Extrem-Speedup für den token-parser. Kleine Änderungen

src/common/kernel/building.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: building.c,v 1.6 2001/02/09 13:53:51 corwin Exp $
+ *	$Id: building.c,v 1.7 2001/02/15 02:41:46 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -747,8 +747,7 @@ findbuildingtype(const char * name, const locale * lang)
 		}
 		bnames = bn;
 	}
-	i = findtoken(&bn->names, name);
+	if (findtoken(&bn->names, name, &i)==E_TOK_NOMATCH) return NULL;
-	if (i==E_TOK_NOMATCH) return NULL;
 	return (const building_type*)i;
 }
 

src/common/kernel/curse.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: curse.c,v 1.4 2001/02/03 13:45:32 enno Exp $
+ *	$Id: curse.c,v 1.5 2001/02/15 02:41:46 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -166,7 +166,8 @@ ct_find(const char *c)
 /* TODO: findet nur curse_types, die auch in curse_data sind.
  * da fehlt noch ene registrierung wie für attrib_type
  */
-	int i = (int)findtoken(&cursenames, c);
+	int i;
+	if (findtoken(&cursenames, c, (void**)&i)==E_TOK_NOMATCH) return NULL;
 	if (i == -1 || cursedaten[i].name[0] == 0) {
 		cursetype_list * ctl = cursetypes;
 		while (ctl) {

src/common/kernel/eressea.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: eressea.c,v 1.14 2001/02/13 02:58:51 enno Exp $
+ *	$Id: eressea.c,v 1.15 2001/02/15 02:41:46 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -1109,16 +1109,20 @@ struct tnode cursenames;
 skill_t
 findskill(const char *s)
 {
-	return (skill_t)findtoken(&tokens[UT_SKILL], s)-1;
+	int i;
+	if (findtoken(&tokens[UT_SKILL], s, (void**)&i)==E_TOK_NOMATCH) return NOSKILL;
+	return (skill_t)i;
 }
 
 keyword_t
 findkeyword(const char *s)
 {
+	int i;
 #ifdef AT_PERSISTENT
 	if(*s == '@') s++;
 #endif
-	return (keyword_t) findtoken(&tokens[UT_KEYWORD], s)-1;
+	if (findtoken(&tokens[UT_KEYWORD], s, (void**)&i)==E_TOK_NOMATCH) return NOKEYWORD;
+	return (keyword_t) i;
 }
 
 keyword_t
@@ -1136,13 +1140,15 @@ getkeyword (void)
 param_t
 findparam(const char *s)
 {
-	int p;
 	const building_type * btype;
-	p = (int)findtoken(&tokens[UT_PARAM], s)-1;
+
-	if (p>=0) return (param_t)p;
+	int i;
-	btype = findbuildingtype(s, NULL);
+	if (findtoken(&tokens[UT_PARAM], s, (void**)&i)==E_TOK_NOMATCH) {
-	if (btype!=NULL) return (param_t) P_BUILDING;
+		btype = findbuildingtype(s, NULL);
-	return NOPARAM;
+		if (btype!=NULL) return (param_t) P_BUILDING;
+		return NOPARAM;
+	}
+	return (param_t)i;
 }
 
 param_t
@@ -1749,7 +1755,7 @@ use_birthdayamulet(region * r, unit * magician, strlist * cmdstrings)
 
 typedef struct t_umlaut {
 	const char *txt;
-	unsigned char id;
+	int id;
 	int typ;
 } t_umlaut;
 
@@ -1801,17 +1807,17 @@ init_tokens(void)
 {
 	int i;
 	for (i=0;i!=MAXPARAMS;++i)
-		addtoken(&tokens[UT_PARAM], parameters[i], (void*)(i+1));
+		addtoken(&tokens[UT_PARAM], parameters[i], (void*)i);
 	for (i=0;i!=MAXSKILLS;++i)
-		addtoken(&tokens[UT_SKILL], skillnames[i], (void*)(i+1));
+		addtoken(&tokens[UT_SKILL], skillnames[i], (void*)i);
 	for (i=0;i!=MAXKEYWORDS;++i)
-		addtoken(&tokens[UT_KEYWORD], keywords[i], (void*)(i+1));
+		addtoken(&tokens[UT_KEYWORD], keywords[i], (void*)i);
 	for (i=0;umlaut[i].txt;++i)
-		addtoken(&tokens[umlaut[i].typ], umlaut[i].txt, (void*)(umlaut[i].id+1));
+		addtoken(&tokens[umlaut[i].typ], umlaut[i].txt, (void*)umlaut[i].id);
 	for (i=0; spelldaten[i].id != SPL_NOSPELL; i++)
-		addtoken(&spellnames, spelldaten[i].name, (void*)(i+1));
+		addtoken(&spellnames, spelldaten[i].name, (void*)i);
 	for (i=0; i!=MAXCURSE; i++){
-		addtoken(&cursenames, cursedaten[i].name, (void*)(i+1));
+		addtoken(&cursenames, cursedaten[i].name, (void*)i);
 	}
 }
 
@@ -2096,44 +2102,46 @@ resolve2(void)
 
 #endif
 
+static void 
+init_directions(tnode * root)
+{
+	/* mit dieser routine kann man mehrere namen für eine direction geben,
+	 * das ist für die hexes ideal. */
+	const struct {
+		const char* name;
+		int direction;
+	} dirs [] = {
+		{ "no", D_NORTHEAST},
+		{ "nw", D_NORTHWEST},
+		{ "nordosten", D_NORTHEAST},
+		{ "nordwesten", D_NORTHWEST},
+		{ "so", D_SOUTHEAST},
+		{ "sw", D_SOUTHWEST},
+		{ "südosten", D_SOUTHEAST},
+		{ "südwesten", D_SOUTHWEST},
+		{ "osten", D_EAST },
+		{ "westen",D_WEST },
+		{ NULL, NODIRECTION}
+	};
+	int i;
+	for (i=0; dirs[i].direction!=NODIRECTION;++i) {
+		addtoken(root, dirs[i].name, (void*)dirs[i].direction);
+	}
+}
+
 direction_t
 finddirection(const char *s)
 {
-#define MAXDIRNAMES 12
+	static boolean init = false;
-	static const char* dirs[MAXDIRNAMES] = {
+	static tnode dirnames;
-		"no",
+	int dir;
-		"nw",
-		"nordosten",
-		"nordwesten",
-		"so",
-		"sw",
-		"suedosten",
-		"südosten",
-		"suedwesten",
-		"südwesten",
-		"osten",
-		"westen"
-		/* mit dieser routine kann man mehrere namen für eine direction geben,
-		 * das ist für die hexes ideal. */
-	};
-	static const direction_t dir_xref[MAXDIRNAMES] = {
-		D_NORTHEAST,
-		D_NORTHWEST,
-		D_NORTHEAST,
-		D_NORTHWEST,
-		D_SOUTHEAST,
-		D_SOUTHWEST,
-		D_SOUTHEAST,
-		D_SOUTHEAST,
-		D_SOUTHWEST,
-		D_SOUTHWEST,
-		D_EAST,
-		D_WEST
-	};
-	int uc = findstr(dirs, s, MAXDIRNAMES);
 
-	if (uc == -1) return NODIRECTION;
+	if (!init) init_directions(&dirnames);
-	return dir_xref[uc];
+
+	if (findtoken(&dirnames, s, (void**)&dir)==E_TOK_SUCCESS) {
+		return (direction_t)dir;
+	}
+	return NODIRECTION;
 }
 
 unit *

src/common/kernel/eressea.h

@@ -1105,7 +1105,7 @@ extern void guard(struct unit * u, unsigned int mask);
 typedef struct local_names {
 	struct local_names * next;
 	const struct locale * lang;
-	tnode names;
+	struct tnode names;
 } local_names;
 
 extern int maxworkingpeasants(const struct region * r);

src/common/kernel/item.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: item.c,v 1.11 2001/02/14 07:44:57 enno Exp $
+ *	$Id: item.c,v 1.12 2001/02/15 02:41:46 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -2031,8 +2031,8 @@ findresourcetype(const char * name, const locale * lang)
 		}
 		rnames = rn;
 	}
-	i = findtoken(&rn->names, name);
+	
-	if (i==E_TOK_NOMATCH) return NULL;
+	if (findtoken(&rn->names, name, &i)==E_TOK_NOMATCH) return NULL;
 	return (const resource_type*)i;
 }
 
@@ -2073,8 +2073,7 @@ finditemtype(const char * name, const locale * lang)
 		}
 		inames = in;
 	}
-	i = findtoken(&in->names, name);
+	if (findtoken(&in->names, name, &i)==E_TOK_NOMATCH) return NULL;
-	if (i==E_TOK_NOMATCH) return NULL;
 	return (const item_type*)i;
 }
 

src/common/kernel/magic.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: magic.c,v 1.7 2001/02/12 22:39:56 enno Exp $
+ *	$Id: magic.c,v 1.8 2001/02/15 02:41:46 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -567,8 +567,8 @@ find_spellbyname(unit *u, char *s)
 	spell *sp;
 	int i;
 
-  m = get_mage(u);
+	m = get_mage(u);
-  if (!m) {
+	if (!m) {
 		return (spell *) NULL;
 	}
 
@@ -579,8 +579,7 @@ find_spellbyname(unit *u, char *s)
 		}
 	}
 
-	i = (int)findtoken(&spellnames, s)-1;
+	if (findtoken(&spellnames, s, (void**)&i) && getspell(u, spelldaten[i].id)) return &spelldaten[i];
-	if (i>=0 && getspell(u, spelldaten[i].id)) return &spelldaten[i];
 
 	return (spell *) NULL;
 }

src/common/modules/gmcmd.c

@@ -169,8 +169,7 @@ gm_command(const char * cmd, struct unit * u)
 	i = min(16, c-cmd);
 	strncpy(zText, cmd, i);
 	zText[i]=0;
-	cm = (command*)findtoken(&g_keys, zText);
+	if (findtoken(&g_keys, zText, (void**)&cm) && cm->perform) cm->perform(++c, u);
-	if (cm && cm->perform) cm->perform(++c, u);
 }
 
 void

src/common/triggers/shock.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: shock.c,v 1.2 2001/01/26 16:19:41 enno Exp $
+ *	$Id: shock.c,v 1.3 2001/02/15 02:41:47 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -49,18 +49,13 @@ static void
 shock_write(const trigger * t, FILE * F)
 {
 	unit * u = (unit*)t->data.v;
-	fprintf(F, "%s ", itoa36(u->no));
+	write_unit_reference(u, F);
 }
 
 static int
 shock_read(trigger * t, FILE * F)
 {
-	char zId[10];
+	read_unit_reference((unit**)&t->data.v, F);
-	int i;
-	fscanf(F, "%s", zId);
-	i = atoi36(zId);
-	t->data.v = findunit(i);
-	if (t->data.v==NULL) ur_add((void*)i, &t->data.v, resolve_unit);
 	return 1;
 }
 

src/common/util/umlaut.c

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: umlaut.c,v 1.4 2001/02/13 02:58:51 enno Exp $
+ *	$Id: umlaut.c,v 1.5 2001/02/15 02:41:47 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -27,6 +27,15 @@
 #include <string.h>
 #include <assert.h>
 
+typedef struct tref {
+	struct tref * nexthash;
+	char c;
+	struct tnode * node;
+} tref;
+
+#define LEAF 1 /* leaf node for a word. always matches */
+#define SHARED 2 /* at least two words share the node */
+
 void
 addtoken(tnode * root, const char* str, void * id)
 {
@@ -44,28 +53,43 @@ addtoken(tnode * root, const char* str, void * id)
 		{'ß', "ss"},
 		{ 0, 0 }
 	};
-	assert(id!=E_TOK_NOMATCH);
-	if (root->id!=E_TOK_NOMATCH && root->id!=id && !root->leaf) root->id=E_TOK_NOMATCH;
 	if (!*str) {
 		root->id = id;
-		root->leaf=1;
+		root->flags |= LEAF;
 	} else {
-		tnode * tk;
+		tref * next;
 		int index, i = 0;
 		char c = *str;
 		if (c<'a' || c>'z') c = (char)tolower((unsigned char)c);
 		index = ((unsigned char)c) % 32;
-		tk = root->next[index];
+		next = root->next[index];
-		if (root->id==E_TOK_NOMATCH) root->id = id;
+		if (!root->flags) root->id = id;
-		while (tk && tk->c != c) tk = tk->nexthash;
+		while (next && next->c != c) next = next->nexthash;
-		if (!tk) {
+		if (!next) {
-			tk = calloc(1, sizeof(tnode));
+			tref * ref;
-			tk->id = E_TOK_NOMATCH;
+			char u = (char)toupper((unsigned char)c);
-			tk->c = c;
+			tnode * node = calloc(1, sizeof(tnode));
-			tk->nexthash=root->next[index];
+
-			root->next[index] = tk;
+			ref = malloc(sizeof(tref));
+			ref->c = c;
+			ref->node = node;
+			ref->nexthash=root->next[index];
+			root->next[index] = ref;
+			
+			if (u!=c) {
+				index = ((unsigned char)u) % 32;
+				ref = malloc(sizeof(tref));
+				ref->c = u;
+				ref->node = node;
+				ref->nexthash = root->next[index];
+				root->next[index] = ref;
+			}
+			next=ref;
+		} else {
+			next->node->flags |= SHARED;
+			if (next->node->flags & LEAF == 0) next->node->id = NULL;
 		}
-		addtoken(tk, str+1, id);
+		addtoken(next->node, str+1, id);
 		while (replace[i].str) {
 			if (*str==replace[i].c) {
 				strcat(strcpy(zText, replace[i].str), str+1);
@@ -77,22 +101,32 @@ addtoken(tnode * root, const char* str, void * id)
 	}
 }
 
-void *
+int
-findtoken(tnode * tk, const char * str)
+findtoken(tnode * tk, const char * str, void **result)
 {
-	if(*str == 0) return E_TOK_NOMATCH;
+	if (*str == 0) return E_TOK_NOMATCH;
 
 	while (*str) {
 		int index;
+		tref * ref;
 		char c = *str;
-		if (c<'a' || c>'z') c = (char)tolower((unsigned char)c);
+
+/*		if (c<'a' || c>'z') c = (char)tolower((unsigned char)c); */
 
 		index = ((unsigned char)c) % 32;
-		tk = tk->next[index];
+		ref = tk->next[index];
-		while (tk && tk->c!=c) tk = tk->nexthash;
+		while (ref && ref->c!=c) ref = ref->nexthash;
 		++str;
-		if (!tk) return E_TOK_NOMATCH;
+		if (!ref) return E_TOK_NOMATCH;
+		tk = ref->node;
+		if (tk && !(tk->flags & SHARED)) {
+			*result = tk->id;
+			return E_TOK_SUCCESS;
+		}
 	}
-	return tk->id;
+	if (tk && (tk->flags & LEAF)) {
+		*result = tk->id;
+		return E_TOK_SUCCESS;
+	}
+	return E_TOK_NOMATCH;
 }
-

src/common/util/umlaut.h

@@ -1,6 +1,6 @@
 /* vi: set ts=2:
  *
- *	$Id: umlaut.h,v 1.2 2001/01/26 16:19:41 enno Exp $
+ *	$Id: umlaut.h,v 1.3 2001/02/15 02:41:47 enno Exp $
  *	Eressea PB(E)M host Copyright (C) 1998-2000
  *      Christian Schlittchen (corwin@amber.kn-bremen.de)
  *      Katja Zedel (katze@felidae.kn-bremen.de)
@@ -14,18 +14,19 @@
 
 #ifndef _UMLAUT_H
 #define _UMLAUT_H
-typedef struct tnode tnode;
+
-struct tnode {
+#define E_TOK_NOMATCH (-1)
-	struct tnode * nexthash;
+#define E_TOK_SUCCESS 0
-	struct tnode * next[32];
+
+struct tref;
+
+typedef struct tnode {
+	struct tref * next[32];
+	unsigned char flags;
 	void * id;
-	char c;
+} tnode;
-	unsigned char leaf;
-};
 
-#define E_TOK_NOMATCH NULL
+int findtoken(struct tnode * tk, const char * str, void** result);
-
-void * findtoken(struct tnode * tk, const char * str);
 void addtoken(struct tnode * root, const char* str, void * id);
 
 #endif