feat: extract batdoc-core library crate from CLI binary

Split the single-binary crate into a Cargo workspace so other Rust
projects (e.g. Rummage MCP server) can depend on document extraction
as a library without pulling in bat/terminal dependencies.

- New batdoc-core crate with 8-item public API: Format, BatdocError,
  Result, detect_format, extract_plain, extract_markdown, to_plain,
  to_markdown
- 13 source modules moved to batdoc-core/src/ (git history preserved)
- CLI rewritten as thin wrapper using batdoc_core::*
- bat and is-terminal remain CLI-only dependencies
- 257 unit tests pass, clippy clean

Author: damon

Cargo.lock

@@ -1,6 +1,9 @@
+[workspace]
+members = [".", "batdoc-core"]
+
 [package]
 name = "batdoc"
-version = "1.3.1"
+version = "1.4.0"
 edition = "2021"
 description = "cat(1) for doc, docx, xls, xlsx, pptx, and pdf -- renders to markdown with bat"
 license = "MIT"
@@ -15,15 +18,9 @@ name = "batdoc"
 path = "src/main.rs"
 
 [dependencies]
-base64 = "0.22"
+batdoc-core = { path = "batdoc-core" }
 bat = { version = "0.26.1", default-features = false, features = ["regex-fancy", "paging"] }
-cfb = "0.13"
-encoding_rs = "0.8"
 is-terminal = "0.4"
-quick-xml = "0.37"
-pdf-extract = "0.10"
-thiserror = "2"
-zip = { version = "2", default-features = false, features = ["deflate"] }
 
 [build-dependencies]
 man = "0.3"

Cargo.toml

@@ -114,8 +114,28 @@ and for formats without OOXML image support (`.doc`, `.xls`, `.pdf`).
 
 ## Dependencies
 
-Eight crates, no C, no system libs: `base64`, `bat`, `cfb`, `encoding_rs`,
-`pdf-extract`, `quick-xml`, `zip`, `is-terminal`.
+The CLI binary depends on `batdoc-core` (document extraction library),
+`bat` (syntax highlighting), and `is-terminal` (tty detection).
+
+The `batdoc-core` library depends on `cfb`, `encoding_rs`, `quick-xml`,
+`zip`, `pdf-extract`, `base64`, and `thiserror`. No C, no system libs.
+
+## Library
+
+The extraction engine is available as a standalone Rust library for
+programmatic use — no CLI or terminal dependencies:
+
+```toml
+[dependencies]
+batdoc-core = { git = "https://github.com/daemonp/batdoc" }
+```
+
+```rust
+let data = std::fs::read("report.docx")?;
+let markdown = batdoc_core::to_markdown(&data, false)?;
+```
+
+See [batdoc-core/README.md](batdoc-core/README.md) for the full API.
 
 ## History
 

README.md

@@ -0,0 +1,21 @@
+[package]
+name = "batdoc-core"
+version = "1.4.0"
+edition = "2021"
+description = "Document text extraction library — DOCX, XLSX, PPTX, DOC, XLS, PDF to plain text or Markdown"
+license = "MIT"
+repository = "https://github.com/daemonp/batdoc"
+readme = "README.md"
+
+[lib]
+name = "batdoc_core"
+path = "src/lib.rs"
+
+[dependencies]
+base64 = "0.22"
+cfb = "0.13"
+encoding_rs = "0.8"
+pdf-extract = "0.10"
+quick-xml = "0.37"
+thiserror = "2"
+zip = { version = "2", default-features = false, features = ["deflate"] }

batdoc-core/Cargo.toml

@@ -0,0 +1,61 @@
+# batdoc-core
+
+Document text extraction library for Rust. Converts `.doc`, `.docx`, `.xls`,
+`.xlsx`, `.pptx`, and `.pdf` files to plain text or Markdown.
+
+Format detection is by magic bytes, not file extension — works on raw byte
+buffers without filesystem access, which makes it suitable for email
+attachments, HTTP uploads, and other contexts where filenames are unreliable.
+
+## Usage
+
+```toml
+[dependencies]
+batdoc-core = { git = "https://github.com/daemonp/batdoc" }
+```
+
+```rust
+use batdoc_core::{detect_format, extract_markdown, extract_plain, to_markdown, Format};
+
+// One-shot: detect format and extract in a single call
+let data: Vec<u8> = std::fs::read("report.docx").unwrap();
+let markdown = batdoc_core::to_markdown(&data, false).unwrap();
+let plain = batdoc_core::to_plain(&data).unwrap();
+
+// Two-step: detect first, then extract (useful for logging or branching)
+let format = batdoc_core::detect_format(&data).unwrap();
+println!("Detected format: {format}");
+let text = batdoc_core::extract_plain(&data, format).unwrap();
+```
+
+## Public API
+
+```rust
+pub enum Format { Doc, Xls, Docx, Xlsx, Pptx, Pdf }
+pub enum BatdocError { Io, Zip, Document, Render }
+pub type Result<T> = std::result::Result<T, BatdocError>;
+
+pub fn detect_format(data: &[u8]) -> Result<Format>;
+pub fn extract_plain(data: &[u8], format: Format) -> Result<String>;
+pub fn extract_markdown(data: &[u8], format: Format, images: bool) -> Result<String>;
+pub fn to_plain(data: &[u8]) -> Result<String>;
+pub fn to_markdown(data: &[u8], images: bool) -> Result<String>;
+```
+
+`extract_markdown` with `images: true` embeds images from DOCX/XLSX/PPTX as
+base64 data URIs. Has no effect on DOC, XLS, or PDF.
+
+## Supported formats
+
+| Format | Detection | Parser |
+|--------|-----------|--------|
+| `.doc` | OLE2 magic + `/WordDocument` stream | Binary Word 97+ (BIFF-like) |
+| `.xls` | OLE2 magic + `/Workbook` stream | BIFF8 (Excel 97+) |
+| `.docx` | ZIP magic + `word/document.xml` | OOXML |
+| `.xlsx` | ZIP magic + `xl/workbook.xml` | OOXML |
+| `.pptx` | ZIP magic + `ppt/presentation.xml` | OOXML |
+| `.pdf` | `%PDF-` header | pdf-extract |
+
+## License
+
+MIT

batdoc-core/README.md

@@ -5,7 +5,7 @@
 
 /// All errors that can occur during document parsing and rendering.
 #[derive(Debug, thiserror::Error)]
-pub(crate) enum BatdocError {
+pub enum BatdocError {
     /// I/O error (file read, stream read, OLE2 compound file).
     #[error("{0}")]
     Io(#[from] std::io::Error),
@@ -24,4 +24,4 @@ pub(crate) enum BatdocError {
 }
 
 /// Convenience alias used throughout the crate.
-pub(crate) type Result<T> = std::result::Result<T, BatdocError>;
+pub type Result<T> = std::result::Result<T, BatdocError>;