interim commit. updating rex1.c

This commit is contained in:
hyung-hwan 2009-11-19 07:47:12 +00:00
parent 6315d1fa9b
commit 9d3084f455
3 changed files with 579 additions and 148 deletions

View File

@ -1,5 +1,5 @@
/* /*
* $Id: rex.h 300 2009-11-13 14:01:57Z hyunghwan.chung $ * $Id: rex.h 302 2009-11-18 13:47:12Z hyunghwan.chung $
* *
Copyright 2006-2009 Chung, Hyung-Hwan. Copyright 2006-2009 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -113,6 +113,8 @@ enum qse_rex_node_id_t
QSE_REX_NODE_START, QSE_REX_NODE_START,
QSE_REX_NODE_END, QSE_REX_NODE_END,
QSE_REX_NODE_NOP, QSE_REX_NODE_NOP,
QSE_REX_NODE_BOL, /* beginning of line */
QSE_REX_NODE_EOL, /* end of line */
QSE_REX_NODE_ANYCHAR, /* dot */ QSE_REX_NODE_ANYCHAR, /* dot */
QSE_REX_NODE_CHAR, /* single character */ QSE_REX_NODE_CHAR, /* single character */
QSE_REX_NODE_CHARSET, /* character set */ QSE_REX_NODE_CHARSET, /* character set */
@ -125,7 +127,7 @@ typedef enum qse_rex_node_id_t qse_rex_node_id_t;
typedef struct qse_rex_node_t qse_rex_node_t; typedef struct qse_rex_node_t qse_rex_node_t;
struct qse_rex_node_t struct qse_rex_node_t
{ {
qse_rex_node_t* link; /* link for management. not used for startnode */ qse_rex_node_t* link; /* for internal management. not used for startnode */
qse_rex_node_t* next; qse_rex_node_t* next;
qse_rex_node_id_t id; qse_rex_node_id_t id;

View File

@ -1,8 +1,21 @@
/* /*
* $Id$ * $Id$
*
Copyright 2006-2009 Chung, Hyung-Hwan.
This file is part of QSE.
{LICENSE HERE} QSE is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as
published by the Free Software Foundation, either version 3 of
the License, or (at your option) any later version.
QSE 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 Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with QSE. If not, see <http://www.gnu.org/licenses/>.
*/ */
#include <qse/cmn/rex.h> #include <qse/cmn/rex.h>
@ -10,7 +23,6 @@
#include <qse/cmn/lda.h> #include <qse/cmn/lda.h>
#include "mem.h" #include "mem.h"
#define GETC(c) do { if getc(c) <= -1) return -1; } while (0)
#define OCC_MAX QSE_TYPE_MAX(qse_size_t) #define OCC_MAX QSE_TYPE_MAX(qse_size_t)
struct qse_rex_t struct qse_rex_t
@ -27,11 +39,22 @@ struct comp_t
qse_rex_t* rex; qse_rex_t* rex;
qse_cstr_t re; qse_cstr_t re;
const qse_char_t* ptr; const qse_char_t* ptr;
const qse_char_t* end; const qse_char_t* end;
qse_cint_t c;
qse_size_t grouplvl;
struct
{
enum
{
CT_NORMAL,
CT_SPECIAL
} type;
qse_cint_t value;
int escaped;
} c;
qse_size_t gdepth; /* group depth */
qse_rex_node_t* start; qse_rex_node_t* start;
}; };
@ -40,12 +63,27 @@ struct exec_t
{ {
qse_rex_t* rex; qse_rex_t* rex;
qse_cstr_t str; struct
qse_cstr_t sub; {
const qse_char_t* ptr;
const qse_char_t* end;
} str;
qse_lda_t cand[2]; /* candidate arrays */ struct
int xxx, yyy; {
qse_size_t matched; const qse_char_t* ptr;
const qse_char_t* end;
} sub;
struct
{
int active;
int pending;
qse_lda_t set[2]; /* candidate arrays */
} cand;
qse_size_t nmatches;
const qse_char_t* matchend; /* 1 character past the match end */
}; };
typedef struct pair_t pair_t; typedef struct pair_t pair_t;
@ -66,9 +104,18 @@ struct group_t
typedef struct cand_t cand_t; typedef struct cand_t cand_t;
struct cand_t struct cand_t
{ {
qse_rex_node_t* node; qse_rex_node_t* node;
qse_size_t occ; qse_size_t occ;
group_t* group;
/* the stack of groups that this candidate belongs to.
* it is in the singliy linked list form */
group_t* group;
/* match pointer. the number of character advancement
* differs across various node types. BOL and EOL don't advance to
* the next character on match while ANYCHAR and CHAR do on match.
* therefore, the match pointer is managed per candidate basis. */
const qse_char_t* mptr;
}; };
qse_rex_t* qse_rex_open (qse_mmgr_t* mmgr, qse_size_t xtn, void* code) qse_rex_t* qse_rex_open (qse_mmgr_t* mmgr, qse_size_t xtn, void* code)
@ -211,12 +258,185 @@ static qse_rex_node_t* newbranchnode (
return n; return n;
} }
static int getc (comp_t* c) #define CHECK_END(builder) \
do { \
if (builder->ptr >= builder->ptn.end) \
{ \
builder->errnum = QSE_REX_EEND; \
return -1; \
} \
} while(0)
#define IS_HEX(c) \
((c >= QSE_T('0') && c <= QSE_T('9')) || \
(c >= QSE_T('A') && c <= QSE_T('F')) || \
(c >= QSE_T('a') && c <= QSE_T('f')))
#define HEX_TO_NUM(c) \
((c >= QSE_T('0') && c <= QSE_T('9'))? c-QSE_T('0'): \
(c >= QSE_T('A') && c <= QSE_T('F'))? c-QSE_T('A')+10: \
c-QSE_T('a')+10)
static int getc (comp_t* com)
{ {
c->c = (c->ptr < c->end)? *c->ptr++: QSE_CHAR_EOF; if (com->ptr >= com->end)
if (c->c == QSE_CHAR_EOF) {
com->c.type = CT_NORMAL;
com->c.value = QSE_CHAR_EOF;
com->c.escaped = 0;
return 0;
}
com->c.type = CT_NORMAL;
com->c.value = *com->ptr++;
com->c.escaped = QSE_FALSE;
if (com->c.value == QSE_T('\\'))
{
qse_char_t c;
CHECK_END (builder);
c = *com->ptr++;
if (c == QSE_T('n')) c = QSE_T('\n');
else if (c == QSE_T('r')) c = QSE_T('\r');
else if (c == QSE_T('t')) c = QSE_T('\t');
else if (c == QSE_T('f')) c = QSE_T('\f');
else if (c == QSE_T('b')) c = QSE_T('\b');
else if (c == QSE_T('v')) c = QSE_T('\v');
else if (c == QSE_T('a')) c = QSE_T('\a');
else if (c >= QSE_T('0') && c <= QSE_T('7'))
{
qse_char_t cx;
c = c - QSE_T('0');
CHECK_END (builder);
cx = *com->ptr++;
if (cx >= QSE_T('0') && cx <= QSE_T('7'))
{
c = c * 8 + cx - QSE_T('0');
CHECK_END (builder);
cx = *com->ptr++;
if (cx >= QSE_T('0') && cx <= QSE_T('7'))
{
c = c * 8 + cx - QSE_T('0');
}
}
}
else if (c == QSE_T('x'))
{
qse_char_t cx;
CHECK_END (builder);
cx = *com->ptr++;
if (IS_HEX(cx))
{
c = HEX_TO_NUM(cx);
CHECK_END (builder);
cx = *com->ptr++;
if (IS_HEX(cx))
{
c = c * 16 + HEX_TO_NUM(cx);
}
}
}
#ifdef QSE_CHAR_IS_WCHAR
else if (c == QSE_T('u') && QSE_SIZEOF(qse_char_t) >= 2)
{
qse_char_t cx;
CHECK_END (builder);
cx = *com->ptr++;
if (IS_HEX(cx))
{
qse_size_t i;
c = HEX_TO_NUM(cx);
for (i = 0; i < 3; i++)
{
CHECK_END (builder);
cx = *com->ptr++;
if (!IS_HEX(cx)) break;
c = c * 16 + HEX_TO_NUM(cx);
}
}
}
else if (c == QSE_T('U') && QSE_SIZEOF(qse_char_t) >= 4)
{
qse_char_t cx;
CHECK_END (builder);
cx = *com->ptr++;
if (IS_HEX(cx))
{
qse_size_t i;
c = HEX_TO_NUM(cx);
for (i = 0; i < 7; i++)
{
CHECK_END (builder);
cx = *com->ptr++;
if (!IS_HEX(cx)) break;
c = c * 16 + HEX_TO_NUM(cx);
}
}
}
#endif
com->c.value = c;
com->c.escaped = QSE_TRUE;
return 0;
}
else
{
if (level == LEVEL_TOP)
{
if (com->c.value == QSE_T('[') ||
com->c.value == QSE_T('|') ||
com->c.value == QSE_T('^') ||
com->c.value == QSE_T('$') ||
(!(com->option & QSE_REX_BUILD_NOBOUND) &&
com->c.value == QSE_T('{')) ||
com->c.value == QSE_T('+') ||
com->c.value == QSE_T('?') ||
com->c.value == QSE_T('*') ||
com->c.value == QSE_T('.') ||
com->c.value == QSE_T('(') ||
com->c.value == QSE_T(')'))
{
com->c.type = CT_SPECIAL;
}
}
else if (level == LEVEL_CHARSET)
{
if (com->c.value == QSE_T(']'))
{
com->c.type = CT_SPECIAL;
}
}
else if (level == LEVEL_RANGE)
{
if (com->c.value == QSE_T(',') ||
com->c.value == QSE_T('}'))
{
com->c.type = CT_SPECIAL;
}
}
}
#if 0
com->c = (com->ptr < com->end)? *com->ptr++: QSE_CHAR_EOF;
if (com->c == QSE_CHAR_EOF)
qse_printf (QSE_T("getc => <EOF>\n")); qse_printf (QSE_T("getc => <EOF>\n"));
else qse_printf (QSE_T("getc => %c\n"), c->c); else qse_printf (QSE_T("getc => %c\n"), com->c);
#endif
return 0; return 0;
} }
@ -226,7 +446,7 @@ static qse_rex_node_t* comp2 (comp_t* c)
{ {
qse_rex_node_t* n; qse_rex_node_t* n;
switch (c->c) switch (c->c.value)
{ {
case QSE_T('('): case QSE_T('('):
{ {
@ -249,7 +469,7 @@ static qse_rex_node_t* comp2 (comp_t* c)
return QSE_NULL; return QSE_NULL;
} }
c->grouplvl++; c->gdepth++;
x = comp0 (c, ge); x = comp0 (c, ge);
if (x == QSE_NULL) if (x == QSE_NULL)
{ {
@ -258,7 +478,7 @@ static qse_rex_node_t* comp2 (comp_t* c)
return QSE_NULL; return QSE_NULL;
} }
if (c->c != QSE_T(')')) if (c->c.value != QSE_T(')'))
{ {
qse_printf (QSE_T("expecting )\n")); qse_printf (QSE_T("expecting )\n"));
// UNBALANCED PAREN. // UNBALANCED PAREN.
@ -267,7 +487,7 @@ qse_printf (QSE_T("expecting )\n"));
return QSE_NULL; return QSE_NULL;
} }
c->grouplvl--; c->gdepth--;
if (getc(c) <= -1) if (getc(c) <= -1)
{ {
// freere (x); // freere (x);
@ -289,6 +509,27 @@ qse_printf (QSE_T("expecting )\n"));
} }
break; break;
case QSE_T('^'):
n = newnode (c, QSE_REX_NODE_BOL);
if (n == QSE_NULL) return QSE_NULL;
if (getc(c) <= -1)
{
// TODO: error handling..
return QSE_NULL;
}
break;
case QSE_T('$'):
n = newnode (c, QSE_REX_NODE_EOL);
if (n == QSE_NULL) return QSE_NULL;
if (getc(c) <= -1)
{
// TODO: error handling..
return QSE_NULL;
}
break;
/* /*
case QSE_T('['): case QSE_T('['):
.... ....
@ -296,7 +537,7 @@ qse_printf (QSE_T("expecting )\n"));
default: default:
/* normal character */ /* normal character */
n = newcharnode (c, c->c); n = newcharnode (c, c->c.value);
if (n == QSE_NULL) return QSE_NULL; if (n == QSE_NULL) return QSE_NULL;
if (getc(c) <= -1) if (getc(c) <= -1)
{ {
@ -363,8 +604,8 @@ static qse_rex_node_t* comp1 (comp_t* c, pair_t* pair)
pair->tail = pair->head; pair->tail = pair->head;
while (c->c != QSE_T('|') && c->c != QSE_CHAR_EOF && while (c->c.value != QSE_T('|') && c->c.value != QSE_CHAR_EOF &&
!(c->grouplvl >= 0 && c->c == QSE_T(')'))) !(c->gdepth >= 0 && c->c.value == QSE_T(')')))
{ {
qse_rex_node_t* tmp = comp2 (c); qse_rex_node_t* tmp = comp2 (c);
if (tmp == QSE_NULL) if (tmp == QSE_NULL)
@ -389,7 +630,7 @@ static qse_rex_node_t* comp0 (comp_t* c, qse_rex_node_t* ge)
if (left == QSE_NULL) return QSE_NULL; if (left == QSE_NULL) return QSE_NULL;
xpair.tail->next = ge; xpair.tail->next = ge;
while (c->c == QSE_T('|')) while (c->c.value == QSE_T('|'))
{ {
if (getc (c) <= -1) if (getc (c) <= -1)
{ {
@ -435,8 +676,10 @@ qse_rex_node_t* qse_rex_comp (
c.ptr = ptr; c.ptr = ptr;
c.end = ptr + len; c.end = ptr + len;
c.c = QSE_CHAR_EOF;
c.grouplvl = 0; c.c.value = QSE_CHAR_EOF;
c.gdepth = 0;
c.start = QSE_NULL; c.start = QSE_NULL;
if (getc(&c) <= -1) return QSE_NULL; if (getc(&c) <= -1) return QSE_NULL;
@ -454,11 +697,11 @@ qse_rex_node_t* qse_rex_comp (
else else
{ {
qse_rex_node_t* tmp; qse_rex_node_t* tmp;
//tmp = comp0 (&c, QSE_NULL); /*tmp = comp0 (&c, QSE_NULL);*/
tmp = comp0 (&c, end); tmp = comp0 (&c, end);
if (tmp == QSE_NULL) if (tmp == QSE_NULL)
{ {
//freenode (c.start, c.rex->mmgr); /*freenode (c.start, c.rex->mmgr);*/
freeallnodes (c.start); freeallnodes (c.start);
c.start = QSE_NULL; c.start = QSE_NULL;
} }
@ -474,29 +717,95 @@ qse_printf (QSE_T("start has tmp...\n"));
return rex->code; return rex->code;
} }
static group_t* pushgroup (exec_t* e, group_t* pg, qse_rex_node_t* gn) static group_t* dupgroups (exec_t* e, group_t* g)
{ {
group_t* g; group_t* yg, * xg = QSE_NULL;
QSE_ASSERT (gn->id == QSE_REX_NODE_GROUP);
g = (group_t*) QSE_MMGR_ALLOC (e->rex->mmgr, QSE_SIZEOF(*g)); QSE_ASSERT (g != QSE_NULL);
if (g == QSE_NULL)
if (g->next != QSE_NULL)
{
/* TODO: make it non recursive or
* implement stack overflow protection */
xg = dupgroups (e, g->next);
if (xg == QSE_NULL) return QSE_NULL;
}
yg = (group_t*) QSE_MMGR_ALLOC (e->rex->mmgr, QSE_SIZEOF(*g));
if (yg == QSE_NULL)
{
/* TODO: freegroups (xg); */
/* TODO: set error info */
return QSE_NULL;
}
QSE_MEMCPY (yg, g, QSE_SIZEOF(*yg));
yg->next = xg;
return yg;
}
static void freegroup (exec_t* e, group_t* group)
{
QSE_ASSERT (group != QSE_NULL);
QSE_MMGR_FREE (e->rex->mmgr, group);
}
static void freegroups (exec_t* e, group_t* group)
{
group_t* next;
while (group != QSE_NULL)
{
next = group->next;
freegroup (e, group);
group = next;
}
}
static group_t* pushgroup (exec_t* e, group_t* group, qse_rex_node_t* newgn)
{
group_t* newg;
QSE_ASSERT (newgn->id == QSE_REX_NODE_GROUP);
newg = (group_t*) QSE_MMGR_ALLOC (e->rex->mmgr, QSE_SIZEOF(*newg));
if (newg == QSE_NULL)
{ {
/* TODO: set error info */ /* TODO: set error info */
return QSE_NULL; return QSE_NULL;
} }
g->node = gn; newg->node = newgn;
g->occ = 0; newg->occ = 0;
g->next = pg; newg->next = group;
return g; return newg;
}
static group_t* pushgroupdup (exec_t* e, group_t* pg, qse_rex_node_t* gn)
{
group_t* gs = QSE_NULL;
/* duplicate the group stack if necessary */
if (pg != QSE_NULL)
{
gs = dupgroups (e, pg);
if (gs == QSE_NULL) return QSE_NULL;
}
/* and push a new group to the stack */
return pushgroup (e, gs, gn);
} }
static int addsimplecand ( static int addsimplecand (
exec_t* e, cand_t* pcand, qse_rex_node_t* node, qse_size_t occ) exec_t* e, group_t* group, qse_rex_node_t* node,
qse_size_t occ, const qse_char_t* mptr)
{ {
QSE_ASSERT ( QSE_ASSERT (
node->id == QSE_REX_NODE_BOL ||
node->id == QSE_REX_NODE_EOL ||
node->id == QSE_REX_NODE_ANYCHAR ||
node->id == QSE_REX_NODE_CHAR || node->id == QSE_REX_NODE_CHAR ||
node->id == QSE_REX_NODE_CHARSET node->id == QSE_REX_NODE_CHARSET
); );
@ -505,16 +814,17 @@ static int addsimplecand (
cand.node = node; cand.node = node;
cand.occ = occ; cand.occ = occ;
cand.group = pcand->group; cand.group = group;
cand.mptr = mptr;
if (node->id == QSE_REX_NODE_CHAR) /*if (node->id == QSE_REX_NODE_CHAR)
qse_printf (QSE_T("adding %d %c\n"), node->id, node->u.c); qse_printf (QSE_T("adding %d %c\n"), node->id, node->u.c);
else else
qse_printf (QSE_T("adding %d NA\n"), node->id); qse_printf (QSE_T("adding %d NA\n"), node->id);*/
if (qse_lda_insert ( if (qse_lda_insert (
&e->cand[e->xxx], &e->cand.set[e->cand.pending],
QSE_LDA_SIZE(&e->cand[e->xxx]), QSE_LDA_SIZE(&e->cand.set[e->cand.pending]),
&cand, 1) == (qse_size_t)-1) &cand, 1) == (qse_size_t)-1)
{ {
/* TODO: set error code: ENOERR */ /* TODO: set error code: ENOERR */
@ -524,116 +834,188 @@ qse_printf (QSE_T("adding %d NA\n"), node->id);
return 0; return 0;
} }
static int addnextcands (exec_t* e, group_t* group, qse_rex_node_t* cur) static int addcands (
exec_t* e, group_t* group, qse_rex_node_t* prevnode,
qse_rex_node_t* candnode, const qse_char_t* mptr)
{ {
/* skip all NOP nodes */ /* skip all NOP nodes */
while (cur && cur->id == QSE_REX_NODE_NOP) cur = cur->next; while (candnode != QSE_NULL && candnode->id == QSE_REX_NODE_NOP)
candnode = candnode->next;
/* nothing to add */ /* nothing to add */
if (cur == QSE_NULL) return 0; if (candnode == QSE_NULL) return 0;
if (cur->id == QSE_REX_NODE_END) if (candnode->id == QSE_REX_NODE_END)
{ {
qse_printf (QSE_T("== ADDING THE END(MATCH) NODE MEANING MATCH FOUND == \n")); qse_printf (QSE_T("== ADDING THE END(MATCH) NODE MEANING MATCH FOUND == \n"));
e->matched++; if (e->matchend == QSE_NULL || mptr >= e->matchend)
e->matchend = mptr;
e->nmatches++;
} }
else if (cur->id == QSE_REX_NODE_BRANCH) else if (candnode->id == QSE_REX_NODE_BRANCH)
{ {
#if 0 group_t* groupdup;
QSE_ASSERT (cur->next == QSE_NULL);
if (addnextcands (e, group, cur->u.b.left) <= -1) return -1; QSE_ASSERT (candnode->next == QSE_NULL);
if (addnextcands (e, group, cur->u.b.right) <= -1) return -1;
#endif groupdup = dupgroups (e, group);
if (groupdup == QSE_NULL) return -1;
if (addcands (e, group, prevnode, candnode->u.b.left, mptr) <= -1) return -1;
if (addcands (e, groupdup, prevnode, candnode->u.b.right, mptr) <= -1) return -1;
} }
else if (cur->id == QSE_REX_NODE_GROUP) else if (candnode->id == QSE_REX_NODE_GROUP)
{ {
group_t* g = pushgroup (e, group, cur); group_t* groupdup;
if (g == QSE_NULL) return -1;
/* add the first node in the group */ if (candnode->occ.min <= 0)
if (addnextcands (e, g, cur->u.g.head) <= -1) return -1;
if (cur->occ.min <= 0)
{ {
/* if the group node is optional, /* if the group node is optional,
* add the next node to the candidate array. * add the next node to the candidate array. */
* branch case => dup group */ if (addcands (e, group, prevnode, candnode->next, mptr) <= -1) return -1;
if (addnextcands (e, group, cur->next) <= -1) return -1;
} }
/* push the candnoderent group node (candnode) to the group
* stack duplicated. */
groupdup = pushgroupdup (e, group, candnode);
if (groupdup == QSE_NULL) return -1;
/* add the first node in the group */
if (addcands (e, groupdup, candnode, candnode->u.g.head, mptr) <= -1) return -1;
} }
else if (cur->id == QSE_REX_NODE_GROUPEND) else if (candnode->id == QSE_REX_NODE_GROUPEND)
{ {
group_t* group;
qse_rex_node_t* node; qse_rex_node_t* node;
qse_size_t occ;
group = cand->group; QSE_ASSERTX (group != QSE_NULL,
QSE_ASSERT (group != QSE_NULL); "GROUPEND reached must be paired up with a GROUP");
node = group->node; if (prevnode != candnode)
QSE_ASSERT (node == cur->u.ge.group); /*if (prevnode == QSE_NULL || prevnode->id != QSE_REX_NODE_GROUPEND)*/
if (group->occ < node->occ.max)
{ {
/* need to repeat itself */
group->occ++; group->occ++;
if (addnextcands (e, cand, node->u.g.head) <= -1) return -1;
}
if (group->occ >= node->occ.min) occ = group->occ;
{ node = group->node;
/* take the next atom as a candidate. QSE_ASSERT (node == candnode->u.ge.group);
* it is actually a branch case. */
if (occ >= node->occ.min)
cand = dupgrouppoppingtop (cand); {
group_t* gx = group->next;
if (addnextcands (e, pg, node->next) <= -1) return -1;
/* take the next atom as a candidate.
* it is actually a branch case. move on. */
if (occ < node->occ.max)
{
/* check if the group will be repeated.
* if so, duplicate the group stack excluding
* the top. it goes along a different path and
* hence requires a duplicated group stack. */
if (group->next != QSE_NULL)
{
gx = dupgroups (e, group->next);
if (gx == QSE_NULL) return -1;
}
}
if (addcands (e, gx, candnode, node->next, mptr) <= -1) return -1;
}
if (occ < node->occ.max)
{
/* need to repeat itself. */
if (addcands (e, group, candnode, node->u.g.head, mptr) <= -1) return -1;
}
} }
} }
else else
{ {
if (addsimplecand (e, cand, cur, 1) <= -1) return -1; group_t* gx = group;
if (cur->occ.min <= 0)
if (candnode->occ.min <= 0)
{ {
/* if the node is optional, /* if the node is optional,
* add the next node to the candidate array */ * add the next node to the candidate array */
if (addnextcands (e, pg, cur->next) <= -1) return -1; if (addcands (e, group, prevnode, candnode->next, mptr) <= -1) return -1;
if (group != QSE_NULL)
{
gx = dupgroups (e, group);
if (gx == QSE_NULL) return -1;
}
} }
if (addsimplecand (e, gx, candnode, 1, mptr) <= -1) return -1;
} }
return 0; return 0;
} }
static int match (exec_t* e, const qse_char_t* curp) static int match (exec_t* e)
{ {
qse_size_t i; qse_size_t i;
qse_char_t curc = *curp;
for (i = 0; i < QSE_LDA_SIZE(&e->cand[e->yyy]); i++) QSE_ASSERT (QSE_LDA_SIZE(&e->cand.set[e->cand.active]) > 0);
for (i = 0; i < QSE_LDA_SIZE(&e->cand.set[e->cand.active]); i++)
{ {
cand_t* cand = QSE_LDA_DPTR(&e->cand[e->yyy],i); cand_t* cand = QSE_LDA_DPTR(&e->cand.set[e->cand.active],i);
qse_rex_node_t* node = cand->node; qse_rex_node_t* node = cand->node;
const qse_char_t* nmptr = QSE_NULL;
if (node->id == QSE_REX_NODE_CHAR) switch (node->id)
{ {
if (node->u.c == curc) case QSE_REX_NODE_BOL:
{ if (cand->mptr == e->str.ptr) nmptr = cand->mptr;
qse_printf (QSE_T("matched %c\n"), node->u.c); break;
case QSE_REX_NODE_EOL:
if (cand->mptr >= e->str.end) nmptr = cand->mptr;
break;
case QSE_REX_NODE_ANYCHAR:
if (cand->mptr < e->sub.end) nmptr = cand->mptr + 1;
break;
case QSE_REX_NODE_CHAR:
if (cand->mptr < e->sub.end && node->u.c == *cand->mptr) nmptr = cand->mptr + 1;
//qse_printf (QSE_T("matched %c\n"), node->u.c);
break;
case QSE_REX_NODE_CHARSET:
qse_printf (QSE_T("charset not implemented...\n"));
break;
default:
// TODO: set error code -> internal error. this should not happen
return -1;
}
if (nmptr != QSE_NULL)
{
if (cand->occ >= node->occ.min)
{
group_t* gx = cand->group;
if (cand->occ < node->occ.max) if (cand->occ < node->occ.max)
{ {
if (addsimplecand (e, cand, node, cand->occ+1) <= -1) return -1; if (cand->group != QSE_NULL)
} {
if (cand->occ >= node->occ.min) gx = dupgroups (e, cand->group);
{ if (gx == QSE_NULL) return -1;
}
if (addnextcands (e, cand, node->next) <= -1) return -1;
} }
/* move on to the next candidate */
if (addcands (e, gx, node, node->next, nmptr) <= -1) return -1;
}
if (cand->occ < node->occ.max)
{
/* repeat itself more */
if (addsimplecand (e, cand->group, node, cand->occ+1, nmptr) <= -1) return -1;
} }
}
else
{
QSE_ASSERT (node->id == QSE_REX_NODE_CHARSET);
qse_printf (QSE_T("charset not implemented...\n"));
} }
} }
@ -642,73 +1024,110 @@ static int match (exec_t* e, const qse_char_t* curp)
static int exec (exec_t* e) static int exec (exec_t* e)
{ {
const qse_char_t* ptr = e->sub.ptr; int n;
const qse_char_t* end = e->sub.ptr + e->sub.len;
e->matched = 0; e->nmatches = 0;
e->xxx = 0; e->matchend = QSE_NULL;
e->yyy = 1;
/* collect the initial candidates to cand[xxx] */ e->cand.pending = 0;
qse_lda_clear (&e->cand[e->xxx]); e->cand.active = 1;
if (addnextcands (e, QSE_NULL, e->rex->code->next) <= -1) return -1; /* empty the pending set to collect the initial candidates */
qse_lda_clear (&e->cand.set[e->cand.pending]);
while (ptr < end) /* the first node must be the START node */
QSE_ASSERT (e->rex->code->id == QSE_REX_NODE_START);
/* addcands() collects a set of candidates into the pending set */
n = addcands (
e, /* execution structure */
QSE_NULL, /* doesn't belong to any groups yet */
e->rex->code, /* dummy previous node, the start node */
e->rex->code->next, /* start from the second node */
e->sub.ptr /* current match pointer */
);
if (n <= -1) return -1;
do
{ {
/* kind of swap cand[xxx] and cand[yyy] by swapping indices */ /* kind of swap the next set and the current set by swapping indices */
int tmp = e->xxx; int tmp = e->cand.pending;
e->xxx = e->yyy; e->cand.pending = e->cand.active;
e->yyy = tmp; e->cand.active = tmp;
/* check if there are any next candidates */ /* check if there are any next candidates */
if (QSE_LDA_SIZE(&e->cand[e->yyy]) <= 0) if (QSE_LDA_SIZE(&e->cand.set[e->cand.active]) <= 0)
{ {
/* if none, break */ /* if no more candidates, break */
break; break;
} }
{
int i;
qse_printf (QSE_T("SET="));
for (i = 0; i < QSE_LDA_SIZE(&e->cand.set[e->cand.active]); i++)
{
cand_t* cand = QSE_LDA_DPTR(&e->cand.set[e->cand.active],i);
qse_rex_node_t* node = cand->node;
if (node->id == QSE_REX_NODE_CHAR)
qse_printf (QSE_T("%c "), node->u.c);
else if (node->id == QSE_REX_NODE_ANYCHAR)
qse_printf (QSE_T(". "), node->u.c);
else if (node->id == QSE_REX_NODE_BOL)
qse_printf (QSE_T("^ "));
else if (node->id == QSE_REX_NODE_EOL)
qse_printf (QSE_T("$ "));
}
qse_printf (QSE_T("\n"));
}
/* clear the array to hold the next candidates */ /* clear the array to hold the next candidates */
qse_lda_clear (&e->cand[e->xxx]); qse_lda_clear (&e->cand.set[e->cand.pending]);
qse_printf (QSE_T("MATCHING %c\n"), *ptr); if (match (e) <= -1) return -1;
if (match (e, ptr) <= -1) return -1;
ptr++;
} }
while (1);
qse_printf (QSE_T("TOTAL MATCHES FOUND... %d\n"), e->matched); if (e->nmatches > 0)
{
qse_printf (QSE_T("MATCH: %d [%.*s]\n"),
(int)(e->matchend - e->sub.ptr),
(int)(e->matchend - e->sub.ptr), e->sub.ptr);
}
qse_printf (QSE_T("TOTAL MATCHES FOUND... %d\n"), e->nmatches);
return 0; return 0;
} }
static int init_exec_dds (exec_t* e, qse_mmgr_t* mmgr) static int init_exec_dds (exec_t* e, qse_mmgr_t* mmgr)
{ {
/* initializes dynamic data structures */ /* initializes dynamic data structures */
if (qse_lda_init (&e->cand[0], mmgr, 100) == QSE_NULL) if (qse_lda_init (&e->cand.set[0], mmgr, 100) == QSE_NULL)
{ {
/* TOOD: set error */ /* TOOD: set error */
return -1; return -1;
} }
if (qse_lda_init (&e->cand[1], mmgr, 100) == QSE_NULL) if (qse_lda_init (&e->cand.set[1], mmgr, 100) == QSE_NULL)
{ {
/* TOOD: set error */ /* TOOD: set error */
qse_lda_fini (&e->cand[0]); qse_lda_fini (&e->cand.set[0]);
return -1; return -1;
} }
qse_lda_setscale (&e->cand[0], QSE_SIZEOF(cand_t)); qse_lda_setscale (&e->cand.set[0], QSE_SIZEOF(cand_t));
qse_lda_setscale (&e->cand[1], QSE_SIZEOF(cand_t)); qse_lda_setscale (&e->cand.set[1], QSE_SIZEOF(cand_t));
qse_lda_setcopier (&e->cand[0], QSE_LDA_COPIER_INLINE); qse_lda_setcopier (&e->cand.set[0], QSE_LDA_COPIER_INLINE);
qse_lda_setcopier (&e->cand[1], QSE_LDA_COPIER_INLINE); qse_lda_setcopier (&e->cand.set[1], QSE_LDA_COPIER_INLINE);
return 0; return 0;
} }
static void fini_exec_dds (exec_t* e) static void fini_exec_dds (exec_t* e)
{ {
qse_lda_fini (&e->cand[1]); qse_lda_fini (&e->cand.set[1]);
qse_lda_fini (&e->cand[0]); qse_lda_fini (&e->cand.set[0]);
} }
int qse_rex_exec (qse_rex_t* rex, int qse_rex_exec (qse_rex_t* rex,
@ -727,14 +1146,13 @@ int qse_rex_exec (qse_rex_t* rex,
QSE_MEMSET (&e, 0, QSE_SIZEOF(e)); QSE_MEMSET (&e, 0, QSE_SIZEOF(e));
e.rex = rex; e.rex = rex;
e.str.ptr = str; e.str.ptr = str;
e.str.len = len; e.str.end = str + len;
e.sub.ptr = substr; e.sub.ptr = substr;
e.sub.len = sublen; e.sub.end = substr + sublen;
if (init_exec_dds (&e, rex->mmgr) <= -1) return -1; if (init_exec_dds (&e, rex->mmgr) <= -1) return -1;
// TOOD: may have to execute exec in case sublen is 0. while (e.sub.ptr <= e.sub.end)
while (e.sub.len > 0)
{ {
n = exec (&e); n = exec (&e);
if (n <= -1) if (n <= -1)
@ -743,10 +1161,9 @@ int qse_rex_exec (qse_rex_t* rex,
break; break;
} }
if (e.matched > 0) break; if (e.nmatches > 0) break;
e.sub.ptr++; e.sub.ptr++;
e.sub.len--;
} }
fini_exec_dds (&e); fini_exec_dds (&e);

View File

@ -248,6 +248,18 @@ void MyFrame::drawNode (wxDC& dc, qse_rex_node_t* n)
{ {
dc.DrawText (_T("<BR>"), nodex, nodey); dc.DrawText (_T("<BR>"), nodex, nodey);
} }
else if (n->id == QSE_REX_NODE_BOL)
{
dc.DrawText (_T("<^>"), nodex, nodey);
}
else if (n->id == QSE_REX_NODE_EOL)
{
dc.DrawText (_T("<$>"), nodex, nodey);
}
else if (n->id == QSE_REX_NODE_ANYCHAR)
{
dc.DrawText (_T("<AY>"), nodex, nodey);
}
else if (n->id == QSE_REX_NODE_CHAR) else if (n->id == QSE_REX_NODE_CHAR)
{ {
qse_char_t x[2]; qse_char_t x[2];
@ -274,7 +286,7 @@ void MyFrame::drawNode (wxDC& dc, qse_rex_node_t* n)
} }
else if (n->id == QSE_REX_NODE_NOP) else if (n->id == QSE_REX_NODE_NOP)
{ {
dc.DrawText (_T("<NOP>"), nodex, nodey); dc.DrawText (_T("<NP>"), nodex, nodey);
} }
} }
@ -287,19 +299,19 @@ void MyFrame::drawChain (wxDC& dc, qse_rex_node_t* n)
if (t->id == QSE_REX_NODE_BRANCH) if (t->id == QSE_REX_NODE_BRANCH)
{ {
drawNode (dc, t); drawNode (dc, t);
nodex += 50; nodex += 40;
int oldx = nodex; int oldx = nodex;
drawChain (dc, t->u.b.left); drawChain (dc, t->u.b.left);
nodex = oldx; nodex = oldx;
nodey += 50; nodey += 40;
drawChain (dc, t->u.b.right); drawChain (dc, t->u.b.right);
} }
else else
{ {
drawNode (dc, t); drawNode (dc, t);
nodex += 50; nodex += 40;
} }
if (t->id == QSE_REX_NODE_GROUP) if (t->id == QSE_REX_NODE_GROUP)