introduce prefix matching

fixes #641
fixes #604

Author: vladak

src/org/opensolaris/opengrok/analysis/AnalyzerGuru.java

@@ -98,6 +98,10 @@ public class AnalyzerGuru {
     private static final Map<String, FileAnalyzerFactory>
         ext = new HashMap<>();
 
+    /** Map from file prefixes to analyzer factories. */
+    private static final Map<String, FileAnalyzerFactory>
+        pre = new HashMap<>();
+
     // @TODO: have a comparator
     /** Map from magic strings to analyzer factories. */
     private static final SortedMap<String, FileAnalyzerFactory>
@@ -141,7 +145,7 @@ public class AnalyzerGuru {
             new CxxAnalyzerFactory(),
             new ShAnalyzerFactory(),
             PlainAnalyzerFactory.DEFAULT_INSTANCE,
-	    new UuencodeAnalyzerFactory(),
+            new UuencodeAnalyzerFactory(),
             new GZIPAnalyzerFactory(),
             new JavaAnalyzerFactory(),
             new JavaScriptAnalyzerFactory(),
@@ -177,6 +181,11 @@ private static void registerAnalyzer(FileAnalyzerFactory factory) {
             assert old == null :
                 "name '" + name + "' used in multiple analyzers";
         }
+        for (String prefix : factory.getPrefixes()) {
+            FileAnalyzerFactory old = pre.put(prefix, factory);
+            assert old == null :
+            "prefix '" + prefix + "' used in multiple analyzers";
+        }
         for (String suffix : factory.getSuffixes()) {
             FileAnalyzerFactory old = ext.put(suffix, factory);
             assert old == null :
@@ -191,6 +200,24 @@ private static void registerAnalyzer(FileAnalyzerFactory factory) {
         factories.add(factory);
     }
 
+   /**
+     *  Instruct the AnalyzerGuru to use a given analyzer for a given
+     *  file prefix.
+     *  @param prefix the file prefix to add
+     *  @param factory   a factory which creates
+     *                   the analyzer to use for the given extension
+     *                  (if you pass null as the analyzer, you will disable
+     *                   the analyzer used for that extension)
+     */
+    public static void addPrefix(String prefix,
+                                    FileAnalyzerFactory factory) {
+        if (factory == null) {
+            pre.remove(prefix);
+        } else {
+            pre.put(prefix, factory);
+        }
+    }
+
     /**
      *  Instruct the AnalyzerGuru to use a given analyzer for a given
      *  file extension.
@@ -445,19 +472,37 @@ public static FileAnalyzerFactory find(InputStream in, String file)
     public static FileAnalyzerFactory find(String file) {
         String path = file;
         int i;
-        if (((i = path.lastIndexOf('/')) > 0 || (i = path.lastIndexOf('\\')) > 0)
+
+        // Get basename of the file first.
+        if (((i = path.lastIndexOf(File.separatorChar)) > 0)
             && (i + 1 < path.length())) {
             path = path.substring(i + 1);
         }
+
         int dotpos = path.lastIndexOf('.');
         if (dotpos >= 0) {
-            FileAnalyzerFactory factory =
+            FileAnalyzerFactory factory;
+
+            // Try matching the prefix.
+            if (dotpos > 0) {
+                factory =
+                    pre.get(path.substring(0, dotpos).toUpperCase(Locale.getDefault()));
+                if (factory != null) {
+                    return factory;
+                }
+            }
+
+            // Now try matching the suffix. We kind of consider this order (first
+            // prefix then suffix) to be workable although for sure there can be
+            // cases when this does not work.
+            factory =
                 ext.get(path.substring(dotpos + 1).toUpperCase(Locale.getDefault()));
             if (factory != null) {
                 return factory;
             }
         }
-        // file doesn't have any of the extensions we know, try full match
+
+        // file doesn't have any of the prefix or extensions we know, try full match
         return FILE_NAMES.get(path.toUpperCase(Locale.getDefault()));
     }
 

src/org/opensolaris/opengrok/analysis/FileAnalyzerFactory.java

@@ -44,6 +44,9 @@ public class FileAnalyzerFactory {
     private final ThreadLocal<FileAnalyzer> cachedAnalyzer;
     /** List of file names on which this kind of analyzer should be used. */
     private final List<String> names;
+    /** List of file prefixes on which this kind of analyzer should be
+     * used. */
+    private final List<String> prefixes;
     /** List of file extensions on which this kind of analyzer should be
      * used. */
     private final List<String> suffixes;
@@ -62,14 +65,15 @@ public class FileAnalyzerFactory {
      * Create an instance of {@code FileAnalyzerFactory}.
      */
     FileAnalyzerFactory() {
-        this(null, null, null, null, null, null);
+        this(null, null, null, null, null, null, null);
     }
 
     /**
      * Construct an instance of {@code FileAnalyzerFactory}. This constructor
      * should be used by subclasses to override default values.
      *
      * @param names list of file names to recognize (possibly {@code null})
+     * @param prefixes list of prefixes to recognize (possibly {@code null})
      * @param suffixes list of suffixes to recognize (possibly {@code null})
      * @param magics list of magic strings to recognize (possibly {@code null})
      * @param matcher a matcher for this analyzer (possibly {@code null})
@@ -78,11 +82,12 @@ public class FileAnalyzerFactory {
      * Genre.DATA} is used)
      */
     protected FileAnalyzerFactory(
-            String[] names, String[] suffixes, String[] magics,
-            Matcher matcher, String contentType,
+            String[] names, String[] prefixes, String[] suffixes,
+            String[] magics, Matcher matcher, String contentType,
             Genre genre) {
         cachedAnalyzer = new ThreadLocal<FileAnalyzer>();
         this.names = asList(names);
+        this.prefixes = asList(prefixes);
         this.suffixes = asList(suffixes);
         this.magics = asList(magics);
         if (matcher == null) {
@@ -117,6 +122,14 @@ final List<String> getFileNames() {
         return names;
     }
 
+    /**
+     * Get the list of file prefixes recognized by this analyzer.
+     * @return list of prefixes
+     */
+    final List<String> getPrefixes() {
+        return prefixes;
+    }
+
     /**
      * Get the list of file extensions recognized by this analyzer.
      * @return list of suffixes

src/org/opensolaris/opengrok/analysis/PathTokenizer.java

@@ -28,7 +28,6 @@
 import org.apache.lucene.analysis.Tokenizer;
 import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
 import org.apache.lucene.analysis.tokenattributes.OffsetAttribute;
-import org.apache.lucene.analysis.tokenattributes.PositionIncrementAttribute;
 
 /**
  * Tokenizer for paths filenames and extensions Input:

src/org/opensolaris/opengrok/analysis/archive/BZip2AnalyzerFactory.java

@@ -36,7 +36,7 @@ public class BZip2AnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public BZip2AnalyzerFactory() {
-        super(null, SUFFIXES, MAGICS, null, null, null);
+        super(null, null, SUFFIXES, MAGICS, null, null, null);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/archive/GZIPAnalyzerFactory.java

@@ -37,7 +37,7 @@ public class GZIPAnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public GZIPAnalyzerFactory() {
-        super(null, SUFFIXES, MAGICS, null, null, null);
+        super(null, null, SUFFIXES, MAGICS, null, null, null);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/archive/TarAnalyzerFactory.java

@@ -34,7 +34,7 @@ public class TarAnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public TarAnalyzerFactory() {
-        super(null, SUFFIXES, null, null, null, Genre.XREFABLE);
+        super(null, null, SUFFIXES, null, null, null, Genre.XREFABLE);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/archive/ZipAnalyzerFactory.java

@@ -99,7 +99,7 @@ public FileAnalyzerFactory isMagic(byte[] contents, InputStream in)
             new ZipAnalyzerFactory();
 
     private ZipAnalyzerFactory() {
-        super(null, SUFFIXES, null, MATCHER, null, Genre.XREFABLE);
+        super(null, null, SUFFIXES, null, MATCHER, null, Genre.XREFABLE);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/c/CAnalyzerFactory.java

@@ -49,7 +49,7 @@ public class CAnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public CAnalyzerFactory() {
-        super(null, SUFFIXES, null, null, "text/plain", Genre.PLAIN);
+        super(null, null, SUFFIXES, null, null, "text/plain", Genre.PLAIN);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/c/CxxAnalyzerFactory.java

@@ -45,7 +45,7 @@ public class CxxAnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public CxxAnalyzerFactory() {
-        super(null, SUFFIXES, null, null, "text/plain", Genre.PLAIN);
+        super(null, null, SUFFIXES, null, null, "text/plain", Genre.PLAIN);
     }
 
     @Override

src/org/opensolaris/opengrok/analysis/csharp/CSharpAnalyzerFactory.java

@@ -43,7 +43,7 @@ public class CSharpAnalyzerFactory extends FileAnalyzerFactory {
     };
 
     public CSharpAnalyzerFactory() {
-        super(null, SUFFIXES, MAGICS, null, "text/plain", Genre.PLAIN);
+        super(null, null, SUFFIXES, MAGICS, null, "text/plain", Genre.PLAIN);
     }
 
     @Override