dvander@372: /** dvander@375: * vim: set ts=4 : dvander@372: * =============================================================== dvander@376: * SourceMod, Copyright (C) 2004-2007 AlliedModders LLC. dvander@376: * All rights reserved. dvander@372: * =============================================================== dvander@372: * dvander@379: * This file is part of the SourceMod/SourcePawn SDK. This file may only be dvander@379: * used or modified under the Terms and Conditions of its License Agreement, dvander@379: * which is found in public/licenses/LICENSE.txt. As of this notice, derivative dvander@379: * works must be licensed under the GNU General Public License (version 2 or dvander@379: * greater). A copy of the GPL is included under public/licenses/GPL.txt. dvander@379: * dvander@379: * To view the latest information, see: http://www.sourcemod.net/license.php dvander@372: * dvander@372: * Version: $Id$ dvander@372: */ dvander@372: dvander@324: #ifndef _INCLUDE_SOURCEMOD_TEXTPARSERS_INTERFACE_H_ dvander@324: #define _INCLUDE_SOURCEMOD_TEXTPARSERS_INTERFACE_H_ dvander@324: dvander@372: /** dvander@372: * @file ITextParsers.h dvander@372: * @brief Defines various text/file parsing functions, as well as UTF-8 support code. dvander@372: */ dvander@372: dvander@324: #include dvander@324: dvander@324: namespace SourceMod dvander@324: { dvander@324: /** dvander@324: * The INI file format is defined as: dvander@324: * WHITESPACE: 0x20, \n, \t, \r dvander@324: * IDENTIFIER: A-Z a-z 0-9 _ - , + . $ ? / dvander@324: * STRING: Any set of symbols dvander@324: * dvander@324: * Basic syntax is comprised of SECTIONs. dvander@324: * A SECTION is defined as: dvander@324: * [SECTIONNAME] dvander@324: * OPTION dvander@324: * OPTION dvander@324: * OPTION... dvander@324: * dvander@324: * SECTIONNAME is an IDENTIFIER. dvander@324: * OPTION can be repeated any number of times, once per line. dvander@324: * OPTION is defined as one of: dvander@324: * KEY = "VALUE" dvander@324: * KEY = VALUE dvander@324: * KEY dvander@324: * Where KEY is an IDENTIFIER and VALUE is a STRING. dvander@324: * dvander@324: * WHITESPACE should always be omitted. ds@729: * COMMENTS should be stripped, and are defined as text occurring in: dvander@324: * ; dvander@324: * dvander@324: * Example file below. Note that dvander@324: * The second line is technically invalid. The event handler dvander@324: * must decide whether this should be allowed. dvander@324: * --FILE BELOW-- dvander@324: * [gaben] dvander@324: * hi = clams dvander@324: * bye = "NO CLAMS" dvander@324: * dvander@324: * [valve] dvander@324: * cannot dvander@324: * maintain dvander@324: * products dvander@324: */ dvander@372: dvander@372: /** dvander@372: * @brief Contains parse events for INI files. dvander@372: */ dvander@324: class ITextListener_INI dvander@324: { dvander@324: public: dvander@324: /** dvander@324: * @brief Called when a new section is encountered in an INI file. dvander@324: * dvander@324: * @param section Name of section in between the [ and ] characters. dvander@324: * @param invalid_tokens True if invalid tokens were detected in the name. dvander@324: * @param close_bracket True if a closing bracket was detected, false otherwise. dvander@324: * @param extra_tokens True if extra tokens were detected on the line. dvander@324: * @param curtok Contains current token in the line where the section name starts. dvander@324: * You can add to this offset when failing to point to a token. dvander@324: * @return True to keep parsing, false otherwise. dvander@324: */ dvander@324: virtual bool ReadINI_NewSection(const char *section, dvander@324: bool invalid_tokens, dvander@324: bool close_bracket, dvander@324: bool extra_tokens, dvander@324: unsigned int *curtok) dvander@324: { dvander@324: return true; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called when encountering a key/value pair in an INI file. dvander@324: * dvander@324: * @param key Name of key. dvander@324: * @param value String containing value (with quotes stripped, if any). dvander@324: * @param invalid_tokens Whether or not the key contained invalid tokens. dvander@324: * @param equal_token There was an '=' sign present (in case the value is missing). dvander@324: * @param quotes Whether value was enclosed in quotes. dvander@372: * @param curtok Contains the token index of the start of the value string. dvander@324: * This can be changed when returning false. dvander@324: * @return True to keep parsing, false otherwise. dvander@324: */ dvander@324: virtual bool ReadINI_KeyValue(const char *key, dvander@324: const char *value, dvander@324: bool invalid_tokens, dvander@324: bool equal_token, dvander@324: bool quotes, dvander@324: unsigned int *curtok) dvander@324: { dvander@324: return true; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called after a line has been preprocessed, if it has text. dvander@324: * dvander@324: * @param line Contents of line. dvander@324: * @param curtok Pointer to optionally store failed position in string. dvander@324: * @return True to keep parsing, false otherwise. dvander@324: */ dvander@372: virtual bool ReadINI_RawLine(const char *line, unsigned int *curtok) dvander@324: { dvander@324: return true; dvander@324: } dvander@324: }; dvander@324: dvander@324: /** dvander@324: * :TODO: write this in CFG (context free grammar) format so it makes sense dvander@324: * dvander@324: * The SMC file format is defined as: dvander@324: * WHITESPACE: 0x20, \n, \t, \r dvander@351: * IDENTIFIER: Any ASCII character EXCLUDING ", {, }, ;, //, / *, or WHITESPACE. dvander@324: * STRING: Any set of symbols enclosed in quotes. dvander@324: * Note: if a STRING does not have quotes, it is parsed as an IDENTIFIER. dvander@324: * dvander@324: * Basic syntax is comprised of SECTIONBLOCKs. dvander@324: * A SECTIONBLOCK defined as: dvander@324: * dvander@324: * SECTIONNAME dvander@324: * { dvander@324: * OPTION dvander@324: * } dvander@324: * dvander@324: * OPTION can be repeated any number of times inside a SECTIONBLOCK. dvander@324: * A new line will terminate an OPTION, but there can be more than one OPTION per line. dvander@324: * OPTION is defined any of: dvander@324: * "KEY" "VALUE" dvander@324: * SECTIONBLOCK dvander@324: * dvander@324: * SECTIONNAME, KEY, VALUE, and SINGLEKEY are strings dvander@324: * SECTIONNAME cannot have trailing characters if quoted, but the quotes can be optionally removed. dvander@324: * If SECTIONNAME is not enclosed in quotes, the entire sectionname string is used (minus surrounding whitespace). dvander@324: * If KEY is not enclosed in quotes, the key is terminated at first whitespace. dvander@324: * If VALUE is not properly enclosed in quotes, the entire value string is used (minus surrounding whitespace). dvander@324: * The VALUE may have inner quotes, but the key string may not. dvander@324: * dvander@324: * For an example, see configs/permissions.cfg dvander@324: * dvander@324: * WHITESPACE should be ignored. ds@729: * Comments are text occurring inside the following tokens, and should be stripped dvander@324: * unless they are inside literal strings: dvander@324: * ; dvander@324: * // dvander@351: * / * */ dvander@324: dvander@372: /** dvander@372: * @brief Lists actions to take when an SMC parse hook is done. dvander@372: */ dvander@324: enum SMCParseResult dvander@324: { dvander@372: SMCParse_Continue, /**< Continue parsing */ dvander@372: SMCParse_Halt, /**< Stop parsing here */ dvander@372: SMCParse_HaltFail /**< Stop parsing and return SMCParseError_Custom */ dvander@324: }; dvander@324: dvander@372: /** dvander@372: * @brief Lists error codes possible from parsing an SMC file. dvander@372: */ dvander@324: enum SMCParseError dvander@324: { dvander@372: SMCParse_Okay = 0, /**< No error */ dvander@372: SMCParse_StreamOpen, /**< Stream failed to open */ dvander@372: SMCParse_StreamError, /**< The stream died... somehow */ dvander@372: SMCParse_Custom, /**< A custom handler threw an error */ dvander@372: SMCParse_InvalidSection1, /**< A section was declared without quotes, and had extra tokens */ dvander@372: SMCParse_InvalidSection2, /**< A section was declared without any header */ dvander@372: SMCParse_InvalidSection3, /**< A section ending was declared with too many unknown tokens */ dvander@372: SMCParse_InvalidSection4, /**< A section ending has no matching beginning */ dvander@372: SMCParse_InvalidSection5, /**< A section beginning has no matching ending */ dvander@372: SMCParse_InvalidTokens, /**< There were too many unidentifiable strings on one line */ dvander@372: SMCParse_TokenOverflow, /**< The token buffer overflowed */ dvander@372: SMCParse_InvalidProperty1, /**< A property was declared outside of any section */ dvander@324: }; dvander@324: dvander@372: /** dvander@372: * @brief Describes the events available for reading an SMC stream. dvander@372: */ dvander@324: class ITextListener_SMC dvander@324: { dvander@324: public: dvander@324: /** dvander@324: * @brief Called when starting parsing. dvander@324: */ dvander@324: virtual void ReadSMC_ParseStart() dvander@324: { dvander@324: }; dvander@324: dvander@324: /** dvander@324: * @brief Called when ending parsing. dvander@324: * dvander@324: * @param halted True if abnormally halted, false otherwise. dvander@324: * @param failed True if parsing failed, false otherwise. dvander@324: */ dvander@324: virtual void ReadSMC_ParseEnd(bool halted, bool failed) dvander@324: { dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called when a warning occurs. dvander@324: * @param error By-reference variable containing the error message of the warning. dvander@324: * @param tokens Pointer to the token stream causing the error. dvander@324: * @return SMCParseResult directive. dvander@324: */ dvander@324: virtual SMCParseResult ReadSMC_OnWarning(SMCParseError &error, const char *tokens) dvander@324: { dvander@324: return SMCParse_HaltFail; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called when entering a new section dvander@324: * dvander@324: * @param name Name of section, with the colon omitted. dvander@324: * @param opt_quotes Whether or not the option string was enclosed in quotes. dvander@324: * @return SMCParseResult directive. dvander@324: */ dvander@324: virtual SMCParseResult ReadSMC_NewSection(const char *name, bool opt_quotes) dvander@324: { dvander@324: return SMCParse_Continue; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called when encountering a key/value pair in a section. dvander@324: * dvander@324: * @param key Key string. dvander@324: * @param value Value string. If no quotes were specified, this will be NULL, dvander@324: and key will contain the entire string. dvander@324: * @param key_quotes Whether or not the key was in quotation marks. dvander@324: * @param value_quotes Whether or not the value was in quotation marks. dvander@324: * @return SMCParseResult directive. dvander@324: */ dvander@324: virtual SMCParseResult ReadSMC_KeyValue(const char *key, dvander@324: const char *value, dvander@324: bool key_quotes, dvander@324: bool value_quotes) dvander@324: { dvander@324: return SMCParse_Continue; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called when leaving the current section. dvander@324: * dvander@324: * @return SMCParseResult directive. dvander@324: */ dvander@324: virtual SMCParseResult ReadSMC_LeavingSection() dvander@324: { dvander@324: return SMCParse_Continue; dvander@324: } dvander@324: dvander@324: /** dvander@324: * @brief Called after an input line has been preprocessed. dvander@324: * dvander@324: * @param line String containing line input. dvander@324: * @param curline Number of line in file. dvander@324: * @return SMCParseResult directive. dvander@324: */ dvander@324: virtual SMCParseResult ReadSMC_RawLine(const char *line, unsigned int curline) dvander@324: { dvander@324: return SMCParse_Continue; dvander@324: } dvander@324: }; dvander@324: dvander@324: #define SMINTERFACE_TEXTPARSERS_NAME "ITextParsers" dvander@324: #define SMINTERFACE_TEXTPARSERS_VERSION 1 dvander@324: dvander@372: /** dvander@372: * @brief Contains various text stream parsing functions. dvander@372: */ dvander@324: class ITextParsers : public SMInterface dvander@324: { dvander@324: public: dvander@324: virtual const char *GetInterfaceName() dvander@324: { dvander@324: return SMINTERFACE_TEXTPARSERS_NAME; dvander@324: } dvander@324: virtual unsigned int GetInterfaceVersion() dvander@324: { dvander@324: return SMINTERFACE_TEXTPARSERS_VERSION; dvander@324: } dvander@324: public: dvander@324: /** dvander@324: * @brief Parses an INI-format file. dvander@324: * dvander@324: * @param file Path to file. dvander@324: * @param ini_listener Event handler for reading file. dvander@324: * @param line If non-NULL, will contain last line parsed (0 if file could not be opened). dvander@324: * @param col If non-NULL, will contain last column parsed (undefined if file could not be opened). ds@729: * @return True if parsing succeeded, false if file couldn't be opened or there was a syntax error. dvander@324: */ dvander@324: virtual bool ParseFile_INI(const char *file, dvander@324: ITextListener_INI *ini_listener, dvander@324: unsigned int *line, dvander@324: unsigned int *col) =0; dvander@324: dvander@324: /** dvander@324: * @brief Parses an SMC-format text file. dvander@324: * Note that the parser makes every effort to obey broken syntax. dvander@324: * For example, if an open brace is missing, but the section name has a colon, dvander@324: * it will let you know. It is up to the event handlers to decide whether to be strict or not. dvander@324: * dvander@324: * @param file Path to file. dvander@324: * @param smc_listener Event handler for reading file. dvander@324: * @param line If non-NULL, will contain last line parsed (0 if file could not be opened). dvander@324: * @param col If non-NULL, will contain last column parsed (undefined if file could not be opened). dvander@324: * @return An SMCParseError result code. dvander@324: */ dvander@324: virtual SMCParseError ParseFile_SMC(const char *file, dvander@324: ITextListener_SMC *smc_listener, dvander@324: unsigned int *line, dvander@324: unsigned int *col) =0; dvander@324: dvander@324: /** ds@729: * @brief Converts an SMCParseError to a string. dvander@324: * dvander@324: * @param err SMCParseError. dvander@324: * @return String error message, or NULL if none. dvander@324: */ dvander@324: virtual const char *GetSMCErrorString(SMCParseError err) =0; dvander@324: dvander@324: public: dvander@324: /** dvander@324: * @brief Returns the number of bytes that a multi-byte character contains in a UTF-8 stream. dvander@324: * If the current character is not multi-byte, the function returns 1. dvander@324: * dvander@324: * @param stream Pointer to multi-byte ANSI character string. dvander@324: * @return Number of bytes in current character. dvander@324: */ dvander@324: virtual unsigned int GetUTF8CharBytes(const char *stream) =0; dvander@324: }; ds@432: } dvander@324: dvander@324: #endif //_INCLUDE_SOURCEMOD_TEXTPARSERS_INTERFACE_H_