JSON config > + agency_id_cleanup_regex

src/main/java/org/mtransit/parser/DefaultAgencyTools.java

@@ -318,6 +318,34 @@ public String getAgencyColor(@NotNull GAgency gAgency, @NotNull GSpec gSpec) {
 		return getAgencyColor();
 	}
 
+	@Nullable
+	@Override
+	public String getAgencyIdCleanupRegex() {
+		if (Configs.getAgencyConfig() != null) {
+			return Configs.getAgencyConfig().getAgencyIdCleanupRegex();
+		}
+		return null; // OPT-IN feature
+	}
+
+	@Nullable
+	private Pattern agencyIdCleanupPattern = null;
+
+	private boolean agencyIdCleanupPatternSet = false;
+
+	private @Nullable Pattern getAgencyIdCleanupPattern() {
+		if (this.agencyIdCleanupPattern == null && !agencyIdCleanupPatternSet) {
+			this.agencyIdCleanupPattern = GTFSCommons.makeIdCleanupPattern(getAgencyIdCleanupRegex());
+			this.agencyIdCleanupPatternSet = true;
+		}
+		return this.agencyIdCleanupPattern;
+	}
+
+	@NotNull
+	@Override
+	public String cleanAgencyId(@NotNull String gAgencyId) {
+		return GTFSCommons.cleanOriginalString(gAgencyId, getAgencyIdCleanupPattern());
+	}
+
 	@Nullable
 	@Override
 	public String getAgencyId() {

src/main/java/org/mtransit/parser/config/gtfs/data/AgencyConfig.kt

@@ -15,6 +15,8 @@ data class AgencyConfig(
      */
     @SerialName("other_agency_ids")
     val otherAgencyIds: List<String> = emptyList(), // OPT-IN filter
+    @SerialName("agency_id_cleanup_regex")
+    val agencyIdCleanupRegex: String? = null, // optional
     /**
      * Route type filter (integer from GTFS Static `routes.txt` > `route_type` field)
      * (useful when multiple route type in same GTFS)

src/main/java/org/mtransit/parser/gtfs/GAgencyTools.java

@@ -73,6 +73,12 @@ public interface GAgencyTools {
 	@Nullable
 	String getAgencyId();
 
+	@Nullable
+	String getAgencyIdCleanupRegex();
+
+	@NotNull
+	String cleanAgencyId(@NotNull String gAgencyId);
+
 	@NotNull
 	Integer getAgencyRouteType();
 

src/main/java/org/mtransit/parser/mt/GenerateMObjectsTask.java

@@ -589,23 +589,16 @@ private void parseGTrips(HashMap<String, MSchedule> mSchedules,
 						&& directionStopTimesHeadsign != null
 						&& !directionStopTimesHeadsign.isEmpty()) {
 					if (mDirectionStopTimesHeadsign.containsKey(mDirection.getId())) {
-						if (!mDirectionStopTimesHeadsign.get(mDirection.getId()).equals(directionStopTimesHeadsign)) {
-							if (!mDirectionStopTimesHeadsign.get(mDirection.getId()).contains(directionStopTimesHeadsign)) {
+						final String cDirectionStopTimesHeadsign = mDirectionStopTimesHeadsign.get(mDirection.getId());
+						if (!cDirectionStopTimesHeadsign.equals(directionStopTimesHeadsign)) {
+							if (!cDirectionStopTimesHeadsign.contains(directionStopTimesHeadsign)) {
 								MTLog.log("%s: Trip stop times head-sign different for same direction ID ('%s'!='%s')", this.routeId,
-										directionStopTimesHeadsign, mDirectionStopTimesHeadsign.get(mDirection.getId()));
+										directionStopTimesHeadsign, cDirectionStopTimesHeadsign);
 							}
-							String mergedHeadsignValue = MDirection.mergeHeadsignValue(
-									mDirectionStopTimesHeadsign.get(mDirection.getId()),
+							final String mergedHeadsignValue = MDirection.mergeHeadsignValue(
+									cDirectionStopTimesHeadsign,
 									directionStopTimesHeadsign
 							);
-							if (mergedHeadsignValue != null) {
-								mergedHeadsignValue = agencyTools.cleanDirectionHeadsign(
-										gRoute,
-										gTrip.getDirectionIdOrDefault(),
-										true,
-										mergedHeadsignValue
-								);
-							}
 							mDirectionStopTimesHeadsign.put(mDirection.getId(), mergedHeadsignValue);
 						}
 					} else {

src/main/java/org/mtransit/parser/mt/MGenerator.java

@@ -315,10 +315,10 @@ public static void dumpFiles(@NotNull GAgencyTools gAgencyTools,
 		// STRINGS
 		dumpStrings(mSpec, fileBase, deleteAll, dataDirF, rawDirF, dbConnection); // AFTER ALL OTHER TABLE W/ STRINGS
 		if (deleteAll) {
-			dumpValues(rawDirF, fileBase, null, null, null, null, null, -1, -1, null, true);
+			dumpValues(gAgencyTools, rawDirF, fileBase, null, null, null, null, null, -1, -1, null, true);
 		} else {
 			dumpCommonValues(rawDirF, gAgencyTools, mSpec, inputUrl);
-			dumpValues(
+			dumpValues(gAgencyTools,
 					rawDirF, fileBase, mSpec,
 					minMaxLatLng.first.first, minMaxLatLng.second.first, minMaxLatLng.first.second, minMaxLatLng.second.second,
 					mSpec.getFirstTimestampInSeconds(), mSpec.getLastTimestampInSeconds(), inputUrl,
@@ -1057,6 +1057,7 @@ private static String getLastModified(String gtfsDir) {
 	private static final String GTFS_RDS_AGENCY_EXTENDED_TYPE = "gtfs_rts_agency_extended_type"; // do not change to avoid breaking compat w/ old modules
 	public static final String GTFS_RDS_TIMEZONE = "gtfs_rts_timezone"; // do not change to avoid breaking compat w/ old modules
 	private static final String GTFS_RDS_COLOR = "gtfs_rts_color"; // do not change to avoid breaking compat w/ old modules
+	private static final String GTFS_RDS_AGENCY_ID_CLEANUP_REGEX = "gtfs_rts_agency_id_cleanup_regex"; // do not change to avoid breaking compat w/ old modules
 	private static final String GTFS_RDS_SERVICE_ID_CLEANUP_REGEX = "gtfs_rts_service_id_cleanup_regex"; // do not change to avoid breaking compat w/ old modules
 	private static final String GTFS_RDS_ROUTE_ID_CLEANUP_REGEX = "gtfs_rts_route_id_cleanup_regex"; // do not change to avoid breaking compat w/ old modules
 	private static final String GTFS_RDS_TRIP_ID_CLEANUP_REGEX = "gtfs_rts_trip_id_cleanup_regex"; // do not change to avoid breaking compat w/ old modules
@@ -1095,7 +1096,7 @@ private static void dumpCommonValues(File dumpDirF, GAgencyTools gAgencyTools, M
 				ow.write(Constants.NEW_LINE);
 			}
 			//noinspection DiscouragedApi
-			ow.write(getRESOURCES_STRING(GTFS_RDS_AGENCY_ID, mSpec.getFirstAgency().getId()));
+			ow.write(getRESOURCES_STRING(GTFS_RDS_AGENCY_ID, mSpec.getFirstAgency().getCleanAgencyId(gAgencyTools)));
 			ow.write(Constants.NEW_LINE);
 			ow.write(getRESOURCES_INTEGER(GTFS_RDS_AGENCY_TYPE, mSpec.getFirstAgency().getType()));
 			ow.write(Constants.NEW_LINE);
@@ -1107,6 +1108,10 @@ private static void dumpCommonValues(File dumpDirF, GAgencyTools gAgencyTools, M
 			ow.write(Constants.NEW_LINE);
 			ow.write(getRESOURCES_STRING(GTFS_RDS_COLOR, mSpec.getFirstAgency().getColor()));
 			ow.write(Constants.NEW_LINE);
+			if (gAgencyTools.getAgencyIdCleanupRegex() != null) {
+				ow.write(getRESOURCES_STRING(GTFS_RDS_AGENCY_ID_CLEANUP_REGEX, escapeResString(gAgencyTools.getAgencyIdCleanupRegex())));
+				ow.write(Constants.NEW_LINE);
+			}
 			if (gAgencyTools.getServiceIdCleanupRegex() != null) {
 				ow.write(getRESOURCES_STRING(GTFS_RDS_SERVICE_ID_CLEANUP_REGEX, escapeResString(gAgencyTools.getServiceIdCleanupRegex())));
 				ow.write(Constants.NEW_LINE);
@@ -1151,7 +1156,7 @@ public static String escapeResString(@NotNull String string) {
 		return string;
 	}
 
-	private static void dumpValues(File dumpDirF, String fileBase, MSpec mSpec, Double minLat, Double maxLat, Double minLng, Double maxLng,
+	private static void dumpValues(GAgencyTools gAgencyTools, File dumpDirF, String fileBase, MSpec mSpec, Double minLat, Double maxLat, Double minLng, Double maxLng,
 								   int firstTimestampInSec, int lastTimestampInSec, @Nullable String inputUrl, boolean deleteAll) {
 		File file;
 		BufferedWriter ow = null;
@@ -1183,7 +1188,7 @@ private static void dumpValues(File dumpDirF, String fileBase, MSpec mSpec, Doub
 					ow.write(Constants.NEW_LINE);
 				}
 				//noinspection DiscouragedApi
-				ow.write(getRESOURCES_STRING(GTFS_RDS_AGENCY_ID, mSpec.getFirstAgency().getId()));
+				ow.write(getRESOURCES_STRING(GTFS_RDS_AGENCY_ID, mSpec.getFirstAgency().getCleanAgencyId(gAgencyTools)));
 				ow.write(Constants.NEW_LINE);
 				ow.write(getRESOURCES_INTEGER(GTFS_RDS_AGENCY_TYPE, mSpec.getFirstAgency().getType()));
 				ow.write(Constants.NEW_LINE);

src/main/java/org/mtransit/parser/mt/data/MAgency.kt

@@ -1,6 +1,6 @@
 package org.mtransit.parser.mt.data
 
-import androidx.annotation.Discouraged
+import org.mtransit.parser.db.SQLUtils.escapeId
 import org.mtransit.parser.gtfs.GAgencyTools
 import org.mtransit.parser.gtfs.data.GAgency
 import org.mtransit.parser.gtfs.data.GIDs
@@ -29,19 +29,11 @@ data class MAgency(
         agencyTools.agencyRouteType,
     )
 
-    @Suppress("unused")
-    @get:Discouraged(message = "Not memory efficient")
-    val id: String get() = _id
+    fun getCleanAgencyId(agencyTools: GAgencyTools): String = _id.convertAgencyId(agencyTools)
 
-    private val _id: String
-        get() {
-            return GIDs.getString(idInt)
-        }
+    private val _id: String get() = GIDs.getString(idInt)
 
-    fun toStringPlus(): String {
-        return toString() +
-                "+(_id:$_id)"
-    }
+    fun toStringPlus() = toString() + "+(_id:$_id)"
 
     override fun compareTo(other: MAgency?): Int {
         return when {
@@ -98,5 +90,14 @@ data class MAgency(
                 }
             return null
         }
+
+        internal fun convert(agencyTools: GAgencyTools, agencyId: String): String =
+            agencyTools.cleanAgencyId(agencyId).escapeId()
     }
-}
+}
+
+fun String.convertAgencyId(agencyTools: GAgencyTools) =
+    MAgency.convert(
+        agencyTools = agencyTools,
+        agencyId = this,
+    )

src/main/java/org/mtransit/parser/mt/data/MDirection.kt

@@ -190,7 +190,7 @@ data class MDirection @JvmOverloads constructor(
             }
         }
 
-        private const val SLASH = " / "
+        private const val SLASH = "/"
 
         @JvmStatic
         fun mergeHeadsignValue(

src/main/java/org/mtransit/parser/mt/data/MServiceIds.kt

@@ -65,8 +65,7 @@ object MServiceIds {
     fun containsAllIdInts(idInts: Iterable<Int>)
         = idInts.all { idIntToId.containsKey(it) }
 
-    @JvmStatic
-    fun convert(agencyTools: GAgencyTools, serviceId: String, keep: Boolean = true, quotesString: Boolean = false): String =
+    internal fun convert(agencyTools: GAgencyTools, serviceId: String, keep: Boolean = true, quotesString: Boolean = false): String =
         if (FeatureFlags.F_EXPORT_SERVICE_ID_INTS) {
             getInt(agencyTools.cleanServiceId(serviceId, keep)).toString()
         } else {

src/main/java/org/mtransit/parser/mt/data/MTripIds.kt

@@ -43,7 +43,7 @@ object MTripIds {
         synchronized(incrementLock) {
             val iterator = idToIdInt.asMutableMap().entries.iterator()
             while (iterator.hasNext()) {
-                val (tripId,tripIdInt) = iterator.next()
+                val (tripId, tripIdInt) = iterator.next()
                 if (tripId !in usedTripIds) {
                     idIntToId.remove(tripIdInt)
                     iterator.remove()
@@ -88,8 +88,7 @@ object MTripIds {
     @JvmStatic
     fun getAllSorted() = getAll().sorted()
 
-    @JvmStatic
-    fun convert(tripId: String, quotesString: Boolean = false) =
+    internal fun convert(tripId: String, quotesString: Boolean = false) =
         if (FeatureFlags.F_EXPORT_TRIP_ID_INTS) {
             getInt(tripId).toString()
         } else {

src/test/java/org/mtransit/parser/mt/MDirectionHeadSignFinderTest.kt

@@ -450,7 +450,7 @@ class MDirectionHeadSignFinderTest {
         // Act
         val result = MDirectionHeadSignFinder.findDirectionHeadSign(RID, gRouteTrips, routeGTFS, directionId, agencyTools)
         // Assert
-        assertEquals("foo foo / trip head-sign", result?.headSign)
+        assertEquals("foo foo/trip head-sign", result?.headSign)
     }
 
     @Test
@@ -622,7 +622,7 @@ class MDirectionHeadSignFinderTest {
         // Act
         val result = MDirectionHeadSignFinder.findDirectionHeadSign(RID, gRouteTrips, routeGTFS, directionId, agencyTools)
         // Assert
-        assertEquals("foo foo / trip head-sign", result?.headSign)
+        assertEquals("foo foo/trip head-sign", result?.headSign)
     }
 
     @Test