04_simeplfold_1

arnaudroger · Mar 23, 2018 · bc6d2e0 · bc6d2e0
1 parent bec631b
commit bc6d2e0
Show file tree

Hide file tree

Showing 4 changed files with 48 additions and 16 deletions.
diff --git a/java/com/google/re2j/Compiler.java b/java/com/google/re2j/Compiler.java
@@ -171,6 +171,24 @@ private Frag rune(int[] runes, int flags) {
         runes.length == 2 &&
         runes[0] == runes[1]) {
       i.op = Inst.RUNE1;
+      i.f0 = runes[0];
+    } else if (runes.length == 1 && (flags & RE2.FOLD_CASE) != 0) {
+      i.op = Inst.RUNE1_FOLD;
+      i.f0 = runes[0];
+      i.f1 = runes[0];
+      i.f2 = runes[0];
+      i.f3 = runes[0];
+
+      int[] fold = Unicode.fold(i.f0);
+      if (fold.length > 0){
+        i.f1 = fold[0];
+        if (fold.length > 1) {
+          i.f2 = fold[1];
+          if (fold.length > 2) {
+            i.f3 = fold[2];
+          }
+        }
+      }
     } else if (runes.length == 2 &&
                runes[0] == 0 &&
                runes[1] == Unicode.MAX_RUNE) {

diff --git a/java/com/google/re2j/Inst.java b/java/com/google/re2j/Inst.java
@@ -24,6 +24,7 @@ class Inst {
   public static final int RUNE1 = 9;
   public static final int RUNE_ANY = 10;
   public static final int RUNE_ANY_NOT_NL = 11;
+  public static final int RUNE1_FOLD = 12;
 
   int op;
   int out;  // all but MATCH, FAIL
@@ -32,12 +33,17 @@ class Inst {
                 // otherwise a list of [lo,hi] pairs.  hi is *inclusive*.
                 // REVIEWERS: why not half-open intervals?
 
+  int f0;
+  int f1;
+  int f2;
+  int f3;
+
   Inst(int op) {
     this.op = op;
   }
 
   boolean isRune() {
-    return op >= RUNE && op <= RUNE_ANY_NOT_NL;
+    return op >= RUNE && op <= RUNE1_FOLD;
   }
 
 
@@ -46,21 +52,12 @@ boolean isRune() {
   boolean matchRune(int r) {
     // Special case: single-rune slice is from literal string, not char
     // class.
-    if (runes.length == 1) {
-      int r0 = runes[0];
-      if (r == r0) {
-        return true;
-      }
-      if ((arg & RE2.FOLD_CASE) != 0) {
-        for (int r1 = Unicode.simpleFold(r0);
-             r1 != r0;
-             r1 = Unicode.simpleFold(r1)) {
-          if (r == r1) {
-            return true;
-          }
-        }
-      }
-      return false;
+    if (op ==RUNE1) {
+      return f0 == r;
+    }
+
+    if (op == RUNE1_FOLD) {
+      return f0 == r || f1 == r || f2 == r || f3 == r;
     }
 
     // Peek at the first few pairs.
@@ -108,6 +105,7 @@ public String toString() {
       case NOP:
         return "nop -> " + out;
       case RUNE:
+      case RUNE1_FOLD:
         if (runes == null) {
           return "rune <null>";  // can't happen
         }

diff --git a/java/com/google/re2j/Machine.java b/java/com/google/re2j/Machine.java
@@ -307,6 +307,7 @@ private void step(Queue runq, Queue nextq, int pos, int nextPos, int c,
           break;
 
         case Inst.RUNE:
+        case Inst.RUNE1_FOLD:
           add = i.matchRune(c);
           break;
 
@@ -385,6 +386,7 @@ private Thread add(Queue q, int pc, int pos, int[] cap, int cond, Thread t) {
 
       case Inst.MATCH:
       case Inst.RUNE:
+      case Inst.RUNE1_FOLD:
       case Inst.RUNE1:
       case Inst.RUNE_ANY:
       case Inst.RUNE_ANY_NOT_NL:

diff --git a/java/com/google/re2j/Unicode.java b/java/com/google/re2j/Unicode.java
@@ -7,6 +7,8 @@
 
 package com.google.re2j;
 
+import java.util.Arrays;
+
 /**
  * Utilities for dealing with Unicode better than Java does.
  *
@@ -227,6 +229,18 @@ static int simpleFold(int r) {
     return toUpper(r);
   }
 
+
+  public static int[] fold(int r0) {
+    int[] folds = new int[3];
+    int foldsSize = 0;
+    int r1 = r0;
+    while((r1 = simpleFold(r1)) != r0) {
+      folds[foldsSize] = r1;
+      foldsSize++;
+    }
+    return Arrays.copyOf(folds, foldsSize);
+  }
+
   private Unicode() {}  // uninstantiable
 
 }