dvander@370: /**
ds@653:  * vim: set ts=4 :
dvander@370:  * ===============================================================
dvander@370:  * SourceMod (C)2004-2007 AlliedModders LLC.  All rights reserved.
dvander@370:  * ===============================================================
dvander@370:  *
dvander@370:  * This file is not open source and may not be copied without explicit
dvander@370:  * written permission of AlliedModders LLC.  This file may not be redistributed 
dvander@370:  * in whole or significant part.
dvander@370:  * For information, see LICENSE.txt or http://www.sourcemod.net/license.php
dvander@370:  *
dvander@370:  * Version: $Id$
dvander@370:  */
dvander@370: 
dvander@189: #include <string.h>
dvander@189: #include <stdlib.h>
dvander@189: #include <assert.h>
dvander@189: #include "sm_trie.h"
dvander@189: 
dvander@189: /**
dvander@189:  * Double Array Trie algorithm, based on:
dvander@189:  * An Efficient Implementation of Trie Structures, by
dvander@189:  *  Jun-ichi Aoe and Katsushi Maromoto, and Takashi Sato
ds@729:  * from Software - Practice and Experience, Vol. 22(9), 695-721 (September 1992)
dvander@189:  *
dvander@189:  *  A Trie is a simple data structure which stores strings as DFAs, with each 
dvander@189:  * transition state being a string entry.  For example, observe the following strings:
dvander@189:  *
dvander@189:  * BAILOPAN, BAT, BACON, BACK
dvander@189:  *  These transition as the follow production rules:
dvander@189:  *  B -> ...                  B
dvander@189:  *       A -> ...             BA
dvander@189:  *            I -> ...        BAI
dvander@189:  *                 LOPAN      BAILOPAN
dvander@189:  *            T -> ...        BAT
dvander@189:  *            C ->            BAC
dvander@189:  *                 O -> ...   BACO
dvander@189:  *                      N     BACON
dvander@189:  *                 K          BACK
dvander@189:  *
dvander@189:  *  The standard implementation for this - using lists - gives a slow linear lookup, somewhere between
dvander@189:  * O(N+M) or O(log n).  A faster implementation is proposed in the paper above, which is based on compacting
dvander@189:  * the transition states into two arrays.  In the paper's implementation, two arrays are used, and thus it is
dvander@189:  * called the "Double Array" algorithm.  However, the CHECK array's size is maintained the same as BASE, 
dvander@189:  * so they can be combined into one structure.  The array seems complex at first, but is very simple: it is a
dvander@189:  * tree structure flattened out into a single vector.  I am calling this implementation the Flat Array Trie.
dvander@189:  *
dvander@189:  *  BASE[] is an array where each member is a node in the Trie.  The node can either be UNUSED (empty), an ARC
dvander@189:  * (containing an offset to the next set of ARCs), or a TERMINATOR (contains the rest of a string).
ds@729:  * Each node has an index which must be interpreted based on the node type.  If the node is a TERMINATOR, then the
dvander@189:  * index is an index into a string table, to find the rest of the string.  
dvander@189:  *  If the node is an ARC, the index is another index into BASE.  For each possible token that can follow the
dvander@189:  * current token, the value of those tokens can be added to the index given in the ARC.  Thus, given a current
dvander@189:  * position and the next desired token, the current arc will jump to another arc which can contain either:
dvander@189:  *   1) An invalid production (collision, no entry exists)
dvander@189:  *   2) An empty production (no entry exists)
dvander@189:  *   3) Another arc label (the string ends here or continues into more productions)
dvander@189:  *   4) A TERMINATOR (the string ends here and contains an unused set of productions)
dvander@189:  * 
dvander@189:  *  So, given current offset N (starting at N=1), jumping to token C means the next offset will be:
dvander@189:  *      offs = BASE[n] + C
dvander@189:  *  Thus, the next node will be at:
dvander@189:  *      BASE[BASE[n] + C]
dvander@189:  * 
dvander@189:  *  This allows each ARC to specify the base offset for any of its ARC children, like a tree.  Each node specifies
dvander@189:  * its parent ARC -- so if an invalid offset is specified, the parent will not match, and thus no such derived 
dvander@189:  * string exists.
dvander@189:  *
dvander@189:  *  This means that arrays can be laid out "sparsely," maximizing their usage.  Note that N need not be related to
dvander@189:  * the range of tokens (1-256).  I.e., a base index does not have to be at 1, 256, 512, et cetera.  This is because
dvander@189:  * insertion comes with a small deal of complexity.  To insert a new set of tokens T, the algorithm finds a new 
dvander@189:  * BASE index N such that BASE[N+T[i]] is unused for each T[i].  Thus, indirection is not necessarily linear; 
dvander@189:  * traversing a chain of ARC nodes can _and will_ jump around BASE.
dvander@189:  *
dvander@189:  *  Of course, given this level of flexibility in the array organization, there are collisions.  This is largely 
dvander@189:  * where insertions become slow, as the old chain must be relocated before the new one is used.  Relocation means 
dvander@189:  * finding one or more new base indexes, and this means traversing BASE until an acceptable index is found, such 
dvander@189:  * that each offset is unused (see description in previous paragraph).
dvander@189:  *
dvander@189:  *  However, it is not insertion time we are concerned about.  The "trie" name comes from reTRIEval.  We are only
dvander@189:  * concerned with lookup and deletion.  Both lookup and deletion are O(k), where k is relative to the length of the 
dvander@189:  * input string.  Note that it is best case O(1) and worst case O(k).  Deleting the entire trie is always O(1).
dvander@189:  */
dvander@189: 
dvander@189: /**
dvander@189:  * Optimization ideas for the future:
dvander@189:  * 1) Store a reference count for each arc, with the count of sub-children.
dvander@189:  *    This could let us break out of "children searches" for case 4 easily.
dvander@189:  * 2) Use a 'free list' so we can easily search the trie for free points.
dvander@189:  *    This would drastically speed up x_check*
dvander@189:  */
dvander@189: 
dvander@189: enum NodeType
dvander@189: {
dvander@189: 	Node_Unused = 0,		/* Node is not being used (sparse) */
dvander@189: 	Node_Arc,				/* Node is part of an arc and does not terminate */
dvander@189: 	Node_Term,				/* Node is a terminator */
dvander@189: };
dvander@189: 
dvander@189: struct TrieNode
dvander@189: {
dvander@189: 	/**
dvander@189: 	 * For Node_Arc, this index stores the 'base' offset to the next arc chain.
dvander@189: 	 *   I.e. to jump from this arc to character C, it will be at base[idx+C].
dvander@189: 	 * For Node_Term, this is an index into the string table.
dvander@189: 	 */
dvander@189: 	unsigned int idx;
dvander@189: 
dvander@189: 	/**
dvander@189: 	 * This contains the prior arc that we must have come from.
dvander@189: 	 * For example, if arc 63 has a base jump of index 12, and we want to see if
dvander@189: 	 * there is a valid character C, the parent of 12+C must be 63.
dvander@189: 	 */
dvander@189: 	unsigned int parent;
dvander@189: 
dvander@189: 	void *value;			/* Value associated with this node */
dvander@189: 	NodeType mode;			/* Current usage type of the node */
dvander@189: 	bool valset;			/* Whether or not a value is set */
dvander@189: };
dvander@189: 
dvander@189: struct Trie
dvander@189: {
dvander@189: 	TrieNode *base;			/* Base array for the sparse tables */
dvander@189: 	char *stringtab;		/* String table pointer */
dvander@189: 	unsigned int baseSize;	/* Size of the base array, in members */
dvander@189: 	unsigned int stSize;	/* Size of the string table, in bytes */
dvander@189: 	unsigned int tail;		/* Current unused offset into the string table */
dvander@189: };
dvander@189: 
dvander@189: inline unsigned char charval(char c)
dvander@189: {
dvander@190: 	return (unsigned char)c;
dvander@189: }
dvander@189: 
dvander@193: bool sm_trie_grow(Trie *trie)
dvander@193: {
dvander@193: 	/* The current # of nodes in the tree is trie->baseSize + 1 */
dvander@193: 	unsigned int curSize = trie->baseSize;
dvander@193: 	unsigned int newSize = curSize * 2;
dvander@193: 
dvander@193: 	//:TODO: Make functions calling this return failure if this fails 
dvander@193: 	trie->base = (TrieNode *)realloc(trie->base, (newSize + 1) * sizeof(TrieNode));
dvander@193: 	if (!trie->base)
dvander@193: 	{
dvander@193: 		return false;
dvander@193: 	}
dvander@193: 
dvander@193: 	memset(&trie->base[curSize+1], 0, (newSize - curSize) * sizeof(TrieNode));
dvander@193: 	trie->baseSize = newSize;
dvander@193: 
dvander@193: 	return true;
dvander@193: }
dvander@193: 
dvander@193: unsigned int x_check(Trie *trie, char c, unsigned int start=1)
dvander@189: {
dvander@189: 	TrieNode *base = trie->base;
dvander@189: 	unsigned char _c = charval(c);
dvander@189: 	unsigned int to_check = trie->baseSize - _c;
dvander@193: 	for (unsigned int i=start; i<=to_check; i++)
dvander@189: 	{
dvander@189: 		if (base[i+_c].mode == Node_Unused)
dvander@189: 		{
dvander@189: 			return i;
dvander@189: 		}
dvander@189: 	}
dvander@189: 
dvander@193: 	sm_trie_grow(trie);
dvander@189: 
dvander@193: 	return x_check(trie, c, to_check+1);
dvander@189: }
dvander@189: 
dvander@193: unsigned int x_check2(Trie *trie, char c1, char c2, unsigned int start=1)
dvander@189: {
dvander@189: 	TrieNode *base = trie->base;
dvander@189: 	unsigned char _c1 = charval(c1);
dvander@189: 	unsigned char _c2 = charval(c2);
dvander@189: 	unsigned int to_check = trie->baseSize - (_c1 > _c2 ? _c1 : _c2);
dvander@193: 	for (unsigned int i=start; i<=to_check; i++)
dvander@189: 	{
dvander@189: 		if (base[i+_c1].mode == Node_Unused
dvander@189: 			&& base[i+_c2].mode == Node_Unused)
dvander@189: 		{
dvander@189: 			return i;
dvander@189: 		}
dvander@189: 	}
dvander@189: 
dvander@193: 	sm_trie_grow(trie);
dvander@189: 
dvander@193: 	return x_check2(trie, c1, c2, to_check+1);
dvander@189: }
dvander@189: 
dvander@189: unsigned int x_check_multi(Trie *trie, 
dvander@189: 							unsigned int offsets[], 
dvander@193: 							unsigned int count,
dvander@193: 							unsigned int start=1)
dvander@189: {
dvander@189: 	TrieNode *base = trie->base;
dvander@189: 	TrieNode *cur;
dvander@189: 	unsigned int to_check = trie->baseSize;
dvander@193: 	unsigned int highest = 0;
dvander@193: 
dvander@193: 	for (unsigned int i=0; i<count; i++)
dvander@193: 	{
dvander@193: 		if (offsets[i] > highest)
dvander@193: 		{
dvander@193: 			highest = offsets[i];
dvander@193: 		}
dvander@193: 	}
dvander@193: 
dvander@193: 	to_check -= highest;
dvander@193: 
dvander@193: 	for (unsigned int i=start; i<=to_check; i++)
dvander@189: 	{
dvander@189: 		bool okay = true;
dvander@189: 		for (unsigned int j=0; j<count; j++)
dvander@189: 		{
dvander@189: 			cur = &base[i+offsets[j]];
dvander@189: 			if (cur->mode != Node_Unused)
dvander@189: 			{
dvander@189: 				okay = false;
dvander@189: 				break;
dvander@189: 			}
dvander@189: 		}
dvander@189: 		if (okay)
dvander@189: 		{
dvander@189: 			return i;
dvander@189: 		}
dvander@189: 	}
dvander@189: 
dvander@193: 	sm_trie_grow(trie);
dvander@189: 
dvander@193: 	return x_check_multi(trie, offsets, count, to_check+1);
dvander@189: }
dvander@189: 
dvander@189: unsigned int x_addstring(Trie *trie, const char *ptr)
dvander@189: {
dvander@189: 	size_t len = strlen(ptr) + 1;
dvander@189: 
dvander@193: 	if (trie->tail + len >= trie->stSize)
dvander@189: 	{		
dvander@189: 		while (trie->tail + len >= trie->stSize)
dvander@189: 		{
dvander@189: 			trie->stSize *= 2;
dvander@189: 		}
dvander@189: 		trie->stringtab = (char *)realloc(trie->stringtab, trie->stSize);
dvander@189: 	}
dvander@189: 
dvander@189: 	unsigned int tail = trie->tail;
dvander@189: 	strcpy(&trie->stringtab[tail], ptr);
dvander@189: 	trie->tail += len;
dvander@189: 
dvander@189: 	return tail;
dvander@189: }
dvander@189: 
dvander@189: Trie *sm_trie_create()
dvander@189: {
dvander@189: 	Trie *t = new Trie;
dvander@189: 
dvander@193: 	t->base = (TrieNode *)malloc(sizeof(TrieNode) * (256 + 1));
dvander@189: 	t->stringtab = (char *)malloc(sizeof(char) * 256);
dvander@189: 	t->baseSize = 256;
dvander@189: 	t->stSize = 256;
dvander@335: 
dvander@335: 	sm_trie_clear(t);
dvander@335: 
dvander@335: 	return t;
dvander@335: }
dvander@189: 
dvander@335: void sm_trie_clear(Trie *trie)
dvander@335: {
dvander@335: 	trie->tail = 0;
dvander@335: 
dvander@335: 	memset(trie->base, 0, sizeof(TrieNode) * (trie->baseSize + 1));
dvander@335: 	memset(trie->stringtab, 0, sizeof(char) * trie->stSize);
dvander@189: 
dvander@189: 	/* Sentinel root node */
dvander@335: 	trie->base[1].idx = 1;
dvander@335: 	trie->base[1].mode = Node_Arc;
dvander@335: 	trie->base[1].parent = 1;
dvander@189: }
dvander@189: 
dvander@189: void sm_trie_destroy(Trie *trie)
dvander@189: {
dvander@189: 	free(trie->base);
dvander@189: 	free(trie->stringtab);
dvander@195: 	delete trie;
dvander@189: }
dvander@189: 
dvander@211: bool sm_trie_delete(Trie *trie, const char *key)
dvander@211: {
dvander@211: 	unsigned int lastidx = 1;		/* the last node index */
dvander@211: 	unsigned int curidx;			/* current node index */
dvander@211: 	const char *keyptr = key;		/* input stream at current token */
dvander@211: 	TrieNode *node = NULL;			/* current node being processed */
dvander@211: 	TrieNode *base = trie->base;
dvander@211: 
dvander@211: 	if (!*key)
dvander@211: 	{
dvander@211: 		return false;
dvander@211: 	}
dvander@211: 
dvander@211: 	/* Start traversing at the root node */
dvander@211: 	do
dvander@211: 	{
dvander@211: 		/* Find where the next character is, then advance */
dvander@211: 		curidx = base[lastidx].idx;
dvander@211: 		node = &base[curidx];
dvander@211: 		curidx += charval(*keyptr);
dvander@211: 		node = &base[curidx];
dvander@211: 		keyptr++;
dvander@211: 
dvander@211: 		/* Check if this slot is supposed to be empty or is a collision */
dvander@211: 		if ((curidx > trie->baseSize) || node->mode == Node_Unused || node->parent != lastidx)
dvander@211: 		{
dvander@211: 			return false;
dvander@211: 		} else if (node->mode == Node_Term) {
dvander@211: 			char *term = &trie->stringtab[node->idx];
dvander@211: 			if (strcmp(keyptr, term) == 0)
dvander@211: 			{
dvander@211: 				break;
dvander@211: 			}
dvander@211: 		}
dvander@211: 		lastidx = curidx;
dvander@211: 	} while (*keyptr != '\0');
dvander@211: 
dvander@211: 	assert(node != NULL);
dvander@211: 
dvander@211: 	if (!node->valset)
dvander@211: 	{
dvander@211: 		return false;
dvander@211: 	}
dvander@211: 
dvander@211: 	node->valset = false;
dvander@211: 	node->value = NULL;
dvander@211: 
dvander@211: 	return true;
dvander@211: }
dvander@211: 
dvander@189: bool sm_trie_retrieve(Trie *trie, const char *key, void **value)
dvander@189: {
dvander@189: 	unsigned int lastidx = 1;		/* the last node index */
dvander@189: 	unsigned int curidx;			/* current node index */
dvander@189: 	const char *keyptr = key;		/* input stream at current token */
dvander@189: 	TrieNode *node = NULL;			/* current node being processed */
dvander@189: 	TrieNode *base = trie->base;
dvander@189: 
dvander@189: 	if (!*key)
dvander@189: 	{
dvander@189: 		return false;
dvander@189: 	}
dvander@189: 
dvander@189: 	/* Start traversing at the root node */
dvander@189: 	do
dvander@189: 	{
dvander@189: 		/* Find where the next character is, then advance */
dvander@189: 		curidx = base[lastidx].idx;
dvander@189: 		node = &base[curidx];
dvander@189: 		curidx += charval(*keyptr);
dvander@189: 		node = &base[curidx];
dvander@189: 		keyptr++;
dvander@189: 
dvander@189: 		/* Check if this slot is supposed to be empty or is a collision */
dvander@194: 		if ((curidx > trie->baseSize) || node->mode == Node_Unused || node->parent != lastidx)
dvander@189: 		{
dvander@189: 			return false;
dvander@189: 		} else if (node->mode == Node_Term) {
dvander@189: 			char *term = &trie->stringtab[node->idx];
dvander@189: 			if (strcmp(keyptr, term) == 0)
dvander@189: 			{
dvander@189: 				break;
dvander@189: 			}
dvander@189: 		}
dvander@189: 		lastidx = curidx;
dvander@189: 	} while (*keyptr != '\0');
dvander@189: 
dvander@189: 	assert(node != NULL);
dvander@189: 
dvander@189: 	if (!node->valset)
dvander@189: 	{
dvander@189: 		return false;
dvander@189: 	}
dvander@189: 
dvander@189: 	if (value)
dvander@189: 	{
dvander@189: 		*value = node->value;
dvander@189: 	}
dvander@189: 
dvander@189: 	return true;
dvander@189: }
dvander@189: 
dvander@189: bool sm_trie_insert(Trie *trie, const char *key, void *value)
dvander@189: {
dvander@189: 	unsigned int lastidx = 1;		/* the last node index */
dvander@189: 	unsigned int curidx;			/* current node index */
dvander@189: 	const char *keyptr = key;		/* input stream at current token */
dvander@189: 	TrieNode *node = NULL;			/* current node being processed */
dvander@189: 	TrieNode *basenode = NULL;		/* current base node being processed */
dvander@189: 	unsigned int q;					/* temporary var for x_check results */
dvander@189: 	TrieNode *base = trie->base;
dvander@189: 	unsigned int curoffs;			/* current offset */
dvander@189: 
dvander@189: 	/* Do not handle empty strings for simplicity */
dvander@189: 	if (!*key)
dvander@189: 	{
dvander@189: 		return false;
dvander@189: 	}
dvander@189: 
dvander@189: 	/* Start traversing at the root node (1) */
dvander@189: 	do
dvander@189: 	{
dvander@189: 		/* Find where the next character is, then advance */
dvander@189: 		curidx = base[lastidx].idx;
dvander@189: 		basenode = &base[curidx];
dvander@189: 		curoffs = charval(*keyptr);
dvander@189: 		curidx += curoffs;
dvander@189: 		node = &base[curidx];
dvander@189: 		keyptr++;
dvander@189: 
dvander@189: 		/* Check if this slot is supposed to be empty.  If so, we need to handle CASES 1/2:
dvander@189: 		 * Insertion without collisions
dvander@189: 		 */
dvander@194: 		if ( (curidx > trie->baseSize) || (node->mode == Node_Unused) )
dvander@189: 		{
dvander@194: 			if (curidx > trie->baseSize)
dvander@194: 			{
dvander@194: 				if (!sm_trie_grow(trie))
dvander@194: 				{
dvander@194: 					return false;
dvander@194: 				}
dvander@194: 				node = &trie->base[curidx];
dvander@194: 			}
dvander@189: 			node->parent = lastidx;
dvander@194: 			if (*keyptr == '\0')
dvander@194: 			{
dvander@194: 				node->mode = Node_Arc;
dvander@194: 			} else {
dvander@194: 				node->idx = x_addstring(trie, keyptr);
dvander@194: 				node->mode = Node_Term;
dvander@194: 			}
dvander@189: 			node->valset = true;
dvander@189: 			node->value = value;
dvander@189: 			
dvander@189: 			return true;
dvander@189: 		} else if (node->parent != lastidx) {
dvander@189: 			/* Collision! We have to split up the tree here.  CASE 4:
dvander@189: 			 * Insertion when a new word is inserted with a collision.
dvander@189: 			 * NOTE: This is the hardest case to handle.  All below examples are based on:
dvander@189: 			 * BACHELOR, BADGE, inserting BABY.
dvander@189: 			 * The problematic production here is A -> B, where B is already being used.
dvander@189: 			 *
dvander@189: 			 * This process has to rotate one half of the 'A' arc.  We generate two lists:
dvander@189: 			 *  Outgoing Arcs - Anything leaving this 'A'
dvander@189: 			 *  Incoming Arcs - Anything going to this 'A'
dvander@189: 			 * Whichever list is smaller will be moved.  Note that this works because the intersection
dvander@189: 			 * affects both arc chains, and moving one will make the slot available to either.
dvander@189: 			 */
dvander@189: 			TrieNode *cur;
dvander@189: 			
dvander@189: 			/* Find every node arcing from the last node.
dvander@189: 			 * I.e. for BACHELOR, BADGE, BABY,
dvander@189: 			 * The arcs leaving A will be C and D, but our current node is B -> *.
dvander@189: 			 * Thus, we use the last index (A) to find the base for arcs leaving A.
dvander@189: 			 */
dvander@189: 			unsigned int outgoing_base = base[lastidx].idx;
dvander@189: 			unsigned int outgoing_list[256];
dvander@189: 			unsigned int outgoing_count = 0;	/* count the current index here */
dvander@189: 			cur = &base[outgoing_base] + 1;
dvander@194: 			unsigned int outgoing_limit = 255;
dvander@194: 
dvander@194: 			if (outgoing_base + outgoing_limit > trie->baseSize)
dvander@194: 			{
dvander@194: 				outgoing_limit = trie->baseSize - outgoing_base;
dvander@194: 			}
dvander@194: 
dvander@194: 			for (unsigned int i=1; i<=outgoing_limit; i++,cur++)
dvander@189: 			{
dvander@189: 				if (cur->mode == Node_Unused || cur->parent != lastidx)
dvander@189: 				{
dvander@189: 					continue;
dvander@189: 				}
dvander@189: 				outgoing_list[outgoing_count++] = i;
dvander@189: 			}
dvander@189: 			outgoing_list[outgoing_count++] = curidx - outgoing_base;
dvander@189: 
dvander@189: 			/* Now we need to find all the arcs leaving our parent...
dvander@189: 			 * Note: the inconsistency is the base of our parent.  
dvander@189: 			 */
dvander@193: 			assert(base[node->parent].mode == Node_Arc);
dvander@189: 			unsigned int incoming_list[256];
dvander@189: 			unsigned int incoming_base = base[node->parent].idx;
dvander@189: 			unsigned int incoming_count = 0;
dvander@194: 			unsigned int incoming_limit = 255;
dvander@189: 			cur = &base[incoming_base] + 1;
dvander@194: 
dvander@194: 			if (incoming_base + incoming_limit > trie->baseSize)
dvander@194: 			{
dvander@195: 				incoming_limit = trie->baseSize - incoming_base;
dvander@194: 			}
dvander@194: 
dvander@195: 			assert(incoming_limit > 0 && incoming_limit <= 255);
dvander@195: 
dvander@189: 			for (unsigned int i=1; i<=255; i++,cur++)
dvander@189: 			{
dvander@189: 				if (cur->mode == Node_Arc || cur->mode == Node_Term)
dvander@189: 				{
dvander@189: 					if (cur->parent == node->parent)
dvander@189: 					{
dvander@189: 						incoming_list[incoming_count++] = i;
dvander@189: 					}
dvander@189: 				}
dvander@189: 			}
dvander@189: 
dvander@189: 			if (incoming_count < outgoing_count + 1)
dvander@189: 			{
dvander@189: 				unsigned int q = x_check_multi(trie, incoming_list, incoming_count);
dvander@189: 
dvander@193: 				base = trie->base;
dvander@193: 				node = &base[curidx];
dvander@193: 
dvander@189: 				/* If we're incoming, we need to modify our parent */
dvander@193: 				base[node->parent].idx = q;
dvander@189: 
dvander@189: 				/* For each node in the "to move" list,
dvander@189: 				 * Relocate the node's info to the new position.
dvander@189: 				 */
dvander@189: 				unsigned int idx, newidx, oldidx;
dvander@189: 				for (unsigned int i=0; i<incoming_count; i++)
dvander@189: 				{
dvander@189: 					idx = incoming_list[i];
dvander@189: 					newidx = q + idx;
dvander@189: 					oldidx = incoming_base + idx;
dvander@189: 					if (oldidx == lastidx)
dvander@189: 					{
dvander@189: 						/* Important! Make sure we're not invalidating our sacred lastidx */
dvander@189: 						lastidx = newidx;
dvander@189: 					}
dvander@189: 					base[newidx] = base[oldidx];
dvander@194: 					assert(base[base[newidx].parent].mode == Node_Arc);
dvander@189: 					memset(&base[oldidx], 0, sizeof(TrieNode));
dvander@189: 					/* If we are not a terminator, we have children we must take care of */
dvander@189: 					if (base[newidx].mode == Node_Arc)
dvander@189: 					{
dvander@189: 						TrieNode *check_base = &base[base[newidx].idx] + 1;
dvander@193: 						for (unsigned int j=1; j<=255; j++, check_base++)
dvander@189: 						{
dvander@189: 							if (check_base->parent == oldidx)
dvander@189: 							{
dvander@189: 								check_base->parent = newidx;
dvander@189: 							}
dvander@189: 						}
dvander@189: 					}
dvander@189: 				}
dvander@189: 			} else {
dvander@189: 				unsigned int q = x_check_multi(trie, outgoing_list, outgoing_count);
dvander@189: 
dvander@193: 				base = trie->base;
dvander@193: 				node = &base[curidx];
dvander@193: 
dvander@189: 				/* If we're outgoing, we need to modify our own base */
dvander@189: 				base[lastidx].idx = q;
dvander@189: 
dvander@189: 				/* Take the last index (curidx) out of the list.  Technically we are not moving this,
dvander@189: 				 * since it's already being used by something else.  
dvander@189: 				 */
dvander@189: 				outgoing_count--;
dvander@189: 
dvander@189: 				/* For each node in the "to move" list,
dvander@189: 				 * Relocate the node's info to the new position.
dvander@189: 				 */
dvander@189: 				unsigned int idx, newidx, oldidx;
dvander@189: 				for (unsigned int i=0; i<outgoing_count; i++)
dvander@189: 				{
dvander@189: 					idx = outgoing_list[i];
dvander@189: 					newidx = q + idx;
dvander@189: 					oldidx = outgoing_base + idx;
dvander@189: 					if (oldidx == lastidx)
dvander@189: 					{
dvander@189: 						/* Important! Make sure we're not invalidating our sacred lastidx */
dvander@189: 						lastidx = newidx;
dvander@189: 					}
dvander@189: 					base[newidx] = base[oldidx];
dvander@194: 					assert(base[base[newidx].parent].mode == Node_Arc);
dvander@189: 					memset(&base[oldidx], 0, sizeof(TrieNode));
dvander@189: 					/* If we are not a terminator, we have children we must take care of */
dvander@189: 					if (base[newidx].mode == Node_Arc)
dvander@189: 					{
dvander@189: 						TrieNode *check_base = &base[base[newidx].idx] + 1;
dvander@193: 						for (unsigned int j=1; j<=255; j++, check_base++)
dvander@189: 						{
dvander@189: 							if (check_base->parent == oldidx)
dvander@189: 							{
dvander@189: 								check_base->parent = newidx;
dvander@189: 							}
dvander@189: 						}
dvander@189: 					}
dvander@189: 				}
dvander@189: 
dvander@189: 				/* Take the invisible node and use it as our new node */
dvander@189: 				node = &base[q + outgoing_list[outgoing_count]];
dvander@189: 			}
dvander@189: 
dvander@189: 			/* We're finally done! */
dvander@189: 			node->parent = lastidx;
dvander@189: 			if (*keyptr == '\0')
dvander@189: 			{
dvander@189: 				node->mode = Node_Arc;
dvander@189: 			} else {
dvander@189: 				node->idx = x_addstring(trie, keyptr);
dvander@189: 				node->mode = Node_Term;
dvander@189: 			}
dvander@189: 			node->valset = true;
dvander@189: 			node->value = value;
dvander@189: 
dvander@189: 			return true;
dvander@189: 		} else {
dvander@189: 			/* See what's in the next node - special case if terminator! */
dvander@189: 			if (node->mode == Node_Term)
dvander@189: 			{
dvander@189: 				/* If we're a terminator, we need to handle CASE 3:
dvander@189: 				 * Insertion when a terminating collision occurs
dvander@189: 				 */
dvander@189: 				char *term = &trie->stringtab[node->idx];
dvander@189: 				/* Do an initial browsing to make sure they're not the same string */
dvander@189: 				if (strcmp(keyptr, term) == 0)
dvander@189: 				{
faluco@216: 					if (!node->valset)
faluco@216: 					{
faluco@216: 						node->valset = true;
faluco@216: 						node->value = value;
faluco@216: 						return true;
faluco@216: 					}
dvander@189: 					/* Same string.  We can't insert. */
dvander@189: 					return false;
dvander@189: 				}
dvander@189: 				/* For each matching character pair, we need to disband the terminator.
dvander@189: 				 * This splits the similar prefix into a single arc path.
dvander@189: 				 * First, save the old values so we can move them to a new node.
dvander@189: 				 * Next, for each loop:
dvander@189: 				 *  Take the current (invalid) node, and point it to the next arc base.
dvander@189: 				 *  Set the current node to the node at the next arc.
dvander@189: 				 */
dvander@189: 				void *oldvalue = node->value;
dvander@189: 				bool oldvalset = node->valset;
dvander@193: 				if (*term == *keyptr)
dvander@193: 				{
dvander@189: 				while (*term == *keyptr)
dvander@189: 				{
dvander@189: 					/* Find the next free slot in the check array.
dvander@189: 					 * This is the "vector base" essentially
dvander@189: 					 */
dvander@189: 					q = x_check(trie, *term);
dvander@193: 						base = trie->base;
dvander@193: 						node = &base[curidx];
dvander@189: 					/* Point the node to the next new base */
dvander@189: 					node->idx = q;
dvander@189: 					node->mode = Node_Arc;
dvander@189: 					node->valset = false;
dvander@189: 					/* Advance the input stream and local variables */
dvander@189: 					lastidx = curidx;
dvander@189: 					curidx = q + charval(*term);
dvander@189: 					node = &base[curidx];
dvander@189: 					/* Make sure the new current node has its parent set. */
dvander@189: 					node->parent = lastidx;
dvander@189: 					node->mode = Node_Arc;	/* Just in case we run x_check again */
dvander@189: 					*term = '\0';	/* Unmark the string table here */
dvander@189: 					term++;
dvander@189: 					keyptr++;
dvander@193: 					}
dvander@193: 				} else {
dvander@193: 					node->valset = false;
dvander@189: 				}
dvander@189: 				/* We're done inserting new pairs.  If one of them is exhausted,
dvander@189: 				 * we take special shortcuts.
dvander@189: 				 */
dvander@189: 				if (*term == '\0')				//EX: BADGERHOUSE added over B -> ADGER.  
dvander@189: 				{
dvander@189: 					/* First backpatch the current node - it ends the newly split terminator.
dvander@189: 					 * In the example, this would mean the node is the production from R -> ?
dvander@189: 					 * This node ends the old BADGER, so we set it here.
dvander@189: 					 */
dvander@189: 					node->valset = oldvalset;
dvander@189: 					node->value = oldvalue;
dvander@189: 
dvander@189: 					/* The terminator was split up, but pieces of keyptr remain.
dvander@189: 					 * We need to generate a new production, in this example, R -> H,
dvander@189: 					 * with H being a terminator to OUSE.  Thus we get:
dvander@189: 					 * B,A,D,G,E,R*,H*->OUSE (* = value set).
dvander@189: 					 * NOTE: parent was last set at the end of the while loop.
dvander@189: 					 */
dvander@193: 					/* Get the new base and apply re-basing */
dvander@189: 					q = x_check(trie, *keyptr);
dvander@193: 					base = trie->base;
dvander@193: 					node = &base[curidx];
dvander@193: 
dvander@189: 					node->idx = q;
dvander@189: 					node->mode = Node_Arc;
dvander@189: 					lastidx = curidx;
dvander@189: 					/* Finish the final node */
dvander@189: 					curidx = q + charval(*keyptr);
dvander@189: 					node = &trie->base[curidx];
dvander@189: 					keyptr++;
dvander@189: 					/* Optimize - don't add to string table if there's nothing more to eat */
dvander@189: 					if (*keyptr == '\0')
dvander@189: 					{
dvander@189: 						node->mode = Node_Arc;
dvander@189: 					} else {
dvander@189: 						node->idx = x_addstring(trie, keyptr);
dvander@189: 						node->mode = Node_Term;
dvander@189: 					}
dvander@189: 					node->parent = lastidx;
dvander@189: 					node->valset = true;
dvander@189: 					node->value = value;
dvander@189: 				} else if (*keyptr == '\0') {	//EX: BADGER added over B -> ADGERHOUSE
dvander@189: 					/* First backpatch the current node - it ends newly split input string.
dvander@189: 					 * This is the exact opposite of the above procedure.
dvander@189: 					 */
dvander@189: 					node->valset = true;
dvander@189: 					node->value = value;
dvander@189: 
dvander@193: 					/* Get the new base and apply re-basing */
dvander@189: 					q = x_check(trie, *term);
dvander@193: 					base = trie->base;
dvander@193: 					node = &base[curidx];
dvander@193: 
dvander@189: 					node->idx = q;
dvander@189: 					node->mode = Node_Arc;
dvander@189: 					lastidx = curidx;
dvander@189: 					/* Finish the final node */
dvander@189: 					curidx = q + charval(*term);
dvander@189: 					node = &trie->base[curidx];
dvander@189: 					term++;
dvander@189: 					/* Optimize - don't add to string table if there's nothing more to eat */
dvander@189: 					if (*term == '\0')
dvander@189: 					{
dvander@189: 						node->mode = Node_Arc;
dvander@189: 					} else {
dvander@189: 						node->idx = (term - trie->stringtab); /* Already in the string table! */
dvander@189: 						node->mode = Node_Term;
dvander@189: 					}
dvander@189: 					node->parent = lastidx;
dvander@189: 					node->valset = oldvalset;
dvander@189: 					node->value = oldvalue;
dvander@189: 				} else {
dvander@189: 					/* Finally, we have to create two new nodes instead of just one. */
dvander@189: 					node->mode = Node_Arc;
dvander@189: 					
dvander@193: 					/* Get the new base and apply re-basing */
dvander@189: 					q = x_check2(trie, *keyptr, *term);
dvander@193: 					base = trie->base;
dvander@193: 					node = &base[curidx];
dvander@193: 
dvander@189: 					node->idx = q;
dvander@189: 					lastidx = curidx;
dvander@189: 
dvander@189: 					/* Re-create the old terminated node */
dvander@189: 					curidx = q + charval(*term);
dvander@189: 					node = &trie->base[curidx];
dvander@189: 					term++;
dvander@189: 					node->valset = oldvalset;
dvander@189: 					node->value = oldvalue;
dvander@189: 					node->parent = lastidx;
dvander@189: 					if (*term == '\0')
dvander@189: 					{
dvander@189: 						node->mode = Node_Arc;
dvander@189: 					} else {
dvander@189: 						node->mode = Node_Term;
dvander@189: 						node->idx = (term - trie->stringtab); /* Already in the string table! */
dvander@189: 					}
dvander@189: 
dvander@189: 					/* Create the new keyed input node */
dvander@189: 					curidx = q + charval(*keyptr);
dvander@189: 					node = &trie->base[curidx];
dvander@189: 					keyptr++;
dvander@189: 					node->valset = true;
dvander@189: 					node->value = value;
dvander@189: 					node->parent = lastidx;
dvander@189: 					if (*keyptr == '\0')
dvander@189: 					{
dvander@189: 						node->mode = Node_Arc;
dvander@189: 					} else {
dvander@189: 						node->mode = Node_Term;
dvander@189: 						node->idx = x_addstring(trie, keyptr);
dvander@189: 					}
dvander@189: 				}
dvander@189: 
dvander@189: 				/* Phew! */
dvander@189: 				return true;
dvander@195: 			} else {
dvander@195: 				assert(node->mode == Node_Arc);
dvander@189: 			}
dvander@189: 		}
dvander@189: 		lastidx = curidx;
dvander@189: 	} while (*keyptr != '\0');
dvander@189: 
dvander@189: 	assert(node);
dvander@189: 
dvander@189: 	/* If we've exhausted the string and we have a valid reached node,
dvander@189: 	 * the production rule already existed.  Make sure it's valid to set first.
dvander@189: 	 */
dvander@189: 
dvander@189: 	/* We have to be an Arc.  If the last result was anything else, we would have returned a new 
dvander@189: 	 * production earlier.
dvander@189: 	 */
dvander@189: 	assert(node->mode == Node_Arc);
dvander@189: 
dvander@189: 	if (!node->valset)
dvander@189: 	{
dvander@189: 		/* Insert is only possible if we have no production */
dvander@189: 		node->valset = true;
dvander@189: 		node->value = value;
dvander@189: 		return true;
dvander@189: 	}
dvander@189: 
dvander@189: 	return false;
dvander@189: }