| # PXF Documentation |
| |
| The Markdown source in this directory is published as part of the Apache |
| Cloudberry website at <https://cloudberry.apache.org/docs>. The site build is |
| owned by [`apache/cloudberry-site`](https://github.com/apache/cloudberry-site) |
| (Docusaurus-based); this repository keeps the source-of-truth Markdown so |
| documentation can be edited alongside the PXF code it describes. |
| |
| ## Authoring guidelines |
| |
| - Files use plain Markdown with Docusaurus-flavoured frontmatter: |
| |
| ```markdown |
| --- |
| title: Reading and Writing HDFS Parquet Data |
| description: Read and write Parquet data in HDFS via PXF. |
| sidebar_position: 6 |
| --- |
| ``` |
| |
| - Internal links use relative `.md` paths (for example, |
| `./hdfs_text.md` or `../administering/cfg_server.md#about-the-pxffsbasepath-property`). |
| Heading anchors are auto-generated by Docusaurus from the heading text. |
| - Images live under [`graphics/`](graphics/) and are referenced with relative |
| paths (for example, `../graphics/pxfarch.png`). |
| - The directory layout mirrors the website sidebar. Each sub-directory carries a |
| `_category_.json` describing its label and position; new categories should |
| follow the same pattern. |
| |
| ## Local preview |
| |
| Clone `apache/cloudberry-site`, point its sync script at this checkout, and run |
| `npm start`. See the cloudberry-site repository for details. |
