This is an automated email from the ASF dual-hosted git repository.
danny0405 pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/hudi.git
The following commit(s) were added to refs/heads/master by this push:
new 6334bf19569 [HUDI-6791] Make some comments look better (#11854)
6334bf19569 is described below
commit 6334bf19569fadd5f54bf26b7c8f33a0a2bcca67
Author: Lin Liu <[email protected]>
AuthorDate: Wed Aug 28 18:34:21 2024 -0700
[HUDI-6791] Make some comments look better (#11854)
---
.../common/table/cdc/HoodieCDCInferenceCase.java | 47 +++++++++++-----------
1 file changed, 23 insertions(+), 24 deletions(-)
diff --git
a/hudi-common/src/main/java/org/apache/hudi/common/table/cdc/HoodieCDCInferenceCase.java
b/hudi-common/src/main/java/org/apache/hudi/common/table/cdc/HoodieCDCInferenceCase.java
index ed2a1c4c185..6722860ad8e 100644
---
a/hudi-common/src/main/java/org/apache/hudi/common/table/cdc/HoodieCDCInferenceCase.java
+++
b/hudi-common/src/main/java/org/apache/hudi/common/table/cdc/HoodieCDCInferenceCase.java
@@ -24,18 +24,19 @@ package org.apache.hudi.common.table.cdc;
*
* AS_IS:
* For this type, there must be a real cdc log file from which we get the
whole/part change data.
- * when `hoodie.table.cdc.supplemental.logging.mode` is {@link
HoodieCDCSupplementalLoggingMode#DATA_BEFORE_AFTER}, it keeps all the fields
about the
- * change data, including `op`, `ts_ms`, `before` and `after`. So read it
and return directly,
- * no more other files need to be loaded.
- * when `hoodie.table.cdc.supplemental.logging.mode` is {@link
HoodieCDCSupplementalLoggingMode#DATA_BEFORE}, it keeps the `op`, the key and
the
- * `before` of the changing record. When `op` is equal to 'i' or 'u', need
to get the current record from the
- * current base/log file as `after`.
- * when `hoodie.table.cdc.supplemental.logging.mode` is 'op_key', it just
keeps the `op` and the key of
- * the changing record. When `op` is equal to 'i', `before` is null and get
the current record
- * from the current base/log file as `after`. When `op` is equal to 'u', get
the previous
- * record from the previous file slice as `before`, and get the current
record from the
- * current base/log file as `after`. When `op` is equal to 'd', get the
previous record from
- * the previous file slice as `before`, and `after` is null.
+ * When `hoodie.table.cdc.supplemental.logging.mode` is {@link
HoodieCDCSupplementalLoggingMode#DATA_BEFORE_AFTER},
+ * it keeps all the fields about the change data, including `op`, `ts_ms`,
`before` and `after`.
+ * So read it and return directly, no more other files need to be loaded.
+ * When `hoodie.table.cdc.supplemental.logging.mode` is {@link
HoodieCDCSupplementalLoggingMode#DATA_BEFORE},
+ * it keeps the `op`, the key and the `before` of the changing record.
+ * When `op` is equal to 'i' or 'u', need to get the current record from
the current base/log file as `after`.
+ * When `hoodie.table.cdc.supplemental.logging.mode` is '{@link
HoodieCDCSupplementalLoggingMode#OP_KEY_ONLY',
+ * it just keeps the `op` and the key of the changing record.
+ * When `op` is equal to 'i', `before` is null and get the current record
+ * from the current base/log file as `after`.
+ * When `op` is equal to 'u', get the previous record from the previous
file slice as `before`,
+ * and get the current record from the current base/log file as `after`.
+ * When `op` is equal to 'd', get the previous record from the previous
file slice as `before`, and `after` is null.
*
* BASE_FILE_INSERT:
* For this type, there must be a base file at the current instant. All the
records from this
@@ -49,18 +50,16 @@ package org.apache.hudi.common.table.cdc;
* the value of `before`. The value of `after` for each record is null.
*
* LOG_FILE:
- * For this type, a normal log file of mor table will be used. First we need
to load the previous
- * file slice(including the base file and other log files in the same file
group). Then for each
- * record from the log file, get the key of this, and execute the following
steps:
- * 1) if the record is deleted,
- * a) if there is a record with the same key in the data loaded, `op` is
'd', 'before' is the
- * record from the data loaded, `after` is null;
- * b) if there is not a record with the same key in the data loaded,
just skip.
- * 2) the record is not deleted,
- * a) if there is a record with the same key in the data loaded, `op` is
'u', 'before' is the
- * record from the data loaded, `after` is the current record;
- * b) if there is not a record with the same key in the data loaded,
`op` is 'i', 'before' is
- * null, `after` is the current record;
+ * For this type, a normal log file of MOR table will be used. First we need
to load the previous
+ * file slice (including the base file and other log files in the same file
group). Then for each
+ * record (called `current record` hereafter) from the log file, get its
key, and execute the following steps:
+ * 1) if the current record is deleted,
+ * a) if there is a record with the same key in the data loaded (called
`loaded record` hereafter),
+ * `op` is 'd', 'before' is the loaded record, `after` is null;
+ * b) if the loaded reocrd does not exist, just skip.
+ * 2) the current record is not deleted,
+ * a) if there is a loaded record, `op` is 'u', 'before' is the loaded
record, `after` is the current record;
+ * b) if the loaded record does not exist, `op` is 'i', 'before' is
null, `after` is the current record;
*
* REPLACE_COMMIT:
* For this type, it must be a replacecommit, like INSERT_OVERWRITE and
DROP_PARTITION. It drops
