[HUDI-5336] Fixing log file pattern match to ignore extraneous files (#…

…7612)

Co-authored-by: Y Ethan Guo <ethan.guoyihua@gmail.com>

hudi-common/src/main/java/org/apache/hudi/common/fs/FSUtils.java

@@ -80,7 +80,7 @@ public class FSUtils {
   // Log files are of this pattern - .b5068208-e1a4-11e6-bf01-fe55135034f3_20170101134598.log.1_1-0-1
   // Archive log files are of this pattern - .commits_.archive.1_1-0-1
   public static final Pattern LOG_FILE_PATTERN =
-      Pattern.compile("\\.(.+)_(.*)\\.(.+)\\.(\\d+)(_((\\d+)-(\\d+)-(\\d+))(.cdc)?)?");
+      Pattern.compile("^\\.(.+)_(.*)\\.(log|archive)\\.(\\d+)(_((\\d+)-(\\d+)-(\\d+))(.cdc)?)?");
   private static final int MAX_ATTEMPTS_RECOVER_LEASE = 10;
   private static final long MIN_CLEAN_TO_KEEP = 10;
   private static final long MIN_ROLLBACK_TO_KEEP = 10;

hudi-common/src/test/java/org/apache/hudi/common/functional/TestHoodieLogFormat.java

@@ -345,7 +345,7 @@ public void testAppendNotSupported(@TempDir java.nio.file.Path tempDir) throws I
 
     for (int i = 0; i < 2; i++) {
       Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(testPath)
-          .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits.archive").overBaseCommit("")
+          .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits").overBaseCommit("")
           .withFs(localFs).build();
       writer.appendBlock(dataBlock);
       writer.close();

hudi-common/src/test/java/org/apache/hudi/common/functional/TestHoodieLogFormatAppendFailure.java

@@ -103,7 +103,7 @@ public void testFailedToGetAppendStreamFromHDFSNameNode()
     HoodieAvroDataBlock dataBlock = new HoodieAvroDataBlock(records, header, HoodieRecord.RECORD_KEY_METADATA_FIELD);
 
     Writer writer = HoodieLogFormat.newWriterBuilder().onParentPath(testPath)
-        .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits.archive")
+        .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits")
         .overBaseCommit("").withFs(fs).build();
 
     writer.appendBlock(dataBlock);
@@ -134,7 +134,7 @@ public void testFailedToGetAppendStreamFromHDFSNameNode()
     // Opening a new Writer right now will throw IOException. The code should handle this, rollover the logfile and
     // return a new writer with a bumped up logVersion
     writer = HoodieLogFormat.newWriterBuilder().onParentPath(testPath)
-        .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits.archive")
+        .withFileExtension(HoodieArchivedLogFile.ARCHIVE_EXTENSION).withFileId("commits")
         .overBaseCommit("").withFs(fs).build();
     header = new HashMap<>();
     header.put(HoodieLogBlock.HeaderMetadataType.COMMAND_BLOCK_TYPE,

hudi-common/src/test/java/org/apache/hudi/common/table/view/TestHoodieTableFileSystemView.java

@@ -306,12 +306,14 @@ protected void testInvalidLogFiles() throws Exception {
     String fileName2 =
         FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 1, TEST_WRITE_TOKEN);
     // create a dummy log file mimicing cloud stores marker files
-    String fileName3 = "_DUMMY_" + fileName1.substring(1, fileName1.length());
+    String fileName3 = "_GCS_SYNCABLE_TEMPFILE_" + fileName1;
+    String fileName4 = "_DUMMY_" + fileName1.substring(1, fileName1.length());
     // this file should not be deduced as a log file.
 
     Paths.get(basePath, partitionPath, fileName1).toFile().createNewFile();
     Paths.get(basePath, partitionPath, fileName2).toFile().createNewFile();
     Paths.get(basePath, partitionPath, fileName3).toFile().createNewFile();
+    Paths.get(basePath, partitionPath, fileName4).toFile().createNewFile();
     HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
 
     HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, instantTime1);