pdfedit

A command-line tool that finds and edits any text in a PDF while matching the original font, size, and color — not a fixed find-and-replace on known form fields. It walks every text span via PyMuPDF, resolves each span's real font (embedded extraction, or the closest Base-14 fallback), then redacts the old glyphs and redraws the replacement in the same font, size, and color at the same position.

The original file is never modified — every edit writes <name>_edited.pdf next to the input.

Status: alpha, pre-1.0 — the API and CLI may still change. White-fill redaction only (colored backgrounds not yet sampled); no paragraph reflow; scanned pages are detected and warned about, not OCR'd.

Install

pip install pdfedit

This puts a pdfedit command on your PATH. You can also run it as python -m pdfedit.

From source (for development):

git clone https://github.com/sachinpandey22/pdfedit && cd pdfedit
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"      # editable install + pytest

Usage

pdfedit inspect your-document.pdf          # read-only: list every text span with its id, font, size, color
pdfedit edit your-document.pdf             # interactive: pick a span id, type a replacement, `save`
pdfedit batch your-document.pdf edits.json  # non-interactive: apply a list of edits

inspect

Prints one row per span. The ID (p<page>-b<block>-l<line>-s<span>) is the handle you pass to edit or a batch file.

edit

A small REPL. Enter a span id, then the replacement text; the edit is queued with a before/after preview (and the resolved font, and an auto-shrink note if the text had to be made smaller to fit). save applies every queued edit at once and writes the output. quit discards everything.

batch

edits.json is a JSON array; each entry is either:

[
  { "id": "p1-b0-l0-s0", "new_text": "New Name" },
  { "find": "Old Company", "new_text": "New Company" }
]

• id targets one exact span.
• find replaces that substring in every span containing it. Multiple find entries against the same span compose in order.

The whole file is validated first — if any entry is malformed or names a missing span id, nothing is written and the command exits non-zero.

Autofit

If a replacement is wider than the original box, the font size shrinks (down to 60% of the original) until it fits. Past that floor it's drawn at the minimum size and a warning is printed rather than overflowing silently.

Encrypted PDFs

You're prompted for the password. Note: the _edited.pdf copy is written without encryption.

Development

pytest                                   # run the test suite
python tests/fixtures/make_sample.py      # regenerate the synthetic test fixture

There's no separate lint config. See CLAUDE.md for the architecture and the two redraw pitfalls the engine works around, and docs/ for longer background.