From 012cf9c86cf91cb8354e229bde335592d41b84b2 Mon Sep 17 00:00:00 2001
From: Roberto Ierusalimschy <roberto@inf.puc-rio.br>
Date: Thu, 27 Apr 2023 10:32:39 -0300
Subject: Compact charsets used in trees, too.

---
 lpcset.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 48 insertions(+), 7 deletions(-)

(limited to 'lpcset.c')

diff --git a/lpcset.c b/lpcset.c
index 9ecf475..2e62d94 100644
--- a/lpcset.c
+++ b/lpcset.c
@@ -16,7 +16,7 @@ static int onlybit (int c, int b) {
 
 /*
 ** Check whether a charset is empty (returns IFail), singleton (IChar),
-** full (IAny), or none of those (ISet). When singleton, 'info.aux1'
+** full (IAny), or none of those (ISet). When singleton, 'info.offset'
 ** returns which character it is. When generic set, 'info' returns
 ** information about its range.
 */
@@ -31,7 +31,7 @@ Opcode charsettype (const byte *cs, charsetinfo *info) {
   if (low1 == high1) {  /* only one byte with 1-bits? */
     int b = cs[low1];
     if ((b & (b - 1)) == 0) {  /* does byte has only one 1-bit? */
-      info->aux1 = onlybit(low1 * BITSPERCHAR, b);  /* get that bit */
+      info->offset = onlybit(low1 * BITSPERCHAR, b);  /* get that bit */
       return IChar;  /* single character */
     }
   }
@@ -42,15 +42,16 @@ Opcode charsettype (const byte *cs, charsetinfo *info) {
   for (high0 = CHARSETSIZE - 1; cs[high0] == 0xFF; high0--)
     /* find highest byte with a 0-bit; low0 is a sentinel */;
   if (high1 - low1 <= high0 - low0) {  /* range of 1s smaller than of 0s? */
-    info->aux1 = low1;
+    info->offset = low1;
     info->size = high1 - low1 + 1;
     info->deflt = 0;  /* all discharged bits were 0 */
   }
   else {
-    info->aux1 = low0;
+    info->offset = low0;
     info->size = high0 - low0 + 1;
     info->deflt = 0xFF;  /* all discharged bits were 1 */
   }
+  info->cs = cs + info->offset;
   return ISet;
 }
 
@@ -60,10 +61,50 @@ Opcode charsettype (const byte *cs, charsetinfo *info) {
 ** range, get the byte from the supporting charset (correcting it
 ** by the offset). Otherwise, return the default for the set.
 */
-byte getbytefromcharset (const byte *cs, const charsetinfo *info,
-                                int index) {
+byte getbytefromcharset (const charsetinfo *info, int index) {
   if (index < info->size)
-    return cs[info->aux1 + index];
+    return info->cs[index];
   else return info->deflt;
 }
 
+
+/*
+** If 'tree' is a 'char' pattern (TSet, TChar, TAny, TFalse), convert it
+** into a charset and return 1; else return 0.
+*/
+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 */
+      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 */
+      return 1;
+    }
+    case TFalse: {
+      loopset(i, cs->cs[i] = 0);  /* empty set */
+      return 1;
+    }
+    case TSet: {  /* fill set */
+      int i;
+      loopset(j, cs->cs[j] = 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;
+    }
+    default: return 0;
+  }
+}
+
+
+void tree2cset (TTree *tree, charsetinfo *info) {
+  assert(tree->tag == TSet);
+  info->offset = tree->u.set.offset;
+  info->size = tree->u.set.size;
+  info->deflt = tree->u.set.deflt;
+  info->cs = treebuffer(tree);
+}
+
-- 
cgit v1.2.3-55-g6feb