Merge pull request #16 from zoho/hawking_dev

Parser Model Enhancement, New bug fixes, EUPL licence removed
zoho · May 11, 2021 · 931ed76 · 931ed76
2 parents 7ad23a7 + 23cb95a
commit 931ed76
Show file tree

Hide file tree

Showing 5 changed files with 35 additions and 11 deletions.
diff --git a/pom.xml b/pom.xml
@@ -4,7 +4,7 @@
 
     <groupId>com.zoho</groupId>
     <artifactId>hawking</artifactId>
-    <version>0.1.1</version>
+    <version>0.1.2</version>
     <packaging>jar</packaging>
     <name>Hawking</name>
     <description>Hawking is a natural language date time parser that extracts date and time from text with context and parse to the required format.</description>
@@ -34,10 +34,6 @@
             <name>GPL-v3.0</name>
             <url>http://www.gnu.org/licenses/gpl-3.0.txt</url>
         </license>
-        <license>
-            <name>EUPL-v1.1</name>
-            <url>http://joinup.ec.europa.eu/system/files/ES/EUPL%20v.1.1%20-%20Licencia.pdf</url>
-        </license>
     </licenses>
 
     <properties>

diff --git a/src/main/java/com/zoho/hawking/language/english/Recognizer.java b/src/main/java/com/zoho/hawking/language/english/Recognizer.java
@@ -44,6 +44,7 @@ public static ParsedDate recognize(String input) {
         }
         parsedDate.setOutputWithoffsets(dateList);
         parsedDate.setTaggedWithXML(taggedWithXML.toString().trim());
+        parsedDate = tagAlternator(input, parsedDate);
         parsedDate = tagShrinker(input, parsedDate);
         LOGGER.info("Recoginzer Regex Tagged Sequence::::"+ parsedDate.getTaggedWithXML()+":::::");
         return parsedDate;
@@ -81,7 +82,7 @@ public static Map<String, String> tagPredictor(String input, List<Triple<String,
 
         return TagUtils.tagRegulator(input, tagList, tagsEach);
     }
-    private static  ParsedDate tagShrinker(String parseText, ParsedDate parserDateCurrent) {
+    private static ParsedDate tagShrinker(String parseText, ParsedDate parserDateCurrent) {
         List<Triple<String, Integer, Integer>> triples = parserDateCurrent.getOutputWithOffsets();
         for (int i = 0; i < triples.size(); i++) {
             Triple<String, Integer, Integer> triple = triples.get(i);
@@ -122,6 +123,27 @@ private static  ParsedDate tagShrinker(String parseText, ParsedDate parserDateCu
         return parserDateCurrent;
     }
 
+    private static ParsedDate tagAlternator(String parseText, ParsedDate parserDateCurrent) {
+        List<Triple<String, Integer, Integer>> triples = parserDateCurrent.getOutputWithOffsets();
+        String tag_xml = parserDateCurrent.getTaggedWithXML();
+        if ((tag_xml.contains("day_of_week") || tag_xml.contains("current_day")) && tag_xml.contains("month_of_year") && tag_xml.contains("exact_number")) {
+            List<Triple<String, Integer, Integer>> triple = parserDateCurrent
+                .getOutputWithOffsets();
+            String date_xml = parserDateCurrent.getTaggedWithXML();
+            for (int i = 0; i < triple.size(); i++) {
+                Triple<String, Integer, Integer> triplet = triples.get(i);
+                String tag = triplet.first();
+                if (tag.equals("day_of_week") || tag.equals("current_day")) {
+                    String text = parseText.substring(triplet.second(), triplet.third());
+                    triple.remove(i);
+                    parserDateCurrent.setOutputWithoffsets(triple);
+                    parserDateCurrent.setTaggedWithXML(date_xml.replace("<day_of_week>" + text + "</day_of_week>", "")); //No I18N
+                }
+            }
+        }
+        return parserDateCurrent;
+    }
+
 
 }
 
diff --git a/src/main/java/com/zoho/hawking/utils/DateTimeProperties.java b/src/main/java/com/zoho/hawking/utils/DateTimeProperties.java
@@ -167,8 +167,10 @@ private String removeTimeZone(String parsedText) {
         for (String timezone : TimeZoneExtractor.timeZoneList) {
             if (isContain(parsedText, timezone)) {
                 returnText = parsedText.replaceAll("\\b(?i)" + timezone + "\\b", ""); //No I18N
-                returnText = returnText.replaceAll(",", " "); //No I18N
+                returnText = returnText.replaceAll("([,“”\"~()@])", " "); //No I18N
+                returnText = returnText.replaceAll("(hrs|hr|Hr|Hrs)", " hrs"); //No I18N
                 returnText = returnText.replaceAll("date", "day"); //No I18N
+                returnText = returnText.replaceAll("nextweek", "next week"); //No I18N
                 returnText = returnText.replaceAll("Year", "year"); //No I18N
                 returnText = returnText.replaceAll("\\b(?i)" + "final", "last"); //No I18N
                 returnText = returnText.replaceAll("\\.$", " "); //No I18N
@@ -178,11 +180,14 @@ private String removeTimeZone(String parsedText) {
                 returnText = returnText.replaceAll("lunch", "1 PM"); //No I18N
                 returnText = returnText.replaceAll("dinner", "8 PM"); //No I18N
                 returnText = returnText.replaceAll("null",""); //No I18N
+                returnText = returnText.replaceAll("\\s{2,}", " ").trim();//No I18N
                 return returnText;
             }
         }
-        returnText = parsedText.replaceAll(",", " "); //No I18N
+        returnText = parsedText.replaceAll("([,“”\"~()@])", " "); //No I18N
+        returnText = returnText.replaceAll("(hrs|Hrs|Hr|hr)", " hrs"); //No I18N
         returnText = returnText.replaceAll("date", "day"); //No I18N
+        returnText = returnText.replaceAll("nextweek", "next week"); //No I18N
         returnText = returnText.replaceAll("Year", "year"); //No I18N
         returnText = returnText.replaceAll("\\b(?i)" + "final", "last"); //No I18N
         returnText = returnText.replaceAll("\\.$", " ").trim(); //No I18N
@@ -193,6 +198,7 @@ private String removeTimeZone(String parsedText) {
         returnText = returnText.replaceAll("lunch", "1 PM"); //No I18N
         returnText = returnText.replaceAll("dinner", "8 PM"); //No I18N
         returnText = returnText.replaceAll("null",""); //No I18N
+        returnText = returnText.replaceAll("\\s{2,}", " ").trim();//No I18N
         returnText = test.length() > 0 ? returnText : test;
         return returnText;
 

diff --git a/src/main/java/com/zoho/hawking/utils/RecognizerTagger.java b/src/main/java/com/zoho/hawking/utils/RecognizerTagger.java
@@ -4,7 +4,7 @@
 public class RecognizerTagger {
 
   private final static Pattern implictPrefix = Pattern.compile(
-      "^(beginning|upcoming|starting|previous|current|between|coming|within|ending|before|until|after|since|start|forty|this|next|last|past|from|till|rest|most|with|the|for|few|end|in|at|on|of|an|a)$");
+      "^(beginning|upcoming|starting|previous|current|between|coming|within|ending|before|until|after|since|start|forty|this|next|last|past|from|till|rest|most|with|the|for|few|end|in|at|on|of|by|an|a)$");
   private final static Pattern implictPostfix = Pattern.compile("^(back|ago)$");
 
   private final static Pattern second_span = Pattern.compile("^(second|seconds|sec|secs)$");
@@ -29,10 +29,10 @@ public class RecognizerTagger {
   private final static Pattern exact_date_2 = Pattern.compile("^(\\d{1,2})[-/]\\d{1,2}$");
 
   private final static Pattern exact_time_1 = Pattern
-      .compile("^(((0[0-9]|1[0-9]|2[0-3]|[0-9])([:.][0-5][0-9])?([:.][0-5][0-9])?)([AaPp][Mm]))$");
+      .compile("^(((0[0-9]|1[0-9]|2[0-3]|[0-9])([:.][0-5][0-9])?([:.][0-5][0-9])?)([AaPp][.]?[Mm]))$");
   private final static Pattern exact_time_2 = Pattern
       .compile("^(((0[0-9]|1[0-9]|2[0-3]|[0-9])([:.][0-5][0-9])([:.][0-5][0-9])?))$");
-  private final static Pattern exact_time_3 = Pattern.compile("^([AaPp][Mm])$");
+  private final static Pattern exact_time_3 = Pattern.compile("^([AaPp][.]?[Mm])$");
 
   private final static Pattern exact_year = Pattern.compile("^\\d{4}$");
   private final static Pattern exact_number_1 = Pattern.compile(

diff --git a/src/main/resources/parser/parser.crf.ser.gz b/src/main/resources/parser/parser.crf.ser.gz