|
|
| version 1.88, 2001/11/14 11:26:17 | version 1.326, 2021/12/22 21:52:49 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: commonly functions. | Parser: commonly functions. |
| Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) | Copyright (c) 2000-2020 Art. Lebedev Studio (http://www.artlebedev.com) |
| Author: Alexander Petrosyan <paf@design.ru>(http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| $Id$ | |
| */ | */ |
| #include "pa_common.h" | #include "pa_common.h" |
| #include "pa_types.h" | |
| #include "pa_exception.h" | #include "pa_exception.h" |
| #include "pa_pool.h" | |
| #include "pa_globals.h" | |
| #include "pa_value.h" | |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_string.h" | #include "pa_globals.h" |
| #include "pa_charsets.h" | |
| #include "pa_http.h" | |
| #include "pa_request_charsets.h" | |
| #include "pcre.h" | |
| #include "pa_request.h" | |
| #include "pa_idna.h" | |
| #include "pa_convert_utf.h" | |
| #ifdef _MSC_VER | |
| #include <windows.h> | |
| #include <direct.h> | |
| #endif | |
| #ifdef _MSC_VER | |
| #define pa_mkdir(path, mode) _mkdir(path) | |
| #else | |
| #define pa_mkdir(path, mode) mkdir(path, mode) | |
| #endif | |
| volatile const char * IDENT_PA_COMMON_C="$Id$" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; | |
| // some maybe-undefined constants | |
| #ifndef _O_TEXT | #ifndef _O_TEXT |
| # define _O_TEXT 0 | # define _O_TEXT 0 |
| Line 22 | Line 39 |
| #ifndef _O_BINARY | #ifndef _O_BINARY |
| # define _O_BINARY 0 | # define _O_BINARY 0 |
| #endif | #endif |
| #ifndef O_TRUNC | |
| # define O_TRUNC 0 | #ifdef HAVE_FTRUNCATE |
| # define PA_O_TRUNC 0 | |
| #else | |
| # ifdef _O_TRUNC | |
| # define PA_O_TRUNC _O_TRUNC | |
| # else | |
| # error you must have either ftruncate function or _O_TRUNC bit declared | |
| # endif | |
| #endif | #endif |
| int __vsnprintf(char *b, size_t s, const char *f, va_list l) { | // defines for globals |
| if(!s) | |
| return 0; | |
| int r; | #define FILE_STATUS_NAME "status" |
| // note: on win32& maybe somewhere else | |
| // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing | |
| --s; | |
| #if _MSC_VER | |
| /* | |
| win32: | |
| 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 | // globals |
| */ | |
| r=_vsnprintf(b, s, f, l); | |
| if(r<0) | |
| r=s; | |
| #else | |
| int r=vsnprintf(b, s, f, l); | |
| /* | |
| solaris: | |
| man vsnprintf | |
| The snprintf() function returns the number of characters | const String file_status_name(FILE_STATUS_NAME); |
| formatted, that is, the number of characters that would have | |
| been written to the buffer if it were large enough. If the | |
| value of n is 0 on a call to snprintf(), an unspecified | |
| value less than 1 is returned. | |
| */ | |
| if(r<0) | String sql_bind_name(SQL_BIND_NAME); |
| r=0; | String sql_limit_name(PA_SQL_LIMIT_NAME); |
| else if(r>s) | String sql_offset_name(PA_SQL_OFFSET_NAME); |
| r=s; | String sql_default_name(SQL_DEFAULT_NAME); |
| #endif | String sql_distinct_name(SQL_DISTINCT_NAME); |
| b[r]=0; | String sql_value_type_name(SQL_VALUE_TYPE_NAME); |
| return r; | |
| // forwards | |
| const UTF16* pa_utf16_encode(const char* in, Charset& source_charset); | |
| // functions | |
| #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 __snprintf(char *b, size_t s, const char *f, ...) { | int pa_open(const char *pathname, int flags, int mode){ |
| va_list l; | const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); |
| va_start(l, f); | return _wopen((const wchar_t *)utf16name, flags, mode); |
| int r=__vsnprintf(b, s, f, l); | |
| va_end(l); | |
| return r; | |
| } | } |
| 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); | |
| } | |
| #endif | |
| static char *strnchr(char *buf, size_t size, char c) { | /// these options were handled but not checked elsewhere, now check them |
| for(; size-->0; buf++) { | int pa_get_valid_file_options_count(HashStringValue& options) { |
| if(*buf==c) | int result=0; |
| return buf; | if(options.get(PA_SQL_LIMIT_NAME)) |
| result++; | |
| if(options.get(PA_SQL_OFFSET_NAME)) | |
| result++; | |
| if(options.get(PA_COLUMN_SEPARATOR_NAME)) | |
| result++; | |
| if(options.get(PA_COLUMN_ENCLOSER_NAME)) | |
| result++; | |
| if(options.get(PA_CHARSET_NAME)) | |
| result++; | |
| return result; | |
| } | |
| #ifndef DOXYGEN | |
| struct File_read_action_info { | |
| char **data; size_t *data_size; | |
| 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<File_read_action_info *>(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) | |
| 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; | |
| } 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; | |
| return; | |
| } | } |
| } | |
| return 0; | 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; | |
| } | } |
| void fix_line_breaks(char *buf, size_t& size) { | File_read_result file_read(Request_charsets& charsets, const String& file_spec, |
| //_asm int 3; | bool as_text, HashStringValue *options, |
| const char * const eob=buf+size; | bool fail_on_read_problem, |
| char *dest=buf; | size_t offset = 0, size_t limit = 0, bool transcode_text_result = true) { |
| // fix DOS: \r\n -> \n | File_read_result result = {false, 0, 0, 0}; |
| // fix Macintosh: \r -> \n | if(options){ |
| char *bol=buf; | int valid_options = pa_get_valid_file_options_count(*options); |
| while(char *eol=strnchr(bol, eob -bol, '\r')) { | if(valid_options != options->count()) |
| size_t len=eol-bol; | throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); |
| if(dest!=bol) | } |
| memcpy(dest, bol, len); | |
| dest+=len; | |
| *dest++='\n'; | |
| if(&eol[1]<eob && eol[1]=='\n') { // \r,\n = DOS | File_read_action_info info = {&result.str, &result.length, 0, offset, limit}; |
| bol=eol+2; | |
| size--; | result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); |
| } else // \r,not \n = Macintosh | |
| bol=eol+1; | if(as_text){ |
| if(result.success){ | |
| 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); | |
| 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); | |
| body=Charset::transcode(body, *asked_charset, charsets.source()); | |
| result.str = const_cast<char*>(body.str); // hacking a little | |
| result.length = body.length; | |
| } | |
| } | |
| if(result.length) | |
| fix_line_breaks(result.str, result.length); | |
| } | } |
| // last piece without \r, including terminating 0 | |
| if(dest!=bol) | return result; |
| memcpy(dest, bol, eob-bol); | |
| } | } |
| char *file_read_text(Pool& pool, const String& file_spec, bool fail_on_read_problem) { | File_read_result file_load(Request& r, const String& file_spec, |
| void *result; size_t size; | bool as_text, HashStringValue *options, |
| return file_read(pool, file_spec, result, size, true, fail_on_read_problem)?(char *)result:0; | bool fail_on_read_problem, |
| } | bool transcode_text_result) { |
| bool file_read(Pool& pool, const String& file_spec, | |
| void*& data, size_t& data_size, bool as_text, | size_t offset = 0; |
| bool fail_on_read_problem, | size_t limit = 0; |
| size_t offset, size_t limit) { | |
| const char *fname=file_spec.cstr(String::UL_FILE_SPEC); | if(options){ |
| //printf("file_read(%s)\n", fname); | 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 | |
| } | |
| if(file_spec.starts_with("http://")) { | |
| 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, options, transcode_text_result); | |
| File_read_result result = {true, http.str, http.length, http.headers}; | |
| return result; | |
| } else | |
| return file_read(r.charsets, file_spec, as_text, options, fail_on_read_problem, offset, limit, transcode_text_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) { | |
| 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() | |
| ); | |
| } | |
| #else | |
| void check_safe_mode(struct stat, const String&, const char*) { | |
| } | |
| #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) { | |
| const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); | |
| int f; | int f; |
| struct stat finfo; | |
| // first open, next stat: | // first open, next stat: |
| // directory update of NTFS hard links performed on open. | // directory update of NTFS hard links performed on open. |
| // ex: | // ex: |
| // a.html:^test[] and b.html hardlink to a.html | // a.html:^test[] and b.html hardlink to a.html |
| // user inserts ! before ^test in a.html | // user inserts ! before ^test in a.html |
| // directory entry of b.html in NTFS not updated at once, | // directory entry of b.html in NTFS not updated at once, |
| // they delay update till open, so we would receive "!^test[" string | // they delay update till open, so we would receive "!^test[" string |
| // if would do stat, next open. | // if would do stat, next open. |
| if( | // later: it seems, even this does not help sometimes |
| (f=open(fname, O_RDONLY|(as_text?_O_BINARY/*_O_TEXT*/:_O_BINARY)))>=0 && | if((f=pa_open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { |
| stat(fname, &finfo)==0) { | try { |
| /*if(exclusive) | if(pa_lock_shared_blocking(f)!=0) |
| flock(f, LOCK_EX);*/ | throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); |
| size_t max_size=limit?min(offset+limit, finfo.st_size)-offset:finfo.st_size; | |
| int read_size; | struct stat finfo; |
| if(!max_size) { // eof | if(pa_fstat(f, &finfo)!=0) |
| if(as_text) { | throw Exception("file.missing", // hardly possible: we just opened it OK |
| data=pool.malloc(1); | &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); |
| *(char*)data=0; | |
| } else | check_safe_mode(finfo, file_spec, fname); |
| data=0; | |
| read_size=0; | action(finfo, f, file_spec, context); |
| } else { | } catch(...) { |
| data=pool.malloc(max_size+(as_text?1:0), 3); | pa_unlock(f);close(f); |
| if(offset) | if(fail_on_read_problem) |
| lseek(f, offset, SEEK_SET); | rethrow; |
| read_size=read(f, data, max_size); | return false; |
| } | } |
| /*if(exclusive) | |
| flock(f, LOCK_UN);*/ | pa_unlock(f);close(f); |
| close(f); | |
| if(!max_size) // eof | |
| return true; | |
| if(read_size<0 || read_size>max_size) | |
| throw Exception(0, 0, | |
| &file_spec, | |
| "read failed: actually read %d bytes count not in [0..%lu] valid range", | |
| read_size, (unsigned long)max_size); //never | |
| data_size=read_size; | |
| if(as_text) { | |
| fix_line_breaks((char *)data, data_size); | |
| // note: after fixing | |
| ((char*&)data)[data_size]=0; | |
| } | |
| return true; | return true; |
| } | } else { |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception(0, 0, | throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0, |
| &file_spec, | &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); |
| "read failed: %s (%d), actual filename '%s'", | return false; |
| strerror(errno), errno, fname); | } |
| 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; | const char *str=file_spec.taint_cstr(String::L_FILE_SPEC); |
| int pos_before; | if(str[0]){ |
| while((pos_before=file_spec.pos("/", 1, pos_after))>=0) { | const char *pos=str+1; |
| mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775); | while((pos=strchr(pos,'/')) && pos[1]) { // to avoid trailing /, see #1166 |
| pos_after=pos_before+1; | pa_mkdir(pa_strdup(str,pos-str), 0775); |
| pos++; | |
| } | |
| } | } |
| } | } |
| void file_write(Pool& pool, | bool file_write_action_under_lock( |
| const String& file_spec, | const String& file_spec, |
| const void *data, size_t size, | const char* action_name, |
| bool as_text, | File_write_action action, |
| bool do_append/*, | void *context, |
| bool exclusive*/) { | bool as_text, |
| const char *fname=file_spec.cstr(String::UL_FILE_SPEC); | bool do_append, |
| bool do_block, | |
| bool fail_on_lock_problem) { | |
| const char* fname=file_spec.taint_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); |
| if((f=open(fname, | if((f=pa_open(fname, |
| O_CREAT|O_RDWR | O_CREAT|O_RDWR |
| |(as_text?_O_TEXT:_O_BINARY) | |(as_text?_O_TEXT:_O_BINARY) |
| |(do_append?O_APPEND:O_TRUNC), 0666))>=0) { | |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { |
| /*if(exclusive) | if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) { |
| flock(f, LOCK_EX);*/ | Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); |
| close(f); | |
| if(fail_on_lock_problem) | |
| throw e; | |
| return false; | |
| } | |
| try { | |
| #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE)) | |
| struct stat finfo; | |
| 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 | |
| #endif | |
| pa_unlock(f);close(f); | |
| rethrow; | |
| } | |
| if(size) write(f, data, size); | #ifdef HAVE_FTRUNCATE |
| #if O_TRUNC==0 | if(!do_append) |
| ftruncate(f, size); | 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 |
| /*if(exclusive) | pa_unlock(f);close(f); |
| flock(f, LOCK_UN);*/ | return true; |
| close(f); | |
| } else | } else |
| throw Exception(0, 0, | throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); |
| &file_spec, | // here should be nothing, see rethrow above |
| "write failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| } | } |
| // throws nothing! [this is required in file_move & file_delete] | #ifndef DOXYGEN |
| static void rmdir(const String& file_spec, size_t pos_after) { | struct File_write_action_info { |
| int pos_before; | const char* str; |
| if((pos_before=file_spec.pos("/", 1, pos_after))>=0) | size_t length; |
| rmdir(file_spec, pos_before+1); | }; |
| #endif | |
| rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC)); | |
| static void file_write_action(int f, void *context) { | |
| File_write_action_info& info=*static_cast<File_write_action_info *>(context); | |
| if(info.length) { | |
| ssize_t written=write(f, info.str, info.length); | |
| if(written<0) | |
| throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno); | |
| if((size_t)written!=info.length) | |
| throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length); | |
| } | |
| } | } |
| void file_delete(Pool& pool, const String& file_spec) { | |
| const char *fname=file_spec.cstr(String::UL_FILE_SPEC); | |
| if(unlink(fname)!=0) | |
| throw Exception(0, 0, | |
| &file_spec, | |
| "unlink failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| rmdir(file_spec, 1); | |
| } | |
| void file_move(Pool& pool, const String& old_spec, const String& new_spec) { | |
| const char *old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC); | |
| const char *new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC); | |
| create_dir_for_file(new_spec); | |
| if(rename(old_spec_cstr, new_spec_cstr)!=0) | void file_write( |
| throw Exception(0, 0, | Request_charsets& charsets, |
| &old_spec, | const String& file_spec, |
| "rename failed: %s (%d), actual filename '%s' to '%s'", | const char* data, |
| strerror(errno), errno, old_spec_cstr, new_spec_cstr); | size_t size, |
| bool as_text, | |
| bool do_append, | |
| Charset* asked_charset) { | |
| if(as_text && asked_charset){ | |
| String::C body=String::C(data, size); | |
| body=Charset::transcode(body, charsets.source(), *asked_charset); | |
| data=body.str; | |
| size=body.length; | |
| }; | |
| File_write_action_info info={data, size}; | |
| file_write_action_under_lock( | |
| file_spec, | |
| "write", | |
| file_write_action, | |
| &info, | |
| as_text, | |
| do_append); | |
| } | |
| static size_t get_dir(char* fname, size_t helper_length){ | |
| bool dir=false; | |
| size_t pos=0; | |
| for(pos=helper_length; pos; pos--){ | |
| char c=fname[pos-1]; | |
| if(c=='/' || c=='\\'){ | |
| fname[pos-1]=0; | |
| dir=true; | |
| } else if(dir) break; | |
| } | |
| return pos; | |
| } | |
| rmdir(old_spec, 1); | 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) { | |
| char c=fname[size-1]; | |
| if(c=='/' || c=='\\') | |
| fname[--size]=0; | |
| else | |
| break; | |
| } | |
| } | |
| static bool entry_readable(const String& file_spec, bool need_dir) { | |
| const char *fname=file_spec.cstr(String::UL_FILE_SPEC); | |
| struct stat finfo; | struct stat finfo; |
| if(access(fname, R_OK)==0 && stat(fname, &finfo)==0) { | if(entry_exists(fname, &finfo)) { |
| bool is_dir=finfo.st_mode&S_IFDIR != 0; | bool is_dir=(finfo.st_mode&S_IFDIR) != 0; |
| return is_dir==need_dir; | return is_dir==need_dir; |
| } | } |
| return false; | return false; |
| } | } |
| bool file_readable(const String& file_spec) { | |
| return entry_readable(file_spec, false); | static bool entry_ifdir(const String& file_spec, bool need_dir) { |
| return entry_ifdir(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir); | |
| } | } |
| bool dir_readable(const String& file_spec) { | |
| return entry_readable(file_spec, true); | // throws nothing! [this is required in file_move & file_delete] |
| static void rmdir(const String& file_spec, size_t pos_after) { | |
| char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC); | |
| size_t length=strlen(dir_spec); | |
| while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){ | |
| #ifdef _MSC_VER | |
| if(!entry_ifdir(dir_spec, true)) | |
| break; | |
| DWORD attrs=GetFileAttributes(dir_spec); | |
| if( | |
| (attrs==INVALID_FILE_ATTRIBUTES) | |
| || !(attrs & FILE_ATTRIBUTE_DIRECTORY) | |
| || (attrs & FILE_ATTRIBUTE_REPARSE_POINT) | |
| ) | |
| break; | |
| #endif | |
| if( rmdir(dir_spec) ) | |
| break; | |
| }; | |
| } | } |
| String *file_readable(const String& path, const String& name) { | |
| String *result=new(path.pool()) String(path); | bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) { |
| *result << "/"; | const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); |
| *result << name; | if(unlink(fname)!=0) { |
| return file_readable(*result)?result: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); | |
| else | |
| return false; | |
| } | |
| if(!keep_empty_dirs) | |
| rmdir(file_spec, 1); | |
| return true; | |
| } | } |
| void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) { | |
| const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC); | |
| const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC); | |
| create_dir_for_file(new_spec); | |
| if(rename(old_spec_cstr, new_spec_cstr)!=0) | |
| throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0, | |
| &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr); | |
| if(!keep_empty_dirs) | |
| rmdir(old_spec, 1); | |
| } | |
| bool file_exist(const String& file_spec) { | |
| return entry_ifdir(file_spec, false); | |
| } | |
| bool dir_exists(const String& file_spec) { | |
| return entry_ifdir(file_spec, true); | |
| } | |
| const String* file_exist(const String& path, const String& name) { | |
| String& result=*new String(path); | |
| if(path.last_char() != '/') | |
| result << "/"; | |
| result << name; | |
| return file_exist(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.taint_cstr(String::L_FILE_SPEC), X_OK)==0; |
| } | } |
| bool file_stat(const String& file_spec, | bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { |
| size_t& rsize, | const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); |
| time_t& ratime, | struct stat finfo; |
| time_t& rmtime, | if(pa_stat(fname, &finfo)!=0) { |
| time_t& rctime, | |
| bool fail_on_read_problem) { | |
| Pool& pool=file_spec.pool(); | |
| const char *fname=file_spec.cstr(String::UL_FILE_SPEC); | |
| struct stat finfo; | |
| if(stat(fname, &finfo)!=0) | |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception(0, 0, | throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname); |
| &file_spec, | |
| "getting file size failed: %s (%d), real filename '%s'", | |
| strerror(errno), errno, fname); | |
| else | else |
| return false; | return false; |
| } | |
| rsize=finfo.st_size; | rsize=finfo.st_size; |
| ratime=finfo.st_atime; | ratime=(time_t)finfo.st_atime; |
| rmtime=finfo.st_mtime; | rmtime=(time_t)finfo.st_mtime; |
| rctime=finfo.st_ctime; | 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 | |
| */ | |
| bool capitalized(const char* s){ | |
| bool upper=true; | |
| for(const char* c=s; *c; c++){ | |
| if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c))) | |
| return false; | |
| upper=strchr("-_ ", *c) != 0; | |
| } | |
| return true; | return true; |
| } | } |
| char *getrow(char **row_ref, char delim) { | const char* capitalize(const char* s){ |
| char *result=*row_ref; | if(!s || capitalized(s)) |
| if(result) { | return s; |
| *row_ref=strchr(result, delim); | |
| char* result=pa_strdup(s); | |
| if(result){ | |
| bool upper=true; | |
| for(char* c=result; *c; c++){ | |
| *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c); | |
| upper=strchr("-_ ", *c) != 0; | |
| } | |
| } | |
| return (const char*)result; | |
| } | |
| 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 length){ | |
| char *result=pa_strdup(s, length); | |
| for(char* c=result; *c; c++) | |
| *c=(char)toupper((unsigned char)*c); | |
| return result; | |
| } | |
| void fix_line_breaks(char *str, size_t& length) { | |
| //_asm int 3; | |
| const char* const eob=str+length; | |
| char* dest=str; | |
| // fix DOS: \r\n -> \n | |
| // fix Macintosh: \r -> \n | |
| char* bol=str; | |
| while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { | |
| size_t len=eol-bol; | |
| if(dest!=bol) | |
| memmove(dest, bol, len); | |
| dest+=len; | |
| *dest++='\n'; | |
| if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS | |
| bol=eol+2; | |
| length--; | |
| } else // \r, not \n = Macintosh | |
| bol=eol+1; | |
| } | |
| // last piece without \r | |
| if(dest!=bol) | |
| memmove(dest, bol, eob-bol); | |
| str[length]=0; // terminating | |
| } | |
| /** | |
| scans for @a delim[default \n] in @a *row_ref, | |
| @return piece of line before it or end of string, if no @a delim found | |
| assigns @a *row_ref to point right after delimiter if there were one | |
| or to zero if no @a delim were found. | |
| */ | |
| char* getrow(char* *row_ref, char delim) { | |
| char* result=*row_ref; | |
| if(result) { | |
| *row_ref=strchr(result, delim); | |
| if(*row_ref) | if(*row_ref) |
| *((*row_ref)++)=0; | *((*row_ref)++)=0; |
| else if(!*result) | else if(!*result) |
| return 0; | return 0; |
| } | } |
| return result; | return result; |
| } | } |
| char *lsplit(char *string, char delim) { | char* lsplit(char* string, char delim) { |
| if(string) { | if(string) { |
| char *v=strchr(string, delim); | char* v=strchr(string, delim); |
| if(v) { | if(v) { |
| *v=0; | *v=0; |
| return v+1; | return v+1; |
| } | } |
| } | } |
| return 0; | return 0; |
| } | } |
| char *lsplit(char **string_ref, char delim) { | char* lsplit(char* *string_ref, char delim) { |
| char *result=*string_ref; | char* result=*string_ref; |
| char *next=lsplit(*string_ref, delim); | char* next=lsplit(*string_ref, delim); |
| *string_ref=next; | *string_ref=next; |
| return result; | return result; |
| } | } |
| char *rsplit(char *string, char delim) { | char* rsplit(char* string, char delim) { |
| if(string) { | if(string) { |
| char *v=strrchr(string, delim); | char* v=strrchr(string, delim); |
| if(v) { | if(v) { |
| *v=0; | *v=0; |
| return v+1; | return v+1; |
| } | } |
| } | } |
| return NULL; | return NULL; |
| } | } |
| /// @todo less stupid type detection | void pa_strncpy(char *dst, const char *src, size_t n){ |
| char *format(Pool& pool, double value, char *fmt) { | size_t left = n; |
| char *result=(char *)pool.malloc(MAX_NUMBER, 4); | |
| if(fmt) | if (left != 0 && src) { |
| if(strpbrk(fmt, "diouxX")) | while (--left != 0) { |
| if(strpbrk(fmt, "ouxX")) | if ((*dst++ = *src++) == '\0') |
| snprintf(result, MAX_NUMBER, fmt, (uint)value ); | return; |
| else | } |
| snprintf(result, MAX_NUMBER, fmt, (int)value ); | } |
| else | if (n != 0) |
| snprintf(result, MAX_NUMBER, fmt, value); | *dst = '\0'; |
| else | } |
| snprintf(result, MAX_NUMBER, "%d", (int)value); | |
| #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 | |
| // width, precision: non negative decimal number | |
| enum FormatType { | |
| FormatInvalid, | |
| FormatInt, | |
| FormatUInt, | |
| FormatDouble | |
| }; | |
| FormatType format_type(const char* fmt){ | |
| enum FormatState { | |
| Percent, | |
| Flags, | |
| Width, | |
| Precision, | |
| Done | |
| } state=Percent; | |
| FormatType result=FormatInvalid; | |
| const char* pos=fmt; | |
| while(char c=*(pos++)){ | |
| switch(state){ | |
| case Percent: | |
| if(c=='%'){ | |
| state=Flags; | |
| } else { | |
| return FormatInvalid; // 1st char must be '%' only | |
| } | |
| break; | |
| case Flags: | |
| if(strchr("-+ #0", c)!=0){ | |
| break; | |
| } | |
| // go to the next step | |
| case Width: | |
| if(c=='.'){ | |
| state=Precision; | |
| break; | |
| } | |
| // go to the next step | |
| case Precision: | |
| if(c>='0' && c<='9'){ | |
| if(state == Flags) state=Width; // no more flags | |
| break; | |
| } else if(c=='d' || c=='i'){ | |
| result=FormatInt; | |
| } else if(strchr("feEgG", c)!=0){ | |
| result=FormatDouble; | |
| } else if(strchr("uoxX", c)!=0){ | |
| result=FormatUInt; | |
| } else { | |
| return FormatInvalid; // invalid char | |
| } | |
| state=Done; | |
| break; | |
| case Done: | |
| return FormatInvalid; // no chars allowed after 'type' | |
| } | |
| } | |
| return result; | return result; |
| } | } |
| const char* format(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); | |
| break; | |
| case FormatUInt: | |
| size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)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); | |
| if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing | |
| throw Exception(PARSER_RUNTIME, 0, "Error occurred white executing snprintf with format string '%s'.", fmt); | |
| } | |
| return pa_strdup(local_buf, (size_t)size); | |
| } | |
| 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; | fflush(stdout); |
| return to_write-size; | |
| #else | #else |
| return fwrite(buf, 1, size, stdout); | size_t result=fwrite(buf, 1, size, stdout); |
| fflush(stdout); | |
| return result; | |
| #endif | #endif |
| } | } |
| char *unescape_chars(Pool& pool, const char *cp, int len) { | enum EscapeState { |
| char *s=(char *)pool.malloc(len + 1, 5); | EscapeRest, |
| enum EscapeState { | EscapeFirst, |
| EscapeRest, | EscapeSecond, |
| EscapeFirst, | EscapeUnicode |
| EscapeSecond | }; |
| } escapeState=EscapeRest; | |
| int escapedValue=0; | // @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; | |
| EscapeState escapeState=EscapeRest; | |
| uint escapedValue=0; | |
| int srcPos=0; | int srcPos=0; |
| int dstPos=0; | short int jsCnt=0; |
| while(srcPos < len) { | while(srcPos<len){ |
| int ch=cp[srcPos]; | uchar c=(uchar)cp[srcPos]; |
| switch(escapeState) { | if(c=='%' || (c=='\\' && js)){ |
| case EscapeRest: | escapeState=EscapeFirst; |
| if(ch=='%') { | } else { |
| escapeState=EscapeFirst; | switch(escapeState) { |
| } else if(ch=='+') { | case EscapeRest: |
| s[dstPos++]=' '; | if(c=='+' && !js){ |
| } else { | *dst++=' '; |
| s[dstPos++]=ch; | } else { |
| *dst++=c; | |
| } | |
| break; | |
| case EscapeFirst: | |
| if(charset && c=='u'){ | |
| // escaped unicode value: %u0430 | |
| jsCnt=0; | |
| escapedValue=0; | |
| escapeState=EscapeUnicode; | |
| } else { | |
| if(isxdigit(c)){ | |
| escapedValue=hex_value[c] << 4; | |
| escapeState=EscapeSecond; | |
| } else { | |
| *dst++=c; | |
| escapeState=EscapeRest; | |
| } | |
| } | |
| break; | |
| case EscapeSecond: | |
| if(isxdigit(c)){ | |
| escapedValue+=hex_value[c]; | |
| *dst++=(char)escapedValue; | |
| } | |
| escapeState=EscapeRest; | |
| break; | |
| case EscapeUnicode: | |
| if(isxdigit(c)){ | |
| 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, '?'); | |
| escapeState=EscapeRest; | |
| } | |
| } else { | |
| // not full unicode value | |
| escapeState=EscapeRest; | |
| } | |
| break; | |
| } | } |
| break; | |
| case EscapeFirst: | |
| escapedValue=hex_value[ch] << 4; | |
| escapeState=EscapeSecond; | |
| break; | |
| case EscapeSecond: | |
| escapedValue +=hex_value[ch]; | |
| s[dstPos++]=escapedValue; | |
| escapeState=EscapeRest; | |
| break; | |
| } | } |
| srcPos++; | srcPos++; |
| } | } |
| s[dstPos]=0; | |
| *dst=0; // zero-termination | |
| return s; | return s; |
| } | } |
| /// used by attributed_meaning_to_string / append_attribute_subattribute | char *search_stop(char*& current, char cstop_at) { |
| struct Attributed_meaning_info { | // sanity check |
| String *header; // header line being constructed | if(!current) |
| String::Untaint_lang lang; // language in which to append to that line | return 0; |
| }; | |
| 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 | |
| *ami.header << "; "; | |
| ami.header->append(akey, ami.lang); | |
| *ami.header << "="; | |
| ami.header->append(static_cast<Value *>(avalue)->as_string(), ami.lang); | |
| } | |
| const String& attributed_meaning_to_string(Value& meaning, | |
| String::Untaint_lang lang) { | |
| String &result=*new(meaning.pool()) String(meaning.pool()); | |
| if(Hash *hash=meaning.get_hash(0)) { | |
| // $value(value) $subattribute(subattribute value) | |
| if(Value *value=static_cast<Value *>(hash->get(*value_name))) | |
| result.append(value->as_string(), lang, true); | |
| Attributed_meaning_info attributed_meaning_info={ | |
| &result, | |
| lang | |
| }; | |
| hash->for_each(append_attribute_subattribute, &attributed_meaning_info); | |
| } else // result value | |
| result.append(meaning.as_string(), lang, true); | |
| // skip leading WS | |
| while(*current==' ' || *current=='\t') | |
| current++; | |
| if(!*current) | |
| return current=0; | |
| char *result=current; | |
| if(char *pstop_at=strchr(current, cstop_at)) { | |
| *pstop_at=0; | |
| current=pstop_at+1; | |
| } else | |
| current=0; | |
| return result; | return result; |
| } | } |
| #ifdef WIN32 | #ifdef WIN32 |
| void back_slashes_to_slashes(char *s) { | void back_slashes_to_slashes(char* s) { |
| if(s) | if(s) |
| for(; *s; s++) | for(; *s; s++) |
| if(*s=='\\') | if(*s=='\\') |
| *s='/'; | *s='/'; |
| } | |
| /* | |
| void slashes_to_back_slashes(char *s) { | |
| if(s) | |
| for(; *s; s++) | |
| if(*s=='/') | |
| *s='\\'; | |
| } | } |
| */ | |
| #endif | #endif |
| bool StrEqNc(const char *s1, const char *s2, bool strict) { | size_t strpos(const char *str, const char *substr) { |
| while(true) { | const char *p = strstr(str, substr); |
| if(!(*s1)) { | return (p==0)?STRING_NOT_FOUND:p-str; |
| if(!(*s2)) | } |
| return true; | |
| else | size_t remove_crlf(char* start, char* end) { |
| return !strict; | char* from=start; |
| } else if(!(*s2)) | char* to=start; |
| return !strict; | bool skip=false; |
| if(isalpha(*s1)) { | while(from < end){ |
| if(tolower(*s1) !=tolower(*s2)) | switch(*from){ |
| return false; | case '\n': |
| } else if((*s1) !=(*s2)) | case '\r': |
| return false; | case '\t': |
| s1++; | case ' ': |
| s2++; | if(!skip){ |
| *to=' '; | |
| to++; | |
| skip=true; | |
| } | |
| break; | |
| default: | |
| if(from != to) | |
| *to=*from; | |
| to++; | |
| skip=false; | |
| } | |
| from++; | |
| } | |
| return to-start; | |
| } | |
| const char* hex_digits="0123456789ABCDEF"; | |
| const char* hex_string(unsigned char* bytes, size_t size, bool upcase) { | |
| char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/]; | |
| unsigned char *src=bytes; | |
| unsigned char *end=bytes+size; | |
| char *dest=bytes_hex; | |
| const char *hex=upcase? hex_digits : "0123456789abcdef"; | |
| for(; src<end; src++) { | |
| *dest++=hex[*src/0x10]; | |
| *dest++=hex[*src%0x10]; | |
| } | |
| *dest=0; | |
| return bytes_hex; | |
| } | |
| 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; | |
| } | |
| static unsigned long crc32Table[256]; | |
| static void InitCrc32Table() | |
| { | |
| if(crc32Table[1] == 0){ | |
| // This is the official polynomial used by CRC32 in PKZip. | |
| // Often times the polynomial shown reversed as 0x04C11DB7. | |
| static const unsigned long dwPolynomial = 0xEDB88320; | |
| for(int i = 0; i < 256; i++) | |
| { | |
| unsigned long dwCrc = i; | |
| for(int j = 8; j > 0; j--) | |
| { | |
| if(dwCrc & 1) | |
| dwCrc = (dwCrc >> 1) ^ dwPolynomial; | |
| else | |
| dwCrc >>= 1; | |
| } | |
| crc32Table[i] = dwCrc; | |
| } | |
| } | |
| } | |
| inline void CalcCrc32(const unsigned char byte, unsigned long &crc32) | |
| { | |
| crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)]; | |
| } | |
| unsigned long pa_crc32(const char *in, size_t in_size){ | |
| unsigned long crc32=0xFFFFFFFF; | |
| InitCrc32Table(); | |
| for(size_t i = 0; i<in_size; i++) | |
| CalcCrc32(in[i], crc32); | |
| return ~crc32; | |
| } | |
| static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) { | |
| unsigned long& crc32=*static_cast<unsigned long *>(context); | |
| if(finfo.st_size) { | |
| InitCrc32Table(); | |
| int nCount=0; | |
| do { | |
| unsigned char buffer[FILE_BUFFER_SIZE]; | |
| nCount = file_block_read(f, buffer, sizeof(buffer)); | |
| for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32); | |
| } while(nCount > 0); | |
| } | } |
| } | } |
| static bool isLeap(int year) { | unsigned long pa_crc32(const String& file_spec){ |
| return !( | unsigned long crc32=0xFFFFFFFF; |
| (year % 4) || ((year % 400) && !(year % 100)) | file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32); |
| ); | return ~crc32; |
| } | } |
| int getMonthDays(int year, int month) { | // content-type: xxx; charset=WE-NEED-THIS |
| int monthDays[]={ | // content-type: xxx; charset="WE-NEED-THIS" |
| 31, | // content-type: xxx; charset="WE-NEED-THIS"; |
| isLeap(year) ? 29 : 28, | Charset* detect_charset(const char* content_type){ |
| 31, | if(content_type){ |
| 30, | char* CONTENT_TYPE=str_upper(content_type); |
| 31, | |
| 30, | if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){ |
| 31, | begin+=8; // skip "CHARSET=" |
| 31, | char* end=0; |
| 30, | if(*begin && (*begin=='"' || *begin =='\'')){ |
| 31, | char quote=*begin; |
| 30, | begin++; |
| 31 | end=(char*)strchr(begin, quote); |
| }; | } |
| return monthDays[month]; | if(!end) |
| } | end=(char*)strchr(begin, ';'); |
| void remove_crlf(char *start, char *end) { | if(end) |
| for(char *p=start; p<end; p++) | *end=0; // terminator |
| switch(*p) { | |
| case '\n': *p='|'; break; | return *begin ? &pa_charsets.get_direct(begin) : 0; |
| case '\r': *p=' '; break; | |
| } | } |
| } | |
| 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++){ | |
| if ((unsigned char)(*in) > 0x7F) | |
| return false; | |
| } | |
| return true; | |
| } | |
| #define MAX_IDNA_LENGTH 256 | |
| const char *pa_idna_encode(const char *in, Charset& source_charset){ | |
| if(!in || is_latin(in)) | |
| return in; | |
| uint32_t utf32[MAX_IDNA_LENGTH]; | |
| uint32_t *utf32_end=utf32; | |
| String::C sIn(in,strlen(in)); | |
| if(!source_charset.isUTF8()) | |
| sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset); | |
| int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status); | |
| *utf32_end=0; | |
| char *result = (char *)pa_malloc(MAX_IDNA_LENGTH); | |
| status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0); | |
| if(status != IDNA_SUCCESS) | |
| throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status)); | |
| return result; | |
| } | |
| const char *pa_idna_decode(const char *in, Charset &asked_charset){ | |
| if(!in || !(*in)) | |
| return in; | |
| uint32_t utf32[MAX_IDNA_LENGTH]; | |
| const uint32_t *utf32_start=utf32; | |
| uint32_t *utf32_end; | |
| int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0); | |
| if(status != IDNA_SUCCESS) | |
| throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status)); | |
| for(utf32_end=utf32; *utf32_end; utf32_end++); | |
| char *result = (char *)pa_malloc(MAX_IDNA_LENGTH); | |
| char *result_end = result; | |
| status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status); | |
| *result_end='\0'; | |
| if(!asked_charset.isUTF8()) | |
| result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr(); | |
| return result; | |
| } | |
| /// must be last in this file | |
| #undef vsnprintf | |
| int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) { | |
| if(!s) | |
| return 0; | |
| int r; | |
| // note: on win32 & maybe somewhere else | |
| // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing | |
| // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010 | |
| --s; | |
| // clients do not check for negative 's', feature: ignore such prints | |
| if((ssize_t)s<0) | |
| return 0; | |
| #ifdef _MSC_VER | |
| // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned | |
| r=_vsnprintf(b, s, f, l); | |
| if(r<0) | |
| r=s; | |
| #else | |
| r=vsnprintf(b, s, f, l); | |
| /* | |
| solaris: man vsnprintf | |
| The snprintf() function returns the number of characters | |
| formatted, that is, the number of characters that would have | |
| been written to the buffer if it were large enough. If the | |
| value of n is 0 on a call to snprintf(), an unspecified | |
| value less than 1 is returned. | |
| */ | |
| if(r<0) | |
| r=0; | |
| else if((size_t)r>s) | |
| r=s; | |
| #endif | |
| b[r]=0; | |
| return r; | |
| } | |
| int pa_snprintf(char* b, size_t s, const char* f, ...) { | |
| va_list l; | |
| va_start(l, f); | |
| int r=pa_vsnprintf(b, s, f, l); | |
| va_end(l); | |
| return r; | |
| } | |