enhanced qse_sed_t

- changed qse_sed_t to support arbitrary numbers of commands
- added qse_set_getmaxdepth() and qse_sed_setmaxdepth()
- added relevant wrapper functions to QSE::Sed
This commit is contained in:
hyung-hwan 2009-05-30 23:29:06 +00:00
parent 7875def835
commit d851798b3c
5 changed files with 315 additions and 165 deletions

View File

@ -48,6 +48,8 @@ public:
typedef qse_sed_io_arg_t io_arg_t; typedef qse_sed_io_arg_t io_arg_t;
/// The option_t type redefines an option type /// The option_t type redefines an option type
typedef qse_sed_option_t option_t; typedef qse_sed_option_t option_t;
/// The depth_t type redefines an depth IDs
typedef qse_sed_depth_t depth_t;
/** /**
* The Sed() function creates an uninitialized stream editor. * The Sed() function creates an uninitialized stream editor.
@ -115,6 +117,19 @@ public:
int opt ///< option code int opt ///< option code
) throw (); ) throw ();
/**
* The getMaxDepth() function gets the maximum processing depth.
*/
size_t getMaxDepth (depth_t id) const throw ();
/**
* The setMaxDepth() function gets the maximum processing depth.
*/
void setMaxDepth (
int ids, ///< 0 or a number OR'ed of depth_t values
size_t depth ///< 0 maximum depth
) throw ();
/** /**
* The getErrorMessage() function gets the description of the last * The getErrorMessage() function gets the description of the last
* error occurred. It returns an empty string if the stream editor * error occurred. It returns an empty string if the stream editor

View File

@ -35,10 +35,6 @@
* qse_sed_close (sed); * qse_sed_close (sed);
* @endcode * @endcode
* *
* @todo
* - allow flexible numbers of commands
* - allow configurable recursion depth for a regular expression
*
* @example sed01.c * @example sed01.c
* This example shows how to embed a basic stream editor. * This example shows how to embed a basic stream editor.
*/ */
@ -123,6 +119,18 @@ enum qse_sed_option_t
}; };
typedef enum qse_sed_option_t qse_sed_option_t; typedef enum qse_sed_option_t qse_sed_option_t;
/**
* The qse_sed_depth_t type defines IDs for qse_sed_getmaxdepth() and
* qse_sed_setmaxdepth().
*/
enum qse_sed_depth_t
{
QSE_SED_DEPTH_REX_BUILD = (1 << 0),
QSE_SED_DEPTH_REX_MATCH = (1 << 1)
};
typedef enum qse_sed_depth_t qse_sed_depth_t;
/** /**
* The qse_sed_io_cmd_t type defines IO command codes. The code indicates * The qse_sed_io_cmd_t type defines IO command codes. The code indicates
* the action to take in an IO handler. * the action to take in an IO handler.
@ -219,6 +227,23 @@ void qse_sed_setoption (
int opt /**< 0 or a number OR'ed of qse_sed_option_t values */ int opt /**< 0 or a number OR'ed of qse_sed_option_t values */
); );
/**
* The qse_sed_getmaxdepth() gets the maximum processing depth.
*/
qse_size_t qse_sed_getmaxdepth (
qse_sed_t* sed, /**< a stream editor */
qse_sed_depth_t id /**< one of qse_sed_depth_t values */
);
/**
* The qse_sed_setmaxdepth() sets the maximum processing depth.
*/
void qse_sed_setmaxdepth (
qse_sed_t* sed, /**< a stream editor */
int ids, /**< 0 or a number OR'ed of qse_sed_depth_t values */
qse_size_t depth /**< maximum depth level */
);
/** /**
* The qse_sed_geterrstr() gets an error string getter. * The qse_sed_geterrstr() gets an error string getter.
*/ */

View File

@ -74,6 +74,18 @@ void Sed::setOption (int opt) throw ()
qse_sed_setoption (sed, opt); qse_sed_setoption (sed, opt);
} }
Sed::size_t Sed::getMaxDepth (depth_t id) const throw ()
{
QSE_ASSERT (sed != QSE_NULL);
return qse_sed_getmaxdepth (sed, id);
}
void Sed::setMaxDepth (int ids, size_t depth) throw ()
{
QSE_ASSERT (sed != QSE_NULL);
qse_sed_setmaxdepth (sed, ids, depth);
}
const Sed::char_t* Sed::getErrorMessage () const throw () const Sed::char_t* Sed::getErrorMessage () const throw ()
{ {
return (sed == QSE_NULL)? QSE_T(""): qse_sed_geterrmsg (sed); return (sed == QSE_NULL)? QSE_T(""): qse_sed_geterrmsg (sed);

View File

@ -24,6 +24,8 @@
QSE_IMPLEMENT_COMMON_FUNCTIONS (sed) QSE_IMPLEMENT_COMMON_FUNCTIONS (sed)
static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd); static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd);
static void free_all_command_blocks (qse_sed_t* sed);
static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr); static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr);
static void qse_sed_fini (qse_sed_t* sed); static void qse_sed_fini (qse_sed_t* sed);
@ -137,27 +139,8 @@ static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
qse_map_setcopier (&sed->tmp.labs, QSE_MAP_KEY, QSE_MAP_COPIER_INLINE); qse_map_setcopier (&sed->tmp.labs, QSE_MAP_KEY, QSE_MAP_COPIER_INLINE);
qse_map_setscale (&sed->tmp.labs, QSE_MAP_KEY, QSE_SIZEOF(qse_char_t)); qse_map_setscale (&sed->tmp.labs, QSE_MAP_KEY, QSE_SIZEOF(qse_char_t));
sed->cmd.len = 256; /* TODO: parameterize this value */
sed->cmd.buf = QSE_MMGR_ALLOC (
sed->mmgr, QSE_SIZEOF(qse_sed_cmd_t) * sed->cmd.len);
if (sed->cmd.buf == QSE_NULL)
{
qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex);
return QSE_NULL;
}
sed->cmd.cur = sed->cmd.buf;
sed->cmd.end = sed->cmd.buf + sed->cmd.len - 1;
#if 0
sed->cmd.lb = &sed->cmd.fb; /* on init, the last points to the first */
sed->fb.len = 0; /* the block has no data yet */
#endif
if (qse_lda_init (&sed->e.txt.appended, mmgr, 32) == QSE_NULL) if (qse_lda_init (&sed->e.txt.appended, mmgr, 32) == QSE_NULL)
{ {
QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf);
qse_map_fini (&sed->tmp.labs); qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab); qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex); qse_str_fini (&sed->tmp.rex);
@ -167,7 +150,6 @@ static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
if (qse_str_init (&sed->e.txt.read, mmgr, 256) == QSE_NULL) if (qse_str_init (&sed->e.txt.read, mmgr, 256) == QSE_NULL)
{ {
qse_lda_fini (&sed->e.txt.appended); qse_lda_fini (&sed->e.txt.appended);
QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf);
qse_map_fini (&sed->tmp.labs); qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab); qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex); qse_str_fini (&sed->tmp.rex);
@ -178,7 +160,6 @@ static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
{ {
qse_str_fini (&sed->e.txt.read); qse_str_fini (&sed->e.txt.read);
qse_lda_fini (&sed->e.txt.appended); qse_lda_fini (&sed->e.txt.appended);
QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf);
qse_map_fini (&sed->tmp.labs); qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab); qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex); qse_str_fini (&sed->tmp.rex);
@ -190,29 +171,30 @@ static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
qse_str_fini (&sed->e.txt.held); qse_str_fini (&sed->e.txt.held);
qse_str_fini (&sed->e.txt.read); qse_str_fini (&sed->e.txt.read);
qse_lda_fini (&sed->e.txt.appended); qse_lda_fini (&sed->e.txt.appended);
QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf);
qse_map_fini (&sed->tmp.labs); qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab); qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex); qse_str_fini (&sed->tmp.rex);
return QSE_NULL; return QSE_NULL;
} }
/* on init, the last points to the first */
sed->cmd.lb = &sed->cmd.fb;
/* the block has no data yet */
sed->cmd.fb.len = 0;
return sed; return sed;
} }
static void qse_sed_fini (qse_sed_t* sed) static void qse_sed_fini (qse_sed_t* sed)
{ {
qse_sed_cmd_t* c; free_all_command_blocks (sed);
qse_str_fini (&sed->e.txt.subst); qse_str_fini (&sed->e.txt.subst);
qse_str_fini (&sed->e.txt.held); qse_str_fini (&sed->e.txt.held);
qse_str_fini (&sed->e.txt.read); qse_str_fini (&sed->e.txt.read);
qse_lda_fini (&sed->e.txt.appended); qse_lda_fini (&sed->e.txt.appended);
/* TODO: use different data structure -> look at qse_sed_init */
for (c = sed->cmd.buf; c != sed->cmd.cur; c++) free_command (sed, c);
QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf);
qse_map_fini (&sed->tmp.labs); qse_map_fini (&sed->tmp.labs);
qse_str_fini (&sed->tmp.lab); qse_str_fini (&sed->tmp.lab);
qse_str_fini (&sed->tmp.rex); qse_str_fini (&sed->tmp.rex);
@ -228,6 +210,18 @@ int qse_sed_getoption (qse_sed_t* sed)
return sed->option; return sed->option;
} }
qse_size_t qse_sed_getmaxdepth (qse_sed_t* sed, qse_sed_depth_t id)
{
return (id & QSE_SED_DEPTH_REX_BUILD)? sed->depth.rex.build:
(id & QSE_SED_DEPTH_REX_MATCH)? sed->depth.rex.match: 0;
}
void qse_sed_setmaxdepth (qse_sed_t* sed, int ids, qse_size_t depth)
{
if (ids & QSE_SED_DEPTH_REX_BUILD) sed->depth.rex.build = depth;
if (ids & QSE_SED_DEPTH_REX_MATCH) sed->depth.rex.match = depth;
}
/* check if c is a space character */ /* check if c is a space character */
#define IS_SPACE(c) ((c) == QSE_T(' ') || (c) == QSE_T('\t') || (c) == QSE_T('\r')) #define IS_SPACE(c) ((c) == QSE_T(' ') || (c) == QSE_T('\t') || (c) == QSE_T('\r'))
#define IS_LINTERM(c) ((c) == QSE_T('\n')) #define IS_LINTERM(c) ((c) == QSE_T('\n'))
@ -275,6 +269,47 @@ static void free_address (qse_sed_t* sed, qse_sed_cmd_t* cmd)
} }
} }
static int add_command_block (qse_sed_t* sed)
{
qse_sed_cmd_blk_t* b;
b = (qse_sed_cmd_blk_t*) QSE_MMGR_ALLOC (sed->mmgr, QSE_SIZEOF(*b));
if (b == QSE_NULL)
{
SETERR0 (sed, QSE_SED_ENOMEM, 0);
return -1;
}
QSE_MEMSET (b, 0, QSE_SIZEOF(*b));
b->next = QSE_NULL;
b->len = 0;
sed->cmd.lb->next = b;
sed->cmd.lb = b;
return 0;
}
static void free_all_command_blocks (qse_sed_t* sed)
{
qse_sed_cmd_blk_t* b;
for (b = &sed->cmd.fb; b != QSE_NULL; )
{
qse_sed_cmd_blk_t* nxt = b->next;
while (b->len > 0) free_command (sed, &b->buf[--b->len]);
if (b != &sed->cmd.fb) QSE_MMGR_FREE (sed->mmgr, b);
b = nxt;
}
QSE_MEMSET (&sed->cmd.fb, 0, QSE_SIZEOF(sed->cmd.fb));
sed->cmd.lb = &sed->cmd.fb;
sed->cmd.lb->len = 0;
sed->cmd.lb->next = QSE_NULL;
}
static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd) static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd)
{ {
free_address (sed, cmd); free_address (sed, cmd);
@ -371,9 +406,9 @@ static void* compile_rex (qse_sed_t* sed, qse_char_t rxend)
} }
} }
/* TODO: maximum depth - optionize the second parameter */
code = qse_buildrex ( code = qse_buildrex (
sed->mmgr, 0, sed->mmgr,
sed->depth.rex.build,
QSE_STR_PTR(&sed->tmp.rex), QSE_STR_PTR(&sed->tmp.rex),
QSE_STR_LEN(&sed->tmp.rex), QSE_STR_LEN(&sed->tmp.rex),
QSE_NULL QSE_NULL
@ -582,7 +617,9 @@ static int get_label (qse_sed_t* sed, qse_sed_cmd_t* cmd)
if (IS_CMDTERM(c)) if (IS_CMDTERM(c))
{ {
if (c != QSE_T('#') && c != QSE_CHAR_EOF) NXTSC (sed); if (c != QSE_T('}') &&
c != QSE_T('#') &&
c != QSE_CHAR_EOF) NXTSC (sed);
} }
return 0; return 0;
@ -888,7 +925,8 @@ static int get_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
QSE_ASSERT (cmd->u.subst.rex == QSE_NULL); QSE_ASSERT (cmd->u.subst.rex == QSE_NULL);
cmd->u.subst.rex = qse_buildrex ( cmd->u.subst.rex = qse_buildrex (
sed->mmgr, 0, sed->mmgr,
sed->depth.rex.build,
QSE_STR_PTR(t[0]), QSE_STR_PTR(t[0]),
QSE_STR_LEN(t[0]), QSE_STR_LEN(t[0]),
QSE_NULL QSE_NULL
@ -1003,10 +1041,10 @@ oops:
return -1; return -1;
} }
static int get_command (qse_sed_t* sed) /* process a command code and following parts into cmd */
static int get_command (qse_sed_t* sed, qse_sed_cmd_t* cmd)
{ {
qse_cint_t c; qse_cint_t c;
qse_sed_cmd_t* cmd = sed->cmd.cur;
c = CURSC (sed); c = CURSC (sed);
cmd->lnum = sed->src.lnum; cmd->lnum = sed->src.lnum;
@ -1028,7 +1066,7 @@ static int get_command (qse_sed_t* sed)
case QSE_T(':'): case QSE_T(':'):
/* label - this is not a command */ /* label - this is not a command */
cmd->type = c; cmd->type = QSE_SED_CMD_NOOP;
if (cmd->a1.type != QSE_SED_ADR_NONE) if (cmd->a1.type != QSE_SED_ADR_NONE)
{ {
/* label cannot have an address */ /* label cannot have an address */
@ -1044,7 +1082,7 @@ static int get_command (qse_sed_t* sed)
c = CURSC (sed); c = CURSC (sed);
while (IS_SPACE(c)) c = NXTSC(sed); while (IS_SPACE(c)) c = NXTSC(sed);
return 0; break;
case QSE_T('{'): case QSE_T('{'):
/* insert a negated branch command at the beginning /* insert a negated branch command at the beginning
@ -1069,6 +1107,8 @@ static int get_command (qse_sed_t* sed)
{ {
qse_sed_cmd_t* tc; qse_sed_cmd_t* tc;
cmd->type = QSE_SED_CMD_NOOP;
if (sed->tmp.grp.level <= 0) if (sed->tmp.grp.level <= 0)
{ {
/* group not balanced */ /* group not balanced */
@ -1080,7 +1120,7 @@ static int get_command (qse_sed_t* sed)
tc->u.branch.target = cmd; tc->u.branch.target = cmd;
NXTSC (sed); NXTSC (sed);
return 0; break;
} }
case QSE_T('q'): case QSE_T('q'):
@ -1187,7 +1227,7 @@ static int get_command (qse_sed_t* sed)
break; break;
} }
return 1; return 0;
} }
int qse_sed_comp (qse_sed_t* sed, const qse_char_t* sptr, qse_size_t slen) int qse_sed_comp (qse_sed_t* sed, const qse_char_t* sptr, qse_size_t slen)
@ -1203,9 +1243,9 @@ int qse_sed_comp (qse_sed_t* sed, const qse_char_t* sptr, qse_size_t slen)
sed->src.cc = (slen > 0)? (*sptr): QSE_CHAR_EOF; sed->src.cc = (slen > 0)? (*sptr): QSE_CHAR_EOF;
/* free all the commands previously compiled */ /* free all the commands previously compiled */
for (cmd = sed->cmd.buf; cmd != sed->cmd.cur; cmd++) free_all_command_blocks (sed);
free_command (sed, cmd); QSE_ASSERT (sed->cmd.lb == &sed->cmd.fb && sed->cmd.lb->len == 0);
cmd = sed->cmd.cur = sed->cmd.buf; cmd = &sed->cmd.lb->buf[sed->cmd.lb->len];
/* clear the label table */ /* clear the label table */
qse_map_clear (&sed->tmp.labs); qse_map_clear (&sed->tmp.labs);
@ -1308,27 +1348,19 @@ int qse_sed_comp (qse_sed_t* sed, const qse_char_t* sptr, qse_size_t slen)
while (IS_SPACE(c)) c = NXTSC (sed); while (IS_SPACE(c)) c = NXTSC (sed);
} }
n = get_command (sed, cmd);
n = get_command (sed);
if (n <= -1) if (n <= -1)
{ {
free_address (sed, cmd); free_address (sed, cmd);
return -1; return -1;
} }
if (n > 0)
{
QSE_ASSERT (n == 1);
if (sed->cmd.cur >= sed->cmd.end) sed->cmd.lb->len++;
if (sed->cmd.lb->len >= QSE_COUNTOF(sed->cmd.lb->buf))
{ {
/* TODO: too many commands. change errnum */ if (add_command_block (sed) <= -1) return -1;
free_command (sed, cmd);
SETERR0 (sed, QSE_SED_ENOMEM, 0);
return -1;
}
cmd = ++sed->cmd.cur;
} }
cmd = &sed->cmd.lb->buf[sed->cmd.lb->len];
} }
if (sed->tmp.grp.level != 0) if (sed->tmp.grp.level != 0)
@ -1814,9 +1846,10 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
{ {
if (max_count == 0 || sub_count < max_count) if (max_count == 0 || sub_count < max_count)
{ {
/* TODO: maximum match depth... */
n = qse_matchrex ( n = qse_matchrex (
sed->mmgr, 0, cmd->u.subst.rex, opt, sed->mmgr,
sed->depth.rex.match,
cmd->u.subst.rex, opt,
str_ptr, str_len, str_ptr, str_len,
cur_ptr, cur_len, cur_ptr, cur_len,
&mat, &errnum &mat, &errnum
@ -1824,7 +1857,7 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
} }
else n = 0; else n = 0;
if (n == -1) if (n <= -1)
{ {
SETERR0 (sed, QSE_SED_EREXMA, cmd->lnum); SETERR0 (sed, QSE_SED_EREXMA, cmd->lnum);
return -1; return -1;
@ -1972,7 +2005,9 @@ static int match_a (qse_sed_t* sed, qse_sed_cmd_t* cmd, qse_sed_adr_t* a)
QSE_STR_CHAR(line,llen-1) == QSE_T('\n')) llen--; QSE_STR_CHAR(line,llen-1) == QSE_T('\n')) llen--;
n = qse_matchrex ( n = qse_matchrex (
sed->mmgr, 0, a->u.rex, 0, sed->mmgr,
sed->depth.rex.match,
a->u.rex, 0,
QSE_STR_PTR(line), llen, QSE_STR_PTR(line), llen,
QSE_STR_PTR(line), llen, QSE_STR_PTR(line), llen,
&match, &errnum); &match, &errnum);
@ -2114,13 +2149,16 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
switch (cmd->type) switch (cmd->type)
{ {
case QSE_SED_CMD_NOOP:
break;
case QSE_SED_CMD_QUIT: case QSE_SED_CMD_QUIT:
n = write_str (sed, n = write_str (sed,
QSE_STR_PTR(&sed->e.in.line), QSE_STR_PTR(&sed->e.in.line),
QSE_STR_LEN(&sed->e.in.line)); QSE_STR_LEN(&sed->e.in.line));
if (n <= -1) return QSE_NULL; if (n <= -1) return QSE_NULL;
case QSE_SED_CMD_QUIT_QUIET: case QSE_SED_CMD_QUIT_QUIET:
jumpto = sed->cmd.cur + 1; jumpto = &sed->cmd.quit;
break; break;
case QSE_SED_CMD_APPEND: case QSE_SED_CMD_APPEND:
@ -2162,7 +2200,7 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
/* move past the last command so as to start /* move past the last command so as to start
* the next cycle */ * the next cycle */
jumpto = sed->cmd.cur; jumpto = &sed->cmd.over;
break; break;
case QSE_SED_CMD_DELETE_FIRSTLN: case QSE_SED_CMD_DELETE_FIRSTLN:
@ -2185,12 +2223,12 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
/* if the pattern space is not empty, /* if the pattern space is not empty,
* arrange to execute from the first * arrange to execute from the first
* command */ * command */
jumpto = sed->cmd.cur + 2; jumpto = &sed->cmd.again;
} }
else else
{ {
/* arrange to start the the next cycle */ /* finish the current cycle */
jumpto = sed->cmd.cur; jumpto = &sed->cmd.over;
} }
break; break;
} }
@ -2200,9 +2238,9 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
case QSE_SED_CMD_DELETE: case QSE_SED_CMD_DELETE:
/* delete the pattern space */ /* delete the pattern space */
qse_str_clear (&sed->e.in.line); qse_str_clear (&sed->e.in.line);
/* move past the last command so as to start
* the next cycle */ /* finish the current cycle */
jumpto = sed->cmd.cur; jumpto = &sed->cmd.over;
break; break;
case QSE_SED_CMD_PRINT_LNNUM: case QSE_SED_CMD_PRINT_LNNUM:
@ -2301,8 +2339,7 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
if (n == 0) if (n == 0)
{ {
/* EOF is reached. */ /* EOF is reached. */
/*jumpto = sed->cmd.cur + 1;*/ jumpto = &sed->cmd.over;
jumpto = sed->cmd.cur;
} }
break; break;
@ -2313,8 +2350,7 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
if (n == 0) if (n == 0)
{ {
/* EOF is reached. */ /* EOF is reached. */
/*jumpto = sed->cmd.cur + 1;*/ jumpto = &sed->cmd.over;
jumpto = sed->cmd.cur;
} }
break; break;
@ -2407,7 +2443,7 @@ static qse_sed_cmd_t* exec_cmd (qse_sed_t* sed, qse_sed_cmd_t* cmd)
} }
} }
if (jumpto == QSE_NULL) jumpto = cmd + 1; if (jumpto == QSE_NULL) jumpto = cmd->state.next;
return jumpto; return jumpto;
} }
@ -2424,9 +2460,112 @@ static void close_outfile (qse_map_t* map, void* dptr, qse_size_t dlen)
} }
} }
static int init_command_block_for_exec (qse_sed_t* sed, qse_sed_cmd_blk_t* b)
{
qse_size_t i;
QSE_ASSERT (b->len <= QSE_COUNTOF(b->buf));
for (i = 0; i < b->len; i++)
{
qse_sed_cmd_t* c = &b->buf[i];
const qse_xstr_t* file = QSE_NULL;
/* clear states */
c->state.a1_matched = 0;
c->state.c_ready = 0;
/* let c point to the next command */
if (i + 1 >= b->len)
{
if (b->next == QSE_NULL || b->next->len <= 0)
c->state.next = &sed->cmd.over;
else
c->state.next = &b->next->buf[0];
}
else
{
c->state.next = &b->buf[i+1];
}
if ((c->type == QSE_SED_CMD_BRANCH ||
c->type == QSE_SED_CMD_BRANCH_COND) &&
c->u.branch.target == QSE_NULL)
{
/* resolve unresolved branch targets */
qse_map_pair_t* pair;
qse_xstr_t* lab = &c->u.branch.label;
if (lab->ptr == QSE_NULL)
{
/* arrange to branch past the last */
c->u.branch.target = &sed->cmd.over;
}
else
{
/* resolve the target */
pair = qse_map_search (
&sed->tmp.labs, lab->ptr, lab->len);
if (pair == QSE_NULL)
{
SETERR1 (
sed, QSE_SED_ELABNF,
c->lnum, lab->ptr, lab->len
);
return -1;
}
c->u.branch.target = QSE_MAP_VPTR(pair);
/* free resolved label name */
QSE_MMGR_FREE (sed->mmgr, lab->ptr);
lab->ptr = QSE_NULL;
lab->len = 0;
}
}
else
{
/* open output files in advance */
if (c->type == QSE_SED_CMD_WRITE_FILE ||
c->type == QSE_SED_CMD_WRITE_FILELN)
{
file = &c->u.file;
}
else if (c->type == QSE_SED_CMD_SUBSTITUTE &&
c->u.subst.file.ptr != QSE_NULL)
{
file = &c->u.subst.file;
}
if (file != QSE_NULL)
{
/* call this function to an open output file */
int n = write_str_to_file (
sed, c, QSE_NULL, 0,
file->ptr, file->len
);
if (n <= -1) return -1;
}
}
}
return 0;
}
static int init_all_commands_for_exec (qse_sed_t* sed)
{
qse_sed_cmd_blk_t* b;
for (b = &sed->cmd.fb; b != QSE_NULL; b = b->next)
{
if (init_command_block_for_exec (sed, b) <= -1) return -1;
}
return 0;
}
int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf) int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
{ {
qse_sed_cmd_t* c, * j;
qse_ssize_t n; qse_ssize_t n;
int ret = 0; int ret = 0;
@ -2505,76 +2644,12 @@ int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
sed->e.out.eof = 1; sed->e.out.eof = 1;
} }
for (c = sed->cmd.buf; c < sed->cmd.cur; c++) if (init_all_commands_for_exec (sed) <= -1)
{ {
const qse_xstr_t* file = QSE_NULL;
/* clear states */
c->state.a1_matched = 0;
c->state.c_ready = 0;
if ((c->type == QSE_SED_CMD_BRANCH ||
c->type == QSE_SED_CMD_BRANCH_COND) &&
c->u.branch.target == QSE_NULL)
{
/* resolve unresolved branch targets */
qse_map_pair_t* pair;
qse_xstr_t* lab = &c->u.branch.label;
if (lab->ptr == QSE_NULL)
{
/* arrange to branch past the last */
c->u.branch.target = sed->cmd.cur;
}
else
{
/* resolve the target */
pair = qse_map_search (
&sed->tmp.labs, lab->ptr, lab->len);
if (pair == QSE_NULL)
{
SETERR1 (
sed, QSE_SED_ELABNF,
c->lnum, lab->ptr, lab->len
);
ret = -1; ret = -1;
goto done; goto done;
} }
c->u.branch.target = QSE_MAP_VPTR(pair);
/* free resolved label name */
QSE_MMGR_FREE (sed->mmgr, lab->ptr);
lab->ptr = QSE_NULL;
lab->len = 0;
}
}
else
{
/* open output files in advance */
if (c->type == QSE_SED_CMD_WRITE_FILE ||
c->type == QSE_SED_CMD_WRITE_FILELN)
{
file = &c->u.file;
}
else if (c->type == QSE_SED_CMD_SUBSTITUTE &&
c->u.subst.file.ptr != QSE_NULL)
{
file = &c->u.subst.file;
}
if (file != QSE_NULL)
{
/* call this function to an open output file */
n = write_str_to_file (
sed, c, QSE_NULL, 0,
file->ptr, file->len
);
if (n <= -1) { ret = -1; goto done; }
}
}
}
while (1) while (1)
{ {
qse_size_t i; qse_size_t i;
@ -2586,9 +2661,14 @@ int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
qse_lda_clear (&sed->e.txt.appended); qse_lda_clear (&sed->e.txt.appended);
qse_str_clear (&sed->e.txt.read); qse_str_clear (&sed->e.txt.read);
if (sed->cmd.fb.len > 0)
{
qse_sed_cmd_t* c, * j;
again: again:
c = sed->cmd.buf; c = &sed->cmd.fb.buf[0];
while (c < sed->cmd.cur)
while (c != &sed->cmd.over)
{ {
n = match_address (sed, c); n = match_address (sed, c);
if (n <= -1) { ret = -1; goto done; } if (n <= -1) { ret = -1; goto done; }
@ -2602,13 +2682,13 @@ int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
j = exec_cmd (sed, c); j = exec_cmd (sed, c);
if (j == QSE_NULL) { ret = -1; goto done; } if (j == QSE_NULL) { ret = -1; goto done; }
if (j == sed->cmd.cur + 1) goto done; if (j == &sed->cmd.quit) goto done;
if (j == sed->cmd.cur + 2) goto again; if (j == &sed->cmd.again) goto again;
QSE_ASSERT (j <= sed->cmd.cur);
/* go to the next command */ /* go to the next command */
c = j; c = j;
} }
}
if (!(sed->option & QSE_SED_QUIET)) if (!(sed->option & QSE_SED_QUIET))
{ {

View File

@ -44,6 +44,7 @@ struct qse_sed_adr_t
} u; } u;
}; };
#define QSE_SED_CMD_NOOP QSE_T('\0')
#define QSE_SED_CMD_QUIT QSE_T('q') #define QSE_SED_CMD_QUIT QSE_T('q')
#define QSE_SED_CMD_QUIT_QUIET QSE_T('Q') #define QSE_SED_CMD_QUIT_QUIET QSE_T('Q')
#define QSE_SED_CMD_APPEND QSE_T('a') #define QSE_SED_CMD_APPEND QSE_T('a')
@ -118,13 +119,17 @@ struct qse_sed_cmd_t
{ {
int a1_matched; int a1_matched;
int c_ready; int c_ready;
/* points to the next command for fast traversal and
* fast random jumps */
qse_sed_cmd_t* next;
} state; } state;
}; };
struct qse_sed_cmd_blk_t struct qse_sed_cmd_blk_t
{ {
qse_size_t len; qse_size_t len;
qse_sed_cmd_t buf[512]; qse_sed_cmd_t buf[256];
qse_sed_cmd_blk_t* next; qse_sed_cmd_blk_t* next;
}; };
@ -142,6 +147,15 @@ struct qse_sed_t
int option; /**< stores options */ int option; /**< stores options */
struct
{
struct
{
qse_size_t build;
qse_size_t match;
} rex;
} depth;
/** source text pointers */ /** source text pointers */
struct struct
{ {
@ -172,6 +186,7 @@ struct qse_sed_t
} tmp; } tmp;
/** compiled commands */ /** compiled commands */
#if 0
struct struct
{ {
qse_size_t len; /**< buffer size */ qse_size_t len; /**< buffer size */
@ -179,14 +194,17 @@ struct qse_sed_t
qse_sed_cmd_t* end; /**< end of the buffer */ qse_sed_cmd_t* end; /**< end of the buffer */
qse_sed_cmd_t* cur; /**< points next to the last command */ qse_sed_cmd_t* cur; /**< points next to the last command */
} cmd; } cmd;
#endif
#if 0
struct struct
{ {
qse_sed_cmd_blk_t fb; /**< the first block is static */ qse_sed_cmd_blk_t fb; /**< the first block is static */
qse_sed_cmd_blk_t* lb; /**< points to the last block */ qse_sed_cmd_blk_t* lb; /**< points to the last block */
qse_sed_cmd_t quit;
qse_sed_cmd_t again;
qse_sed_cmd_t over;
} cmd; } cmd;
#endif
/** data for execution */ /** data for execution */
struct struct