Convert REGNODE_{SIMPLE,VARIES} to a bitmask lookup, from a strchr() lookup.

This is O(1) with no branching, instead of O(n) with branching.
Deprecate the old implementation's externally visible variables
PL_simple and PL_varies. Google codesearch suggests that nothing outside the
core regexp code was using these.

diff --git a/regcomp.pl b/regcomp.pl
index 6a61f98..9370487 100644 (file)
--- a/regcomp.pl
+++ b/regcomp.pl
@@ -85,22 +85,46 @@ sub process_flags {
 
   $ind = 0;
   my @selected;
+  my $bitmap = '';
   do {
-    push @selected, $name[$ind] if $flags[$ind] && $flags[$ind] eq $flag;
+    my $set = $flags[$ind] && $flags[$ind] eq $flag ? 1 : 0;
+    # Whilst I could do this with vec, I'd prefer to do longhand the arithmetic
+    # ops in the C code.
+    my $current = do {
+      no warnings 'uninitialized';
+      ord do {
+       no warnings 'substr';
+       substr $bitmap, ($ind >> 3);
+      }
+    };
+    substr $bitmap, ($ind >> 3), 1, chr($current | ($set << ($ind & 7)));
+
+    push @selected, $name[$ind] if $set;
   } while (++$ind < $lastregop);
   my $out_string = join ', ', @selected, 0;
   $out_string =~ s/(.{1,70},) /$1\n    /g;
+
+  my $out_mask = join ', ', map {sprintf "0x%02X", ord $_} split '', $bitmap;
+
   return $comment . <<"EOP";
-#define REGNODE_\U$varname\E(node) strchr((const char *)PL_${varname}, (node))
+#define REGNODE_\U$varname\E(node) (PL_${varname}_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 
 #ifndef DOINIT
-EXTCONST U8 PL_${varname}[];
+EXTCONST U8 PL_${varname}[] __attribute__deprecated__;
 #else
-EXTCONST U8 PL_${varname}[] = {
+EXTCONST U8 PL_${varname}[] __attribute__deprecated__ = {
     $out_string
 };
 #endif /* DOINIT */
 
+#ifndef DOINIT
+EXTCONST U8 PL_${varname}_bitmask[];
+#else
+EXTCONST U8 PL_${varname}_bitmask[] = {
+    $out_mask
+};
+#endif /* DOINIT */
+
 EOP
 }
 

diff --git a/regnodes.h b/regnodes.h
index c9ba109..348410c 100644 (file)
--- a/regnodes.h
+++ b/regnodes.h
@@ -662,26 +662,34 @@ EXTCONST char * const PL_reg_extflags_name[] = {
 #endif /* DOINIT */
 
 /* The following have no fixed length. U8 so we can do strchr() on it. */
-#define REGNODE_VARIES(node) strchr((const char *)PL_varies, (node))
+#define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 
 #ifndef DOINIT
-EXTCONST U8 PL_varies[];
+EXTCONST U8 PL_varies[] __attribute__deprecated__;
 #else
-EXTCONST U8 PL_varies[] = {
+EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
     CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX, WHILEM,
     REF, REFF, REFFL, SUSPEND, IFTHEN, BRANCHJ, NREF, NREFF, NREFFL,
     0
 };
 #endif /* DOINIT */
 
+#ifndef DOINIT
+EXTCONST U8 PL_varies_bitmask[];
+#else
+EXTCONST U8 PL_varies_bitmask[] = {
+    0x00, 0x00, 0x00, 0xC0, 0xC1, 0x9F, 0x33, 0x01, 0x38, 0x00, 0x00, 0x00
+};
+#endif /* DOINIT */
+
 /* The following always have a length of 1. U8 we can do strchr() on it. */
 /* (Note that length 1 means "one character" under UTF8, not "one octet".) */
-#define REGNODE_SIMPLE(node) strchr((const char *)PL_simple, (node))
+#define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 
 #ifndef DOINIT
-EXTCONST U8 PL_simple[];
+EXTCONST U8 PL_simple[] __attribute__deprecated__;
 #else
-EXTCONST U8 PL_simple[] = {
+EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
     REG_ANY, SANY, CANY, ANYOF, ALNUM, ALNUML, NALNUM, NALNUML, SPACE,
     SPACEL, NSPACE, NSPACEL, DIGIT, NDIGIT, VERTWS, NVERTWS, HORIZWS,
     NHORIZWS,
@@ -689,4 +697,12 @@ EXTCONST U8 PL_simple[] = {
 };
 #endif /* DOINIT */
 
+#ifndef DOINIT
+EXTCONST U8 PL_simple_bitmask[];
+#else
+EXTCONST U8 PL_simple_bitmask[] = {
+    0x00, 0xC0, 0xFF, 0x17, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xF0, 0x00
+};
+#endif /* DOINIT */
+
 /* ex: set ro: */