Add specialized Byte/CharTrieOfOneSingleChar.

wrandelshofer · Oct 19, 2024 · 3c2f3fe · 3c2f3fe
1 parent 7cdd157
commit 3c2f3fe
Show file tree

Hide file tree

Showing 9 changed files with 109 additions and 12 deletions.
diff --git a/.idea/runConfigurations/Main_Dev_formatted_de_CH.xml b/.idea/runConfigurations/Main_Dev_formatted_de_CH.xml
diff --git a/.idea/runConfigurations/Main_Java17_canada_hex.xml b/.idea/runConfigurations/Main_Java17_canada_hex.xml
diff --git a/.../java/ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrie.java b/.../java/ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrie.java
@@ -4,6 +4,7 @@
  */
 package ch.randelshofer.fastdoubleparser.bte;
 
+import java.nio.charset.StandardCharsets;
 import java.util.HashSet;
 import java.util.LinkedHashSet;
 import java.util.Set;
@@ -39,20 +40,29 @@ static ByteTrie copyOf(Set<String> set, boolean ignoreCase) {
             case 0:
                 return new ByteTrieOfNone();
             case 1:
+                String str = set.iterator().next();
                 if (ignoreCase) {
-                    String str = set.iterator().next();
                     switch (str.length()) {
                         case 0:
                             return new ByteTrieOfNone();
                         case 1:
                             LinkedHashSet<String> newSet = new LinkedHashSet<>();
                             newSet.add(str.toLowerCase());
                             newSet.add(str.toUpperCase());
-                            return newSet.size() == 1 ? new ByteTrieOfOne(newSet) : new ByteTrieOfFew(newSet);
+                            if (newSet.size() == 1) {
+                                if (newSet.iterator().next().getBytes(StandardCharsets.UTF_8).length == 1) {
+                                    return new ByteTrieOfOneSingleByte(newSet);
+                                }
+                                return new ByteTrieOfOne(newSet);
+                            }
+                            return new ByteTrieOfFew(newSet);
                         default:
                             return new ByteTrieOfOneIgnoreCase(set);
                     }
                 }
+                if (set.iterator().next().getBytes(StandardCharsets.UTF_8).length == 1) {
+                    return new ByteTrieOfOneSingleByte(set);
+                }
                 return new ByteTrieOfOne(set);
             default:
                 if (ignoreCase) {

diff --git a/.../ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrieOfOne.java b/.../ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrieOfOne.java
@@ -15,10 +15,6 @@ public ByteTrieOfOne(Set<String> set) {
         chars = set.iterator().next().getBytes(StandardCharsets.UTF_8);
     }
 
-    public ByteTrieOfOne(byte[] chars) {
-        this.chars = chars;
-    }
-
 
     @Override
     public int match(byte[] str) {
@@ -27,6 +23,10 @@ public int match(byte[] str) {
 
     @Override
     public int match(byte[] str, int startIndex, int endIndex) {
+        /*
+        int mismatch = Arrays.mismatch(chars, 0, chars.length, str, startIndex,startIndex+ Math.min(endIndex - startIndex, chars.length));
+        return mismatch<0?chars.length:0;
+        */
         int i = 0;
         int limit = Math.min(endIndex - startIndex, chars.length);
         while (i < limit && str[i + startIndex] == chars[i]) {

diff --git a/...shofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrieOfOneSingleByte.java b/...shofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/bte/ByteTrieOfOneSingleByte.java
@@ -0,0 +1,34 @@
+/*
+ * @(#)ByteTrieOfOne.java
+ * Copyright © 2024 Werner Randelshofer, Switzerland. MIT License.
+ */
+package ch.randelshofer.fastdoubleparser.bte;
+
+import java.nio.charset.StandardCharsets;
+import java.util.Set;
+
+class ByteTrieOfOneSingleByte implements ByteTrie {
+    private final byte ch;
+
+    public ByteTrieOfOneSingleByte(Set<String> set) {
+        if (set.size() != 1) throw new IllegalArgumentException("set size must be 1, size=" + set.size());
+        byte[] chars = set.iterator().next().getBytes(StandardCharsets.UTF_8);
+        if (chars.length != 1) throw new IllegalArgumentException("char size must be 1, size=" + set.size());
+        ch = chars[0];
+    }
+
+    public ByteTrieOfOneSingleByte(byte ch) {
+        this.ch = ch;
+    }
+
+
+    @Override
+    public int match(byte[] str) {
+        return match(str, 0, str.length);
+    }
+
+    @Override
+    public int match(byte[] str, int startIndex, int endIndex) {
+        return startIndex < endIndex && str[startIndex] == ch ? 1 : 0;
+    }
+}
diff --git a/.../java/ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrie.java b/.../java/ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrie.java
@@ -58,6 +58,9 @@ public static CharTrie of(Set<String> set, boolean ignoreCase) {
             case 0:
                 return new CharTrieOfNone();
             case 1:
+                if (set.iterator().next().length() == 1) {
+                    return ignoreCase ? new CharTrieOfOneIgnoreCase(set) : new CharTrieOfOneSingleChar(set);
+                }
                 return ignoreCase ? new CharTrieOfOneIgnoreCase(set) : new CharTrieOfOne(set);
             default:
                 return ignoreCase ? new CharTrieOfFewIgnoreCase(set) : new CharTrieOfFew(set);

diff --git a/.../ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrieOfOne.java b/.../ch.randelshofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrieOfOne.java
@@ -40,11 +40,16 @@ public int match(char[] str) {
 
     @Override
     public int match(char[] str, int startIndex, int endIndex) {
+        /*
+        int mismatch = Arrays.mismatch(chars, 0, chars.length, str, startIndex,startIndex+ Math.min(endIndex - startIndex, chars.length));
+        return mismatch<0?chars.length:0;
+        */
         int i = 0;
         int limit = Math.min(endIndex - startIndex, chars.length);
         while (i < limit && str[i + startIndex] == chars[i]) {
             i++;
         }
         return i == chars.length ? chars.length : 0;
+
     }
 }
diff --git a/...shofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrieOfOneSingleChar.java b/...shofer.fastdoubleparser/ch/randelshofer/fastdoubleparser/chr/CharTrieOfOneSingleChar.java
@@ -0,0 +1,37 @@
+/*
+ * @(#)ByteTrieOfOne.java
+ * Copyright © 2024 Werner Randelshofer, Switzerland. MIT License.
+ */
+package ch.randelshofer.fastdoubleparser.chr;
+
+import java.util.Set;
+
+class CharTrieOfOneSingleChar implements CharTrie {
+    private final char ch;
+
+    public CharTrieOfOneSingleChar(Set<String> set) {
+        if (set.size() != 1) throw new IllegalArgumentException("set size must be 1, size=" + set.size());
+        char[] chars = set.iterator().next().toCharArray();
+        if (chars.length != 1) throw new IllegalArgumentException("char size must be 1, size=" + set.size());
+        ch = chars[0];
+    }
+
+    public CharTrieOfOneSingleChar(char ch) {
+        this.ch = ch;
+    }
+
+    @Override
+    public int match(CharSequence str, int startIndex, int endIndex) {
+        return startIndex < endIndex && str.charAt(startIndex) == ch ? 1 : 0;
+    }
+
+    @Override
+    public int match(char[] str) {
+        return match(str, 0, str.length);
+    }
+
+    @Override
+    public int match(char[] str, int startIndex, int endIndex) {
+        return startIndex < endIndex && str[startIndex] == ch ? 1 : 0;
+    }
+}
diff --git a/.../java/ch.randelshofer.fastdoubleparserdemo/ch/randelshofer/fastdoubleparserdemo/Main.java b/.../java/ch.randelshofer.fastdoubleparserdemo/ch/randelshofer/fastdoubleparserdemo/Main.java
@@ -229,13 +229,21 @@ public void loadFile(String filename) throws IOException {
     }
 
     private VarianceStatistics measure(Supplier<? extends Number> func, int numberOfTrials,
-                                       double confidenceLevel, double confidenceIntervalWidth, int minTrials) {
+                                       double confidenceLevel, double confidenceIntervalWidth, int minTrials, String name) {
         long t1;
         double confidenceWidth;
         long t2;
         double elapsed;
         VarianceStatistics stats = new VarianceStatistics();
 
+        // sleep to cooldown the processor
+        try {
+            System.out.println(name);
+            Thread.sleep(100);
+        } catch (InterruptedException e) {
+            // ignore
+        }
+
         // measure
         int trials = 0;
         do {
@@ -308,6 +316,7 @@ private void printStatsMarkdown(List<String> lines, double volumeMB, String name
         String reference = functions.get(name).reference;
         boolean isBaseline = reference.equals(name);
         String speedupOrBaseline = isBaseline ? "=" : "*";
+        Character first = baselines.isEmpty() ? null : baselines.values().iterator().next();
         System.out.printf("|%-40s|%7.2f|%4.1f %%|%7.2f|%9.2f|%7.2f%s%s|%-8s|\n",
                 name,
                 volumeMB * 1e9 / stats.getAverage(),
@@ -316,7 +325,7 @@ private void printStatsMarkdown(List<String> lines, double volumeMB, String name
                 stats.getAverage() / lines.size(),
                 speedup,
                 speedupOrBaseline,
-                baselines.getOrDefault(reference, baselines.values().iterator().next()),
+                baselines.getOrDefault(reference, first),
                 System.getProperty("java.version")
         );
     }
@@ -333,7 +342,7 @@ private void process(List<String> lines, Map<String, BenchmarkFunction> function
         for (Map.Entry<String, BenchmarkFunction> entry : entries) {
             System.out.print(".");
             System.out.flush();
-            VarianceStatistics warmup = measure(entry.getValue().supplier, WARMUP_NUMBER_OF_TRIALS, WARMUP_CONFIDENCE_LEVEL, WARMUP_CONFIDENCE_INTERVAL_WIDTH, WARMUP_MIN_TRIALS);
+            VarianceStatistics warmup = measure(entry.getValue().supplier, WARMUP_NUMBER_OF_TRIALS, WARMUP_CONFIDENCE_LEVEL, WARMUP_CONFIDENCE_INTERVAL_WIDTH, WARMUP_MIN_TRIALS, entry.getKey());
             warmupResults.put(entry.getKey(), warmup);
         }
         System.out.println();
@@ -352,7 +361,7 @@ private void process(List<String> lines, Map<String, BenchmarkFunction> function
             System.out.print(".");
             System.out.flush();
 
-            VarianceStatistics stats = measure(entry.getValue().supplier, MEASUREMENT_NUMBER_OF_TRIALS, MEASUREMENT_CONFIDENCE_LEVEL, MEASUREMENT_CONFIDENCE_INTERVAL_WIDTH, 1);
+            VarianceStatistics stats = measure(entry.getValue().supplier, MEASUREMENT_NUMBER_OF_TRIALS, MEASUREMENT_CONFIDENCE_LEVEL, MEASUREMENT_CONFIDENCE_INTERVAL_WIDTH, 1, entry.getKey());
             results.put(entry.getKey(), stats);
         }
         System.out.println();