# rpptx-cli
`rpptx-cli` makes complete PPTX workflows available to shell scripts. It
inspects and extracts deck content, validates package invariants, compares or
replaces text, and produces deterministic fixed output.
## Capabilities
- Human-readable inspection, and JSON inspection with per-shape kind,
placeholder, geometry, rotation, autofit, and text detail.
- Slide-order text extraction and recursive outlines, as plain text or schema-1
JSON, with speaker notes on request.
- PDF, PNG, JPEG, and multi-page TIFF conversion.
- Selected-slide rendering, thumbnails, text diff, replacement, and validation.
- Modern comment thread listing, addition, replies, resolution, and removal.
- Scriptable output covers slide order, notes, comments, recursive group
content, relationship-backed media, and deterministic rendering diagnostics.
## Measured footprint and speed
| Crates.io archive: rpptx-cli | 40,881 compressed bytes, 179,264 member bytes, 8 members | 0.13.1 | macOS 26.6.2, Apple M5 Max, arm64 | `cargo package --locked --no-verify` | Tracked `rpptx-cli` package inventory | `python3 scripts/readme_doctests.py --record-measurements` | gzip archive bytes, tar member bytes, tar member count | 2026-10-04 |
## Use it when
Use the CLI for shell automation. Use `rpptx` when the same operations belong inside a Rust application.
## Relationship
It uses `oxml-cli-support` for shared command conventions and the real `rpptx` facade for document behavior.
## Example
```sh
cargo install rpptx-cli --version '^0.13.1'
rpptx inspect deck.pptx --json
rpptx text deck.pptx --json
rpptx outline deck.pptx --notes
rpptx comment add deck.pptx --slide 2 --author Reviewer \
--text "Check the figures" --date 2026-09-25T10:00:00Z -o reviewed.pptx
rpptx comment list reviewed.pptx --json
rpptx convert deck.pptx --to pdf -o deck.pdf
rpptx thumbnail deck.pptx -o thumbnail.png
```
`inspect --json` keeps its existing keys and adds `shape_details` beside each
slide's shape count. Each shape reports its z-order index, id, name, kind,
placeholder type and index, direct position and size in EMU, rotation in
degrees, direct autofit mode, paragraphs, table size, and children. A
placeholder that inherits its geometry from its layout reports `null` geometry,
and a placeholder without an explicit type reports a `null` type.
Both inspection forms report every core property the deck sets. Beside title,
creator, subject, description, keywords, last modified by, created and
modified, they report category, content status, identifier, language, last
printed, revision and version, which `inspect --json` adds to its `metadata`
object. Human-readable inspection prints `(none)` when the deck sets no core
property.
`text --json` reports each slide's one-based number, slide id, paragraphs, and
speaker notes, or `null` notes when the slide has no notes part. Each paragraph
has a typed zero-based `path` of shape, table row, table cell, and paragraph
positions, the id of the shape that owns it, its level, its visible text, and
its regular runs. Run indexes are the facade's run positions, so fields and line
breaks appear only in the paragraph text, where a line break is U+000B. Run
`formatting` is `null` when a run has no direct properties. Otherwise it records
nullable direct bold, italic, underline, font, point size, and sRGB colour
values.
`outline --json` reports each slide's title, its outline items with their
levels, and its speaker notes. Notes are the plain text of the notes body, with
paragraphs and line breaks both written as newlines. `--notes` prints one
`Notes:` line for each non-empty notes line after a slide's plain text or
outline. JSON output always carries the notes.
`comment` works on modern PowerPoint comments. Legacy comments stay preserved
but are not listed. `list` shows each comment and reply as `open`, `resolved`,
or `closed`, and its JSON carries the `resolved` flag beside the raw `status`
token. `add --slide` takes a one-based slide number. `add` and `reply` require
an RFC 3339 `--date`, reuse an existing author with the same name, and add the
author otherwise. New ids are sequential GUIDs, so the output depends on
neither a clock nor a random source. `resolve` takes a thread id. `remove`
takes a thread id, which also removes its replies, or a reply id. Every
mutation requires `-o/--output`, refuses an existing output, publishes only a
complete presentation, and supports a schema-1 record through `--json`.
`convert`, `render`, and `thumbnail` refuse an output file that already exists
unless `--force` is given. Even with `--force` they refuse their own input file
under any spelling of its path, and an output that is not a regular file, such
as a directory, a symbolic link, a FIFO, or a device like `/dev/null`. A run
checks every file it would write before it writes the first one, and publishes
each file only once it is complete, so a failed run leaves no truncated output.
The output extension of `replace` and of every comment mutation selects the
package class the output declares. A `.potx` template written to `deck.pptx`
becomes a presentation, and a deck that carries a VBA project cannot change to
a macro-free extension.
Run `rpptx --help` or `rpptx <command> --help` for the complete command surface.