zhangshenghang opened a new pull request, #13390:
URL: https://github.com/apache/seatunnel/pull/13390

   Yes, I am contributing to **Apache SeaTunnel**.
   
   ## What documentation issues were found
   
   While comparing connector documentation against the actual Java option 
definitions and runtime code on the current `dev` branch (commit 7029c0969a), 
the following verified inaccuracies were found:
   
   1. **Csv option section headers in file sinks** (`en`/`zh` × CosFile, 
FtpFile, HdfsFile, LocalFile, OssFile, OssJindoFile, S3File, SftpFile): the 
`### csv_string_quote_mode [string]` section header contradicted both the 
option table (`enum`) and the code (`FileBaseSinkOptions.CSV_STRING_QUOTE_MODE` 
is `.enumType(CsvStringQuoteMode.class)`). Changed all 16 headers to `[enum]`.
   2. **Cassandra sink** (`en`/`zh`): (a) the `batch_size` section header said 
`[number]` while the table and code (`.intType()`) say int → `[int]`; (b) the 
notes claimed a sync batch holds "at most `batch_size` statements", but 
`CassandraSinkWriter.write()` appends the row before checking the counter 
(`getAndIncrement() >= batchSize`), so a flush actually executes `batch_size + 
1` statements. Reworded both.
   3. **Sentry sink** (`en`/`zh`): `maxCacheItems` / `maxQueueSize` headers 
said `[number]`; code declares `.intType()` → `[int]`.
   4. **Paimon source** (`en`/`zh`): `paimon.hadoop.conf` header said 
`[string]`; code is `.mapType()` (`PaimonBaseOptions`) → `[map]`.
   5. **SqlServer CDC** (`en`/`zh`): `table-names` and `table-pattern` were 
both marked plain required while being documented as mutually exclusive 
(self-contradictory). `SqlServerIncrementalSourceFactory.optionRule()` 
registers them only as `.exclusive(...)` (neither `.required()`), and 
`SqlServerSourceConfigFactory` omits `table.include.list` when both are unset, 
capturing all CDC-enabled tables of `database-names`. Changed to "Conditionally 
required"/"条件必填" with the both-omitted behavior documented, matching the 
wording already used by `MySQL-CDC.md` for the identical code pattern.
   6. **Vitess CDC** (`en`/`zh`): the option table omitted `username` / 
`password`, which `VitessSourceOptions` defines and 
`VitessSourceFactory.optionRule()` accepts (optional VTGate gRPC credentials). 
Added both rows.
   7. **DataHub sink FAQ** (`en`/`zh`): claimed failures beyond the retry 
budget "are surfaced to the job". `DataHubWriter.write()` catches 
`DatahubClientException` and only logs it; `retry()` also only logs on final 
failure. Corrected to state records are logged and dropped (best-effort), so 
checkpointing does not provide at-least-once replay.
   8. **Console sink** (`en`/`zh`): claimed single-table jobs print 
`tableId=-1`. `SeaTunnelRow.tableId` is a `String` (default empty), and sources 
set it to the table path (e.g. FakeSource sets 
`catalogTable.getTableId().toTablePath().toString()`). `-1` is the 
pre-String-era output. Reworded accordingly.
   9. **FakeSource** (`en`/`zh`): the feature list marked parallelism 
unchecked, contradicting `FakeSource implements SupportParallelism`, 
`FakeSourceSplitEnumerator` (`splitId % currentParallelism`), and the doc's own 
split-assignment note. Checked the box.
   10. **Assert sink** (`en`/`zh`): described `MIN_ROW`/`MAX_ROW` as evaluated 
against "the cumulative row count observed by that writer instance". 
`AssertSinkWriter` keeps JVM-static `LongAccumulator` counters shared by every 
parallel writer of a table, evaluated by the last writer of the table to close. 
Corrected the scope description.
   11. **InfluxDB source** (`zh` only): the `where` reserved-option note was 
missing the `tables_configs` paragraph that the `en` doc already has 
(configuring `where` in a table entry fails at build time with 
`OptionValidationException`; the multi-table split parser is case-insensitive 
on `where`).
   
   ## What was changed / which areas were updated
   
   35 documentation files under `docs/en/connectors` and `docs/zh/connectors` 
(source + sink): CDC (SqlServer, Vitess), file sinks (8 connectors × csv option 
header), Paimon, Cassandra, Sentry, DataHub, Console, FakeSource, Assert, 
InfluxDB. Every change is backed by the cited code location; no formatting-only 
churn, no unrelated edits.
   
   ## Were English and Chinese docs both checked?
   
   Yes. Every issue was fixed in both the `en` and `zh` versions where the 
content exists. The only single-language change is item 11 (`zh` InfluxDB), 
where the `en` doc already contained the correct paragraph and `zh` was lagging 
behind.
   
   ## How duplicate PRs from the last 7 days were checked
   
   Before implementing and again before opening this PR, I listed all PRs 
created in the last 7 days (`gh pr list` with author `zhangshenghang` plus 
repo-wide doc searches) and inspected the changed-file lists of #13380, #13379, 
#13327, #13292, #13289, #13284, #13302, #13274, #13282. None of them covers the 
connectors or issues fixed here: #13380 touches 
Clickhouse/Redis/Elasticsearch/Pulsar/RocketMQ/Maxcompute/IoTDB/HugeGraph/Neo4j/LLM/Replace/Zeta
 REST; #13327 touches CDC MongoDB/PostgreSQL/Opengauss + Http/Kafka + 
transforms; #13302 touches JDBC-family + getting-started; #13289 aligns 
file-connector option tables (its `csv_string_quote_mode` change is on the 
table rows, not the section headers fixed here — checked its diff explicitly, 
no heading lines touched); #13284 touches FtpFile/LocalFile/Paimon sink 
examples. The remaining open `@site` link in 
`docs/zh/engines/zeta/rest-api-v2.md` was deliberately left untouched because 
#13380 already fixes it.
   
   ## How the change was verified
   
   - Every documented claim above was checked against the code before editing 
(option definitions, factory option rules, writer/enumerator runtime paths), 
with file/line evidence collected for each issue.
   - Re-ran a repo-wide relative-link checker over `docs/en` and `docs/zh`: no 
new broken links (the only remaining hit is the pre-existing `@site` line owned 
by #13380).
   - Ran a heading-vs-table type-consistency checker over all connector docs: 
the mismatches fixed here are resolved; the only remaining hits are 
`cursor.stop.mode`/`auth_type` (owned by #13380) and an `int`/`Integer` wording 
difference that is semantically identical.
   - Ran a markdown table structure checker over all 35 edited files: 0 newly 
malformed rows (the checker's few hits are pre-existing escaped-pipe cells that 
also exist on `upstream/dev`).
   - The Docusaurus website build was not run locally because it requires 
checking out and installing the separate `apache/seatunnel-website` repository; 
the link, table, and consistency checks above cover the changed markdown.
   
   ## Scope
   
   35 documentation files, 46 insertions / 38 deletions — all real content 
corrections satisfying the multi-file scope.


-- 
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]

Reply via email to