Implement streamdown plugin `ReadPdfFile` (text-layer PDF only) for `ReadFile`

[gravity-api]

Create a streamdown plugin ReadPdfFile that extracts text from PDF files with a text layer. This plugin is executed via the ReadFile meta plugin when FileFormat=Pdf.

OCR is explicitly out of scope.

Inputs

Name	Type	Mandatory	Description
Path	String	Expression	❌
Url	String	Expression	❌
Base64	String	Expression	❌

Outputs (for meta normalization)

Return/emit values that the meta plugin can normalize:

Output Key	Description
Text	Extracted text
PagesCount	Total pages (optional but recommended)

Behavior / Error Handling

• If the PDF is valid but no text can be extracted, fail with a clear message indicating:

  • the PDF may be scanned / image-only
  • OCR is not supported in current version

• Fail on invalid base64 / missing file / download failures / non-PDF content.

Implementation Notes

• Resolve bytes from source:

  • Base64 → bytes
  • Path → File.ReadAllBytes
  • Url → HTTP GET bytes

• Extract text using a PDF text extraction library (no OCR).

[gravity-api]

Resolved
Implemented as isolated plugin instead of stream-down