Optimizes some of the character patterns

petitparser · Dec 27, 2014 · 7dbc8f1 · 7dbc8f1
1 parent c327d20
commit 7dbc8f1
Show file tree

Hide file tree

Showing 2 changed files with 41 additions and 24 deletions.
diff --git a/petitparser-core/src/main/java/org/petitparser/parser/primitive/CharacterPredicate.java b/petitparser-core/src/main/java/org/petitparser/parser/primitive/CharacterPredicate.java
@@ -2,7 +2,6 @@
 
 import org.petitparser.parser.Parser;
 
-import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.List;
 import java.util.stream.Collectors;
@@ -25,7 +24,7 @@ static CharacterPredicate any() {
    */
   static CharacterPredicate anyOf(String string) {
     List<CharacterRange> ranges = string.chars()
-        .mapToObj((value) -> new CharacterRange((char) value))
+        .mapToObj((value) -> new CharacterRange((char) value, (char) value))
         .collect(Collectors.toList());
     return CharacterRange.toCharacterPredicate(ranges);
   }
@@ -42,7 +41,7 @@ static CharacterPredicate none() {
    */
   static CharacterPredicate noneOf(String string) {
     List<CharacterRange> ranges = string.chars()
-        .mapToObj((value) -> new CharacterRange((char) value))
+        .mapToObj((value) -> new CharacterRange((char) value, (char) value))
         .collect(Collectors.toList());
     return CharacterRange.toCharacterPredicate(ranges).not();
   }
@@ -55,12 +54,35 @@ static CharacterPredicate of(char character) {
   }
 
   /**
-   * Returns a character predicate that matches any character between {@code start} and {@code stop}.
+   * Returns a character predicate that matches any character between {@code start} and {@code
+   * stop}.
    */
   static CharacterPredicate range(char start, char stop) {
     return value -> start <= value && value <= stop;
   }
 
+  /**
+   * Returns a character predicate that matches character ranges between {@code starts} and {@code
+   * stops}.
+   */
+  static CharacterPredicate ranges(char[] starts, char[] stops) {
+    if (starts.length != stops.length) {
+      throw new IllegalArgumentException("Invalid ranges.");
+    }
+    for (int i = 0; i < starts.length; i++) {
+      if (starts[i] > stops[i]) {
+        throw new IllegalArgumentException("Invalid range: " + starts[i] + "-" + stops[i]);
+      }
+      if (i + 1 < starts.length && starts[i + 1] <= stops[i]) {
+        throw new IllegalArgumentException("Invalid sequence.");
+      }
+    }
+    return value -> {
+      int index = Arrays.binarySearch(starts, value);
+      return index >= 0 || index < -1 && value <= stops[-index - 2];
+    };
+  }
+
   /**
    * Returns a character predicate that matches the provided pattern.
    */
@@ -70,7 +92,7 @@ static CharacterPredicate pattern(String pattern) {
 
   class PatternParser {
     static final Parser PATTERN_SIMPLE = CharacterParser.any()
-        .map((Character value) -> new CharacterRange(value));
+        .map((Character value) -> new CharacterRange(value, value));
     static final Parser PATTERN_RANGE = CharacterParser.any()
         .seq(CharacterParser.of('-'))
         .seq(CharacterParser.any())
@@ -156,5 +178,4 @@ public CharacterPredicate or(CharacterPredicate... others) {
       return new AltCharacterPredicate(array);
     }
   }
-
 }
diff --git a/petitparser-core/src/main/java/org/petitparser/parser/primitive/CharacterRange.java b/petitparser-core/src/main/java/org/petitparser/parser/primitive/CharacterRange.java
@@ -38,32 +38,28 @@ public int compare(CharacterRange first, CharacterRange second) {
       }
     }
 
-    // 3. build the corresponding predicates
-    List<CharacterPredicate> predicates = new ArrayList<>();
-    for (CharacterRange range : mergedRanges) {
-      if (range.stop - range.start > 1) {
-        predicates.add(CharacterPredicate.range(range.start, range.stop));
-      } else {
-        for (char value = range.start; value <= range.stop; value++) {
-          predicates.add(CharacterPredicate.of(value));
-        }
+    // 3. build the best resulting predicates
+    if (mergedRanges.isEmpty()) {
+      return CharacterPredicate.none();
+    } else if (mergedRanges.size() == 1) {
+      return mergedRanges.get(0).start == mergedRanges.get(0).stop
+          ? CharacterPredicate.of(mergedRanges.get(0).start)
+          : CharacterPredicate.range(mergedRanges.get(0).start, mergedRanges.get(0).stop);
+    } else {
+      char[] starts = new char[mergedRanges.size()];
+      char[] stops = new char[mergedRanges.size()];
+      for (int i = 0; i < mergedRanges.size(); i++) {
+        starts[i] = mergedRanges.get(i).start;
+        stops[i] = mergedRanges.get(i).stop;
       }
+      return CharacterPredicate.ranges(starts, stops);
     }
 
-    // 4. when necessary build a composite predicate
-    return predicates.size() == 1
-        ? predicates.get(0)
-        : new CharacterPredicate.AltCharacterPredicate(
-            predicates.toArray(new CharacterPredicate[predicates.size()]));
   }
 
   private final char start;
   private final char stop;
 
-  CharacterRange(char start) {
-    this(start, start);
-  }
-
   CharacterRange(char start, char stop) {
     this.start = start;
     this.stop = stop;