mirror of
https://github.com/lua/lua
synced 2025-02-23 16:44:14 +03:00
Standard library for strings and pattern-matching
This commit is contained in:
Roberto Ierusalimschy
parent
9f3785a2f3
commit
75ac0d2172
@ -1,17 +1,17 @@
|
||||
/*
|
||||
** strlib.c
|
||||
** String library to LUA
|
||||
** $Id: lstrlib.c,v 1.1 1997/08/14 19:47:57 roberto Exp roberto $
|
||||
** Standard library for strings and pattern-matching
|
||||
** See Copyright Notice in lua.h
|
||||
*/
|
||||
|
||||
char *rcs_strlib="$Id: strlib.c,v 1.46 1997/06/19 18:49:40 roberto Exp roberto $";
|
||||
|
||||
#include <string.h>
|
||||
#include <ctype.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <ctype.h>
|
||||
#include <string.h>
|
||||
|
||||
#include "lauxlib.h"
|
||||
#include "lua.h"
|
||||
#include "auxlib.h"
|
||||
#include "lualib.h"
|
||||
|
||||
|
||||
@ -37,12 +37,14 @@ static char *strbuffer (unsigned long size)
|
||||
return lbuffer.b;
|
||||
}
|
||||
|
||||
|
||||
static char *openspace (unsigned long size)
|
||||
{
|
||||
char *buff = strbuffer(lbuffer.size+size);
|
||||
return buff+lbuffer.size;
|
||||
}
|
||||
|
||||
|
||||
char *luaI_addchar (int c)
|
||||
{
|
||||
if (lbuffer.size >= lbuffer.max)
|
||||
@ -51,6 +53,7 @@ char *luaI_addchar (int c)
|
||||
return lbuffer.b;
|
||||
}
|
||||
|
||||
|
||||
void luaI_emptybuff (void)
|
||||
{
|
||||
lbuffer.size = 0; /* prepare for next string */
|
||||
@ -64,23 +67,19 @@ static void addnchar (char *s, int n)
|
||||
lbuffer.size += n;
|
||||
}
|
||||
|
||||
|
||||
static void addstr (char *s)
|
||||
{
|
||||
addnchar(s, strlen(s));
|
||||
}
|
||||
|
||||
|
||||
/*
|
||||
** Return the string length
|
||||
*/
|
||||
static void str_len (void)
|
||||
{
|
||||
lua_pushnumber(strlen(luaL_check_string(1)));
|
||||
}
|
||||
|
||||
/*
|
||||
** Return the substring of a string
|
||||
*/
|
||||
|
||||
static void str_sub (void)
|
||||
{
|
||||
char *s = luaL_check_string(1);
|
||||
@ -97,9 +96,7 @@ static void str_sub (void)
|
||||
else lua_pushstring("");
|
||||
}
|
||||
|
||||
/*
|
||||
** Convert a string to lower case.
|
||||
*/
|
||||
|
||||
static void str_lower (void)
|
||||
{
|
||||
char *s;
|
||||
@ -109,9 +106,7 @@ static void str_lower (void)
|
||||
lua_pushstring(luaI_addchar(0));
|
||||
}
|
||||
|
||||
/*
|
||||
** Convert a string to upper case.
|
||||
*/
|
||||
|
||||
static void str_upper (void)
|
||||
{
|
||||
char *s;
|
||||
@ -131,9 +126,7 @@ static void str_rep (void)
|
||||
lua_pushstring(luaI_addchar(0));
|
||||
}
|
||||
|
||||
/*
|
||||
** get ascii value of a character in a string
|
||||
*/
|
||||
|
||||
static void str_ascii (void)
|
||||
{
|
||||
char *s = luaL_check_string(1);
|
||||
@ -143,74 +136,12 @@ static void str_ascii (void)
|
||||
}
|
||||
|
||||
|
||||
/* pattern matching */
|
||||
|
||||
#define ESC '%'
|
||||
#define SPECIALS "^$*?.([%-"
|
||||
|
||||
static char *bracket_end (char *p)
|
||||
{
|
||||
return (*p == 0) ? NULL : strchr((*p=='^') ? p+2 : p+1, ']');
|
||||
}
|
||||
|
||||
char *luaL_item_end (char *p)
|
||||
{
|
||||
switch (*p++) {
|
||||
case '\0': return p-1;
|
||||
case ESC:
|
||||
if (*p == 0) luaL_verror("incorrect pattern (ends with `%c')", ESC);
|
||||
return p+1;
|
||||
case '[': {
|
||||
char *end = bracket_end(p);
|
||||
if (end == NULL) lua_error("incorrect pattern (missing `]')");
|
||||
return end+1;
|
||||
}
|
||||
default:
|
||||
return p;
|
||||
}
|
||||
}
|
||||
|
||||
static int matchclass (int c, int cl)
|
||||
{
|
||||
int res;
|
||||
switch (tolower((unsigned char)cl)) {
|
||||
case 'a' : res = isalpha((unsigned char)c); break;
|
||||
case 'c' : res = iscntrl((unsigned char)c); break;
|
||||
case 'd' : res = isdigit((unsigned char)c); break;
|
||||
case 'l' : res = islower((unsigned char)c); break;
|
||||
case 'p' : res = ispunct((unsigned char)c); break;
|
||||
case 's' : res = isspace((unsigned char)c); break;
|
||||
case 'u' : res = isupper((unsigned char)c); break;
|
||||
case 'w' : res = isalnum((unsigned char)c); break;
|
||||
default: return (cl == c);
|
||||
}
|
||||
return (islower((unsigned char)cl) ? res : !res);
|
||||
}
|
||||
|
||||
int luaL_singlematch (int c, char *p)
|
||||
{
|
||||
if (c == 0) return 0;
|
||||
switch (*p) {
|
||||
case '.': return 1;
|
||||
case ESC: return matchclass(c, *(p+1));
|
||||
case '[': {
|
||||
char *end = bracket_end(p+1);
|
||||
int sig = *(p+1) == '^' ? (p++, 0) : 1;
|
||||
while (++p < end) {
|
||||
if (*p == ESC) {
|
||||
if (((p+1) < end) && matchclass(c, *++p)) return sig;
|
||||
}
|
||||
else if ((*(p+1) == '-') && (p+2 < end)) {
|
||||
p+=2;
|
||||
if (*(p-2) <= c && c <= *p) return sig;
|
||||
}
|
||||
else if (*p == c) return sig;
|
||||
}
|
||||
return !sig;
|
||||
}
|
||||
default: return (*p == c);
|
||||
}
|
||||
}
|
||||
/*
|
||||
** =======================================================
|
||||
** PATTERN MATCHING
|
||||
** =======================================================
|
||||
*/
|
||||
|
||||
#define MAX_CAPT 9
|
||||
|
||||
@ -222,6 +153,10 @@ static struct {
|
||||
static int num_captures; /* only valid after a sucessful call to match */
|
||||
|
||||
|
||||
#define ESC '%'
|
||||
#define SPECIALS "^$*?.([%-"
|
||||
|
||||
|
||||
static void push_captures (void)
|
||||
{
|
||||
int i;
|
||||
@ -235,6 +170,7 @@ static void push_captures (void)
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
static int check_cap (int l, int level)
|
||||
{
|
||||
l -= '1';
|
||||
@ -243,6 +179,7 @@ static int check_cap (int l, int level)
|
||||
return l;
|
||||
}
|
||||
|
||||
|
||||
static int capture_to_close (int level)
|
||||
{
|
||||
for (level--; level>=0; level--)
|
||||
@ -251,6 +188,71 @@ static int capture_to_close (int level)
|
||||
return 0; /* to avoid warnings */
|
||||
}
|
||||
|
||||
|
||||
static char *bracket_end (char *p)
|
||||
{
|
||||
return (*p == 0) ? NULL : strchr((*p=='^') ? p+2 : p+1, ']');
|
||||
}
|
||||
|
||||
|
||||
static int matchclass (int c, int cl)
|
||||
{
|
||||
int res;
|
||||
if (c == 0) return 0;
|
||||
switch (tolower((unsigned char)cl)) {
|
||||
case 'a' : res = isalpha((unsigned char)c); break;
|
||||
case 'c' : res = iscntrl((unsigned char)c); break;
|
||||
case 'd' : res = isdigit((unsigned char)c); break;
|
||||
case 'l' : res = islower((unsigned char)c); break;
|
||||
case 'p' : res = ispunct((unsigned char)c); break;
|
||||
case 's' : res = isspace((unsigned char)c); break;
|
||||
case 'u' : res = isupper((unsigned char)c); break;
|
||||
case 'w' : res = isalnum((unsigned char)c); break;
|
||||
default: return (cl == c);
|
||||
}
|
||||
return (islower((unsigned char)cl) ? res : !res);
|
||||
}
|
||||
|
||||
|
||||
int luaI_singlematch (int c, char *p, char **ep)
|
||||
{
|
||||
switch (*p) {
|
||||
case '\0':
|
||||
*ep = p;
|
||||
return 0;
|
||||
case '.':
|
||||
*ep = p+1;
|
||||
return (c != 0);
|
||||
case ESC:
|
||||
if (*(++p) == '\0')
|
||||
luaL_verror("incorrect pattern (ends with `%c')", ESC);
|
||||
*ep = p+1;
|
||||
return matchclass(c, *p);
|
||||
case '[': {
|
||||
char *end = bracket_end(p+1);
|
||||
int sig = *(p+1) == '^' ? (p++, 0) : 1;
|
||||
if (end == NULL) lua_error("incorrect pattern (missing `]')");
|
||||
*ep = end+1;
|
||||
if (c == 0) return 0;
|
||||
while (++p < end) {
|
||||
if (*p == ESC) {
|
||||
if (((p+1) < end) && matchclass(c, *++p)) return sig;
|
||||
}
|
||||
else if ((*(p+1) == '-') && (p+2 < end)) {
|
||||
p+=2;
|
||||
if (*(p-2) <= c && c <= *p) return sig;
|
||||
}
|
||||
else if (*p == c) return sig;
|
||||
}
|
||||
return !sig;
|
||||
}
|
||||
default:
|
||||
*ep = p+1;
|
||||
return (*p == c);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
static char *matchbalance (char *s, int b, int e)
|
||||
{
|
||||
if (*s != b) return NULL;
|
||||
@ -266,6 +268,31 @@ static char *matchbalance (char *s, int b, int e)
|
||||
return NULL; /* string ends out of balance */
|
||||
}
|
||||
|
||||
|
||||
static char *matchitem (char *s, char *p, int level, char **ep)
|
||||
{
|
||||
if (*p == ESC) {
|
||||
p++;
|
||||
if (isdigit((unsigned char)*p)) { /* capture */
|
||||
int l = check_cap(*p, level);
|
||||
*ep = p+1;
|
||||
if (strncmp(capture[l].init, s, capture[l].len) == 0)
|
||||
return s+capture[l].len;
|
||||
else return NULL;
|
||||
}
|
||||
else if (*p == 'b') { /* balanced string */
|
||||
p++;
|
||||
if (*p == 0 || *(p+1) == 0)
|
||||
lua_error("bad balanced pattern specification");
|
||||
*ep = p+2;
|
||||
return matchbalance(s, *p, *(p+1));
|
||||
}
|
||||
else p--; /* and go through */
|
||||
}
|
||||
return (luaI_singlematch(*s, p, ep) ? s+1 : NULL);
|
||||
}
|
||||
|
||||
|
||||
static char *match (char *s, char *p, int level)
|
||||
{
|
||||
init: /* using goto's to optimize tail recursion */
|
||||
@ -283,38 +310,19 @@ static char *match (char *s, char *p, int level)
|
||||
capture[l].len = -1; /* undo capture */
|
||||
return res;
|
||||
}
|
||||
case ESC:
|
||||
if (isdigit((unsigned char)(*(p+1)))) { /* capture */
|
||||
int l = check_cap(*(p+1), level);
|
||||
if (strncmp(capture[l].init, s, capture[l].len) == 0) {
|
||||
/* return match(p+2, s+capture[l].len, level); */
|
||||
p+=2; s+=capture[l].len; goto init;
|
||||
}
|
||||
else return NULL;
|
||||
}
|
||||
else if (*(p+1) == 'b') { /* balanced string */
|
||||
if (*(p+2) == 0 || *(p+3) == 0)
|
||||
lua_error("bad balanced pattern specification");
|
||||
s = matchbalance(s, *(p+2), *(p+3));
|
||||
if (s == NULL) return NULL;
|
||||
else { /* return match(p+4, s, level); */
|
||||
p+=4; goto init;
|
||||
}
|
||||
}
|
||||
else goto dflt;
|
||||
case '\0': case '$': /* (possibly) end of pattern */
|
||||
if (*p == 0 || (*(p+1) == 0 && *s == 0)) {
|
||||
num_captures = level;
|
||||
return s;
|
||||
}
|
||||
else goto dflt;
|
||||
default: dflt: { /* it is a pattern item */
|
||||
int m = luaL_singlematch(*s, p);
|
||||
char *ep = luaL_item_end(p); /* get what is next */
|
||||
/* else go through */
|
||||
default: { /* it is a pattern item */
|
||||
char *ep; /* get what is next */
|
||||
char *s1 = matchitem(s, p, level, &ep);
|
||||
switch (*ep) {
|
||||
case '*': { /* repetition */
|
||||
char *res;
|
||||
if (m && (res = match(s+1, p, level)))
|
||||
if (s1 && (res = match(s1, p, level)))
|
||||
return res;
|
||||
p=ep+1; goto init; /* else return match(s, ep+1, level); */
|
||||
}
|
||||
@ -322,34 +330,35 @@ static char *match (char *s, char *p, int level)
|
||||
char *res;
|
||||
if ((res = match(s, ep+1, level)) != 0)
|
||||
return res;
|
||||
else if (m) {
|
||||
s++;
|
||||
goto init; /* return match(s+1, p, level); */
|
||||
else if (s1) {
|
||||
s = s1;
|
||||
goto init; /* return match(s1, p, level); */
|
||||
}
|
||||
else
|
||||
return NULL;
|
||||
}
|
||||
case '?': { /* optional */
|
||||
char *res;
|
||||
if (m && (res = match(s+1, ep+1, level)))
|
||||
if (s1 && (res = match(s1, ep+1, level)))
|
||||
return res;
|
||||
p=ep+1; goto init; /* else return match(s, ep+1, level); */
|
||||
}
|
||||
default:
|
||||
if (m) { s++; p=ep; goto init; } /* return match(s+1, ep, level); */
|
||||
if (s1) { s=s1; p=ep; goto init; } /* return match(s1, ep, level); */
|
||||
else return NULL;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
static void str_find (void)
|
||||
{
|
||||
char *s = luaL_check_string(1);
|
||||
char *p = luaL_check_string(2);
|
||||
long init = (long)luaL_opt_number(3, 1) - 1;
|
||||
luaL_arg_check(0 <= init && init <= strlen(s), 3, "out of range");
|
||||
if (lua_getparam(4) != LUA_NOOBJECT ||
|
||||
if (lua_getparam(4) != LUA_NOOBJECT ||
|
||||
strpbrk(p, SPECIALS) == NULL) { /* no special caracters? */
|
||||
char *s2 = strstr(s+init, p);
|
||||
if (s2) {
|
||||
@ -372,7 +381,8 @@ static void str_find (void)
|
||||
}
|
||||
}
|
||||
|
||||
static void add_s (lua_Object newp, lua_Object table, int n)
|
||||
|
||||
static void add_s (lua_Object newp)
|
||||
{
|
||||
if (lua_isstring(newp)) {
|
||||
char *news = lua_getstring(newp);
|
||||
@ -390,12 +400,8 @@ static void add_s (lua_Object newp, lua_Object table, int n)
|
||||
struct lbuff oldbuff;
|
||||
int status;
|
||||
lua_beginblock();
|
||||
if (lua_istable(table)) {
|
||||
lua_pushobject(table);
|
||||
lua_pushnumber(n);
|
||||
}
|
||||
push_captures();
|
||||
/* function may use lbuffer, so save it and create a new one */
|
||||
/* function may use lbuffer, so save it and create a luaM_new one */
|
||||
oldbuff = lbuffer;
|
||||
lbuffer.b = NULL; lbuffer.max = lbuffer.size = 0;
|
||||
status = lua_callfunction(newp);
|
||||
@ -411,13 +417,13 @@ static void add_s (lua_Object newp, lua_Object table, int n)
|
||||
else luaL_arg_check(0, 3, NULL);
|
||||
}
|
||||
|
||||
|
||||
static void str_gsub (void)
|
||||
{
|
||||
char *src = luaL_check_string(1);
|
||||
char *p = luaL_check_string(2);
|
||||
lua_Object newp = lua_getparam(3);
|
||||
lua_Object table = lua_getparam(4);
|
||||
int max_s = (int)luaL_opt_number(lua_istable(table)?5:4, strlen(src)+1);
|
||||
int max_s = (int)luaL_opt_number(4, strlen(src)+1);
|
||||
int anchor = (*p == '^') ? (p++, 1) : 0;
|
||||
int n = 0;
|
||||
luaI_emptybuff();
|
||||
@ -425,7 +431,7 @@ static void str_gsub (void)
|
||||
char *e = match(src, p, 0);
|
||||
if (e) {
|
||||
n++;
|
||||
add_s(newp, table, n);
|
||||
add_s(newp);
|
||||
}
|
||||
if (e && e>src) /* non empty match? */
|
||||
src = e; /* skip it */
|
||||
@ -439,18 +445,6 @@ static void str_gsub (void)
|
||||
lua_pushnumber(n); /* number of substitutions */
|
||||
}
|
||||
|
||||
static void str_set (void)
|
||||
{
|
||||
char *item = luaL_check_string(1);
|
||||
int i;
|
||||
luaL_arg_check(*luaL_item_end(item) == 0, 1, "wrong format");
|
||||
luaI_emptybuff();
|
||||
for (i=1; i<256; i++) /* 0 cannot be part of a set */
|
||||
if (luaL_singlematch(i, item))
|
||||
luaI_addchar(i);
|
||||
lua_pushstring(luaI_addchar(0));
|
||||
}
|
||||
|
||||
|
||||
void luaI_addquoted (char *s)
|
||||
{
|
||||
@ -505,7 +499,7 @@ static void str_format (void)
|
||||
case 'u': case 'x': case 'X':
|
||||
sprintf(buff, form, (int)luaL_check_number(arg));
|
||||
break;
|
||||
case 'e': case 'E': case 'f': case 'g':
|
||||
case 'e': case 'E': case 'f': case 'g': case 'G':
|
||||
sprintf(buff, form, luaL_check_number(arg));
|
||||
break;
|
||||
default: /* also treat cases 'pnLlh' */
|
||||
@ -521,7 +515,6 @@ static void str_format (void)
|
||||
static struct luaL_reg strlib[] = {
|
||||
{"strlen", str_len},
|
||||
{"strsub", str_sub},
|
||||
{"strset", str_set},
|
||||
{"strlower", str_lower},
|
||||
{"strupper", str_upper},
|
||||
{"strrep", str_rep},
|
||||
Loading…
x
Reference in New Issue
Block a user