From 3fa1c0eaa27593edf2f05b30f661b2726c7cdcbe Mon Sep 17 00:00:00 2001
From: Patrick Tang <36460266+patricksptang@users.noreply.github.com>
Date: Mon, 9 Oct 2023 00:12:53 -0700
Subject: [PATCH] separate out the Greek / Hebriew accent and vowel filter

---
 .../jsword/impl/JSwordPassageServiceImpl.java | 18 ++++++-----
 .../core/utils/StringConversionUtils.java     | 30 ++++++++++++++++++-
 .../step/core/service/jsword/impl/default.xsl | 12 ++++----
 .../core/service/jsword/impl/interlinear.xsl  | 10 ++++---
 4 files changed, 50 insertions(+), 20 deletions(-)

diff --git a/step-core/src/main/java/com/tyndalehouse/step/core/service/jsword/impl/JSwordPassageServiceImpl.java b/step-core/src/main/java/com/tyndalehouse/step/core/service/jsword/impl/JSwordPassageServiceImpl.java
index e369aced4e..08ef7364da 100644
--- a/step-core/src/main/java/com/tyndalehouse/step/core/service/jsword/impl/JSwordPassageServiceImpl.java
+++ b/step-core/src/main/java/com/tyndalehouse/step/core/service/jsword/impl/JSwordPassageServiceImpl.java
@@ -1370,19 +1370,18 @@ protected void setOptions(final TransformingSAXEventProvider tsep, final List<Lo
                         break;
                     case GREEK_ACCENTS:
                         if (isGreek) {
-                            tsep.setParameter("RemovePointing", "false");
-                            tsep.setParameter("RemoveVowels", "false");
+                            tsep.setParameter("RemoveGreekAccent", "false");
                         }
                         break;
                     case HEBREW_VOWELS:
                         if (isHebrew) {
-                            tsep.setParameter("RemoveVowels", "false");
+                            tsep.setParameter("RemoveHebrewVowels", "false");
                         }
                         break;
                     case HEBREW_ACCENTS:
                         if (isHebrew) {
-                            tsep.setParameter("RemovePointing", "false");
-                            tsep.setParameter("RemoveVowels", "false");
+                            tsep.setParameter("RemoveHebrewPointing", "false");
+                            tsep.setParameter("RemoveHebrewVowels", "false");
                         }
                         break;
                 }
@@ -1390,9 +1389,12 @@ protected void setOptions(final TransformingSAXEventProvider tsep, final List<Lo
         }
 
         //if no greek or hebrew, then override to false
-        if (!isGreek && !isHebrew) {
-            tsep.setParameter("RemovePointing", false);
-            tsep.setParameter("RemoveVowels", false);
+        if (!isGreek) {
+            tsep.setParameter("RemoveGreekAccent", "false");
+        }
+        if (!isHebrew) {
+            tsep.setParameter("RemoveHebrewPointing", "false");
+            tsep.setParameter("RemoveHebrewVowels", "false");
         }
 // SM Verse Per Line Issue for RTL
 //        if (!books[0].getBookMetaData().isLeftToRight())        {
diff --git a/step-core/src/main/java/com/tyndalehouse/step/core/utils/StringConversionUtils.java b/step-core/src/main/java/com/tyndalehouse/step/core/utils/StringConversionUtils.java
index 313ded552b..64402add6e 100644
--- a/step-core/src/main/java/com/tyndalehouse/step/core/utils/StringConversionUtils.java
+++ b/step-core/src/main/java/com/tyndalehouse/step/core/utils/StringConversionUtils.java
@@ -13,6 +13,8 @@
 import java.util.Iterator;
 import java.util.List;
 import java.util.Locale;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import static com.tyndalehouse.step.core.utils.StringUtils.isEmpty;
 import static com.tyndalehouse.step.core.utils.language.GreekUtils.removeGreekTranslitMarkUpForIndexing;
@@ -237,6 +239,33 @@ public static String unAccent(final String word) {
         return unAccent(unAccent(word, true), false);
     }
 
+    public static String unAccentOrVowels(final String word, final String removeGreekAccent,
+                                       final String removeHebrewVowel, final String removeHebrewAccent) {
+        if (removeGreekAccent.equals("true")) {
+            String regex = "\\p{InGreek}";
+            Pattern pattern = Pattern.compile(regex, Pattern.CASE_INSENSITIVE);
+            Matcher matcher = pattern.matcher(word);
+            if (matcher.find())
+//                return unAccent(unAccent(word, true), false);
+                return unAccent(word, true);
+        }
+        if (removeHebrewVowel.equals("true") || removeHebrewAccent.equals("true")) {
+            String regexHebrew = "\\p{InHebrew}";
+            Pattern patternHebrew = Pattern.compile(regexHebrew, Pattern.CASE_INSENSITIVE);
+            Matcher matcherHebrew = patternHebrew.matcher(word);
+            if (matcherHebrew.find()) {
+                String updatedWord = word;
+                if (removeHebrewVowel.equals("true"))
+//                    updatedWord =  unAccent(unAccent(updatedWord, false), false);
+                    updatedWord =  unAccent(updatedWord, false);
+                if (removeHebrewAccent.equals("true"))
+                    updatedWord = unAccentHebrewLeavingVowels(updatedWord);
+                return updatedWord;
+            }
+        }
+        return word;
+    }
+
     /**
      * takes accents and other punctuation off the word
      *
@@ -248,7 +277,6 @@ public static String unAccent(final String word, final boolean isGreek) {
         return unAccent(word, isGreek, true);
     }
 
-
     /**
      * takes accents and other punctuation off the word
      *
diff --git a/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/default.xsl b/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/default.xsl
index 9304d57bdc..e20448e16c 100644
--- a/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/default.xsl
+++ b/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/default.xsl
@@ -108,8 +108,9 @@
 
     <!--  true to display color coding information -->
     <xsl:param name="ColorCoding" select="'false'"/>
-    <xsl:param name="RemovePointing" select="'true'"/>
-    <xsl:param name="RemoveVowels" select="'true'"/>
+    <xsl:param name="RemoveGreekAccent" select="'true'"/>
+    <xsl:param name="RemoveHebrewPointing" select="'true'"/>
+    <xsl:param name="RemoveHebrewVowels" select="'true'"/>
     <xsl:param name="HideCompareHeaders" select="'false'"/>
 
     <xsl:param name="HideXGen" select="'false'"/>
@@ -2394,11 +2395,8 @@
             <xsl:when test="./ancestor::cell/@xml:lang != 'grc' and ./ancestor::cell/@xml:lang != 'he' and ./ancestor::cell/@xml:lang != 'hbo'">
                 <xsl:value-of select="."/>
             </xsl:when>
-            <xsl:when test="$RemoveVowels = 'true'">
-                <xsl:value-of select="conversion:unAccent(string(.))"/>
-            </xsl:when>
-            <xsl:when test="$RemovePointing = 'true'">
-                <xsl:value-of select="conversion:unAccentLeavingVowels(string(.))"/>
+            <xsl:when test="$RemoveGreekAccent = 'true' or $RemoveHebrewVowels = 'true' or $RemoveHebrewPointing = 'true'">
+                <xsl:value-of select="conversion:unAccentOrVowels(string(.), $RemoveGreekAccent, $RemoveHebrewVowels, $RemoveHebrewPointing)"/>
             </xsl:when>
             <xsl:otherwise>
                 <xsl:value-of select="."/>
diff --git a/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/interlinear.xsl b/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/interlinear.xsl
index 2d7bac1e14..113ceb94aa 100644
--- a/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/interlinear.xsl
+++ b/step-core/src/main/resources/com/tyndalehouse/step/core/service/jsword/impl/interlinear.xsl
@@ -105,8 +105,10 @@
   <!--  true to display color coding information -->
   <xsl:param name="ColorCoding" select="'false'" />
   <xsl:param name="DivideHebrew" select="'false'" />
-  <xsl:param name="RemovePointing" select="'true'" />
-  <xsl:param name="RemoveVowels" select="'true'" />
+  <xsl:param name="RemoveGreekPointing" select="'true'" />
+  <xsl:param name="RemoveGreekVowels" select="'true'" />
+  <xsl:param name="RemoveHebrewPointing" select="'true'" />
+  <xsl:param name="RemoveHebrewVowels" select="'true'" />
 
 
   <xsl:param name="baseVersion" select="''" />
@@ -1956,8 +1958,8 @@
   <xsl:template name="outputPunctuatedText">
     <xsl:param name="text" />
     <xsl:choose>
-        <xsl:when test="$RemoveVowels = 'true'"><xsl:value-of select="conversion:unAccent(string($text))" /></xsl:when>
-        <xsl:when test="$RemovePointing = 'true'"><xsl:value-of select="conversion:unAccentLeavingVowels(string($text))" /></xsl:when>
+        <xsl:when test="$RemoveGreekAccent = 'true' or $RemoveHebrewVowels = 'true' or $RemoveHebrewPointing = 'true'">
+          <xsl:value-of select="conversion:unAccentOrVowels(string($text), $RemoveGreekAccent, $RemoveHebrewVowels, $RemoveHebrewPointing)" />
         <xsl:otherwise><xsl:value-of select="." /></xsl:otherwise>
     </xsl:choose>
   </xsl:template>