TengHuo commented on code in PR #6000:
URL: https://github.com/apache/hudi/pull/6000#discussion_r953286608
##########
hudi-common/src/main/java/org/apache/hudi/common/table/timeline/HoodieActiveTimeline.java:
##########
@@ -75,16 +75,56 @@ public class HoodieActiveTimeline extends
HoodieDefaultTimeline {
REQUESTED_REPLACE_COMMIT_EXTENSION, INFLIGHT_REPLACE_COMMIT_EXTENSION,
REPLACE_COMMIT_EXTENSION,
REQUESTED_INDEX_COMMIT_EXTENSION, INFLIGHT_INDEX_COMMIT_EXTENSION,
INDEX_COMMIT_EXTENSION,
REQUESTED_SAVE_SCHEMA_ACTION_EXTENSION,
INFLIGHT_SAVE_SCHEMA_ACTION_EXTENSION, SAVE_SCHEMA_ACTION_EXTENSION));
+
+ private static final Set<String> NOT_PARSABLE_TIMESTAMPS = new
HashSet<String>(3) {{
+ add(HoodieTimeline.INIT_INSTANT_TS);
+ add(HoodieTimeline.METADATA_BOOTSTRAP_INSTANT_TS);
+ add(HoodieTimeline.FULL_BOOTSTRAP_INSTANT_TS);
+ }};
+
private static final Logger LOG =
LogManager.getLogger(HoodieActiveTimeline.class);
protected HoodieTableMetaClient metaClient;
/**
* Parse the timestamp of an Instant and return a {@code Date}.
+ * Throw ParseException if timestamp not valid format as
+ * {@link
org.apache.hudi.common.table.timeline.HoodieInstantTimeGenerator#SECS_INSTANT_TIMESTAMP_FORMAT}.
+ *
+ * @param timestamp a timestamp String which follow pattern as
+ * {@link
org.apache.hudi.common.table.timeline.HoodieInstantTimeGenerator#SECS_INSTANT_TIMESTAMP_FORMAT}.
+ * @return Date of instant timestamp
*/
public static Date parseDateFromInstantTime(String timestamp) throws
ParseException {
return HoodieInstantTimeGenerator.parseDateFromInstantTime(timestamp);
}
+ /**
+ * The same format method as above, but this method will mute ParseException
+ * if the gaven timestamp is invalid and return Date(0), or a corresponding
Date if these timestamp provided
+ * {@link
org.apache.hudi.common.table.timeline.HoodieTimeline#INIT_INSTANT_TS},
+ * {@link
org.apache.hudi.common.table.timeline.HoodieTimeline#METADATA_BOOTSTRAP_INSTANT_TS},
+ * {@link
org.apache.hudi.common.table.timeline.HoodieTimeline#FULL_BOOTSTRAP_INSTANT_TS}.
+ * This method is useful when parse timestamp for metrics
+ *
+ * @param timestamp a timestamp String which follow pattern as
+ * {@link
org.apache.hudi.common.table.timeline.HoodieInstantTimeGenerator#SECS_INSTANT_TIMESTAMP_FORMAT}.
+ * @return Date of instant timestamp
+ */
+ public static Date parseDateFromInstantTimeSafely(String timestamp) {
+ Date parsedDate;
+ try {
+ parsedDate =
HoodieInstantTimeGenerator.parseDateFromInstantTime(timestamp);
+ } catch (ParseException e) {
+ LOG.warn("Failed to parse timestamp " + timestamp + " because of " +
e.getMessage());
+ if (NOT_PARSABLE_TIMESTAMPS.contains(timestamp)) {
+ parsedDate = new Date(Integer.parseInt(timestamp));
+ } else {
+ parsedDate = new Date(0);
Review Comment:
It's the old logic in `HoodieInstantTimeGenerator.parseDateFromInstantTime`,
if it catch the error and the timestamp is all zero, it will return `Date(0)`,
so I keep it.
```java
// Special handling for all zero timestamp which is not parsable by
DateTimeFormatter
if (timestamp.equals(ALL_ZERO_TIMESTAMP)) {
return new Date(0);
}
throw e;
```
but I agree with you, it will return a dirty value, which is bad for the
code where it uses this method. `parseDateFromInstantTimeSafely` should return
an optional value, then the code who use this method can decide how to deal
with Option.empty.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
