Changeset 6828

Timestamp:

2004-02-18T13:12:08+13:00 (20 years ago)

Author:

mdewsnip

Message:

Added a transformation from plain text to a regular expression that matches it. This involves escaping special regular expression characters such as \().[]{}.

File:

• trunk/gli/src/org/greenstone/gatherer/util/Codec.java (modified) (4 diffs)

trunk/gli/src/org/greenstone/gatherer/util/Codec.java

@@ -48 +48 @@
     static final public String TEXT_TO_DOM = "TEXT_TO_DOM";
     static final public String TEXT_TO_GREENSTONE = "TEXT_TO_GREENSTONE";
+    static final public String TEXT_TO_REGEXP = "TEXT_TO_REGEXP";
     static final public String TEXT_TO_SHELL_UNIX = "TEXT_TO_SHELL_UNIX";
     static final public String TEXT_TO_SHELL_WINDOWS = "TEXT_TO_SHELL_WINDOWS";
@@ -102 +103 @@
     dom_to_text = null;
 
+    // Transform text into a regular expression that will match it
+    String[] text_to_regexp = {
+        "\\\\", "\\\\\\\\",
+        "\\(", "\\\\(",
+        "\\)", "\\\\)",
+        "\\[", "\\\\[",
+        "\\]", "\\\\]",
+        "\\{", "\\\\{",
+        "\\}", "\\\\}",
+        "\\.", "\\\\."
+    };
+    TRANSFORMS.put(TEXT_TO_REGEXP, text_to_regexp);
+    text_to_regexp = null;
+
     String[] encode_path = {
         "\\|",  "|",
@@ -191 +206 @@
     }
 
-    static public String transform(String raw, String transform) {
-        if(raw == null) {
-             return raw;
-        }
-    ///ystem.err.println("Transforming by "+transform+":\n" + raw);
+    static public String transform(String raw, String transform) {
+    if(raw == null) {
+        return raw;
+    }
+    // System.err.println("Transforming by "+transform+":\n" + raw);
     String processed = (String) CACHE.get(transform, raw);
     if(processed == null) {
@@ -207 +222 @@
         }
         }
-        //Gatherer.println("\n*** Transform: " + transform + " ***");
-        //Gatherer.println("*** Raw      : '" + raw + "'");
-        //Gatherer.println("*** Processed: '" + processed + "'");
+        Gatherer.println("\n*** Transform: " + transform + " ***");
+        Gatherer.println("*** Raw      : '" + raw + "'");
+        Gatherer.println("*** Processed: '" + processed + "'");
         // If cache is at maximum size, empty it and start again
         if(CACHE.size() == MAX_CACHE_SIZE) {