Improve list detection

veraPDF · Jul 26, 2024 · ea01bf5 · ea01bf5
1 parent d20ce32
commit ea01bf5
Show file tree

Hide file tree

Showing 8 changed files with 104 additions and 54 deletions.
diff --git a/.../org/verapdf/wcag/algorithms/semanticalgorithms/consumers/LinesPreprocessingConsumer.java b/.../org/verapdf/wcag/algorithms/semanticalgorithms/consumers/LinesPreprocessingConsumer.java
@@ -78,14 +78,6 @@ private List<TableBorderBuilder> findTableBorders(Integer pageNumber) {
                 i++;
             }
         }
-        for (TableBorderBuilder border : tableBorders) {
-            for (LineChunk lineChunk : border.getVerticalLines()) {
-                StaticContainers.getLinesCollection().getVerticalLines(pageNumber).remove(lineChunk);
-            }
-            for (LineChunk lineChunk : border.getHorizontalLines()) {
-                StaticContainers.getLinesCollection().getHorizontalLines(pageNumber).remove(lineChunk);
-            }
-        }
         for (int i = 0; i < tableBorders.size();) {
             TableBorderBuilder border = tableBorders.get(i);
             if ((border.getHorizontalLinesNumber() <= 2 && border.getVerticalLinesNumber() <= 1) ||
@@ -95,6 +87,10 @@ private List<TableBorderBuilder> findTableBorders(Integer pageNumber) {
                 i++;
             }
         }
+        for (TableBorderBuilder border : tableBorders) {
+            StaticContainers.getLinesCollection().getVerticalLines(pageNumber).removeAll(border.getVerticalLines());
+            StaticContainers.getLinesCollection().getHorizontalLines(pageNumber).removeAll(border.getHorizontalLines());
+        }
         return tableBorders;
     }
 

diff --git a/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/utils/ListLabelsUtils.java b/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/utils/ListLabelsUtils.java
@@ -14,7 +14,7 @@ public class ListLabelsUtils {
 
 	private static final Set<Character> labels = new HashSet<>(
 			Arrays.asList('\u002D', '\u2022', '\u25CF', '\u2714', '\u2717', '\u2794', '\u27A2', '\uE00A', '\uE00C',
-			              '\uF076', '\u2588', '\u25A0', '\u2013', '\uF0B7', '\uF0A7', '\u25A1', '\uF0A1', '\u25AA', '\u25FC', '\u25CB', '\u203B')); //office labels examples (-, •, ✔, ✗, ●, ➔, ➢), pdf files labels examples (█, ■, , □, , ▪, ◼, ○, ※)
+			              '\uF076', '\u2588', '\u25A0', '\u2013', '\uF0B7', '\uF0A7', '\u25A1', '\uF0A1', '\u25AA', '\u25FC', '\u25CB', '\u203B', '\u274D')); //office labels examples (-, •, ✔, ✗, ●, ➔, ➢), pdf files labels examples (█, ■, , □, , ▪, ◼, ○, ※, ❍)
 	private static final Character o =  '\u006F';
 
 	public static boolean isListLabel(String value) {

diff --git a/...semanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm.java b/...semanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm.java
@@ -26,4 +26,9 @@ protected String getLowerCaseRegex() {
     protected String getUpperCaseRegex() {
         return UPPER_CASE_ENGLISH_LETTER_REGEX;
     }
+
+    @Override
+    protected List<Character> getLetters() {
+        return letters;
+    }
 }
diff --git a/...emanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm1.java b/...emanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm1.java
@@ -12,22 +12,22 @@ protected Integer getNumberFromString(String string) {
         return getNumberFromLetters1(string);
     }
 
-    private static String getLetters1FromNumber(int integer) {
+    private String getLetters1FromNumber(int integer) {
         integer--;
-        int n = integer / letters.size();
-        char c = letters.get(integer % letters.size());
+        int n = integer / getLetters().size();
+        char c = getLetters().get(integer % getLetters().size());
         StringBuilder str = new StringBuilder();
         for (int i = 0; i <= n; i++) {
             str.append(c);
         }
         return str.toString();
     }
 
-    private static Integer getNumberFromLetters1(String s) {
+    private Integer getNumberFromLetters1(String s) {
         if (s.isEmpty()) {
             return null;
         }
-        int num = letters.indexOf(s.charAt(0));
+        int num = getLetters().indexOf(s.charAt(0));
         if (num < 0) {
             return null;
         }
@@ -36,6 +36,6 @@ private static Integer getNumberFromLetters1(String s) {
                 return null;
             }
         }
-        return letters.size() * (s.length() - 1) + num + 1;
+        return getLetters().size() * (s.length() - 1) + num + 1;
     }
 }
diff --git a/...emanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm2.java b/...emanticalgorithms/utils/listLabelsDetection/AlfaLettersListLabelsDetectionAlgorithm2.java
@@ -12,28 +12,28 @@ protected Integer getNumberFromString(String string) {
         return getNumberFromLetters2(string);
     }
 
-    private static String getLetters2FromNumber(int integer) {
+    private String getLetters2FromNumber(int integer) {
         StringBuilder str = new StringBuilder();
         while (integer > 0) {
             integer--;
-            int k = integer % letters.size();
-            str.insert(0, letters.get(k));
-            integer /= letters.size();
+            int k = integer % getLetters().size();
+            str.insert(0, getLetters().get(k));
+            integer /= getLetters().size();
         }
         return str.toString();
     }
 
-    private static Integer getNumberFromLetters2(String s) {
+    private Integer getNumberFromLetters2(String s) {
         if (s.isEmpty()) {
             return null;
         }
         int result = 0;
         for (char c : s.toCharArray()) {
-            int num = letters.indexOf(c);
+            int num = getLetters().indexOf(c);
             if (num < 0) {
                 return null;
             }
-            result = result * letters.size() + num + 1;
+            result = result * getLetters().size() + num + 1;
         }
         return result;
     }

diff --git a/...lgorithms/utils/listLabelsDetection/CircledArabicNumbersListLabelsDetectionAlgorithm.java b/...lgorithms/utils/listLabelsDetection/CircledArabicNumbersListLabelsDetectionAlgorithm.java
@@ -0,0 +1,39 @@
+package org.verapdf.wcag.algorithms.semanticalgorithms.utils.listLabelsDetection;
+
+import java.util.Arrays;
+import java.util.List;
+
+public class CircledArabicNumbersListLabelsDetectionAlgorithm extends LettersListLabelsDetectionAlgorithm {
+
+    protected static final List<Character> letters = Arrays.asList(
+            '\u24EA','\u2460','\u2461','\u2462','\u2463','\u2464','\u2465','\u2466','\u2467','\u2468',
+            '\u2469','\u246A','\u246B','\u246C','\u246D','\u246E','\u246F','\u2470','\u2471','\u2472',
+            '\u2473','\u3251','\u3252','\u3253','\u3254','\u3255','\u3256','\u3257','\u3258','\u3259',
+            '\u325A','\u325B','\u325C','\u325D','\u325E','\u325F','\u32B1','\u32B2','\u32B3','\u32B4',
+            '\u32B5','\u32B6','\u32B7','\u32B8','\u32B9','\u32BA','\u32BB','\u32BC','\u32BD','\u32BE',
+            '\u32BF');
+
+    private static final String KOREAN_LETTER_REGEX = "[\u24EA\u2460-\u2473\u3251-\u325F\u32B1-\u32BF]+";
+    private static final String UPPER_CASE_KOREAN_LETTER_REGEX = KOREAN_LETTER_REGEX;
+    private static final String LOWER_CASE_KOREAN_LETTER_REGEX = KOREAN_LETTER_REGEX;
+
+    @Override
+    protected String getRegex() {
+        return KOREAN_LETTER_REGEX;
+    }
+
+    @Override
+    protected String getLowerCaseRegex() {
+        return LOWER_CASE_KOREAN_LETTER_REGEX;
+    }
+
+    @Override
+    protected List<Character> getLetters() {
+        return letters;
+    }
+
+    @Override
+    protected String getUpperCaseRegex() {
+        return UPPER_CASE_KOREAN_LETTER_REGEX;
+    }
+}
diff --git a/...manticalgorithms/utils/listLabelsDetection/KoreanLettersListLabelsDetectionAlgorithm.java b/...manticalgorithms/utils/listLabelsDetection/KoreanLettersListLabelsDetectionAlgorithm.java
@@ -5,11 +5,12 @@
 
 public class KoreanLettersListLabelsDetectionAlgorithm extends LettersListLabelsDetectionAlgorithm {
 
-    protected static final List<Character> letters = Arrays.asList('가', '나', '다', '라', '마', '바', '사', '아', '자', '차', '카', '타', '파', '하');
+    protected static final List<Character> letters = Arrays.asList('가', '나', '다', '라', '마', '바', '사', '아', '자', '차', 
+            '카', '타', '파', '하', '거', '너', '더', '러', '머', '버', '서', '어', '저', '처', '커', '터', '퍼', '허');
 
-    private static final String UPPER_CASE_KOREAN_LETTER_REGEX = "[가나다라마바사아자차카타파하]+";
-    private static final String LOWER_CASE_KOREAN_LETTER_REGEX = "[가나다라마바사아자차카타파하]+";
-    private static final String KOREAN_LETTER_REGEX = "[가나다라마바사아자차카타파하]+";
+    private static final String KOREAN_LETTER_REGEX = "[가나다라마바사아자차카타파하거너더러머버서어저처커터퍼허]+";
+    private static final String UPPER_CASE_KOREAN_LETTER_REGEX = KOREAN_LETTER_REGEX;
+    private static final String LOWER_CASE_KOREAN_LETTER_REGEX = KOREAN_LETTER_REGEX;
 
     @Override
     protected String getRegex() {
@@ -22,36 +23,12 @@ protected String getLowerCaseRegex() {
     }
 
     @Override
-    protected String getUpperCaseRegex() {
-        return UPPER_CASE_KOREAN_LETTER_REGEX;
-    }
-
-    @Override
-    protected String getStringFromNumber(Integer number) {
-        return getLettersFromNumber(number);
+    protected List<Character> getLetters() {
+        return letters;
     }
 
     @Override
-    protected Integer getNumberFromString(String string) {
-        return getNumberFromLetters(string);
-    }
-
-    private static String getLettersFromNumber(int integer) {
-        integer--;
-        if (integer < letters.size()) {
-            return letters.get(integer).toString();
-        }
-        return null;
-    }
-
-    private static Integer getNumberFromLetters(String s) {
-        if (s.length() != 1) {
-            return null;
-        }
-        int num = letters.indexOf(s.charAt(0));
-        if (num < 0) {
-            return null;
-        }
-        return num + 1;
+    protected String getUpperCaseRegex() {
+        return UPPER_CASE_KOREAN_LETTER_REGEX;
     }
 }
diff --git a/...hms/semanticalgorithms/utils/listLabelsDetection/LettersListLabelsDetectionAlgorithm.java b/...hms/semanticalgorithms/utils/listLabelsDetection/LettersListLabelsDetectionAlgorithm.java
@@ -123,6 +123,10 @@ private boolean isBadItem(ListItemTextInfo listItem, String item, String s, int
     }
 
     protected abstract String getLowerCaseRegex();
+
+    protected List<Character> getLetters() {
+        return null;
+    }
 
     protected abstract String getUpperCaseRegex();
 
@@ -136,4 +140,33 @@ private static boolean isCharMatchRegex(String s, int index, String regex) {
         }
         return s.substring(index, index + 1).matches(regex);
     }
+
+    @Override
+    protected String getStringFromNumber(Integer number) {
+        return getLettersFromNumber(number);
+    }
+
+    @Override
+    protected Integer getNumberFromString(String string) {
+        return getNumberFromLetters(string);
+    }
+
+    private String getLettersFromNumber(int integer) {
+        integer--;
+        if (integer < getLetters().size()) {
+            return getLetters().get(integer).toString();
+        }
+        return null;
+    }
+
+    private Integer getNumberFromLetters(String s) {
+        if (s.length() != 1) {
+            return null;
+        }
+        int num = getLetters().indexOf(s.charAt(0));
+        if (num < 0) {
+            return null;
+        }
+        return num + 1;
+    }
 }