Document Metadata and Annotations

Three separate things, three separate options

This lesson turns on three independent options that are easy to conflate:

DocumentMetadata: about the file, not the content

parser = liteparse.LiteParse(extract_document_metadata=True, ...)
result = parser.parse(SAMPLE_PDF)
meta = result.doc_meta

doc_meta has nothing to do with what the document says. It's provenance: creation_date/mod_date from the PDF's /Info dictionary, file_version (an encoded integer, 14 means PDF 1.4), is_encrypted, and raw_file_size. result.creator and result.producer (not nested under doc_meta) report the authoring tool: here, producer shows "ReportLab PDF Library", the real library used to generate this course's sample PDFs.

DocumentAnnotation: content sitting on top of the page

page = result.pages[0]
for annotation in page.annotations:
    ...

vendor_memo.pdf contains a real link annotation: the phrase "Procurement Policy" is backed by a link-subtype annotation with a uri (https://example.com/procurement-policy) and a rect in the same top-left, 72-DPI coordinate space every other bounding box in this course uses (TextItem, FormField, ScreenshotRect). Annotations also cover comments, highlights, and stamps on richer PDFs, though this sample only exercises the link case.

extract_links: a rendering switch, not an extraction switch

extract_links (on by default) only matters when output_format="markdown" (Lesson 3): it decides whether a hyperlink becomes Markdown syntax ([Procurement Policy](https://example.com/procurement-policy)) or stays as plain anchor text. It does not control whether the link data itself is available, that's extract_annotations's job. Plain result.text never includes URLs inline either way; it just reads as prose, as shown in the last line of this lesson's output.