scummvm/engines/sci/vocabulary.h

/* ScummVM - Graphic Adventure Engine
 *
 * ScummVM is the legal property of its developers, whose names
 * are too numerous to list here. Please refer to the COPYRIGHT
 * file distributed with this source distribution.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version 2
 * of the License, or (at your option) any later version.

 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.

 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
 *
 * $URL$
 * $Id$
 *
 */

#ifndef SCI_SCICORE_VOCABULARY_H
#define SCI_SCICORE_VOCABULARY_H

#include "common/str.h"
#include "common/hashmap.h"
#include "common/hash-str.h"
#include "common/list.h"

#include "sci/sci.h"

namespace Sci {

class ResourceManager;

/*#define VOCABULARY_DEBUG */

/** Number of bytes allocated on the heap to store bad words if parsing fails */
#define PARSE_HEAP_SIZE 64

enum {
	VOCAB_RESOURCE_SELECTORS = 997,

	VOCAB_RESOURCE_SCI0_MAIN_VOCAB = 0,
	VOCAB_RESOURCE_SCI0_PARSE_TREE_BRANCHES = 900,
	VOCAB_RESOURCE_SCI0_SUFFIX_VOCAB = 901,

	VOCAB_RESOURCE_SCI1_MAIN_VOCAB = 900,
	VOCAB_RESOURCE_SCI1_PARSE_TREE_BRANCHES = 901,
	VOCAB_RESOURCE_SCI1_SUFFIX_VOCAB = 902
};


enum {
	VOCAB_CLASS_PREPOSITION = 0x01,
	VOCAB_CLASS_ARTICLE = 0x02,
	VOCAB_CLASS_ADJECTIVE = 0x04,
	VOCAB_CLASS_PRONOUN = 0x08,
	VOCAB_CLASS_NOUN = 0x10,
	VOCAB_CLASS_INDICATIVE_VERB = 0x20,
	VOCAB_CLASS_ADVERB = 0x40,
	VOCAB_CLASS_IMPERATIVE_VERB = 0x80,
	VOCAB_CLASS_NUMBER = 0x001
};

enum {
	kParseEndOfInput = 0,
	kParseOpeningParenthesis = 1,
	kParseClosingParenthesis = 2,
	kParseNil = 3,
	kParseNumber = 4
};

#define VOCAB_CLASS_ANYWORD 0xff
/* Anywords are ignored by the parser */

#define VOCAB_MAGIC_NUMBER_GROUP 0xffd /* 0xffe ? */
/* This word class is used for numbers */

#define VOCAB_TREE_NODES 500
/* Number of nodes for each parse_tree_node structure */

#define VOCAB_TREE_NODE_LAST_WORD_STORAGE 0x140
#define VOCAB_TREE_NODE_COMPARE_TYPE 0x146
#define VOCAB_TREE_NODE_COMPARE_GROUP 0x14d
#define VOCAB_TREE_NODE_FORCE_STORAGE 0x154

#define SAID_COMMA   0xf0
#define SAID_AMP     0xf1
#define SAID_SLASH   0xf2
#define SAID_PARENO  0xf3
#define SAID_PARENC  0xf4
#define SAID_BRACKO  0xf5
#define SAID_BRACKC  0xf6
#define SAID_HASH    0xf7
#define SAID_LT      0xf8
#define SAID_GT      0xf9
#define SAID_TERM    0xff

#define SAID_FIRST SAID_COMMA

/* There was no 'last matching word': */
#define SAID_FULL_MATCH 0xffff
#define SAID_NO_MATCH 0xfffe
#define SAID_PARTIAL_MATCH 0xfffd

#define SAID_LONG(x) ((x) << 8)

struct ResultWord {
	int _class; /* Word class */
	int _group; /* Word group */
};

typedef Common::List<ResultWord> ResultWordList;

typedef Common::HashMap<Common::String, ResultWord, Common::IgnoreCase_Hash, Common::IgnoreCase_EqualTo> WordMap;


struct parse_rule_t {
	int id; /* non-terminal ID */
	int first_special; /* first terminal or non-terminal */
	int specials_nr; /* number of terminals and non-terminals */
	int length;
	int data[1]; /* actual data (size 1 to avoid compiler warnings) */
};


struct parse_rule_list_t {
	int terminal; /* Terminal character this rule matches against or 0 for a non-terminal rule */
	parse_rule_t *rule;
	parse_rule_list_t *next;
};


struct suffix_t {

	int class_mask; /* the word class this suffix applies to */
	int result_class; /* the word class a word is morphed to if it doesn't fail this check */

	int alt_suffix_length; /* String length of the suffix */
	int word_suffix_length; /* String length of the other suffix */

	const char *alt_suffix; /* The alternative suffix */
	const char *word_suffix; /* The suffix as used in the word vocabulary */

};

typedef Common::List<suffix_t> SuffixList;


struct synonym_t {
	int replaceant; /* The word group to replace */
	int replacement; /* The replacement word group for this one */
};

typedef Common::List<synonym_t> SynonymList;

struct parse_tree_branch_t {
	int id;
	int data[10];
};

enum ParseTypes {
	kParseTreeLeafNode = 0,
	kParseTreeBranchNode = 1
};

struct parse_tree_node_t {
	short type;  /* leaf or branch */
	union {
		int value;  /* For leaves */
		short branches[2]; /* For branches */
	} content;
};

enum VocabularyVersions {
	kVocabularySCI0 = 0,
	kVocabularySCI1 = 1
};

class Vocabulary {
public:
	Vocabulary(ResourceManager *resMan);
	~Vocabulary();

	/**
	 * Gets any word from the specified group. For debugging only.
	 * @param group		Group number
	 */
	const char *getAnyWordFromGroup(int group);


	/**
	 * Looks up a single word in the words and suffixes list.
	 * @param word		pointer to the word to look up
	 * @param word_len	length of the word to look up
	 * @return the matching word (or (-1,-1) if there was no match)
	 */
	ResultWord lookupWord(const char *word, int word_len);


	/* Tokenizes a string and compiles it into word_ts.
	** Parameters: (char *) sentence: The sentence to examine
	**             (char **) error: Points to a malloc'd copy of the offending text or to NULL on error
	**             (ResultWordList) retval: A list of word_ts containing the result, or NULL.
	** Returns   : true on success, false on failure
	** On error, NULL is returned. If *error is NULL, the sentence did not contain any useful words;
	** if not, *error points to a malloc'd copy of the offending word.
	** The returned list may contain anywords.
	*/
	bool tokenizeString(ResultWordList &retval, const char *sentence, char **error);

	/* Builds a parse tree from a list of words, using a set of Greibach Normal Form rules
	** Parameters: 
	**             (const ResultWordList &) words: The words to build the tree from
	**             bool verbose: Set to true for debugging
	** Returns   : 0 on success, 1 if the tree couldn't be built in VOCAB_TREE_NODES nodes
	**             or if the sentence structure in 'words' is not part of the language
	**             described by the grammar passed in 'rules'.
	*/
	int parseGNF(const ResultWordList &words, bool verbose = false);

	/* Constructs the Greibach Normal Form of the grammar supplied in 'branches'
	**             bool verbose: Set to true for debugging.
	**             If true, the list is freed before the function ends
	** Returns   : (parse_rule_list_t *): Pointer to a list of singly linked
	**                                    GNF rules describing the same language
	**                                    that was described by 'branches'
	** The original SCI rules are in almost-CNF (Chomsky Normal Form). Note that
	** branch[0] is used only for a few magical incantations, as it is treated
	** specially by the SCI parser.
	*/
	parse_rule_list_t *buildGNF(bool verbose = false);

	/**
	 * Deciphers a said block and dumps its content via printf.
	 * For debugging only.
	 * @param pos	pointer to the data to dump
	 */
	void decipherSaidBlock(byte *pos);

	/**
	 * Prints the parser suffixes to the debug console.
	 */
	void printSuffixes() const;

	/**
	 * Prints the parser words to the debug console.
	 */
	void printParserWords() const;

	uint getParserBranchesSize() const { return _parserBranches.size(); }
	const parse_tree_branch_t &getParseTreeBranch(int number) const { return _parserBranches[number]; }

	/**
	 * Adds a new synonym to the list
	 */
	void addSynonym(synonym_t syn) { _synonyms.push_back(syn); }
	
	/**
	 * Clears the list of synonyms
	 */
	void clearSynonyms() { _synonyms.clear(); }
	
	/** 
	 * Synonymizes a token list
	 * Parameters: (ResultWordList &) words: The word list to synonymize
	 */
	void synonymizeTokens(ResultWordList &words);

	void printParserNodes(int num);

	void dumpParseTree();

	int parseNodes(int *i, int *pos, int type, int nr, int argc, const char **argv);

	// Accessed by said()
	parse_tree_node_t _parserNodes[VOCAB_TREE_NODES]; /**< The parse tree */

private:
	/**
	 * Loads all words from the main vocabulary.
	 * @return true on success, false on failure
	 */
	bool loadParserWords();

	/**
	 * Loads all suffixes from the suffix vocabulary.
	 * @return true on success, false on failure
	 */
	bool loadSuffixes();

	/**
	 * Frees all suffixes in the given list.
	 * @param suffixes: The suffixes to free
	 */
	void freeSuffixes();

	/**
	 * Retrieves all grammar rules from the resource data.
	 * @param branches		The rules are stored into this Array
	 * @return true on success, false on error
	 */
	bool loadBranches();

	/* Frees a parser rule list as returned by vocab_build_gnf()
	** Parameters: (parse_rule_list_t *) rule_list: The rule list to free
	*/
	void freeRuleList(parse_rule_list_t *rule_list);

	ResourceManager *_resMan;
	VocabularyVersions _vocabVersion;

	// Parser-related lists
	SuffixList _parserSuffixes;
	parse_rule_list_t *_parserRules; /**< GNF rules used in the parser algorithm */
	Common::Array<parse_tree_branch_t> _parserBranches;
	WordMap _parserWords;
	SynonymList _synonyms; /**< The list of synonyms */
};

/* Prints a parse tree
** Parameters: (const char *) tree_name: Name of the tree to dump (free-form)
**             (parse_tree_node_t *) nodes: The nodes containing the parse tree
*/
void vocab_dump_parse_tree(const char *tree_name, parse_tree_node_t *nodes);



/* Builds a parse tree from a spec and compares it to a parse tree
** Parameters: (EngineState *) s: The affected state
**             (byte *) spec: Pointer to the spec to build
**             (bool) verbose: Whether to display the parse tree after building it
** Returns   : (int) 1 on a match, 0 otherwise
*/
int said(EngineState *s, byte *spec, bool verbose);

int getAllocatedRulesCount();

} // End of namespace Sci

#endif // SCI_SCICORE_VOCABULARY_H