winix/core/confparser.h

/*
 * This file is a part of Winix
 * and is not publicly distributed
 * 
 * Copyright (c) 2008-2010, Tomasz Sowa
 * All rights reserved.
 * 
 */
 
#ifndef headerfileconfparser
#define headerfileconfparser

#include <fstream>
#include <string>
#include <vector>
#include <map>




/*
A parser for parsing config files.

A config file can look like this:
	variable1 = value 1
	variable2 = " value 2 "
	variable3 = (value 1, value 2)
	variable4 = (" value 1 " , "value2", value 3)

sample of use:
	ConfParser parser;
	parser.Parse("/path/to/config");

	if( parser.status == ConfParser::ok )
	{
		// the whole config we have in parser.table (parser.table_single)
	}

config syntax:
	option = list

	list can consists of any number of items, if you're using more than one item you should
	use brackets ()

	for one item the brackets can be ommited:
		option = value
	white characters at the beginning of the value (and at the end) will be trimmed,
	or you can use quotes:
		option  = "value"
		option2 = "value with spaces at the end   "

	the form without quotes:
		option = value
	should be written in one line, so this is not allowed:
		option =
				 value
	you can use a new line characters only between brackets and quotes:
		option = "this is
				  a multiline string"
		option = ( value1, 
				   value2 )

	but there is one requirement: the first character " or ( should be in the same line,
	so this is not allowed
		option =
				"this is wrong"
		but this is ok:
		option = "
		         that is ok"

	empty lists:
		option = ()
		this creates an empty list: parser.table['option'].empty() == true

		option = 
		this creates an empty list too (the same as previously)

		option = ""
		but this doesn't create an empty list, it creates a list with one (empty) item

	commentaries:
		# this is a commentary (until the end of the line)
		option = value # this is a commentary too
		
		commentaries are treated as white characters, other example:
		option = ( # this is my list
			"value 1" # this is a value one
			value 2 # and this is a value two
			) # end of my list

	overwriting:
		option1 = some value
		option1 = other value
		# always the last option is used so option1 is "other value"

	list delimiter:
		option1 = (value1, value2, value3)
		option2 = ("value1", "value2", "value3")
		above we're using a comma ',' as a list delimiter but when using quotes (second line)
		the commas can be omitted:
		option2 = ("value1" "value2" "value3")

	white characters:
		the name of an option cannot consist of white characters
		some option = value # this is wrong
		some_option = value # this is ok

		which characters are allowed in an option name is defined by IsVariableChar() method

		you can use white characters in values
		option = value with spaces or tabs
		white characters at the beginning and at the end will be trimmed,
		so if you want them use quotes:
		option = "  other value with spaces  "

	special characters in quoted strings:
		option = "this is a string with \" a quote inside"
		the option will be: this is a string with " a quote inside
		\\ - means one \
		basically: \char produces char
		so:
		"\a" gives "a"
		"\\" gives "\"
		"\Z" gives "Z" and so on


*/
class ConfParser
{
public:

	ConfParser();


	/*
		status of parsing
	*/
	enum Status { ok, cant_open_file, syntax_error };


	/*
		the last status of parsing, set by Parse() methods
	*/
	Status status;


	/*
		the main methods used to parse
		file_name is the path to a file
	*/
	Status Parse(const char * file_name);
	Status Parse(const std::string & file_name);

	
	/*
		a number of a line in which there is a syntax_error
	*/
	int line;


	/*
		this is the table which represents your config file
		in the Table map: the first (key) is your 'option' and the second is 'list'
	*/
	typedef std::vector<std::string>     Value;
	typedef std::map<std::string, Value> Table;
	Table table;


	/*
		if your config file consists mainly of single forms such as:
			option = value
			option2 = value2
		then you can call SplitSingle(true) for not inserting single values to 
		previous 'table' but instead to 'table_single'
		table_single as the second parameter takes only std::string (instead of the whole std::vector)
		so you can save a little memory from not using std::vector
	*/
	typedef std::map<std::string, std::string> TableSingle;
	TableSingle table_single;


	/*
		if your list consists of only one item, e.g:
			option1 = value 1
			option2 = "value 2"
			option3 = ( "value 3" )
		then if you call SplitSingle(true) then such values will be stored in 
		'table_single' instead of 'table' map
		default: false
	*/
	void SplitSingle(bool split);


	/*
		if true then empty lists, e.g:
			option = 
			option2 = ()
		will be omitted (not inserted to 'table' or 'table_single')
		default: false	
	*/
	void SkipEmpty(bool skip);


	/*
		those methods are used to extract information from table or table_single
		as a parameter they take the name of an option
		and a default value (if there is no such a parameter),
		they return appropriate value (either text, int or boolean)
		(in lists they return the first item if exists)
	*/
	std::string Text(const char * name);
	std::string Text(const char * name, const char * def);
	std::string Text(const std::string & name, const std::string & def);
	int Int(const char *);
	int Int(const char * name, int def);
	int Int(const std::string & name, int def);
	size_t Size(const char *);
	size_t Size(const char * name, size_t def);
	size_t Size(const std::string & name, size_t def);
	bool Bool(const char *);
	bool Bool(const char * name, bool def);
	bool Bool(const std::string & name, bool def);


	/*
		some default values
		used in Text() Int() or Bool() when you don't explicitly set the default value

		if you don't set it directly then:
		default text is: "" (empty)
		default int or size is:  0
		default bool is: false
	*/
	void SetDefaultText(const std::string & def);
	void SetDefaultInt(int def);
	void SetDefaultSize(size_t def);
	void SetDefaultBool(bool def);


	/*
		those methods are used to extract lists
		note: if there is one option in table_single they will return it
	*/
	void ListText(const char * name, std::vector<std::string> & list);
	void ListText(const std::string & name, std::vector<std::string> & list);


private:

	/*
		last read variable (option)
	*/
	std::string variable;
	

	/*
		last read list item
	*/
	std::string value_item;


	/*
		last read list
	*/
	Value value;


	/*
		separator between a variable and a value, default: '='
	*/
	int separator;	


	/*
		commentary char, default: '#'
	*/
	int commentary;


	/*
		list starting character, default: '('
	*/
	int list_start;


	/*
		list ending character, default: ')'
	*/
	int list_end;


	/*
		list delimiter, default: ','
	*/
	int list_delimiter;


	/*
		last read char
	*/
	int lastc;


	/*
		current file
	*/
	std::ifstream file;


	/*
		if true then lists with one item will be put into 'table_single' table
		default: false
	*/
	bool split_single;


	/*
		if true then empty lists, e.g:
			option = 
			option2 = ()
		will be omitted (not inserted to 'table' or 'table_single')
		default: false	
	*/
	bool skip_empty;



	std::string default_str;
	int default_int;
	size_t default_size;
	bool default_bool;

	int ToInt(const std::string & value);
	size_t ToSize(const std::string & value);
	bool ToBool(const std::string & value);

	Status ParseFile();
	void AddOption();

	void DeleteFromTable(const std::string & var);
	void DeleteFromTableSingle(const std::string & var);

	bool ReadVariable();
	bool ReadValue();
	bool ReadValueList();
	bool ReadValueNoList(bool use_list_delimiter = false);
	bool ReadValueQuoted();
	bool ReadValueSimple(bool use_list_delimiter = false);
	
	int  ReadChar();
	bool IsWhite(int c);
	bool IsVariableChar(int c);
	void SkipWhite();
	void SkipWhiteLines();
	void SkipLine();
	void Trim(std::string & s);	
};



#endif