--- parser3/src/classes/file.C 2004/07/26 10:44:21 1.126 +++ parser3/src/classes/file.C 2016/07/21 18:30:10 1.243 @@ -1,16 +1,12 @@ /** @file Parser: @b file parser class. - Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_FILE_C="$Date: 2004/07/26 10:44:21 $"; - #include "pa_config_includes.h" -#include "pcre.h" - #include "classes.h" #include "pa_vmethod_frame.h" @@ -25,34 +21,54 @@ static const char * const IDENT_FILE_C=" #include "pa_charset.h" #include "pa_charsets.h" #include "pa_sql_connection.h" -#include "pa_vresponse.h" -#include "pa_vcookie.h" +#include "pa_md5.h" +#include "pa_vregex.h" +#include "pa_version.h" + +volatile const char * IDENT_FILE_C="$Id: file.C,v 1.243 2016/07/21 18:30:10 moko Exp $"; // defines -#define TEXT_MODE_NAME "text" -#define BINARY_MODE_NAME "binary" #define STDIN_EXEC_PARAM_NAME "stdin" #define CHARSET_EXEC_PARAM_NAME "charset" +#define NAME_NAME "name" +#define KEEP_EMPTY_DIRS_NAME "keep-empty-dirs" +#define SUPPRESS_EXCEPTION_NAME "exception" + +// externs + +extern String sql_limit_name; +extern String sql_offset_name; + +// helpers + +class File_list_table_template_columns: public ArrayString { +public: + File_list_table_template_columns() { + *this+=new String("name"); + *this+=new String("dir"); + *this+=new String("size"); + *this+=new String("cdate"); + *this+=new String("mdate"); + *this+=new String("adate"); + } +}; + +Table file_list_table_template(new File_list_table_template_columns); + // class class MFile: public Methoded { public: // VStateless_class - Value* create_new_value(Pool&) { return new VFile(); } - -public: // Methoded - bool used_directly() { return true; } - public: MFile(); - }; // global variable -DECLARE_CLASS_VAR(file, new MFile, 0); +DECLARE_CLASS_VAR(file, new MFile); // consts @@ -108,83 +124,246 @@ static const String::Body cdate_name("cd // methods -static bool is_text_mode(const String& mode) { - if(mode==TEXT_MODE_NAME) - return true; - if(mode==BINARY_MODE_NAME) - return false; - throw Exception("parser.runtime", - &mode, - "is invalid mode, must be either '"TEXT_MODE_NAME"' or '"BINARY_MODE_NAME"'"); -} - static void _save(Request& r, MethodParams& params) { - Value& vmode_name=params. as_no_junction(0, "mode must not be code"); - Value& vfile_name=params.as_no_junction(1, "file name must not be code"); + bool is_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); + Value& vfile_name=params.as_no_junction(1, FILE_NAME_MUST_NOT_BE_CODE); + + Charset* asked_charset=0; + if(params.count()>2) + if(HashStringValue* options=params.as_hash(2)){ + int valid_options=0; + if(Value* vcharset_name=options->get(PA_CHARSET_NAME)){ + asked_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } // save - GET_SELF(r, VFile).save(r.absolute(vfile_name.as_string()), - is_text_mode(vmode_name.as_string())); + GET_SELF(r, VFile).save(r.charsets, r.absolute(vfile_name.as_string()), is_text, asked_charset); } static void _delete(Request& r, MethodParams& params) { - Value& vfile_name=params.as_no_junction(0, "file name must not be code"); + const String& file_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); + bool keep_empty_dirs=false; + bool fail_on_problem=true; + + if(params.count()>1) + if(HashStringValue* options=params.as_hash(1)){ + int valid_options=0; + if(Value* vkeep_empty_dirs=options->get(KEEP_EMPTY_DIRS_NAME)){ + keep_empty_dirs=r.process_to_value(*vkeep_empty_dirs).as_bool(); + valid_options++; + } + if(Value* vsuppress_exception=options->get(SUPPRESS_EXCEPTION_NAME)){ + fail_on_problem=r.process_to_value(*vsuppress_exception).as_bool(); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } // unlink - file_delete(r.absolute(vfile_name.as_string())); + file_delete(r.absolute(file_name), fail_on_problem, keep_empty_dirs); } static void _move(Request& r, MethodParams& params) { Value& vfrom_file_name=params.as_no_junction(0, "from file name must not be code"); Value& vto_file_name=params.as_no_junction(1, "to file name must not be code"); + bool keep_empty_dirs=false; + + if(params.count()>2) + if(HashStringValue* options=params.as_hash(2)){ + int valid_options=0; + if(Value* vkeep_empty_dirs=options->get(KEEP_EMPTY_DIRS_NAME)){ + keep_empty_dirs=r.process_to_value(*vkeep_empty_dirs).as_bool(); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } // move file_move( r.absolute(vfrom_file_name.as_string()), - r.absolute(vto_file_name.as_string())); + r.absolute(vto_file_name.as_string()), + keep_empty_dirs); +} + +static void copy_process_source(struct stat& , int from_file, const String&, void *context) { + int& to_file=*static_cast(context); + + int nCount=0; + do { + unsigned char buffer[FILE_BUFFER_SIZE]; + nCount = file_block_read(from_file, buffer, sizeof(buffer)); + int written=write(to_file, buffer, nCount); + if( written < 0 ) + throw Exception("file.access", + 0, + "write failed: %s (%d)", strerror(errno), errno); + + } while(nCount > 0); +} + +static void copy_open_target(int f, void *from_spec) { + String& file_spec=*static_cast(from_spec); + file_read_action_under_lock(file_spec, "copy", copy_process_source, &f); +} + +static void _copy(Request& r, MethodParams& params) { + Value& vfrom_file_name=params.as_no_junction(0, "from file name must not be code"); + Value& vto_file_name=params.as_no_junction(1, "to file name must not be code"); + + String from_spec = r.absolute(vfrom_file_name.as_string()); + const String& to_spec = r.absolute(vto_file_name.as_string()); + + file_write_action_under_lock( + to_spec, + "copy", + copy_open_target, + &from_spec); } static void _load_pass_param( - HashStringValue::key_type key, - HashStringValue::value_type value, - HashStringValue *dest) { + HashStringValue::key_type key, + HashStringValue::value_type value, + HashStringValue *dest) { dest->put(key, value); } + static void _load(Request& r, MethodParams& params) { - Value& vmode_name=params. as_no_junction(0, "mode must not be code"); - const String& lfile_name=r.absolute(params.as_no_junction(1, "file name must not be code").as_string()); - Value* third_param=params.count()>2?¶ms.as_no_junction(2, "filename or options must not be code") - :0; - HashStringValue* third_param_hash=third_param?third_param->get_hash():0; - size_t alt_filename_param_index=2; - if(third_param_hash) - alt_filename_param_index++; - - File_read_result file=file_read(r.charsets, lfile_name, - is_text_mode(vmode_name.as_string()), - third_param_hash - ); + bool as_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); + const String& lfile_name=r.absolute(params.as_string(1, FILE_NAME_MUST_NOT_BE_CODE)); + + size_t param_index=params.count()-1; + Value* param_value=param_index>1?¶ms.as_no_junction(param_index, "file name or options must not be code"):0; + + HashStringValue* options=0; + const String* user_file_name=0; - const char *user_file_name=params.count()>alt_filename_param_index? - params.as_string(alt_filename_param_index, "filename must be string").cstr() - :lfile_name.cstr(String::L_FILE_SPEC); + if(param_value){ + options=param_value->get_hash(); + if(options || param_index>2) + param_index--; + if(param_index>1){ + const String& luser_file_name=params.as_string(param_index, FILE_NAME_MUST_BE_STRING); + if(!luser_file_name.is_empty()) + user_file_name=&luser_file_name; + } + } + if(!user_file_name) + user_file_name=&lfile_name; + + size_t offset=0; + size_t limit=0; + + if(options){ + options=new HashStringValue(*options); + if(Value *voffset=(Value *)options->get(sql_offset_name)){ + offset=r.process_to_value(*voffset).as_int(); + } + if(Value *vlimit=(Value *)options->get(sql_limit_name)){ + limit=r.process_to_value(*vlimit).as_int(); + } + // no check on options count here, see file_read + } + File_read_result file=file_load(r, lfile_name, + as_text, options, true, 0, offset, limit + ); Value* vcontent_type=0; - if(file.headers) - vcontent_type=file.headers->get(content_type_name); - if(!vcontent_type) - vcontent_type=new VString(r.mime_type_of(user_file_name)); + if(file.headers){ + if(Value* remote_content_type=file.headers->get(HTTP_CONTENT_TYPE_UPPER)) + vcontent_type=new VString(*new String(remote_content_type->as_string().cstr())); + } + + VFile& self=GET_SELF(r, VFile); + self.set(true/*tainted*/, as_text, file.str, file.length, user_file_name, vcontent_type, &r); + + if(file.headers){ + file.headers->for_each(_load_pass_param, &self.fields()); + } else { + size_t size; + time_t atime, mtime, ctime; + + file_stat(lfile_name, size, atime, mtime, ctime); + HashStringValue& ff=self.fields(); + ff.put(adate_name, new VDate((pa_time_t)atime)); + ff.put(mdate_name, new VDate((pa_time_t)mtime)); + ff.put(cdate_name, new VDate((pa_time_t)ctime)); + } +} + +static void _create(Request& r, MethodParams& params) { + const String* mode=0; + const String* file_name=0; + bool is_text=true; + + // new format: ^file::create[string-or-file-content[;$.mode[text|binary] $.name[...] $.content-type[...] $.charset[...] ]] + size_t content_index=0; + size_t options_index=1; + bool extended_options=true; + + if(params.count()>=3){ + // old format: ^file::create[text|binary;file-name;string-or-file-content[;options]] + mode=¶ms.as_string(0, MODE_MUST_NOT_BE_CODE); + is_text=VFile::is_text_mode(*mode); + file_name=¶ms.as_string(1, FILE_NAME_MUST_NOT_BE_CODE); + content_index=2; + options_index=3; + extended_options=false; + } + + VString* vcontent_type=0; + Charset* asked_charset=0; + if(params.count()>options_index) + if(HashStringValue* options=params.as_hash(options_index)) { + int valid_options=0; + if(extended_options) { + if(Value* vmode=options->get(MODE_NAME)) { + mode=&vmode->as_string(); + is_text=VFile::is_text_mode(*mode); + valid_options++; + } + if(Value* vfile_name=options->get(NAME_NAME)) { + file_name=&vfile_name->as_string(); + valid_options++; + } + } + if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) { + asked_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + valid_options++; + } + if(Value* value=options->get(CONTENT_TYPE_NAME)) { + vcontent_type=new VString(value->as_string()); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + + Value& vcontent=params.as_no_junction(content_index, "content must be string or file"); + VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, file.str, file.length, user_file_name, vcontent_type); - if(file.headers) - file.headers->for_each(_load_pass_param, &self.fields()); + + if(const String* content_str=vcontent.get_string()){ + String::Body body=content_str->cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false), &r.charsets); // explode content, honor tainting changes + if(asked_charset && is_text) + body=Charset::transcode(body, r.charsets.source(), *asked_charset); + self.set(true/*tainted*/, is_text, body.cstrm(), body.length(), file_name, vcontent_type, &r); + } else { + if(asked_charset) + throw Exception(PARSER_RUNTIME, 0, "charset option can not be used with file-content"); + self.set(*vcontent.as_vfile(String::L_AS_IS), mode != 0, is_text, file_name, vcontent_type, &r); + } + } static void _stat(Request& r, MethodParams& params) { - Value& vfile_name=params.as_no_junction(0, "file name must not be code"); - - const String& lfile_name=vfile_name.as_string(); + const String& lfile_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); size_t size; time_t atime, mtime, ctime; @@ -193,20 +372,21 @@ static void _stat(Request& r, MethodPara atime, mtime, ctime); VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, 0/*no bytes*/, size); + + self.set_binary(true/*tainted*/, 0/*no bytes*/, size, &lfile_name, 0, &r); HashStringValue& ff=self.fields(); - ff.put(adate_name, new VDate(atime)); - ff.put(mdate_name, new VDate(mtime)); - ff.put(cdate_name, new VDate(ctime)); - ff.put(content_type_name, new VString(r.mime_type_of(lfile_name.cstr(String::L_FILE_SPEC)))); + ff.put(adate_name, new VDate((pa_time_t)atime)); + ff.put(mdate_name, new VDate((pa_time_t)mtime)); + ff.put(cdate_name, new VDate((pa_time_t)ctime)); } static bool is_safe_env_key(const char* key) { for(const char* validator=key; *validator; validator++) { char c=*validator; - if(!(c>='A' && c<='Z' || c>='0' && c<='9' || c=='_' || c=='-')) + if(!( (c>='A' && c<='Z') || (c>='0' && c<='9') || (c=='_' || c=='-') )) return false; } +#ifdef PA_SAFE_MODE if(strncasecmp(key, "HTTP_", 5)==0) return true; if(strncasecmp(key, "CGI_", 4)==0) @@ -216,28 +396,31 @@ static bool is_safe_env_key(const char* return true; } return false; +#else + return true; +#endif } #ifndef DOXYGEN struct Append_env_pair_info { + Request_charsets* charsets; HashStringString* env; Value* vstdin; - Value* vcharset; }; #endif static void append_env_pair( - HashStringValue::key_type akey, - HashStringValue::value_type avalue, - Append_env_pair_info *info) { + HashStringValue::key_type akey, + HashStringValue::value_type avalue, + Append_env_pair_info *info) { if(akey==STDIN_EXEC_PARAM_NAME) { info->vstdin=avalue; } else if(akey==CHARSET_EXEC_PARAM_NAME) { - info->vcharset=avalue; + // ignore, already processed } else { if(!is_safe_env_key(akey.cstr())) - throw Exception("parser.runtime", + throw Exception(PARSER_RUNTIME, new String(akey, String::L_TAINTED), "not safe environment variable"); - info->env->put(akey, avalue->as_string().cstr(String::L_UNSPECIFIED)); + info->env->put(akey, avalue->as_string().cstr_to_string_body_untaint(String::L_AS_IS, 0, info->charsets)); } } #ifndef DOXYGEN @@ -248,44 +431,48 @@ struct Pass_cgi_header_attribute_info { }; #endif static void pass_cgi_header_attribute( - ArrayString::element_type astring, - Pass_cgi_header_attribute_info* info) { + ArrayString::element_type astring, + Pass_cgi_header_attribute_info* info) { size_t colon_pos=astring->pos(':'); - if(colon_pos==STRING_NOT_FOUND) { + if(colon_pos!=STRING_NOT_FOUND) { const String& key=astring->mid(0, colon_pos).change_case( *info->charset, String::CC_UPPER); - Value* value=new VString(astring->mid(colon_pos+1, astring->length())); + Value* value=new VString(astring->mid(colon_pos+1, astring->length()).trim()); info->fields->put(key, value); - if(key=="CONTENT-TYPE") + if(key==HTTP_CONTENT_TYPE_UPPER) info->content_type=value; } } + +static void append_to_argv(Request& r, ArrayString& argv, const String* str){ + if(!str->is_empty()) + argv+=new String(str->cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false), &r.charsets), String::L_AS_IS); +} + /// @todo fix `` in perl - they produced flipping consoles and no output to perl -static void _exec_cgi(Request& r, MethodParams& params, - bool cgi) { +static void _exec_cgi(Request& r, MethodParams& params, bool cgi) { + bool is_text=true; + size_t param_index=0; + const String& mode=params.as_string(0, FIRST_ARG_MUST_NOT_BE_CODE); + if(VFile::is_valid_mode(mode)) { + is_text=VFile::is_text_mode(mode); + param_index++; + } - Value& vfile_name=params.as_no_junction(0, "file name must not be code"); + if(param_index>=params.count()) + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); - const String& script_name=r.absolute(vfile_name.as_string()); + const String& script_name=r.absolute(params.as_string(param_index++, FILE_NAME_MUST_NOT_BE_CODE)); HashStringString env; - #define ECSTR(name, value_cstr) \ - if(value_cstr) \ - env.put( \ - String::Body(#name), \ - String::Body(value_cstr, 0)); \ - // passing SAPI::environment - if(const char *const *pairs=SAPI::environment(r.sapi_info)) { - while(const char* pair=*pairs++) - if(const char* eq_at=strchr(pair, '=')) - if(eq_at[1]) // has value - env.put( - pa_strdup(pair, eq_at-pair), - pa_strdup(eq_at+1, 0)); - } + #define ECSTR(name, value_cstr) if(value_cstr) env.put(#name, value_cstr); + // passing environment + for(SAPI::Env::Iterator i(r.sapi_info); i; i.next() ) + env.put(i.key(), i.value() ); // const ECSTR(GATEWAY_INTERFACE, "CGI/1.1"); + ECSTR(PARSER_VERSION, PARSER_VERSION); // from Request.info ECSTR(DOCUMENT_ROOT, r.request_info.document_root); ECSTR(PATH_TRANSLATED, r.request_info.path_translated); @@ -293,48 +480,69 @@ static void _exec_cgi(Request& r, Method ECSTR(QUERY_STRING, r.request_info.query_string); ECSTR(REQUEST_URI, r.request_info.uri); ECSTR(CONTENT_TYPE, r.request_info.content_type); - char content_length_cstr[MAX_NUMBER]; - snprintf(content_length_cstr, MAX_NUMBER, "%u", r.request_info.content_length); - //String content_length(content_length_cstr); - ECSTR(CONTENT_LENGTH, content_length_cstr); + ECSTR(CONTENT_LENGTH, format(r.request_info.content_length, "%u")); // SCRIPT_* - env.put(String::Body("SCRIPT_NAME"), script_name); - //env.put(String::Body("SCRIPT_FILENAME"), ??&script_name); + env.put("SCRIPT_NAME", script_name); - bool stdin_specified=false; // environment & stdin from param String *in=new String(); Charset *charset=0; // default script works raw_in 'source' charset = no transcoding needed - if(params.count()>1) { - Value& venv=params.as_no_junction(1, "env must not be code"); - if(HashStringValue* user_env=venv.get_hash()) { - Append_env_pair_info info={&env, 0, 0}; - user_env->for_each(append_env_pair, &info); + if(param_index < params.count()) { + if(HashStringValue* user_env=params.as_hash(param_index++, "env")) { + // $.charset [previewing to handle URI pieces] + if(Value* vcharset=user_env->get(CHARSET_EXEC_PARAM_NAME)) + charset=&charsets.get(vcharset->as_string() + .change_case(r.charsets.source(), String::CC_UPPER)); + + // $.others + Append_env_pair_info info={&r.charsets, &env, 0}; + { + // influence tainting + // main target -- $.QUERY_STRING -- URLencoding of tainted pieces to String::L_URI lang + Temp_client_charset temp(r.charsets, charset? *charset: r.charsets.source()); + user_env->for_each(append_env_pair, &info); + } // $.stdin if(info.vstdin) { - stdin_specified=true; if(const String* sstdin=info.vstdin->get_string()) { - in->append(*sstdin, String::L_CLEAN, true); + // untaint stdin + in = new String(sstdin->cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false), &r.charsets), String::L_AS_IS); } else - if(VFile* vfile=static_cast(info.vstdin->as("file", false))) + if(VFile* vfile=static_cast(info.vstdin->as("file"))) in->append_know_length((const char* )vfile->value_ptr(), vfile->value_size(), String::L_TAINTED); else - throw Exception("parser.runtime", + throw Exception(PARSER_RUNTIME, 0, STDIN_EXEC_PARAM_NAME " parameter must be string or file"); } - // $.charset - if(info.vcharset) - charset=&charsets.get(info.vcharset->as_string() - .change_case(r.charsets.source(), String::CC_UPPER)); } } // argv from params ArrayString argv; - if(params.count()>2) { - for(size_t i=2; icount(); i++) { + append_to_argv(r, argv, table->get(i)->get(0)); + } + } else { + throw Exception(PARSER_RUNTIME, + 0, + "param must be string or table"); + } + } + } + } } // transcode if necessary @@ -348,86 +556,105 @@ static void _exec_cgi(Request& r, Method // match silent conversion in OS // exec! - PA_exec_result execution= - pa_exec(false/*forced_allow*/, script_name, &env, argv, *in); + PA_exec_result execution=pa_exec(false/*forced_allow*/, script_name, &env, argv, *in); - String *real_out=&execution.out; + File_read_result *file_out=&execution.out; String *real_err=&execution.err; - // transcode if necessary - if(charset) { - real_out=&Charset::transcode(*real_out, *charset, r.charsets.source()); + + // transcode err if necessary (@todo: need fix line breaks in err as well ) + if(charset) real_err=&Charset::transcode(*real_err, *charset, r.charsets.source()); + + if(file_out->length && is_text){ + fix_line_breaks(file_out->str, file_out->length); + // treat output as string + String *real_out = new String(file_out->str); + + // transcode out if necessary + if(charset) + real_out=&Charset::transcode(*real_out, *charset, r.charsets.source()); + + // FIXME: unsafe cast + file_out->str=const_cast(real_out->cstr()); // hacking a little + file_out->length = real_out->length(); } VFile& self=GET_SELF(r, VFile); - const String* body=real_out; // ^file:exec - const char* eol_marker=0; size_t eol_marker_size; - const String* header=0; - if(cgi) { // ^file:cgi + if(cgi) { // ^file::cgi + const char* eol_marker=0; + size_t eol_marker_size; + // construct with 'out' body and header - size_t dos_pos=real_out->pos("\r\n\r\n", 4); - size_t unix_pos=real_out->pos("\n\n", 2); + size_t dos_pos=(file_out->length)?strpos(file_out->str, "\r\n\r\n"):STRING_NOT_FOUND; + size_t unix_pos=(file_out->length)?strpos(file_out->str, "\n\n"):STRING_NOT_FOUND; bool unix_header_break; switch((dos_pos!=STRING_NOT_FOUND?10:00) + (unix_pos!=STRING_NOT_FOUND?01:00)) { - case 10: // dos - unix_header_break=false; - break; - case 01: // unix - unix_header_break=true; - break; - case 11: // dos & unix - unix_header_break=unix_poslength(), real_out->cstr(), - (uint)real_err->length(), real_err->cstr()); - break; //never reached + case 10: // dos + unix_header_break=false; + break; + case 01: // unix + unix_header_break=true; + break; + case 11: // dos & unix + unix_header_break=unix_poslength, (file_out->length) ? (file_out->str) : "", + real_err->length(), real_err->cstr()); + break; //never reached } - int header_break_pos; + size_t header_break_pos; if(unix_header_break) { header_break_pos=unix_pos; - eol_marker="\n"; eol_marker_size=1; + eol_marker="\n"; + eol_marker_size=1; } else { header_break_pos=dos_pos; - eol_marker="\r\n"; eol_marker_size=2; + eol_marker="\r\n"; + eol_marker_size=2; } - header=&real_out->mid(0, header_break_pos); - body=&real_out->mid(header_break_pos+eol_marker_size*2, real_out->length()); - } - // body - self.set(false/*not tainted*/, body->cstr(), body->length()); - - // $fields << header - if(header && eol_marker) { - ArrayString rows; - size_t pos_after=0; - header->split(rows, pos_after, eol_marker); - Pass_cgi_header_attribute_info info={0, 0, 0}; - info.charset=&r.charsets.source(); - info.fields=&self.fields(); - rows.for_each(pass_cgi_header_attribute, &info); - if(info.content_type) - self.fields().put(content_type_name, info.content_type); + file_out->str[header_break_pos] = 0; + String *header=new String(file_out->str); + unsigned long headersize = header_break_pos+eol_marker_size*2; + file_out->str += headersize; + file_out->length -= headersize; + + // $body + self.set(false/*not tainted*/, is_text, file_out->str, file_out->length); + + // $fields << header + if(header) { + ArrayString rows; + size_t pos_after=0; + header->split(rows, pos_after, eol_marker); + Pass_cgi_header_attribute_info info={0, 0, 0}; + info.charset=&r.charsets.source(); + info.fields=&self.fields(); + rows.for_each(pass_cgi_header_attribute, &info); + if(info.content_type) + self.fields().put(content_type_name, info.content_type); + } + } else { // ^file::exec + // $body + self.set(false/*not tainted*/, is_text, file_out->str, file_out->length); } // $status self.fields().put(file_status_name, new VInt(execution.status)); // $stderr - if(real_err->length()) - self.fields().put( - String::Body("stderr"), - new VString(*real_err)); + if(!real_err->is_empty()) + self.fields().put("stderr", new VString(*real_err)); } static void _exec(Request& r, MethodParams& params) { _exec_cgi(r, params, false); @@ -439,66 +666,71 @@ static void _cgi(Request& r, MethodParam static void _list(Request& r, MethodParams& params) { Value& relative_path=params.as_no_junction(0, "path must not be code"); - const String* regexp; - pcre *regexp_code; - const int ovecsize=(1/*match*/)*3; - int ovector[ovecsize]; - if(params.count()>1) { - regexp=¶ms.as_no_junction(1, "regexp must not be code").as_string(); - - const char* pattern=regexp->cstr(); - const char* errptr; - int erroffset; - regexp_code=pcre_compile(pattern, PCRE_EXTRA | PCRE_DOTALL, - &errptr, &erroffset, - r.charsets.source().pcre_tables); - - if(!regexp_code) - throw Exception(0, - ®exp->mid(erroffset, regexp->length()), - "regular expression syntax error - %s", errptr); - } else { - regexp=0; // not used, just to calm down compiler - regexp_code=0; + bool stat=false; + VRegex* vregex=0; + VRegexCleaner vrcleaner; + + if(params.count()>1){ + Value& voption=params.as_no_junction(1, "option must not be code"); + if(voption.is_defined()) { + Value* vfilter=0; + if(HashStringValue* options=voption.get_hash()) { + int valid_options=0; + if(Value* vstat=options->get("stat")) { + stat=r.process_to_value(*vstat).as_bool(); + valid_options++; + } + if(Value* value=options->get("filter")) { + vfilter=value; + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } else { + vfilter=&voption; + } + if(vfilter) { + if(Value* value=vfilter->as(VREGEX_TYPE)) { + vregex=static_cast(value); + } else if(vfilter->is_string()) { + if(!vfilter->get_string()->trim().is_empty()) { + vregex=new VRegex(r.charsets.source(), &vfilter->as_string(), 0/*options*/); + vregex->study(); + vrcleaner.vregex=vregex; + } + } else { + throw Exception(PARSER_RUNTIME, 0, "filter must be regex or string"); + } + } + } } + const char* absolute_path_cstr=r.absolute(relative_path.as_string()).taint_cstr(String::L_FILE_SPEC); - const char* absolute_path_cstr=r.absolute(relative_path.as_string()).cstr(String::L_FILE_SPEC); + Table::Action_options table_options; + Table& table=*new Table(file_list_table_template, table_options); - Table::columns_type columns(new ArrayString); - *columns+=new String("name"); - Table& table=*new Table(columns); + const int ovector_size=(1/*match*/)*3; + int ovector[ovector_size]; LOAD_DIR(absolute_path_cstr, const char* file_name_cstr=ffblk.ff_name; size_t file_name_size=strlen(file_name_cstr); - bool suits=true; - if(regexp_code) { - int exec_result=pcre_exec(regexp_code, 0, - ffblk.ff_name, file_name_size, 0, - 0, ovector, ovecsize); - - if(exec_result==PCRE_ERROR_NOMATCH) - suits=false; - else if(exec_result<0) { - (*pcre_free)(regexp_code); - throw Exception(0, - regexp, - "regular expression execute (%d)", - exec_result); - } - } - if(suits) { + if(!vregex || vregex->exec(ffblk.ff_name, file_name_size, ovector, ovector_size)>=0) { Table::element_type row(new ArrayString); - *row+=new String(pa_strdup(file_name_cstr, file_name_size), file_name_size, true); + *row+=new String(pa_strdup(file_name_cstr, file_name_size), String::L_TAINTED); + *row+=new String(String::Body::Format(ffblk.is_dir(stat) ? 1 : 0), String::L_CLEAN); + if(stat) { + *row+=VDouble(ffblk.size()).get_string(); + *row+=new String(String::Body::Format((int)ffblk.c_timestamp()), String::L_CLEAN); + *row+=new String(String::Body::Format((int)ffblk.m_timestamp()), String::L_CLEAN); + *row+=new String(String::Body::Format((int)ffblk.a_timestamp()), String::L_CLEAN); + } table+=row; } ); - if(regexp_code) - pcre_free(regexp_code); - // write out result r.write_no_lang(*new VTable(&table)); } @@ -509,38 +741,42 @@ struct Lock_execute_body_info { Value* body_code; }; #endif + static void lock_execute_body(int , void *ainfo) { Lock_execute_body_info& info=*static_cast(ainfo); // execute body info.r->write_assign_lang(info.r->process(*info.body_code)); -}; +} + static void _lock(Request& r, MethodParams& params) { - const String& file_spec=r.absolute(params.as_string(0, "file name must be string")); + const String& file_spec=r.absolute(params.as_string(0, FILE_NAME_MUST_BE_STRING)); Lock_execute_body_info info={ &r, ¶ms.as_junction(1, "body must be code") }; - file_write_action_under_lock(file_spec, "lock", lock_execute_body, &info); + file_write_action_under_lock( + file_spec, + "lock", + lock_execute_body, + &info); } -static int lastposafter(const String& s, size_t after, const char* substr, size_t substr_size, bool beforelast=false) { - size_t size=0; // just to calm down compiler - if(beforelast) - size=s.length(); - size_t at; - while((at=s.pos(String::Body(substr, substr_size), after))!=STRING_NOT_FOUND) { - size_t newafter=at+substr_size/*skip substr*/; - if(beforelast && newafter==size) - break; - after=newafter; - } +static size_t afterlastslash(const String& str) { + size_t pos=str.strrpbrk("/\\"); + return pos!=STRING_NOT_FOUND?pos+1:0; +} - return after; +static size_t afterlastslash(const String& str, size_t right) { + size_t pos=str.strrpbrk("/\\", 0, right); + return pos!=STRING_NOT_FOUND?pos+1:0; } static void _find(Request& r, MethodParams& params) { - const String& file_name=params.as_no_junction(0, "file name must not be code").as_string(); + const String& file_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); + + Value* not_found_code=(params.count()==2)?¶ms.as_junction(1, "not-found param must be code"):0; + const String* file_spec; if(file_name.first_char()=='/') file_spec=&file_name; @@ -548,69 +784,118 @@ static void _find(Request& r, MethodPara file_spec=&r.relative(r.request_info.uri, file_name); // easy way - if(file_readable(r.absolute(*file_spec))) { + if(file_exist(r.absolute(*file_spec))) { r.write_assign_lang(*file_spec); return; } // monkey way - int after_base_slash=lastposafter(*file_spec, 0, "/", 1); - const String* dirname=&file_spec->mid(0, after_base_slash); - const String& basename=file_spec->mid(after_base_slash, file_spec->length()); - - int after_monkey_slash; - while((after_monkey_slash=lastposafter(*dirname, 0, "/", 1, true))>0) { + size_t last_slash=file_spec->strrpbrk("/\\"); + const String& dirname=file_spec->mid(0, last_slash!=STRING_NOT_FOUND?last_slash:0); + const String& basename=file_spec->mid(last_slash!=STRING_NOT_FOUND?last_slash+1:0, file_spec->length()); + + size_t rpos=dirname.is_empty()?0:dirname.length()-1; + while((rpos=dirname.rskipchars("/\\", 0, rpos))!=STRING_NOT_FOUND){ + size_t slash=dirname.strrpbrk("/\\", 0, rpos); + if(slash==STRING_NOT_FOUND) + break; String test_name; - test_name<<*(dirname=&dirname->mid(0, after_monkey_slash)); - test_name< /a - // /a/b/ > /a - int afterslash=lastposafter(file_spec, 0, "/", 1, true); - if(afterslash>0) - r.write_assign_lang(file_spec.mid(0, afterslash==1?1:afterslash-1)); - else - r.write_assign_lang(String(".", 1)); + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + // works as *nix dirname + + // empty > . + // / > / + // /a > / + // /a/ > / + // /a/some.tar.gz > /a + // /a/b/ > /a + // /a///b/ > /a + // /a/b/// > /a + // file > . + + if(file_spec.is_empty()) { + r.write_assign_lang(String(".")); + return; + } + + size_t p; + size_t slash; + if((p=file_spec.rskipchars("/\\"))==STRING_NOT_FOUND) + r.write_assign_lang(String("/")); + else { + if((slash=file_spec.strrpbrk("/\\", 0, p))!=STRING_NOT_FOUND) { + if((p=file_spec.rskipchars("/\\", 0, slash))==STRING_NOT_FOUND) + p=slash; + r.write_assign_lang(file_spec.mid(0, p+1)); + return; + } + r.write_assign_lang(String(".")); + } } static void _basename(Request& r, MethodParams& params) { - const String& file_spec=params.as_string(0, "file name must be string"); - // /a/some.tar.gz > some.tar.gz - int afterslash=lastposafter(file_spec, 0, "/", 1); - r.write_assign_lang(file_spec.mid(afterslash, file_spec.length())); + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + // works as *nix basename + + // empty > . + // / > / + // /a > a + // /a/ > a + // /a/some.tar.gz > some.tar.gz + // /a/b/ > b + // /a///b/ > b + // /a/b/// > b + // file > file + + if(file_spec.is_empty()) { + r.write_assign_lang(String(".")); + return; + } + + size_t p=file_spec.rskipchars("/\\"); + if(p==STRING_NOT_FOUND) + r.write_assign_lang(String("/")); + else + r.write_assign_lang(file_spec.mid(afterlastslash(file_spec, p), p+1)); } static void _justname(Request& r, MethodParams& params) { - const String& file_spec=params.as_string(0, "file name must be string"); - // /a/some.tar.gz > some.tar - int afterslash=lastposafter(file_spec, 0, "/", 1); - int afterdot=lastposafter(file_spec, afterslash, ".", 1); - r.write_assign_lang(file_spec.mid(afterslash, afterdot!=afterslash?afterdot-1:file_spec.length())); + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + // /a/some.tar.gz > some.tar + // /a/b.c/ > empty + // /a/b.c > b + size_t pos=afterlastslash(file_spec); + size_t dotpos=file_spec.strrpbrk(".", pos); + r.write_assign_lang(file_spec.mid(pos, dotpos!=STRING_NOT_FOUND?dotpos:file_spec.length())); } + static void _justext(Request& r, MethodParams& params) { - const String& file_spec=params.as_string(0, "file name must be string"); - // /a/some.tar.gz > gz - int afterdot=lastposafter(file_spec, 0, ".", 1); - if(afterdot>0) - r.write_assign_lang(file_spec.mid(afterdot, file_spec.length())); + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + // /a/some.tar.gz > gz + // /a/b.c/ > empty + size_t pos=afterlastslash(file_spec); + size_t dotpos=file_spec.strrpbrk(".", pos); + if(dotpos!=STRING_NOT_FOUND) + r.write_assign_lang(file_spec.mid(dotpos+1, file_spec.length())); } static void _fullpath(Request& r, MethodParams& params) { - const String& file_spec=params.as_string(0, "file name must be string"); + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); const String* result; if(file_spec.first_char()=='/') result=&file_spec; @@ -643,8 +928,8 @@ class File_sql_event_handlers: public SQ int got_cells; public: String::C value; - String* user_file_name; - String* user_content_type; + const String* user_file_name; + const String* user_content_type; public: File_sql_event_handlers( const String& astatement_string, const char* astatement_cstr): @@ -656,7 +941,7 @@ public: bool add_column(SQL_Error& error, const char* /*str*/, size_t /*length*/) { if(got_columns++==3) { - error=SQL_Error("parser.runtime", "result must contain not more then 3 columns"); + error=SQL_Error(PARSER_RUNTIME, "result must contain not more then 3 columns"); return true; } return false; @@ -670,13 +955,15 @@ public: value=String::C(str, length); break; case 1: - user_file_name=new String(str, length, true); + if(!user_file_name) // user not specified? + user_file_name=new String(str, String::L_TAINTED); break; case 2: - user_content_type=new String(str, length, true); + if(!user_content_type) // user not specified? + user_content_type=new String(str, String::L_TAINTED); break; default: - error=SQL_Error("parser.runtime", "result must not contain more then one row, three rows"); + error=SQL_Error(PARSER_RUNTIME, "result must not contain more then one row, three columns"); return true; } return false; @@ -688,360 +975,281 @@ public: }; #endif static void _sql(Request& r, MethodParams& params) { - const String* user_file_name=0; - if(params.get(0)->is_string()) - user_file_name=¶ms.get(0)->as_string(); - - Value& statement=params.as_junction(params.count()-1, "statement must be code"); + Value& statement=params.as_junction(0, "statement must be code"); Temp_lang temp_lang(r, String::L_SQL); const String& statement_string=r.process_to_string(statement); - const char* statement_cstr= - statement_string.cstr(String::L_UNSPECIFIED, r.connection()); + const char* statement_cstr=statement_string.untaint_cstr(r.flang, r.connection()); + File_sql_event_handlers handlers(statement_string, statement_cstr); + + ulong limit=SQL_NO_LIMIT; + ulong offset=0; + + if(params.count()>1) + if(HashStringValue* options=params.as_hash(1, "sql options")) { + int valid_options=0; + if(Value* vfilename=options->get(NAME_NAME)) { + valid_options++; + handlers.user_file_name=&vfilename->as_string(); + } + if(Value* vcontent_type=options->get(CONTENT_TYPE_NAME)) { + valid_options++; + handlers.user_content_type=&vcontent_type->as_string(); + } + if(Value* vlimit=options->get(sql_limit_name)) { + valid_options++; + limit=(ulong)r.process_to_value(*vlimit).as_double(); + } + if(Value* voffset=options->get(sql_offset_name)) { + valid_options++; + offset=(ulong)r.process_to_value(*voffset).as_double(); + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + + r.connection()->query( statement_cstr, 0, 0, - 0, 0, + offset, limit, handlers, statement_string); - if(!handlers.value) - throw Exception("parser.runtime", - 0, - "produced no result"); + if(!handlers.value.str) + throw Exception(PARSER_RUNTIME, 0, "produced no result"); - if(!user_file_name) -class send_attr_info -{ -public: - send_attr_info(Request *t) : r(t), add_content_type(true), add_last_modified(true), add_content_disposition(true) {} - Request *r; - bool add_content_type; - bool add_last_modified; - bool add_content_disposition; -}; + VFile& self=GET_SELF(r, VFile); -static void send_add_header_attribute( - HashStringValue::key_type aattribute, - HashStringValue::value_type ameaning, - send_attr_info *r) -{ - const char *a = aattribute.cstr(); - SAPI::add_header_attribute(r->r->sapi_info, - a, - attributed_meaning_to_string(*ameaning, String::L_HTTP_HEADER, false). - cstr(String::L_UNSPECIFIED)); - if(strcasecmp(a, "content-type")==0) - r->add_content_type = false; - else if(strcasecmp(a, "last-modified")==0) - r->add_last_modified = false; - else if(strcasecmp(a, "content-disposition")==0) - r->add_content_disposition = false; -} + self.set_binary(true/*tainted*/, handlers.value.str, handlers.value.length, handlers.user_file_name + , handlers.user_content_type ? new VString(*handlers.user_content_type) : 0 + , &r); +} + +static void _base64(Request& r, MethodParams& params) { + bool dynamic=!(&r.get_self() == file_class); + if(dynamic) { + VFile& self=GET_SELF(r, VFile); + if(params.count()) { + // decode: + // ^file::base64[encoded] // backward + // ^file::base64[mode;user-file-name;encoded[;$.content-type[...] $.strict(true|false)]] + bool is_text=false; + bool strict=false; + VString* vcontent_type=0; + const String* user_file_name=0; + size_t param_index=0; + + if(params.count() > 1) { + if(params.count() < 3) + throw Exception(PARSER_RUNTIME, + 0, + "constructor can not have less then 3 parameters (has %d parameters)", + params.count()); // actually it accepts 1 parameter (backward) + + is_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); + user_file_name=¶ms.as_string(1, FILE_NAME_MUST_BE_STRING); + + if(params.count() == 4) + if(HashStringValue* options=params.as_hash(3)) { + int valid_options=0; + if(Value* value=options->get(CONTENT_TYPE_NAME)) { + vcontent_type=new VString(value->as_string()); + valid_options++; + } + if(Value* vstrict=options->get(BASE64_STRICT_OPTION_NAME)) { + strict=r.process_to_value(*vstrict).as_bool(); + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } -struct RANGE -{ - size_t start; - size_t end; -}; + param_index=2; + } -static void parse_range(const String* s, Array &ar) -{ - const char *p = s->cstr(); - if(s->starts_with("bytes=")) - p += 6; - RANGE r; - while(*p){ - r.start = (size_t)-1; - r.end = (size_t)-1; - if(*p >= '0' && *p <= '9'){ - r.start = atol(p); - while(*p>='0' && *p<='9') ++p; - } - if(*p++ != '-') break; - if(*p >= '0' && *p <= '9'){ - r.end = atol(p); - while(*p>='0' && *p<='9') ++p; - } - if(*p == ',') ++p; - ar += r; - } -} + const char* encoded=params.as_string(param_index, PARAMETER_MUST_BE_STRING).cstr(); -class auto_file -{ -protected: - FILE *f; -public: - auto_file(FILE *t){ - f = t; - } - ~auto_file(){ - if(f != 0){ - fclose(f); - f = 0; + char* decoded=0; + size_t length=0; + pa_base64_decode(encoded, strlen(encoded), decoded, length, strict); + + self.set(true/*tainted*/, is_text, decoded, length, user_file_name, vcontent_type, &r); + } else { + // encode: ^f.base64[] + const char* encoded=pa_base64_encode(self.value_ptr(), self.value_size()); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed**/)); } + } else { + // encode: ^file:base64[filespec] + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + const char* encoded=pa_base64_encode(r.absolute(file_spec)); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed*/)); } - operator FILE*(){ - return f; - } -}; +} -// ^file:send[filename] -// ^file:send[filename;options hash] -// ^file:send[local_filename;remote_filename] -// ^file:send[local_filename;remote_filename;options hash] -static void _send(Request& r, MethodParams& params) { - SAPI::add_header_attribute(r.sapi_info, "Accept-Ranges", "bytes"); - if(r.response.fields().get("ignore")!=0) throw Exception("parser.runtime", 0, "^file:send not allowed here"); - Value *to_file_name = 0; - Value *options = 0; - Value *from_file_name = params.get(0); - const char *c_from_file_name=0, *disposition=0; - if(!from_file_name->is("string")) throw Exception("parser.runtime", 0, "filename must be string"); - - size_t count = params.count(); - if(count > 1){ - to_file_name = params.get(1); - if(to_file_name->is("hash")){ - options = to_file_name; - to_file_name = 0; - }else if(count > 2){ - options = params.get(2); - if(!options->is("hash")) throw Exception("parser.runtime", 0, "options parameter must be hash"); +static void _crc32(Request& r, MethodParams& params) { + unsigned long crc32 = 0; + if(&r.get_self() == file_class) { + // ^file:crc32[file-name] + if(params.count()) { + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + crc32=pa_crc32(r.absolute(file_spec)); + } else { + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); } + } else { + // ^file.crc32[] + VFile& self=GET_SELF(r, VFile); + crc32=pa_crc32(self.value_ptr(), self.value_size()); } + r.write_no_lang(*new VInt(crc32)); +} - c_from_file_name=r.absolute(from_file_name->as_string()).cstr(); - - size_t offset = 0; - size_t limit = (size_t)-1; - send_attr_info info(&r); - VDate *date = 0; - if(options){ - HashStringValue *opts = options->get_hash(); - if(opts == 0) - throw Exception("parser.runtime", 0, "options must be hash"); - Value *v; - int valid_options = 0; - if(v = opts->get("offset")){ - ++valid_options; - offset = v->as_int(); - } - if(v = opts->get("limit")){ - ++valid_options; - limit = v->as_int(); - } - if(v = opts->get("headers")){ - ++valid_options; - HashStringValue *headers = v->get_hash(); - if(headers == 0) - throw Exception("parser.runtime", 0, "headers must be hash"); - headers->for_each(send_add_header_attribute, &info); - } - if(v = opts->get("mdate")){ - ++valid_options; - if(Value* vdate=v->as(VDATE_TYPE, false)) - date=static_cast(vdate); - else throw Exception("parser.runtime", 0, "mdate must be a date"); - } - if(v = opts->get("disposition")){ - ++valid_options; - if(!v->is("string")) throw Exception("parser.runtime", 0, "disposition must be a string"); - disposition = v->get_string()->cstr(); - if(strcmp(disposition, "inline") && strcmp(disposition, "attachment")) throw Exception("parser.runtime", 0, "disposition can be only 'inline' or 'attachment'"); - } - if(valid_options != opts->count()) - throw Exception("parser.runtime", 0, "invalid option passed"); - } - - auto_file f = fopen(c_from_file_name, "rb"); - if(f == 0) - throw Exception("parser.runtime", 0, "Can't open file"); - - if(fseek(f, 0, SEEK_END)!=0) - throw Exception("parser.runtime", 0, "Can't seek file"); - - size_t file_length = (size_t)ftell(f); - if(file_length == (size_t)-1) - throw Exception("parser.runtime", 0, "can't get file size"); - if(file_length <= offset) - throw Exception("parser.runtime", 0, "offset too big"); - - size_t content_length = file_length-offset; - if(limit != (size_t)-1) - content_length = limit ar; - parse_range(new String(range), ar); - size_t count = ar.count(); - if(count == 1){ - RANGE &rg = ar.get_ref(0); - if(rg.start == (size_t)-1 && rg.end == (size_t)-1){ - SAPI::add_header_attribute(r.sapi_info, "status", "416 Requested Range Not Satisfiable"); - return; - } - if(rg.start == (size_t)-1 && rg.end != (size_t)-1){ - rg.start = content_length - rg.end; - rg.end = content_length; - offset += rg.start; - part_length = rg.end-rg.start; - }else if(rg.start != (size_t)-1 && rg.end == (size_t)-1){ - rg.end = content_length-1; - offset += rg.start; - part_length -= rg.start; - } - if(part_length == 0){ - SAPI::add_header_attribute(r.sapi_info, "status", "204 No Content"); - return; +static void file_md5_file_action(struct stat& finfo, int f, const String&, void *context) +{ + PA_MD5_CTX& md5context=*static_cast(context); + if(finfo.st_size) { + int nCount=0; + do { + unsigned char buffer[FILE_BUFFER_SIZE]; + nCount = file_block_read(f, buffer, sizeof(buffer)); + if ( nCount ){ + pa_MD5Update(&md5context, (const unsigned char*)buffer, nCount); } - SAPI::add_header_attribute(r.sapi_info, "status", "206 Partial Content"); - snprintf((char*)buf, BUFSIZE, "bytes %u-%u/%u", rg.start, rg.end, content_length); - SAPI::add_header_attribute(r.sapi_info, "Content-Range", (char*)buf); - }else if(count != 0){ - SAPI::add_header_attribute(r.sapi_info, "status", "501 Not Implemented"); - return; - } + } while(nCount > 0); } +} - fseek(f, offset, SEEK_SET); - snprintf((char*)buf, BUFSIZE, "%u", part_length); - SAPI::add_header_attribute(r.sapi_info, "Content-Length", (char*)buf); - - if(info.add_content_disposition && disposition){ - const char *fname = 0; - if(to_file_name){ - fname = to_file_name->as_string().cstr(); - }else{ - const char *fname = c_from_file_name; - const char *p1 = strrchr(fname, '/'); - const char *p2 = strrchr(fname, '\\'); - if(p1 || p2) - fname = max(p1, p2)+1; - } - - snprintf((char*)buf, BUFSIZE, "%s; filename=\"%s\"", disposition, fname); - SAPI::add_header_attribute(r.sapi_info, "Content-Disposition", (char*)buf); - } - if(info.add_content_type) - SAPI::add_header_attribute(r.sapi_info, "Content-Type", r.mime_type_of(c_from_file_name).cstr()); - if(info.add_last_modified){ - if(date == 0){ - struct stat st; - if(stat(c_from_file_name, &st)!=0) throw Exception("parser.runtime", 0, "can't get file stat"); - date = new VDate(st.st_mtime); - } - const String &s = attributed_meaning_to_string(*date, String::L_AS_IS, true); - SAPI::add_header_attribute(r.sapi_info, "Last-Modified", s.cstr()); - } - r.cookie.output_result(r.sapi_info); - SAPI::send_header(r.sapi_info); - - const char* request_method=getenv("REQUEST_METHOD"); - bool header_only=request_method && strcasecmp(request_method, "HEAD")==0; - size_t sent = 0; - if(!header_only){ - size_t to_read = 0; - size_t size = 0; - do{ - to_read = part_lengthcstr(): 0; +static void _md5(Request& r, MethodParams& params) { + const char* md5; + if(&r.get_self() == file_class) { + // ^file:md5[file-name] + if(params.count()) { + const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); + md5=pa_md5(r.absolute(file_spec)); + } else { + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); + } + } else { + // ^file.md5[] + VFile& self=GET_SELF(r, VFile); + md5=pa_md5(self.value_ptr(), self.value_size()); - VString* vcontent_type=handlers.user_content_type? - new VString(*handlers.user_content_type) - : user_file_name_cstr? - new VString(r.mime_type_of(user_file_name_cstr)) - : 0; - VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, handlers.value.str, handlers.value.length, user_file_name_cstr, vcontent_type); + } + r.write_no_lang(*new String(md5)); } // constructor MFile::MFile(): Methoded("file") { - // ^save[mode;file-name] - add_native_method("save", Method::CT_DYNAMIC, _save, 2, 2); + // ^file::create[text|binary;file-name;string-or-file[;options hash]] + // ^file::create[string-or-file[;options hash]] + add_native_method("create", Method::CT_DYNAMIC, _create, 1, 4); + + // ^file.save[mode;file-name] + // ^file.save[mode;file-name;$.charset[...]] + add_native_method("save", Method::CT_DYNAMIC, _save, 2, 3); + + // ^file:delete[file-name] + // ^file:delete[file-name;$.keep-empty-dir(true)$.exception(false)] + add_native_method("delete", Method::CT_STATIC, _delete, 1, 2); + + // ^file:move[from-file-name;to-file-name] + // ^file:move[from-file-name;to-file-name;$.keep-empty-dir(true)] + add_native_method("move", Method::CT_STATIC, _move, 2, 3); + + // ^file::load[mode;disk-name] + // ^file::load[mode;disk-name;user-name] + // ^file::load[mode;disk-name;user-name;options hash] + // ^file::load[mode;disk-name;options hash] + add_native_method("load", Method::CT_DYNAMIC, _load, 2, 4); - // ^delete[file-name] - add_native_method("delete", Method::CT_STATIC, _delete, 1, 1); - - // ^move[from-file-name;to-file-name] - add_native_method("move", Method::CT_STATIC, _move, 2, 2); - - // ^load[mode;disk-name] - // ^load[mode;disk-name;user-name] - add_native_method("load", Method::CT_DYNAMIC, _load, 2, 3); - - // ^stat[disk-name] + // ^file::stat[disk-name] add_native_method("stat", Method::CT_DYNAMIC, _stat, 1, 1); - // ^cgi[file-name] - // ^cgi[file-name;env hash] - // ^cgi[file-name;env hash;1cmd;2line;3ar;4g;5s] - add_native_method("cgi", Method::CT_DYNAMIC, _cgi, 1, 2+10); - - // ^exec[file-name] - // ^exec[file-name;env hash] - // ^exec[file-name;env hash;1cmd;2line;3ar;4g;5s] - add_native_method("exec", Method::CT_DYNAMIC, _exec, 1, 2+10); + // ^file::cgi[mode;file-name] + // ^file::cgi[mode;file-name;env hash] + // ^file::cgi[mode;file-name;env hash;1cmd;2line;3ar;4g;5s] + add_native_method("cgi", Method::CT_DYNAMIC, _cgi, 1, 3+50); + + // ^file::exec[mode;file-name] + // ^file::exec[mode;file-name;env hash] + // ^file::exec[mode;file-name;env hash;1cmd;2line;3ar;4g;5s] + add_native_method("exec", Method::CT_DYNAMIC, _exec, 1, 3+50); // ^file:list[path] // ^file:list[path][regexp] + // ^file:list[path][$.filter[regexp] $.stat(true)] add_native_method("list", Method::CT_STATIC, _list, 1, 2); // ^file:lock[path]{code} add_native_method("lock", Method::CT_STATIC, _lock, 2, 2); - // ^find[file-name] - // ^find[file-name]{when-not-found} + // ^file:find[file-name] + // ^file:find[file-name]{when-not-found} add_native_method("find", Method::CT_STATIC, _find, 1, 2); - // ^file:dirname[/a/some.tar.gz]=/a + // ^file:dirname[/a/some.tar.gz]=/a // ^file:dirname[/a/b/]=/a add_native_method("dirname", Method::CT_STATIC, _dirname, 1, 1); - // ^file:basename[/a/some.tar.gz]=some.tar.gz - - // ^file:send[filename] - // ^file:send[filename;options hash] - // ^file:send[filename;new_filename] - // ^file:send[filename;new_filename;options hash] - add_native_method("send", Method::CT_STATIC, _send, 1, 3); - add_native_method("basename", Method::CT_STATIC, _basename, 1, 1); - // ^file:justname[/a/some.tar.gz]=some.tar + // ^file:basename[/a/some.tar.gz]=some.tar.gz + add_native_method("basename", Method::CT_STATIC, _basename, 1, 1); + // ^file:justname[/a/some.tar.gz]=some.tar add_native_method("justname", Method::CT_STATIC, _justname, 1, 1); - // ^file:justext[/a/some.tar.gz]=gz + // ^file:justext[/a/some.tar.gz]=gz add_native_method("justext", Method::CT_STATIC, _justext, 1, 1); - // /some/page.html: ^file:fullpath[a.gif] => /some/a.gif + // /some/page.html: ^file:fullpath[a.gif] => /some/a.gif add_native_method("fullpath", Method::CT_STATIC, _fullpath, 1, 1); - // ^file.sql-string[] + // ^file.sql-string[] add_native_method("sql-string", Method::CT_DYNAMIC, _sql_string, 0, 0); - // ^file::sql[[alt_name]]{} + // ^file::sql{} + // ^file::sql{}[options hash] add_native_method("sql", Method::CT_DYNAMIC, _sql, 1, 2); + + // encode: + // ^file.base64[] + // ^file:base64[file-name] + // decode: + // ^file::base64[encoded] // backward + // ^file::base64[mode;user-file-name;encoded] + // ^file::base64[mode;user-file-name;encoded;$.content-type[...]] + add_native_method("base64", Method::CT_ANY, _base64, 0, 4); + + // ^file.crc32[] + // ^file:crc32[file-name] + add_native_method("crc32", Method::CT_ANY, _crc32, 0, 1); + + // ^file.md5[] + // ^file:md5[file-name] + add_native_method("md5", Method::CT_ANY, _md5, 0, 1); + + // ^file:copy[from-file-name;to-file-name] + add_native_method("copy", Method::CT_STATIC, _copy, 2, 2); }