adjust date in text detection to ignore some program version strings

like "3.1.2.0102" see http://mantis.tokeek.de/view.php?id=650
+ expand test case

2 files changed, 23 insertions, 1 deletions

diff --git a/source/net/yacy/document/DateDetection.java b/source/net/yacy/document/DateDetection.java
index 781101665..c5ecb0a1c 100644
--- a/source/net/yacy/document/DateDetection.java
+++ b/source/net/yacy/document/DateDetection.java
@@ -127,7 +127,7 @@ public class DateDetection {
     private final static Date TODAY = new Date();
     private final static int CURRENT_YEAR  = Integer.parseInt(CONFORM.format(TODAY).substring(0, 4)); // we need that to parse dates without given years, see the ShortStyle class
 
-    private final static String BODNCG = "(?:\\b|^)"; // begin of date non-capturing group
+    private final static String BODNCG = "(?:\\s|^)"; // begin of date non-capturing group
     private final static String EODNCG = "(?:[).:;! ]|$)"; // end of date non-capturing group
     private final static String SEPARATORNCG = "(?:/|-| - |\\.\\s|,\\s|\\.|,|\\s)"; // separator non-capturing group
     private final static String DAYCAPTURE = "(\\d{1,2})";
diff --git a/test/java/net/yacy/document/DateDetectionTest.java b/test/java/net/yacy/document/DateDetectionTest.java
index cedcea6de..eba322d7a 100644
--- a/test/java/net/yacy/document/DateDetectionTest.java
+++ b/test/java/net/yacy/document/DateDetectionTest.java
@@ -28,6 +28,9 @@ public class DateDetectionTest {
         testtext.add("1.1.2016");

         testtext.add("1. Januar 2016");

         testtext.add("2016, January 1.");

+

+        testtext.add("beginning text 1.1.2016");

+        testtext.add("line break\n1.1.2016");

         for (String text : testtext) {

             Date d = DateDetection.parseLine(text, 0);

 

@@ -82,4 +85,23 @@ public class DateDetectionTest {
         }

     }

 

+    /**

+     * Negative test of parseLine method, of class DateDetection

+     * with cases that represent NOT a date

+     */

+    @Test

+    public void testParseLineNoDate() {

+

+        // test input representations

+        Set<String> testtext = new LinkedHashSet();

+        testtext.add("3.1.2.0102"); // example of a program version string

+        // testtext.add("3.1.20.0102"); // date end-capture not working (on modification conflict with YMD parser)

+        testtext.add("v3.1.21");

+        testtext.add("v3.1.22.");

+

+        for (String text : testtext) {

+            Date d = DateDetection.parseLine(text, 0);

+            assertNull("not a date: " + text, d);

+        }

+    }

 }