[Feature][File] Add word parser for RAG support

[joonseolee]

Purpose of this pull request

• Add and refine Word (.docx) reading via WordReadStrategy.
• Output schema (10 fields): element_id, element_type, text_content, font_style, underline_style, font_size, font_family, text_color, alignment, hyperlink_url.
• Process document elements in natural order (paragraphs and tables). Footnote text is included within the referencing paragraph’s text_content.
• Due to Apache POI limitations, the minimal extractable unit is a paragraph. Run-level styles are aggregated at the paragraph level:
  • font_style: NORMAL/BOLD/ITALIC/BOLD_ITALIC
  • underline_style: null or concrete style (e.g., SINGLE)
  • font_size, font_family: first encountered values or null
  • text_color: defaults to "000000" when absent
  • hyperlink_url: all links in a paragraph concatenated with commas

Does this PR introduce any user-facing change?

Yes. The Word reader’s output schema is simplified to 10 fields above. Some formatting attributes now return null when not explicitly present; text_color defaults to "000000". Elements are emitted in document order, and hyperlinks are aggregated per paragraph.

How was this patch tested?

• Added WordReadStrategyTest to validate all 10 fields against a sample .docx.
• Verified:
  • Paragraph rows contain text and aggregated formatting/links.
  • Table rows produce a single text blob per table; formatting-related fields are null.

Check list

• If any new Jar binary package adding in your PR, please add License Notice according
  New License Guide
• If necessary, please update the documentation to describe the new feature. https://github.com/apache/seatunnel/tree/dev/docs
• If you are contributing the connector code, please check that the following files are updated:
  1. Update plugin-mapping.properties and add new connector information in it
  2. Update the pom file of seatunnel-dist
  3. Add ci label in label-scope-conf
  4. Add e2e testcase in seatunnel-e2e
  5. Update connector plugin_config

Related Issue

#9715

[Hisoka-X]

cc @liugddx could you help to review it?

[Hisoka-X]

how about .doc?

[joonseolee]

I forgot it so I will add doc too :)

[Hisoka-X]

we need those too.
"heading_level",
"text",
"page_number",
"position_index",
"parent_id",
"child_ids"
Just like markdown. We should keep most fields are same.

[joonseolee]

@Hisoka-X

I wanted to add those values as well, but I couldn't due to technical issues like the ones below.

• heading_level: Not reliably derivable. Style names/IDs vary by template, language, and custom styles; there is no stable, standard mapping. Outline level may be absent, so consistent extraction isn’t guaranteed.
• text: Same as text_content
• page_number: Not available. Pagination is a rendering/layout result (paper size, margins, fonts, images, printer settings); Apache POI does not perform layout computation.
• position_index: A hierarchical index (Markdown-style) isn’t feasible without a reliable document tree and heading detection; Word’s body is effectively flat and heading detection is inconsistent. A simple global ordinal is possible.
• parent_id: Not available. WordprocessingML does not provide an explicit AST with parent-child relationships for paragraphs.
• child_ids: Not available for the same reason as parent_id; there’s no stable logical child list per element in Word.

[liugddx]

Please add documentation.

[liugddx]

Please fix the test case issue.