convert/

Utilities for converting DOCX and PDF files to Markdown with YAML frontmatter, on the root export, and two thin runnables that convert one file each.

Scripts

import {
  convertDocxToMd,
  convertPdfToMd,
} from '@karmaniverous/jeeves-scripts-core';

const { converted, skipped, failed } = await convertPdfToMd([
  '/base/content/docs',
]);
Copy

Each resolves to { converted, skipped, failed } and logs one line per file (CONVERT, SKIP, FAIL) and a summary. A file that fails is logged and counted, never thrown.

Runnables

To convert one file without writing code, run a thin runnable with the input file and the Markdown file to write:

node $JEEVES_SCRIPTS_DIR/bin/jeeves-scripts.js run convert-pdf-to-md /base/content/docs/spec.pdf /base/content/docs/spec.pdf.md
node $JEEVES_SCRIPTS_DIR/bin/jeeves-scripts.js run convert-docx-to-md report.docx report.md
Copy

Each takes exactly <in> <out> (anything else fails the run: a CRASH line with the usage error on stderr, exit 1), writes <out> with the same frontmatter as the utilities (source, converted, and pages for a PDF), overwriting it, and exits 1 when the input can't be read or converted. Neither has a manifest; an instance that wants one scheduled adds a manifest in its own jobs/*.json (runnables.md).

Data Flow

flowchart LR
  paths["dirs"] --> find["findFiles(ext)"]
  find --> check["needsConversion?\n(skip if up-to-date)"]
  check --> extract["extract content\n(mammoth/pdf-parse)"]
  extract --> output[".docx.md / .pdf.md\n(with YAML frontmatter)"]
Copy

• Both take the directories to scan recursively (missing ones are skipped).
• Output files are written alongside the source (e.g., report.docx → report.docx.md).
• Staleness checking: skips files whose .md output is newer than the source.
• The utilities scan directories and skip up-to-date outputs; the runnables convert exactly the one file named and always write it.

Prerequisites

No external prerequisites. Uses bundled npm dependencies (mammoth, pdf-parse).

Key Files