[WIP]

bkiers · Dec 22, 2024 · 957e039 · 957e039
1 parent 7e66c66
commit 957e039
Show file tree

Hide file tree

Showing 14 changed files with 80 additions and 83 deletions.
diff --git a/src/main/java/liqp/filters/date/fuzzy/LookupResult.java b/src/main/java/liqp/filters/date/fuzzy/LookupResult.java
@@ -4,12 +4,13 @@
 
 class LookupResult {
 
+    private final String name;
     final List<Part> parts;
     final boolean found;
 
-    LookupResult(List<Part> parts, boolean found) {
+    LookupResult(String name, List<Part> parts, boolean found) {
+        this.name = name;
         this.parts = parts;
         this.found = found;
     }
-
 }
diff --git a/src/main/java/liqp/filters/date/fuzzy/Part.java b/src/main/java/liqp/filters/date/fuzzy/Part.java
@@ -103,11 +103,13 @@ class RecognizedPart implements Part {
         final int start;
         final int end;
         private final List<String> patterns;
+        public final String source;
 
-        RecognizedPart(int start, int end, List<String> patterns) {
+        RecognizedPart(int start, int end, List<String> patterns, String source) {
             this.start = start;
             this.end = end;
             this.patterns = patterns;
+            this.source = source;
         }
 
         @Override

diff --git a/src/main/java/liqp/filters/date/fuzzy/PartRecognizer.java b/src/main/java/liqp/filters/date/fuzzy/PartRecognizer.java
@@ -125,19 +125,19 @@ private LookupResult lookup(List<Part> parts, PartExtractor partExtractor) {
                         parts.add(i, after);
                     }
 
-                    RecognizedPart recognized = new RecognizedPart(part.start() + per.start, part.start() + per.end, per.formatterPatterns);
+                    RecognizedPart recognized = new RecognizedPart(part.start() + per.start, part.start() + per.end, per.formatterPatterns, source.substring(per.start, per.end));
                     parts.add(i, recognized);
 
                     if (per.start != 0) {
                         NewPart before = new NewPart(part.start(), part.start() + per.start, source.substring(0, per.start));
                         parts.add(i, before);
                     }
 
-                    return new LookupResult(parts, true);
+                    return new LookupResult(per.extractorName, parts, true);
                 }
             }
         }
-        return new LookupResult(parts, false);
+        return new LookupResult("<none>", parts, false);
     }
 
     private List<Part> markAsUnrecognized(List<Part> parts) {

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/AllYMDPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/AllYMDPatternExtractor.java
@@ -15,52 +15,39 @@ 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);
+        extractors.add(new AnyYMDPatternExtractor("iSO8601Y4MDPatternExtractor",
+                pY4(), pp("-"), pM(), pp("-"), pD())); // yyyy-MM-dd
+
+        extractors.add(new AnyYMDPatternExtractor("americanY4MDPatternExtractor",
+                pM(), pp("/"), pD(), pp("/"), pY4())); // MM/dd/yyyy
         // 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);
+        extractors.add(new AnyYMDPatternExtractor("indianY4MDPatternExtractor",
+                pD(), pp("-"), pM(), pp("-"), pY4())); // d-M-yyyy
+        extractors.add(new AnyYMDPatternExtractor("chineseY4MDPatternExtractor",
+                pY4(), pp("/"), pM(), pp("/"), pD())); // yyyy/M/d
+        extractors.add(new AnyYMDPatternExtractor("englishY4MDPatternExtractor",
+                pD(), pp("/"), pM(), pp("/"), pY4())); // d/M/yyyy
+        extractors.add(new AnyYMDPatternExtractor("slavicY4MDPatternExtractor",
+                pD(), pp("."), pM(), pp("."), pY4())); // dd.MM.yyyy
+        extractors.add(new AnyYMDPatternExtractor("coldEuropeY4MDPatternExtractor",
+                pY4(), pp("-"), pM(), pp("-"), pD())); // yyyy-MM-dd
+        extractors.add(new AnyYMDPatternExtractor("espanaY4MDPatternExtractor",
+                pY4(), pp("-"), pM(), pp("-"), pD())); // yyyy/MM/dd
+
+        extractors.add(new AnyYMDPatternExtractor("americanY2MDPatternExtractor",
+                pM(), pp("/"), pD(), pp("/"), pY2())); // MM/dd/yy
+        extractors.add(new AnyYMDPatternExtractor("indianY2MDPatternExtractor",
+                pD(), pp("-"), pM(), pp("-"), pY2())); // d-M-yy
+        extractors.add(new AnyYMDPatternExtractor("chineseY2MDPatternExtractor",
+                pY2(), pp("/"), pM(), pp("/"), pD())); // yy/M/d
+        extractors.add(new AnyYMDPatternExtractor("englishY2MDPatternExtractor",
+                pD(), pp("/"), pM(), pp("/"), pY2())); // d/M/yy
+        extractors.add(new AnyYMDPatternExtractor("slavicY2MDPatternExtractor",
+                pD(), pp("."), pM(), pp("."), pY2())); // dd.MM.yy
+        extractors.add(new AnyYMDPatternExtractor("coldEuropeY2MDPatternExtractor",
+                pY2(), pp("-"), pM(), pp("-"), pD())); // yy-MM-dd
+        extractors.add(new AnyYMDPatternExtractor("espanaY2MDPatternExtractor",
+                pY2(), pp("-"), pM(), pp("-"), pD())); // yy/MM/dd
     }
 
     @Override
@@ -71,6 +58,6 @@ public PartExtractorResult extract(String source) {
                 return result;
             }
         }
-        return new PartExtractorResult();
+        return new PartExtractorResult("AllYMDPatternExtractor");
     }
 }
diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/AnyYMDPatternExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/AnyYMDPatternExtractor.java
@@ -44,9 +44,10 @@ static RulePart pM() {
     static RulePart pD() {
         return new RulePart(RuleType.D, (Integer)null);
     }
+
     private final RulePart[] partsInOrder;
-    protected AnyYMDPatternExtractor(RulePart... partsInOrder) {
-        super(reconstructPattern(partsInOrder), null);
+    protected AnyYMDPatternExtractor(String name, RulePart... partsInOrder) {
+        super(name, reconstructPattern(partsInOrder), null);
         this.partsInOrder = partsInOrder;
     }
 
@@ -80,14 +81,14 @@ private static String reconstructPattern(RulePart[] partsInOrder) {
     public PartExtractorResult extract(String source) {
         Matcher matcher = pattern.matcher(source);
         if (matcher.find()) {
-            PartExtractorResult result = new PartExtractorResult();
+            PartExtractorResult result = new PartExtractorResult(name);
             result.found = true;
             result.start = matcher.start(findFirstGroupName());
             result.end = matcher.end(findLastGroupName());
             result.formatterPatterns = getPatterns(matcher);
             return result;
         }
-        return new PartExtractorResult();
+        return new PartExtractorResult(name);
     }
 
     private String findLastGroupName() {

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/EnumExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/EnumExtractor.java
@@ -6,13 +6,13 @@
 
 abstract class EnumExtractor extends PartExtractorDelegate {
 
-    public EnumExtractor(Locale locale, String formatterPattern) {
+    public EnumExtractor(String name, Locale locale, String formatterPattern) {
         if (locale == null || Locale.ROOT.equals(locale)) {
             locale = Locale.US;
         }
         String[] values = withoutNulls(getEnumValues(locale), locale);
         String valuesPattern = String.join("|", values);
-        super.delegate = new RegexPartExtractor("(?:^|.*?[^\\w_])(" + valuesPattern + ")(?:$|[^\\w_].*?)",
+        super.delegate = new RegexPartExtractor(name,"(?:^|.*?[^\\w_])(" + valuesPattern + ")(?:$|[^\\w_].*?)",
                 formatterPattern);
     }
 

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/Extractors.java b/src/main/java/liqp/filters/date/fuzzy/extractors/Extractors.java
@@ -11,7 +11,7 @@ public enum Extractors {
         private final Map<Locale, PartExtractor> extractors = new HashMap<>();
         @Override
         public PartExtractor get(Locale locale) {
-            return extractors.computeIfAbsent(locale, l -> new EnumExtractor(locale, "EEEE") {
+            return extractors.computeIfAbsent(locale, l -> new EnumExtractor("fullWeekdaysExtractor", locale, "EEEE") {
                 @Override
                 protected String[] getEnumValues(Locale locale) {
                     return new DateFormatSymbols(locale).getWeekdays();
@@ -23,7 +23,7 @@ protected String[] getEnumValues(Locale locale) {
         private final Map<Locale, PartExtractor> extractors = new HashMap<>();
         @Override
         public PartExtractor get(Locale locale) {
-            return extractors.computeIfAbsent(locale, l -> new EnumExtractor(locale, "EEE") {
+            return extractors.computeIfAbsent(locale, l -> new EnumExtractor("shortWeekdaysExtractor", locale, "EEE") {
                 @Override
                 protected String[] getEnumValues(Locale locale) {
                     return new DateFormatSymbols(locale).getShortWeekdays();
@@ -46,7 +46,7 @@ public PartExtractor get(Locale locale) {
         }
     },
     plainYearExtractor{
-        private final PartExtractor partExtractor = new RegexPartExtractor(".*\\b?(\\d{4})\\b?.*", "yyyy");
+        private final PartExtractor partExtractor = new RegexPartExtractor("plainYearExtractor", ".*\\b?(\\d{4})\\b?.*", "yyyy");
         @Override
         public PartExtractor get(Locale locale) {
             return partExtractor;

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/FullMonthExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/FullMonthExtractor.java
@@ -6,7 +6,7 @@
 class FullMonthExtractor extends EnumExtractor {
 
     public FullMonthExtractor(Locale locale) {
-        super(locale, "MMMM");
+        super("FullMonthExtractor", locale, "MMMM");
     }
 
     @Override

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/PartExtractorResult.java b/src/main/java/liqp/filters/date/fuzzy/extractors/PartExtractorResult.java
@@ -4,11 +4,15 @@
 import java.util.List;
 
 public class PartExtractorResult {
-    public PartExtractorResult(){}
-    public PartExtractorResult(List<String> formatterPatterns){
-        this.formatterPatterns = formatterPatterns;
+
+    public final String extractorName;
+
+    public PartExtractorResult(String extractorName){
+        this.extractorName = extractorName;
+        this.formatterPatterns = new ArrayList<>();
     }
-    public PartExtractorResult(String formatterPattern){
+    public PartExtractorResult(String extractorName, String formatterPattern){
+        this.extractorName = extractorName;
         this.formatterPatterns = new ArrayList<>();
         this.formatterPatterns.add(formatterPattern);
     }

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/RegexPartExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/RegexPartExtractor.java
@@ -6,10 +6,12 @@
 
 class RegexPartExtractor implements PartExtractor {
 
+    protected final String name;
     protected final Pattern pattern;
     protected final String formatterPattern;
 
-    public RegexPartExtractor(String regex, String formatterPattern) {
+    public RegexPartExtractor(String name, String regex, String formatterPattern) {
+        this.name = name;
         this.pattern = Pattern.compile(regex, Pattern.CASE_INSENSITIVE);
         this.formatterPattern = formatterPattern;
     }
@@ -18,12 +20,12 @@ public RegexPartExtractor(String regex, String formatterPattern) {
     public PartExtractorResult extract(String source) {
         Matcher matcher = pattern.matcher(source);
         if (matcher.find()) {
-            PartExtractorResult result = new PartExtractorResult(formatterPattern);
+            PartExtractorResult result = new PartExtractorResult(name, formatterPattern);
             result.found = true;
             result.start = matcher.start(1);
             result.end = matcher.end(1);
             return result;
         }
-        return new PartExtractorResult();
+        return new PartExtractorResult(name);
     }
 }
diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/RegularTimeExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/RegularTimeExtractor.java
@@ -5,7 +5,7 @@
 class RegularTimeExtractor extends RegexPartExtractor {
 
     public RegularTimeExtractor() {
-        super("(?:^|.*?\\D)"
+        super("RegularTimeExtractor", "(?:^|.*?\\D)"
                 + "("
                 + "(?<hours>(\\d|0\\d|1\\d|2[0-3]))"
                 + ":"
@@ -23,7 +23,7 @@ public RegularTimeExtractor() {
     public PartExtractorResult extract(String source) {
         Matcher m = pattern.matcher(source);
         if (m.matches()) {
-            PartExtractorResult r = new PartExtractorResult();
+            PartExtractorResult r = new PartExtractorResult("RegularTimeExtractor");
             r.found = true;
 
             String ampmPart = "";
@@ -70,7 +70,7 @@ public PartExtractorResult extract(String source) {
             r.formatterPatterns = newList(resPattern);
             return r;
         }
-        return new PartExtractorResult();
+        return new PartExtractorResult("RegularTimeExtractor");
     }
 
     static String repeat(String key, int count) {

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/ShortMonthExtractor.java b/src/main/java/liqp/filters/date/fuzzy/extractors/ShortMonthExtractor.java
@@ -6,7 +6,7 @@
 class ShortMonthExtractor extends EnumExtractor {
 
     public ShortMonthExtractor(Locale locale) {
-        super(locale, "MMM");
+        super("ShortMonthExtractor", locale, "MMM");
     }
 
     @Override

diff --git a/src/main/java/liqp/filters/date/fuzzy/extractors/YearWithEra.java b/src/main/java/liqp/filters/date/fuzzy/extractors/YearWithEra.java
@@ -8,15 +8,15 @@
 class YearWithEra extends RegexPartExtractor {
 
     public YearWithEra() {
-        super("(?:^|.*?\\D)(?<year>\\d+)(?<eraSeparator>\\s*)(?<era>AD|BC|Anno Domini|Before Christ)(?:$|\\D.*?)",
+        super("YearWithEra", "(?:^|.*?\\D)(?<year>\\d+)(?<eraSeparator>\\s*)(?<era>AD|BC|Anno Domini|Before Christ)(?:$|\\D.*?)",
                 null);
     }
 
     @Override
     public PartExtractorResult extract(String source) {
         Matcher matcher = pattern.matcher(source);
         if (matcher.find()) {
-            PartExtractorResult result = new PartExtractorResult();
+            PartExtractorResult result = new PartExtractorResult("YearWithEra");
             result.found = true;
             result.start = matcher.start("year");
             String resPattern = repeat("y", matcher.group("year").length());
@@ -38,6 +38,6 @@ public PartExtractorResult extract(String source) {
             result.formatterPatterns = newList(resPattern);
             return result;
         }
-        return new PartExtractorResult();
+        return new PartExtractorResult("YearWithEra");
     }
 }
diff --git a/src/test/java/liqp/filters/date/fuzzy/FuzzyDateParserParametrizedTest.java b/src/test/java/liqp/filters/date/fuzzy/FuzzyDateParserParametrizedTest.java
@@ -102,13 +102,13 @@ public static Collection<Object[]> data() {
                 {null, "2024-1-5 08:15 ", "yyyy-M-d HH:mm "},
                 {null, "2024-12-25 14:45 ", "yyyy-MM-dd HH:mm "},
                 {null, "2024-12-25 14:45:30 ", "yyyy-MM-dd HH:mm:ss "},
-                {null, "1/1/23 ", "d/M/yy "},
-                {null, "1/1/2023 ", "d/M/yyyy "},
-                {null, "01/01/23 ", "dd/MM/yy "},
-                {null, "01/01/2023 ", "dd/MM/yyyy "},
-                {null, "1/1/23 12:34 ", "d/M/yy HH:mm "},
-                {null, "1/1/2023 12:34 ", "d/M/yyyy HH:mm "},
-                {null, "01/01/23 12:34 ", "dd/MM/yy HH:mm "},
+                {null, "1/1/23 ", "M/d/yy "},
+                {null, "1/1/2023 ", "M/d/yyyy "},
+                {null, "01/01/23 ", "MM/dd/yy "},
+                {null, "01/01/2023 ", "MM/dd/yyyy "},
+                {null, "1/1/23 12:34 ", "M/d/yy HH:mm "},
+                {null, "1/1/2023 12:34 ", "M/d/yyyy HH:mm "},
+                {null, "01/01/23 12:34 ", "MM/dd/yy HH:mm "},
 
         });
     }