--- parser3/src/main/pa_common.C 2010/05/17 22:54:49 1.258 +++ parser3/src/main/pa_common.C 2016/12/24 23:02:13 1.297 @@ -1,7 +1,7 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2000-2015 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) * BASE64 part @@ -26,8 +26,6 @@ * */ -static const char * const IDENT_COMMON_C="$Date: 2010/05/17 22:54:49 $"; - #include "pa_common.h" #include "pa_exception.h" #include "pa_hash.h" @@ -38,6 +36,22 @@ static const char * const IDENT_COMMON_C #include "pcre.h" #include "pa_request.h" +#include "pa_idna.h" +#include "pa_convert_utf.h" + +#ifdef _MSC_VER +#include +#include +#endif + +#ifdef _MSC_VER +#define pa_mkdir(path, mode) _mkdir(path) +#else +#define pa_mkdir(path, mode) mkdir(path, mode) +#endif + +volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.297 2016/12/24 23:02:13 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; + // some maybe-undefined constants #ifndef _O_TEXT @@ -67,74 +81,13 @@ const String file_status_name(FILE_STATU // 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; -} - -const char* capitalize(const char* s){ - if(!s || capitalized(s)) - return s; - - 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; -} - -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) - memcpy(dest, bol, len); - dest+=len; - *dest++='\n'; - - if(&eol[1](context); size_t to_read_size=info.count; if(!to_read_size) to_read_size=(size_t)finfo.st_size; - assert( !(info.buf && as_text) ); - if(to_read_size) { + 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); - *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) - throw Exception("file.read", - &file_spec, - "read failed: actually read %u bytes count not in [0..%u] valid range", - *info.data_size, to_read_size); + *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); @@ -200,36 +144,28 @@ File_read_result file_read(Request_chars if(params){ int valid_options=pa_get_valid_file_options_count(*params); if(valid_options!=params->count()) - throw Exception(PARSER_RUNTIME, - 0, - INVALID_OPTION_PASSED); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } 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, - 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){ - if(result.length>=3 && strncmp(result.str, "\xEF\xBB\xBF", 3)==0){ - // skip UTF-8 signature (BOM code) - result.str+=3; - result.length-=3; - } - - if(result.length && transcode_text_result && params){ // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below - if(Value* vcharset_name=params->get(PA_CHARSET_NAME)){ - Charset asked_charset=::charsets.get(vcharset_name->as_string(). - change_case(charsets.source(), String::CC_UPPER)); + Charset* asked_charset=0; + if(params) + if(Value* vcharset_name=params->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()); - 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) @@ -247,9 +183,7 @@ File_read_result file_load(Request& r, c File_read_result result={false, 0, 0, 0}; 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"); + 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); @@ -265,22 +199,23 @@ File_read_result file_load(Request& r, c } -void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { #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()); + 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, @@ -302,37 +237,29 @@ bool file_read_action_under_lock(const S if((f=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); + throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); struct stat finfo; if(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); + &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); check_safe_mode(finfo, file_spec, fname); - action(finfo, f, file_spec, fname, as_text, context); + action(finfo, f, file_spec, context); } catch(...) { pa_unlock(f);close(f); if(fail_on_read_problem) rethrow; - return false; + return false; } pa_unlock(f);close(f); return true; } else { if(fail_on_read_problem) - throw Exception(errno==EACCES?"file.access" - :(errno==ENOENT || errno==ENOTDIR || errno==ENODEV)?"file.missing":0, - &file_spec, - "%s failed: %s (%d), actual filename '%s'", - action_name, strerror(errno), errno, fname); + throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0, + &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); return false; } } @@ -341,7 +268,7 @@ void create_dir_for_file(const String& f size_t pos_after=1; size_t pos_before; while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { - mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775); + pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775); pos_after=pos_before+1; } } @@ -365,10 +292,7 @@ bool file_write_action_under_lock( |(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); + 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; @@ -398,10 +322,7 @@ bool file_write_action_under_lock( 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); + throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); // here should be nothing, see rethrow above } @@ -411,16 +332,18 @@ struct File_write_action_info { size_t length; }; #endif + static void file_write_action(int f, void *context) { File_write_action_info& info=*static_cast(context); if(info.length) { - int written=write(f, info.str, info.length); + ssize_t written=write(f, info.str, info.length); if(written<0) - throw Exception("file.access", - 0, - "write failed: %s (%d)", strerror(errno), errno); + 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_write( Request_charsets& charsets, const String& file_spec, @@ -448,43 +371,92 @@ void file_write( 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; +} + +static bool entry_readable(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; + } + } + + struct stat finfo; + if(access(fname, R_OK)==0 && 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); +} + // throws nothing! [this is required in file_move & file_delete] static void rmdir(const String& file_spec, size_t pos_after) { - size_t pos_before; - if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) - rmdir(file_spec, pos_before+1); - - rmdir(file_spec.mid(0, pos_after-1/* / */).taint_cstr(String::L_FILE_SPEC)); + 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_readable(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; + }; } -bool file_delete(const String& file_spec, bool fail_on_problem) { +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(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); + 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); - rmdir(file_spec, 1); return true; } -void file_move(const String& old_spec, const String& new_spec) { +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); + 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); - rmdir(old_spec, 1); + if(!keep_empty_dirs) + rmdir(old_spec, 1); } @@ -501,26 +473,6 @@ bool entry_exists(const String& file_spe return entry_exists(fname, 0); } -static bool entry_readable(const String& file_spec, bool need_dir) { - char* fname=file_spec.taint_cstrm(String::L_FILE_SPEC); - if(need_dir) { - size_t size=strlen(fname); - while(size) { - char c=fname[size-1]; - if(c=='/' || c=='\\') - fname[--size]=0; - else - break; - } - } - struct stat finfo; - if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) { - bool is_dir=(finfo.st_mode&S_IFDIR) != 0; - return is_dir==need_dir; - } - return false; -} - bool file_exist(const String& file_spec) { return entry_readable(file_spec, false); } @@ -531,7 +483,8 @@ bool dir_exists(const String& file_spec) const String* file_exist(const String& path, const String& name) { String& result=*new String(path); - result << "/"; + if(path.last_char() != '/') + result << "/"; result << name; return file_exist(result)?&result:0; } @@ -540,22 +493,15 @@ bool file_executable(const String& file_ return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0; } -bool file_stat(const String& file_spec, - size_t& rsize, - time_t& ratime, - time_t& rmtime, - time_t& rctime, - bool fail_on_read_problem) { +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(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); + 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; @@ -563,6 +509,82 @@ bool file_stat(const String& file_spec, return true; } +/** + 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; +} + +const char* capitalize(const char* s){ + if(!s || capitalized(s)) + return s; + + 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 helper_length){ + char *result=pa_strdup(s, helper_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); + 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]= 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 occure white executing snprintf with format string '%s'.", fmt); } return pa_strdup(local_buf, (size_t)size); @@ -740,7 +758,7 @@ char* unescape_chars(const char* cp, int } else { switch(escapeState) { case EscapeRest: - if(!js && c=='+'){ + if(c=='+' && !js){ *dst++=' '; } else { *dst++=c; @@ -792,6 +810,26 @@ char* unescape_chars(const char* cp, int return s; } +char *search_stop(char*& current, char cstop_at) { + // sanity check + if(!current) + return 0; + + // 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; +} + #ifdef WIN32 void back_slashes_to_slashes(char* s) { if(s) @@ -799,96 +837,13 @@ void back_slashes_to_slashes(char* s) { if(*s=='\\') *s='/'; } -/* -void slashes_to_back_slashes(char* s) { - if(s) - for(; *s; s++) - if(*s=='/') - *s='\\'; -} -*/ #endif -bool StrStartFromNC(const char* str, const char* substr, bool equal){ - while(true) { - if(!(*substr)){ - if(!(*str)) - return true; - else - return !equal; - } - if(!(*str)) - return false; - if(isalpha((unsigned char)*str)) { - if(tolower((unsigned char)*str)!=tolower((unsigned char)*substr)) - return false; - } else if((*str) != (*substr)) - return false; - str++; - substr++; - } -} - size_t strpos(const char *str, const char *substr) { const char *p = strstr(str, substr); return (p==0)?STRING_NOT_FOUND:p-str; } -// content-type: xxx; charset=WE-NEED-THIS -// content-type: xxx; charset="WE-NEED-THIS" -// content-type: xxx; charset="WE-NEED-THIS"; -Charset* detect_charset(const char* content_type){ - if(content_type){ - char* CONTENT_TYPE=pa_strdup(content_type); - - for(char *p=CONTENT_TYPE; *p; p++) - *p=(char)toupper((unsigned char)*p); - - if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){ - begin+=8; // skip "CHARSET=" - char* end=0; - if(*begin && (*begin=='"' || *begin =='\'')){ - char quote=*begin; - begin++; - end=(char*)strchr(begin, quote); - } - if(!end) - end=(char*)strchr(begin, ';'); - - if(end) - *end=0; // terminator - - return *begin?&charsets.get(begin):0; - } - } - return 0; -} - - -static bool isLeap(int year) { - return !( - (year % 4) || ((year % 400) && !(year % 100)) - ); -} - -int getMonthDays(int year, int month) { - static int monthDays[]={ - 31, - 28, - 31, - 30, - 31, - 30, - 31, - 31, - 30, - 31, - 30, - 31 - }; - return (month == 1 /* january -- 0 */ && isLeap(year)) ? 29 : monthDays[month]; -} - int remove_crlf(char* start, char* end) { char* from=start; char* to=start; @@ -916,60 +871,23 @@ int remove_crlf(char* start, char* end) return to-start; } +const char* hex_digits="0123456789ABCDEF"; -/// must be last in this file -#undef vsnprintf -int __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 - --s; - - // clients do not check for negative 's', feature: ignore such prints - if((ssize_t)s<0) - return 0; +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; -#if _MSC_VER - /* - win32: - mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm + const char *hex=upcase? hex_digits : "0123456789abcdef"; - if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 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; -} + for(; src> 16); - *outptr++ = (unsigned char)(saved >> 8); - *outptr++ = (unsigned char)(saved); - i = 0; - } + 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; + } } } @@ -1205,7 +1130,7 @@ g_mime_utils_base64_decode_step (const u i = 2; while (inptr > in && i) { inptr--; - if (gmime_base64_rank[*inptr] != 0xff) { + if (gmime_base64_rank[*inptr] <= 0xfe) { if (*inptr == '=' && outptr > out) outptr--; i--; @@ -1234,23 +1159,11 @@ char* pa_base64_encode(const char *in, s return result; } +struct File_base64_action_info { + unsigned char** base64; +}; -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; -} - - -static void file_base64_file_action( - struct stat& finfo, - int f, - const String&, const char* /*fname*/, bool, - void *context) { +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); @@ -1271,29 +1184,69 @@ static void file_base64_file_action( } } -void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size) { - /* wont go to more than had (overly conservative) */ - result=new(PointerFreeGC) char[in_size+1/*terminator*/]; +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); - assert(result_size <= in_size); + 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); if (nCount < 0) - throw Exception("file.read", - 0, - "read failed: %s (%d)", strerror(errno), errno); + throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno); return nCount; } -const unsigned long pa_crc32(const char *in, size_t in_size){ +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(); @@ -1303,17 +1256,7 @@ const unsigned long pa_crc32(const char return ~crc32; } -const 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; -} - -static void file_crc32_file_action( - struct stat& finfo, - int f, - const String&, const char* /*fname*/, bool, - void *context) { +static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) { unsigned long& crc32=*static_cast(context); if(finfo.st_size) { InitCrc32Table(); @@ -1326,3 +1269,152 @@ static void file_crc32_file_action( } } +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; +} + +// content-type: xxx; charset=WE-NEED-THIS +// content-type: xxx; charset="WE-NEED-THIS" +// content-type: xxx; charset="WE-NEED-THIS"; +Charset* detect_charset(const char* content_type){ + if(content_type){ + char* CONTENT_TYPE=str_upper(content_type); + + if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){ + begin+=8; // skip "CHARSET=" + char* end=0; + if(*begin && (*begin=='"' || *begin =='\'')){ + char quote=*begin; + begin++; + end=(char*)strchr(begin, quote); + } + if(!end) + end=(char*)strchr(begin, ';'); + + if(end) + *end=0; // terminator + + return *begin ? &pa_charsets.get_direct(begin) : 0; + } + } + return 0; +} + + +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; +} +