From 97a4ca3b4078f581cdc8cebc4fa4cf39d5ff8125 Mon Sep 17 00:00:00 2001
From: Roberto Ierusalimschy <roberto@inf.puc-rio.br>
Date: Thu, 27 Apr 2023 11:03:30 -0300
Subject: New macros 'fillset' and 'clearset'

---
 lpcode.c  | 2 +-
 lpcset.c  | 8 ++++----
 lpprint.c | 4 ++--
 lptree.c  | 9 +++++----
 lptypes.h | 4 ++++
 test.lua  | 6 ++++++
 6 files changed, 22 insertions(+), 11 deletions(-)

diff --git a/lpcode.c b/lpcode.c
index 9289bd3..b68f32d 100644
--- a/lpcode.c
+++ b/lpcode.c
@@ -221,7 +221,7 @@ static int getfirst (TTree *tree, const Charset *follow, Charset *firstset) {
     }
     case TUTFR: {
       int c;
-      loopset(i, firstset->cs[i] = 0);  /* erase all chars */
+      clearset(firstset->cs);  /* erase all chars */
       for (c = tree->key; c <= sib1(tree)->key; c++)
         setchar(firstset->cs, c);
       return 0;
diff --git a/lpcset.c b/lpcset.c
index 2e62d94..2dcffd9 100644
--- a/lpcset.c
+++ b/lpcset.c
@@ -76,21 +76,21 @@ int tocharset (TTree *tree, Charset *cs) {
   switch (tree->tag) {
     case TChar: {  /* only one char */
       assert(0 <= tree->u.n && tree->u.n <= UCHAR_MAX);
-      loopset(i, cs->cs[i] = 0);  /* erase all chars */
+      clearset(cs->cs);  /* erase all chars */
       setchar(cs->cs, tree->u.n);  /* add that one */
       return 1;
     }
     case TAny: {
-      loopset(i, cs->cs[i] = 0xFF);  /* add all characters to the set */
+      fillset(cs->cs, 0xFF);  /* add all characters to the set */
       return 1;
     }
     case TFalse: {
-      loopset(i, cs->cs[i] = 0);  /* empty set */
+      clearset(cs->cs);  /* empty set */
       return 1;
     }
     case TSet: {  /* fill set */
       int i;
-      loopset(j, cs->cs[j] = tree->u.set.deflt);
+      fillset(cs->cs, tree->u.set.deflt);
       for (i = 0; i < tree->u.set.size; i++)
         cs->cs[tree->u.set.offset + i] = treebuffer(tree)[i];
       return 1;
diff --git a/lpprint.c b/lpprint.c
index a432263..3e7e7f2 100644
--- a/lpprint.c
+++ b/lpprint.c
@@ -37,7 +37,7 @@ static void printIcharset (const Instruction *inst, const byte *buff) {
   byte cs[CHARSETSIZE];
   int i;
   printf("(%02x-%d) ", inst->i.aux2.set.offset, inst->i.aux2.set.size);
-  loopset(j, cs[j] = 0);
+  clearset(cs);
   for (i = 0; i < CHARSETSIZE * 8; i++) {
     if (charinset(inst, buff, i))
       setchar(cs, i);
@@ -50,7 +50,7 @@ static void printTcharset (TTree *tree) {
   byte cs[CHARSETSIZE];
   int i;
   printf("(%02x-%d) ", tree->u.set.offset, tree->u.set.size);
-  loopset(j, cs[j] = tree->u.set.deflt);
+  fillset(cs, tree->u.set.deflt);
   for (i = 0; i < tree->u.set.size; i++)
     cs[tree->u.set.offset + i] = treebuffer(tree)[i];
   printcharset(cs);
diff --git a/lptree.c b/lptree.c
index f9e170b..c61a5db 100644
--- a/lptree.c
+++ b/lptree.c
@@ -670,7 +670,7 @@ static int lp_set (lua_State *L) {
   size_t l;
   const char *s = luaL_checklstring(L, 1, &l);
   byte buff[CHARSETSIZE];
-  loopset(i, buff[i] = 0);
+  clearset(buff);
   while (l--) {
     setchar(buff, (byte)(*s));
     s++;
@@ -684,7 +684,7 @@ static int lp_range (lua_State *L) {
   int arg;
   int top = lua_gettop(L);
   byte buff[CHARSETSIZE];
-  loopset(i, buff[i] = 0);
+  clearset(buff);
   for (arg = 1; arg <= top; arg++) {
     int c;
     size_t l;
@@ -734,7 +734,7 @@ static int lp_utfr (lua_State *L) {
   if (to <= 0x7f) {  /* ascii range? */
     unsigned int f;
     byte buff[CHARSETSIZE];  /* code it as a regular charset */
-    loopset(i, buff[i] = 0);
+    clearset(buff);
     for (f = (int)from; f <= to; f++)
       setchar(buff, f);
     newcharset(L, buff);
@@ -1298,7 +1298,7 @@ int lp_gc (lua_State *L) {
 static void createcat (lua_State *L, const char *catname, int (catf) (int)) {
   int c;
   byte buff[CHARSETSIZE];
-  loopset(i, buff[i] = 0);
+  clearset(buff);
   for (c = 0; c <= UCHAR_MAX; c++)
     if (catf(c)) setchar(buff, c);
   newcharset(L, buff);
@@ -1373,6 +1373,7 @@ static struct luaL_Reg metareg[] = {
 
 int luaopen_lpeg (lua_State *L);
 int luaopen_lpeg (lua_State *L) {
+printf("%ld\n", sizeof(TTree));
   luaL_newmetatable(L, PATTERN_T);
   lua_pushnumber(L, MAXBACK);  /* initialize maximum backtracking */
   lua_setfield(L, LUA_REGISTRYINDEX, MAXSTACKIDX);
diff --git a/lptypes.h b/lptypes.h
index 7793e76..17d406e 100644
--- a/lptypes.h
+++ b/lptypes.h
@@ -10,6 +10,7 @@
 
 #include <assert.h>
 #include <limits.h>
+#include <string.h>
 
 #include "lua.h"
 
@@ -97,6 +98,9 @@ typedef struct Charset {
 
 #define loopset(v,b)    { int v; for (v = 0; v < CHARSETSIZE; v++) {b;} }
 
+#define fillset(s,c)	memset(s,c,CHARSETSIZE)
+#define clearset(s)	fillset(s,0)
+
 /* access to charset */
 #define treebuffer(t)      ((byte *)((t) + 1))
 
diff --git a/test.lua b/test.lua
index 9f8d226..d31a69f 100755
--- a/test.lua
+++ b/test.lua
@@ -1227,6 +1227,12 @@ do   -- a few typical UTF-8 ranges
          t[5] == "cjk: " .. cjk and t[6] == "other: —" and
          t[7] == "ascii: " .. ascii and t[8] == "other: " .. last and
          t[9] == nil)
+
+  -- failing UTF-8 matches and borders
+  assert(not m.match(m.utfR(10, 0x2000), "\9"))
+  assert(not m.match(m.utfR(10, 0x2000), "\226\128\129"))
+  assert(m.match(m.utfR(10, 0x2000), "\10") == 2)
+  assert(m.match(m.utfR(10, 0x2000), "\226\128\128") == 4)
 end
 
 
-- 
cgit v1.2.3-55-g6feb