while(size) { @@ -397,15 +431,15 @@ static bool entry_readable(char* fname, } struct stat finfo; - if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) { + if(entry_exists(fname, &finfo)) { bool is_dir=(finfo.st_mode&S_IFDIR) != 0; return is_dir==need_dir; } return false; } -static bool entry_readable(const String& file_spec, bool need_dir) { - return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir); +static bool entry_ifdir(const String& file_spec, bool need_dir) { + return entry_ifdir(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir); } // throws nothing! [this is required in file_move & file_delete] @@ -414,7 +448,7 @@ static void rmdir(const String& file_spe size_t length=strlen(dir_spec); while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){ #ifdef _MSC_VER - if(!entry_readable(dir_spec, true)) + if(!entry_ifdir(dir_spec, true)) break; DWORD attrs=GetFileAttributes(dir_spec); if( @@ -431,7 +465,7 @@ static void rmdir(const String& file_spe bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) { const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); - if(unlink(fname)!=0) { + if(pa_unlink(fname)!=0) { if(fail_on_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); @@ -460,25 +494,12 @@ void file_move(const String& old_spec, c } -bool entry_exists(const char* fname, struct stat *afinfo) { - struct stat lfinfo; - bool result=stat(fname, &lfinfo)==0; - if(afinfo) - *afinfo=lfinfo; - return result; -} - -bool entry_exists(const String& file_spec) { - const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); - return entry_exists(fname, 0); -} - bool file_exist(const String& file_spec) { - return entry_readable(file_spec, false); + return entry_ifdir(file_spec, false); } bool dir_exists(const String& file_spec) { - return entry_readable(file_spec, true); + return entry_ifdir(file_spec, true); } const String* file_exist(const String& path, const String& name) { @@ -496,19 +517,25 @@ bool file_executable(const String& file_ bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); struct stat finfo; - if(stat(fname, &finfo)!=0) { + if(pa_stat(fname, &finfo)!=0) { if(fail_on_read_problem) throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname); else return false; } rsize=finfo.st_size; - ratime=finfo.st_atime; - rmtime=finfo.st_mtime; - rctime=finfo.st_ctime; + ratime=(time_t)finfo.st_atime; + rmtime=(time_t)finfo.st_mtime; + rctime=(time_t)finfo.st_ctime; return true; } +size_t check_file_size(uint64_t size, const String* file_spec){ + if(size > (uint64_t)pa_file_size_limit) + throw Exception(PARSER_RUNTIME, file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit); + return (size_t)size; +} + /** String related functions */ @@ -538,15 +565,15 @@ const char* capitalize(const char* s){ return (const char*)result; } -char *str_lower(const char *s, size_t helper_length){ - char *result=pa_strdup(s, helper_length); +char *str_lower(const char *s, size_t length){ + char *result=pa_strdup(s, length); for(char* c=result; *c; c++) *c=(char)tolower((unsigned char)*c); return result; } -char *str_upper(const char *s, size_t helper_length){ - char *result=pa_strdup(s, helper_length); +char *str_upper(const char *s, size_t length){ + char *result=pa_strdup(s, length); for(char* c=result; *c; c++) *c=(char)toupper((unsigned char)*c); return result; @@ -623,9 +650,46 @@ char* rsplit(char* string, char delim) { return v+1; } } - return NULL; + return NULL; +} + +void pa_strncpy(char *dst, const char *src, size_t n){ + size_t left = n; + + if (left != 0 && src) { + while (--left != 0) { + if ((*dst++ = *src++) == '\0') + return; + } + } + if (n != 0) + *dst = '\0'; +} + +#define STRCAT_STEP(str, len) if(str) { memcpy(ptr, str, len); ptr += len; } + +char *pa_strcat(const char *a, const char *b, const char *c) { + size_t len_a = a ? strlen(a) : 0; + size_t len_b = b ? strlen(b) : 0; + size_t len_c = c ? strlen(c) : 0; + char *result=new(PointerFreeGC) char[len_a + len_b + len_c +1/*0*/]; + char *ptr=result; + STRCAT_STEP(a, len_a); + STRCAT_STEP(b, len_b); + STRCAT_STEP(c, len_c); + *ptr='\0'; + return result; } +const char *pa_filename(const char *path){ + if(!path) + return NULL; + for(const char *c = path + strlen(path) - 1; c >= path; c--) { + if(*c == '/' || *c == '\\') + return c+1; + } + return path; +} // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx @@ -691,29 +755,31 @@ FormatType format_type(const char* fmt){ } -const char* format(double value, const char* fmt) { +const char* format_double(double value, const char* fmt) { char local_buf[MAX_NUMBER]; int size=-1; if(fmt && strlen(fmt)){ switch(format_type(fmt)){ case FormatDouble: - size=snprintf(local_buf, sizeof(local_buf), fmt, value); - break; - case FormatInt: - size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value); + size=snprintf(local_buf, sizeof(local_buf), fmt, value); break; case FormatUInt: - size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); + if(value >= 0){ // on Apple M1 (uint) is 0 + size=snprintf(local_buf, sizeof(local_buf), fmt, clip2uint(value)); + break; + } + case FormatInt: + size=snprintf(local_buf, sizeof(local_buf), fmt, clip2int(value)); break; case FormatInvalid: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt); } } else - size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); + return pa_itoa(clip2int(value)); if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing - throw Exception(PARSER_RUNTIME, 0, "Error occure white executing snprintf with format string '%s'.", fmt); + throw Exception(PARSER_RUNTIME, 0, "Error occurred white executing snprintf with format string '%s'.", fmt); } return pa_strdup(local_buf, (size_t)size); @@ -723,16 +789,19 @@ size_t stdout_write(const void *buf, siz #ifdef WIN32 size_t to_write = size; do{ - int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); + int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); if(chunk_written<=0) break; size-=chunk_written; buf=((const char*)buf)+chunk_written; - } while(size>0); + } while(size>0); + fflush(stdout); return to_write-size; #else - return fwrite(buf, 1, size, stdout); + size_t result=fwrite(buf, 1, size, stdout); + fflush(stdout); + return result; #endif } @@ -746,7 +815,7 @@ enum EscapeState { // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually) char* unescape_chars(const char* cp, int len, Charset* charset, bool js){ char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes) - char* dst=s; + XMLByte* dst=(XMLByte *)s; EscapeState escapeState=EscapeRest; uint escapedValue=0; int srcPos=0; @@ -792,7 +861,7 @@ char* unescape_chars(const char* cp, int escapedValue=(escapedValue << 4) + hex_value[c]; if(++jsCnt==4){ // transcode utf8 char to client charset (we can lost some chars here) - charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?'); + charset->store_Char(dst, (XMLCh)escapedValue, '?'); escapeState=EscapeRest; } } else { @@ -844,7 +913,7 @@ size_t strpos(const char *str, const cha return (p==0)?STRING_NOT_FOUND:p-str; } -int remove_crlf(char* start, char* end) { +size_t remove_crlf(char* start, char* end) { char* from=start; char* to=start; bool skip=false; @@ -890,328 +959,8 @@ const char* hex_string(unsigned char* by return bytes_hex; } -/* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */ -/* - * Authors: Michael Zucchi - * Jeffrey Stedfast - * - * Copyright 2000 Helix Code, Inc. (www.helixcode.com) - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * 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 const 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. - **/ - -#define BASE64_GROUPS_IN_LINE 19 - -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) >= BASE64_GROUPS_IN_LINE) { - *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 = *(char *)save+(char)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); -} - -/** - * g_mime_utils_base64_encode_close: - * @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 the input stream to the output stream. Call this - * when finished encoding data with g_mime_utils_base64_encode_step to - * flush off the last little bit. - * - * Returns the number of bytes encoded. - **/ -static size_t -g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) -{ - unsigned char *outptr = out; - int c1, c2; - - if (inlen > 0) - outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save); - - c1 = ((unsigned char *)save)[1]; - c2 = ((unsigned char *)save)[2]; - - switch (((unsigned char *)save)[0]) { - case 2: - outptr[2] = base64_alphabet [(c2 & 0x0f) << 2]; - goto skip; - case 1: - outptr[2] = '='; - skip: - outptr[0] = base64_alphabet [c1 >> 2]; - outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)]; - outptr[3] = '='; - outptr += 4; - break; - } - - *outptr++ = 0; - - *save = 0; - *state = 0; - - return (outptr - out); -} - -static unsigned char gmime_base64_rank[256] = { - 255,255,255,255,255,255,255,255,255,254,254,255,255,254,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 254,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63, - 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255, - 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, - 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255, - 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, - 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, - 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, -}; - -/** - * g_mime_utils_base64_decode_step: - * @in: input stream - * @inlen: max length of data to decode - * @out: output stream - * @state: holds the number of bits that are stored in @save - * @save: leftover bits that have not yet been decoded - * @strict: only base64 and whitespace chars are allowed - * - * Decodes a chunk of base64 encoded data. - * - * Returns the number of bytes decoded (which have been dumped in @out). - **/ -size_t -g_mime_utils_base64_decode_step(const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save, bool strict=false) -{ - const unsigned char *inptr; - unsigned char *outptr; - const unsigned char *inend; - int saved; - unsigned char c; - int i; - - inend = in + inlen; - outptr = out; - - /* convert 4 base64 bytes to 3 normal bytes */ - saved = *save; - i = *state; - inptr = in; - while (inptr < inend) { - c = gmime_base64_rank[*inptr++]; - switch(c) { - case 0xff: // non-base64 and non-whitespace chars. not allowed in strict mode - if(strict) - throw Exception(BASE64_FORMAT, 0, "Invalid base64 char on position %d is detected", inptr-in-1); - case 0xfe: // whitespace chars 0x09, 0x0A, 0x0D, 0x20 are allowed in any mode - break; - default: - saved = (saved << 6) | c; - i++; - if (i == 4) { - *outptr++ = (unsigned char)(saved >> 16); - *outptr++ = (unsigned char)(saved >> 8); - *outptr++ = (unsigned char)(saved); - i = 0; - } - } - } - - *save = saved; - *state = i; - - /* quick scan back for '=' on the end somewhere */ - /* fortunately we can drop 1 output char for each trailing = (upto 2) */ - i = 2; - while (inptr > in && i) { - inptr--; - if (gmime_base64_rank[*inptr] <= 0xfe) { - if (*inptr == '=' && outptr > out) - outptr--; - i--; - } - } - - /* if i != 0 then there is a truncation error! */ - return (outptr - out); -} - - -char* pa_base64_encode(const char *in, size_t in_size){ - size_t new_size = ((in_size / 3 + 1) * 4); - new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/; - char* result = new(PointerFreeGC) char[new_size]; - int state=0; - int save=0; -#ifndef NDEBUG - size_t filled= -#endif - g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save); - - //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled); - assert(filled <= new_size); - - return result; -} - -struct File_base64_action_info { - unsigned char** base64; -}; - -static void file_base64_file_action(struct stat& finfo, int f, const String&, void *context) { - - if(finfo.st_size) { - File_base64_action_info& info=*static_cast(context); - *info.base64=new(PointerFreeGC) unsigned char[finfo.st_size * 2 + 6]; - unsigned char* base64 = *info.base64; - int state=0; - int save=0; - int nCount; - do { - unsigned char buffer[FILE_BUFFER_SIZE]; - nCount = file_block_read(f, buffer, sizeof(buffer)); - if( nCount ){ - size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save); - base64+=filled; - } - } while(nCount > 0); - g_mime_utils_base64_encode_close (0, 0, base64, &state, &save); - } -} - -char* pa_base64_encode(const String& file_spec){ - unsigned char* base64=0; - File_base64_action_info info={&base64}; - - file_read_action_under_lock(file_spec, "pa_base64_encode", file_base64_file_action, &info); - - return (char*)base64; -} - -void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size, bool strict) { - // every 4 base64 bytes are converted into 3 normal bytes - // not full set (tail) of 4-bytes set is ignored - size_t new_size=in_size/4*3; - result=new(PointerFreeGC) char[new_size+1/*terminator*/]; - - int state=0; - int save=0; - result_size=g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save, strict); - assert(result_size <= new_size); - result[result_size]=0; // for text files - - if(strict && state!=0) - throw Exception(BASE64_FORMAT, 0, "Unexpected end of chars"); -} - - -int file_block_read(const int f, unsigned char* buffer, const size_t size){ - int nCount = read(f, buffer, size); +ssize_t file_block_read(const int f, void* buffer, const size_t size){ + ssize_t nCount = read(f, buffer, size); if (nCount < 0) throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno); return nCount; @@ -1246,7 +995,7 @@ inline void CalcCrc32(const unsigned cha } -const unsigned long pa_crc32(const char *in, size_t in_size){ +unsigned long pa_crc32(const char *in, size_t in_size){ unsigned long crc32=0xFFFFFFFF; InitCrc32Table(); @@ -1269,7 +1018,7 @@ static void file_crc32_file_action(struc } } -const unsigned long pa_crc32(const String& file_spec){ +unsigned long pa_crc32(const String& file_spec){ unsigned long crc32=0xFFFFFFFF; file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32); return ~crc32; @@ -1302,6 +1051,51 @@ Charset* detect_charset(const char* cont return 0; } +const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){ + if(!in) + return 0; + + String::C sIn(in,strlen(in)); + + UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2); + UTF16* utf16_end=utf16; + + if(!source_charset.isUTF8()) + sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset); + + int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion); + if(status != conversionOK) + throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status); + + *utf16_end=0; + + return utf16; +} + +const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){ + if(!in) + return 0; + + const UTF16* utf16_start=in; + const UTF16* utf16_end; + + for(utf16_end=in; *utf16_end; utf16_end++); + + char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1); + char *result_end = result; + + int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion); + + if(status != conversionOK) + throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status); + + *result_end='\0'; + + if(asked_charset.isUTF8()) + return result; + + return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr(); +} static bool is_latin(const char *in){ for(; *in; in++){ @@ -1313,7 +1107,7 @@ static bool is_latin(const char *in){ #define MAX_IDNA_LENGTH 256 -const char *pa_idna_encode(const char *in, Charset &source_charset){ +const char *pa_idna_encode(const char *in, Charset& source_charset){ if(!in || is_latin(in)) return in;