Merge pull request #102 from furfurylic/timidity

Add "timid" attribute to Output sink to avoid unnecessary touching

UsersGuide.asciidoc

@@ -324,6 +324,8 @@ The XPath expression can include names which belong some namespaces only when th
 |mkdirs|Whether this sink creates parent directories of the destination file if needed.| No; defaults to +yes+
 
 |force|Whether this sink creates output files even if existing files seem new enough.| No; defaults to +no+
+
+|timid|Whether this sink avoids overwriting existing files which already have identical contents to be written.| No; defaults to +no+
 |=================
 
 ==== Nested elements

src/net/furfurylic/chionographis/CachingResolver.java

@@ -40,8 +40,6 @@ final class CachingResolver implements EntityResolver, URIResolver {
     private static final NetResourceCache<byte[]> BYTES = new NetResourceCache<>();
     private static final NetResourceCache<Source> TREES = new NetResourceCache<>();
 
-    private static final Pool<byte[]> BUFFER = new Pool<>(() -> new byte[4096]);
-
     private Consumer<URI> listenStored_;
     private Consumer<URI> listenHit_;
 
@@ -76,7 +74,7 @@ public InputSource resolveEntity(String publicId, String systemId)
                     return Files.readAllBytes(Paths.get(u));
                 } else {
                     ByteArrayOutputStream bytes = new ByteArrayOutputStream();
-                    byte[] buffer = BUFFER.get();
+                    byte[] buffer = Pool.BYTES.get();
                     try {
                         try (InputStream in = u.toURL().openStream()) {
                             int length;
@@ -85,8 +83,9 @@ public InputSource resolveEntity(String publicId, String systemId)
                             }
                         }
                     } finally {
-                        BUFFER.release(buffer);
+                        Pool.BYTES.release(buffer);
                     }
+                    // TODO: Can make more efficient (by avoiding copy)
                     return bytes.toByteArray();
                 }
 

src/net/furfurylic/chionographis/Output.java

@@ -40,14 +40,15 @@
  * An <i>Output</i> {@linkplain Sink sink} writes each source document into an filesystem file.
  */
 public final class Output extends Sink {
-    private static final Pool<OutputStream> BUFFER =
+    private static final Pool<ExposingByteArrayOutputStream> BUFFER =
         new Pool<>(() -> new ExposingByteArrayOutputStream());
 
     private Path destDir_ = null;
     private Path dest_ = null;
     private boolean mkDirs_ = true;
     private String referent_ = null;
     private boolean force_ = false;
+    private boolean timid_ = false;
     private FileNameMapper mapper_ = null;
 
     private Logger logger_;
@@ -137,7 +138,7 @@ public void setRefer(String refer) {
      * if necessary. Defaulted to "no".
      *
      * @param mkDirs
-     *      {@code true} if make parent directories; {@code false} otherwise.
+     *      {@code true} if makes parent directories; {@code false} otherwise.
      */
     public void setMkDirs(boolean mkDirs) {
         mkDirs_ = mkDirs;
@@ -154,6 +155,19 @@ public void setForce(boolean force) {
         force_ = force;
     }
 
+    /**
+     * Sets whether this sink should compare existing destination files with the contents
+     * about to be written and avoid overwriting them if not necessary.
+     *
+     * @param timid
+     *      {@code true} if avoids unnecessary overwriting; {@code false} otherwise.
+     *
+     * @since 1.1
+     */
+    public void setTimid(boolean timid) {
+        timid_ = timid;
+    }
+
     /**
      * Installs a file mapper.
      * The file mapper maps a source file name to an destination file name.
@@ -352,14 +366,27 @@ void finishOne(Result result) {
                         Files.createDirectories(parent);
                     }
                 }
+
                 Path absolute = mapped.toAbsolutePath();
+
+                if (timid_) {
+                    File file = absolute.toFile();
+                    if (file.exists() && (file.length() == out.size())
+                     && hasIdenticalContent(file, out.buffer())) {
+                        logger_.log(this, "No need to overwrite the output file: " + absolute,
+                            Logger.Level.FINE);
+                        continue;
+                    }
+                }
+
                 logger_.log(this, "Creating " + absolute, Logger.Level.FINE);
                 // We take advantage of FileChannel for its capability to be interrupted
                 try {
                     try (FileChannel channel = FileChannel.open(
                             absolute, StandardOpenOption.WRITE, StandardOpenOption.CREATE)) {
                         channel.write(ByteBuffer.wrap(out.buffer(), 0, out.size()));
                     }
+                    countInBundle_.incrementAndGet();
                 } catch (IOException e) {
                     logger_.log(this, "Failed to create " + absolute, Logger.Level.WARN);
                     logger_.log(this, e, "  Cause: ", Logger.Level.INFO, Logger.Level.VERBOSE);
@@ -371,16 +398,35 @@ void finishOne(Result result) {
         } finally {
             placeBackBuffer(out);
         }
+    }
 
-        countInBundle_.incrementAndGet();
+    private boolean hasIdenticalContent(File file, byte[] content) throws IOException {
+        byte[] bytes = Pool.BYTES.get();
+        try (FileChannel in = FileChannel.open(file.toPath(), StandardOpenOption.READ)) {
+            ByteBuffer buffer = ByteBuffer.wrap(bytes);
+            int length;
+            int head = 0;
+            while ((length = in.read(buffer)) > -1) {
+                buffer.limit(length);
+                buffer.rewind();
+                if (!buffer.equals(ByteBuffer.wrap(content, head, length))) {
+                    return false;
+                }
+                head += length;
+            }
+            return true;
+        } finally {
+            Pool.BYTES.release(bytes);
+        }
     }
 
     @Override
     void abortOne(Result result) {
-        placeBackBuffer((ByteArrayOutputStream) ((OutputStreamResult) result).getOutputStream());
+        placeBackBuffer(
+            (ExposingByteArrayOutputStream) ((OutputStreamResult) result).getOutputStream());
     }
 
-    private void placeBackBuffer(ByteArrayOutputStream buffer) {
+    private void placeBackBuffer(ExposingByteArrayOutputStream buffer) {
         buffer.reset();
         BUFFER.release(buffer);
     }

src/net/furfurylic/chionographis/Pool.java

@@ -23,6 +23,8 @@
  */
 final class Pool<T> {
 
+    public static final Pool<byte[]> BYTES = new Pool<>(() -> new byte[4096]);
+
     private final ReentrantLock lock_ = new ReentrantLock();
     private Supplier<? extends T> create_;
     private SoftReference<Queue<T>> pool_;

test/crossing/input-timid/input.xml

@@ -0,0 +1 @@
+<X><Y><Z message="abc"/></Y></X>

test/crossing/input-timid/placeholder1.txt

@@ -0,0 +1 @@
+[X:[Y:[Z(message=abc):]]]

test/crossing/input-timid/placeholder2.txt

@@ -0,0 +1 @@
+[X:[Y:[Z(message=abc):]]x

test/crossing/input-timid/placeholder3.txt

@@ -0,0 +1 @@
+[X:[Y:[Z(message=abc):]]

test/test.xml

@@ -35,6 +35,7 @@
     <antcall target="snip-ns"/>
     <antcall target="crossing-cache"/>
     <antcall target="crossing-cache-refer-content"/>
+    <antcall target="crossing-timid"/>
     <antcall target="crossing-parallel-sinks"/>
     <antcall target="crossing-abort-sources"/>
     <antcall target="crossing-abort-sinks-dom"/>
@@ -1111,6 +1112,55 @@
     <delete dir="${dir.output}"/>
   </target>
 
+  <target name="crossing-timid">
+    <property name="test.prefix" value="crossing"/>
+    <property name="test.title" value="timid"/>
+    <property name="test.name" value="${test.prefix}-${test.title}"/>
+    <property name="dir.input" location="${test.prefix}/input-${test.title}"/>
+    <property name="dir.output" location="${test.prefix}/output-${test.title}"/>
+
+    <delete dir="${dir.output}"/>
+    <mkdir dir="${dir.output}"/>
+
+    <copy file="${dir.input}/input.xml" tofile="${dir.output}/input1.xml"/>
+    <copy file="${dir.input}/input.xml" tofile="${dir.output}/input2.xml"/>
+    <copy file="${dir.input}/input.xml" tofile="${dir.output}/input3.xml"/>
+    <copy file="${dir.input}/input.xml" tofile="${dir.output}/input4.xml"/>
+    <copy todir="${dir.output}">
+      <fileset dir="${dir.input}" includes="placeholder*.txt"/>
+      <mapper type="glob" from="placeholder*.txt" to="actual*.txt"/>
+    </copy>
+    <touch file="${dir.output}/input1.xml" datetime="01/01/2001 00:01:10 AM"/>  <!-- newer -->
+    <touch file="${dir.output}/input2.xml" datetime="01/01/2001 00:01:10 AM"/>  <!-- newer -->
+    <touch file="${dir.output}/input3.xml" datetime="01/01/2001 00:01:10 AM"/>  <!-- newer -->
+    <touch file="${dir.output}/actual1.txt" datetime="01/01/2001 00:01:00 AM"/> <!-- older -->
+    <touch file="${dir.output}/actual2.txt" datetime="01/01/2001 00:01:00 AM"/> <!-- older -->
+    <touch file="${dir.output}/actual3.txt" datetime="01/01/2001 00:01:00 AM"/> <!-- older -->
+
+    <chionographis srcdir="${dir.output}" includes="input*.xml" cache="no">
+      <transform style="flatten.xsl" cache="no">
+        <output destdir="${dir.output}" timid="yes">
+          <globmapper from="input*.xml" to="actual*.txt"/>
+        </output>
+      </transform>
+    </chionographis>
+
+    <!-- 1: identical: not touched -->
+    <assertfilelastmodified name="${test.name} - 1"
+      file="${dir.output}/actual1.txt" datetime="01/01/2001 00:01:00 AM"/>
+    <!-- 2: equal length and different content: touched -->
+    <assertfileeq name="${test.name} - 2"
+      expected="${dir.input}/placeholder1.txt" actual="${dir.output}/actual2.txt"/>
+    <!-- 3: different length: touched -->
+    <assertfileeq name="${test.name} - 3"
+      expected="${dir.input}/placeholder1.txt" actual="${dir.output}/actual3.txt"/>
+    <!-- 4: missing in destination: touched -->
+    <assertfileeq name="${test.name} - 4"
+      expected="${dir.input}/placeholder1.txt" actual="${dir.output}/actual4.txt"/>
+
+    <delete dir="${dir.output}"/>
+  </target>
+
   <target name="crossing-parallel-sinks">
     <property name="test.prefix" value="crossing"/>
     <property name="test.title" value="parallel-sinks"/>