[WIP]

bkiers · Dec 21, 2024 · 7e66c66 · 7e66c66
1 parent b184c83
commit 7e66c66
Show file tree

Hide file tree

Showing 7 changed files with 97 additions and 122 deletions.
diff --git a/src/main/java/liqp/filters/date/fuzzy/PartRecognizer.java b/src/main/java/liqp/filters/date/fuzzy/PartRecognizer.java
@@ -1,7 +1,6 @@
 package liqp.filters.date.fuzzy;
 
-import static liqp.filters.date.fuzzy.extractors.Extractors.ISO8601YMDPatternExtractor;
-import static liqp.filters.date.fuzzy.extractors.Extractors.englishDateExtractor;
+import static liqp.filters.date.fuzzy.extractors.Extractors.allYMDPatternExtractor;
 import static liqp.filters.date.fuzzy.extractors.Extractors.fullMonthExtractor;
 import static liqp.filters.date.fuzzy.extractors.Extractors.fullWeekdaysExtractor;
 import static liqp.filters.date.fuzzy.extractors.Extractors.plainYearExtractor;
@@ -55,14 +54,7 @@ List<Part> recognizePart(List<Part> parts, DatePatternRecognizingContext ctx) {
             ctx.hasTime = false;
         }
         if (notSet(ctx.hasYear)) {
-            LookupResult result = lookup(parts, ISO8601YMDPatternExtractor.get(ctx.locale));
-            if (result.found) {
-                ctx.hasYear = true;
-                ctx.hasMonth = true;
-                ctx.hasDay = true;
-                return result.parts;
-            }
-            result = lookup(parts, englishDateExtractor.get(ctx.locale));
+            LookupResult result = lookup(parts, allYMDPatternExtractor.get(ctx.locale));
             if (result.found) {
                 ctx.hasYear = true;
                 ctx.hasMonth = true;

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/AllYMDPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/AllYMDPatternExtractor.java
@@ -0,0 +1,76 @@
+package liqp.filters.date.fuzzy.extractors;
+
+import static liqp.filters.date.fuzzy.extractors.AnyYMDPatternExtractor.pD;
+import static liqp.filters.date.fuzzy.extractors.AnyYMDPatternExtractor.pM;
+import static liqp.filters.date.fuzzy.extractors.AnyYMDPatternExtractor.pY2;
+import static liqp.filters.date.fuzzy.extractors.AnyYMDPatternExtractor.pY4;
+import static liqp.filters.date.fuzzy.extractors.AnyYMDPatternExtractor.pp;
+
+import java.util.ArrayList;
+import java.util.List;
+import liqp.filters.date.fuzzy.PartExtractor;
+
+public class AllYMDPatternExtractor implements PartExtractor {
+
+    private final List<AnyYMDPatternExtractor> extractors = new ArrayList<>();
+
+    public AllYMDPatternExtractor() {
+        AnyYMDPatternExtractor iSO8601Y4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY4(), pp("-"), pM(), pp("-"), pD()); // yyyy-MM-dd
+        extractors.add(iSO8601Y4MDPatternExtractor);
+        AnyYMDPatternExtractor americanY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pM(), pp("/"), pD(), pp("/"), pY4()); // MM/dd/yyyy
+        extractors.add(americanY4MDPatternExtractor);
+        // next are top-rated locale formats, according to gpt
+        AnyYMDPatternExtractor indianY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("-"), pM(), pp("-"), pY4()); // d-M-yyyy
+        extractors.add(indianY4MDPatternExtractor);
+        AnyYMDPatternExtractor chineseY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY4(), pp("/"), pM(), pp("/"), pD()); // yyyy/M/d
+        extractors.add(chineseY4MDPatternExtractor);
+        AnyYMDPatternExtractor englishY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("/"), pM(), pp("/"), pY4()); // d/M/yyyy
+        extractors.add(englishY4MDPatternExtractor);
+        AnyYMDPatternExtractor slavicY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("."), pM(), pp("."), pY4());
+        extractors.add(slavicY4MDPatternExtractor);
+        AnyYMDPatternExtractor coldEuropeY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY4(), pp("-"), pM(), pp("-"), pD()); // yyyy-MM-dd
+        extractors.add(coldEuropeY4MDPatternExtractor);
+        AnyYMDPatternExtractor espanaY4MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY4(), pp("-"), pM(), pp("-"), pD()); // yyyy/MM/dd
+        extractors.add(espanaY4MDPatternExtractor);
+        AnyYMDPatternExtractor americanY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pM(), pp("/"), pD(), pp("/"), pY4()); // MM/dd/yy
+        extractors.add(americanY2MDPatternExtractor);
+        AnyYMDPatternExtractor indianY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("-"), pM(), pp("-"), pY2()); // d-M-yy
+        extractors.add(indianY2MDPatternExtractor);
+        AnyYMDPatternExtractor chineseY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY2(), pp("/"), pM(), pp("/"), pD()); // yyyy/M/d
+        extractors.add(chineseY2MDPatternExtractor);
+        AnyYMDPatternExtractor englishY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("/"), pM(), pp("/"), pY2()); // d/M/yy
+        extractors.add(englishY2MDPatternExtractor);
+        AnyYMDPatternExtractor slavicY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pD(), pp("."), pM(), pp("."), pY2());
+        extractors.add(slavicY2MDPatternExtractor);
+        AnyYMDPatternExtractor coldEuropeY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY2(), pp("-"), pM(), pp("-"), pD()); // yy-MM-dd
+        extractors.add(coldEuropeY2MDPatternExtractor);
+        AnyYMDPatternExtractor espanaY2MDPatternExtractor = new AnyYMDPatternExtractor(
+                pY2(), pp("-"), pM(), pp("-"), pD()); // yy/MM/dd
+        extractors.add(espanaY2MDPatternExtractor);
+    }
+
+    @Override
+    public PartExtractorResult extract(String source) {
+        for (AnyYMDPatternExtractor extractor : extractors) {
+            PartExtractorResult result = extractor.extract(source);
+            if (result.found) {
+                return result;
+            }
+        }
+        return new PartExtractorResult();
+    }
+}
diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/AnyYMDPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/AnyYMDPatternExtractor.java
@@ -7,22 +7,22 @@
 import java.util.Optional;
 import java.util.regex.Matcher;
 
-abstract class AnyYMDPatternExtractor extends RegexPartExtractor {
+class AnyYMDPatternExtractor extends RegexPartExtractor {
 
     public enum RuleType {
         Y, M, D, PUNCTUATION;
     }
     public static class RulePart {
         private final RuleType type;
-        private final Integer[] length;
+        private final Integer length;
         private final String content;
         private RulePart(RuleType type, String content) {
             this.type = type;
             this.content = content;
-            this.length = new Integer[0];
+            this.length = null;
         }
 
-        private RulePart(RuleType type, Integer[] length) {
+        private RulePart(RuleType type, Integer length) {
             this.type = type;
             this.length = length;
             this.content = null;
@@ -32,17 +32,17 @@ private RulePart(RuleType type, Integer[] length) {
     static RulePart pp(String content) {
         return new RulePart(RuleType.PUNCTUATION, content);
     }
-    static RulePart pY(Integer length) {
-        return new RulePart(RuleType.Y, new Integer[]{length});
+    static RulePart pY4() {
+        return new RulePart(RuleType.Y, 4);
     }
-    static RulePart pY(Integer length1, Integer length2) {
-        return new RulePart(RuleType.Y, new Integer[]{length1, length2});
+    static RulePart pY2() {
+        return new RulePart(RuleType.Y, 2);
     }
     static RulePart pM() {
-        return new RulePart(RuleType.M, (Integer[])null);
+        return new RulePart(RuleType.M, (Integer)null);
     }
     static RulePart pD() {
-        return new RulePart(RuleType.D, (Integer[])null);
+        return new RulePart(RuleType.D, (Integer)null);
     }
     private final RulePart[] partsInOrder;
     protected AnyYMDPatternExtractor(RulePart... partsInOrder) {
@@ -54,18 +54,17 @@ private static String reconstructPattern(RulePart[] partsInOrder) {
         StringBuilder sb = new StringBuilder("(?:^|.*?\\D)");
         for (RulePart part : partsInOrder) {
             if (part.type == RuleType.PUNCTUATION) {
-                sb.append(part.content);
+                if (".".equals(part.content)) {
+                    sb.append("\\.");
+                } else {
+                    sb.append(part.content);
+                }
             } else {
                 if (part.type == RuleType.Y) {
-                    if (part.length == null || part.length.length == 0) {
+                    if (part.length == null) {
                         throw new IllegalArgumentException("Year part must have length");
                     }
-                    if (part.length.length == 1) {
-                        sb.append("(?<year>\\d{").append(part.length[0]).append("})");
-                    } else {
-                        sb.append("(?<year>\\d{").append(part.length[0]).append("}|\\d{")
-                                .append(part.length[1]).append("})");
-                    }
+                    sb.append("(?<year>\\d{").append(part.length).append("})");
                 } else if (part.type == RuleType.M) {
                     sb.append("(?<month>0?[1-9]|1[0-2])");
                 } else if (part.type == RuleType.D) {

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/EnglishDMYPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/EnglishDMYPatternExtractor.java
diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/Extractors.java b/src/main/java/liqp/filters/date/fuzzy/extractors/Extractors.java
@@ -82,21 +82,14 @@ public PartExtractor get(Locale locale) {
      * BASIC_ISO_DATE = 20111203 (yyyyMMdd)
      *
      */
-    ISO8601YMDPatternExtractor {
-        private final PartExtractor partExtractor = new ISO8601YMDPatternExtractor();
+    allYMDPatternExtractor {
+        private final PartExtractor partExtractor = new AllYMDPatternExtractor();
         @Override
         public PartExtractor get(Locale locale) {
             return partExtractor;
         }
     },
 
-    englishDateExtractor {
-        private final PartExtractor partExtractor = new EnglishDMYPatternExtractor();
-        @Override
-        public PartExtractor get(Locale locale) {
-            return partExtractor;
-        }
-    },
     ;
 
     public abstract PartExtractor get(Locale locale);

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/ISO8601YMDPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/ISO8601YMDPatternExtractor.java
diff --git a/src/test/java/liqp/filters/date/fuzzy/extractors/EnglishDMYPatternExtractorTest.java b/src/test/java/liqp/filters/date/fuzzy/extractors/EnglishDMYPatternExtractorTest.java