Strings cleaner improvements

Author: mmathieum

src/main/java/org/mtransit/commons/CharUtils.kt

@@ -14,6 +14,12 @@ object CharUtils {
     @JvmStatic
     fun countUpperCase(charArray: CharArray?) = charArray?.count { it.isUpperCase() } ?: 0
 
+    @JvmStatic
+    fun countLowerCase(string: String?) = countLowerCase(string?.toCharArray())
+
+    @JvmStatic
+    fun countLowerCase(charArray: CharArray?) = charArray?.count { it.isLowerCase() } ?: 0
+
     @JvmStatic
     fun isDigitsOnly(str: CharSequence, notEmpty: Boolean): Boolean {
         if (str.isEmpty()) {

src/main/java/org/mtransit/commons/CleanUtils.java

@@ -548,10 +548,17 @@ public static String fixMcXCase(@NotNull String string) { // Mccowan -> McCowan
 			Pattern.CASE_INSENSITIVE | RegexUtils.fUNICODE_CHARACTER_CLASS() | RegexUtils.fCANON_EQ());
 
 	@NotNull
-	public static String toLowerCaseUpperCaseWords(@NotNull Locale locale, @NotNull String string, @NotNull String... ignoreWords) {
-		if (string.isEmpty()) {
-			return string;
+	public static String toLowerCaseUpperCaseStrings(@NotNull Locale locale, @NotNull String string) {
+		if (string.isEmpty()) return string;
+		if (CharUtils.isUppercaseOnly(string, true, true)) {
+			return string.toLowerCase(locale);
 		}
+		return string;
+	}
+
+	@NotNull
+	public static String toLowerCaseUpperCaseWords(@NotNull Locale locale, @NotNull String string, @NotNull String... ignoreWords) {
+		if (string.isEmpty()) return string;
 		final float charCount = string.length();
 		final float upperCaseCount = CharUtils.countUpperCase(string);
 		final float percent = upperCaseCount / charCount;

src/main/java/org/mtransit/commons/StringsCleaner.kt

@@ -4,10 +4,17 @@ import java.util.Locale
 
 object StringsCleaner {
 
+    @JvmOverloads
     @JvmStatic
-    fun cleanRouteLongName(originalRouteLongName: String, languages: List<Locale>?): String {
+    fun cleanRouteLongName(
+        originalRouteLongName: String,
+        languages: List<Locale>?,
+        lowerUCStrings: Boolean = false,
+        lowerUCWords: Boolean = false,
+        vararg ignoredUCWords: String = emptyArray(),
+    ): String {
         var routeLongName = originalRouteLongName
-        routeLongName = cleanString(routeLongName, languages, short = false)
+        routeLongName = cleanString(routeLongName, languages, lowerUCStrings, lowerUCWords, *ignoredUCWords, short = false)
         return routeLongName
     }
 
@@ -16,6 +23,9 @@ object StringsCleaner {
     fun cleanTripHeadsign(
         originalTripHeadsign: String,
         languages: List<Locale>?,
+        lowerUCStrings: Boolean = false,
+        lowerUCWords: Boolean = false,
+        vararg ignoredUCWords: String = emptyArray(),
         removeVia: Boolean = false,
     ): String {
         var tripHeadsign = originalTripHeadsign
@@ -29,19 +39,40 @@ object StringsCleaner {
         if (languages?.contains(Locale.FRENCH) == true) {
             tripHeadsign = CleanUtils.keepToFR(tripHeadsign)
         }
-        tripHeadsign = cleanString(tripHeadsign, languages, short = true)
+        tripHeadsign = cleanString(tripHeadsign, languages, lowerUCStrings, lowerUCWords, *ignoredUCWords, short = true)
         return tripHeadsign
     }
 
+    @JvmOverloads
     @JvmStatic
-    fun cleanStopName(originalStopName: String, languages: List<Locale>?): String {
+    fun cleanStopName(
+        originalStopName: String,
+        languages: List<Locale>?,
+        lowerUCStrings: Boolean = false,
+        lowerUCWords: Boolean = false,
+        vararg ignoredUCWords: String = emptyArray(),
+    ): String {
         var stopName = originalStopName
-        stopName = cleanString(stopName, languages, short = true)
+        stopName = cleanString(stopName, languages, lowerUCStrings, lowerUCWords, *ignoredUCWords, short = true)
         return stopName
     }
 
-    private fun cleanString(originalString: String, languages: List<Locale>?, short: Boolean): String {
+    private fun cleanString(
+        originalString: String,
+        languages: List<Locale>?,
+        lowerUCStrings: Boolean = false,
+        lowerUCWords: Boolean = false,
+        vararg ignoredUCWords: String = emptyArray(),
+        short: Boolean
+    ): String {
         var string = originalString
+        languages?.forEach { language ->
+            if (lowerUCWords) {
+                string = CleanUtils.toLowerCaseUpperCaseWords(language, string, *ignoredUCWords)
+            } else if (lowerUCStrings) {
+                string = CleanUtils.toLowerCaseUpperCaseStrings(language, string)
+            }
+        }
         if (!short) {
             string = CleanUtils.cleanSlashes(string)
         }
@@ -68,4 +99,4 @@ object StringsCleaner {
         }
         return string
     }
-}
+}

src/main/java/org/mtransit/scratch/RegexScratch.kt

@@ -14,20 +14,19 @@ internal object RegexScratch {
         var string: String
         string = ""
         println("string: '$string'.")
-        println("-------")
+        println("=======")
 
         val matcher = pattern.matcher(string)
         while (matcher.find()) {
             println("-------")
-            println("group: '" + matcher.group() + "'.")
-            println("groupCount: '" + matcher.groupCount() + "'.")
+            println("- Found group: '" + matcher.group() + "' (count: '" + matcher.groupCount() + "'):")
             for (g in 0..matcher.groupCount()) {
-                println("group[" + g + "]: '" + matcher.group(g) + "'.")
+                println("  - group[" + g + "]: '" + matcher.group(g) + "'.")
             }
             println("-------")
         }
 
-        println("-------")
+        println("=======")
         var replaceAll: String
         replaceAll = ""
         println("replaceAll: '$replaceAll'.")