[package]
edition = "2024"
rust-version = "1.88"
name = "doc_extractor"
version = "0.1.1"
authors = ["DarkSky <darksky2048@gmail.com>"]
build = false
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "Extract text chunks from common document formats."
homepage = "https://affine.pro"
readme = false
keywords = [
"document",
"extract",
"markdown",
]
categories = ["text-processing"]
license = "AGPL-3.0-only"
repository = "https://github.com/toeverything/AFFiNE"
[lib]
name = "doc_extractor"
path = "src/lib.rs"
[dependencies.infer]
version = "0.19.0"
[dependencies.office_oxide]
version = "0.1.3"
[dependencies.path-ext]
version = "0.1.2"
[dependencies.pdf_oxide]
version = "0.3.65"
default-features = false
[dependencies.readability]
version = "0.3.0"
default-features = false
[dependencies.serde]
version = "1.0"
features = ["derive"]
[dependencies.serde_json]
version = "1.0"
[dependencies.strum_macros]
version = "0.28.0"
[dependencies.text-splitter]
version = "0.32"
features = [
"markdown",
"tiktoken-rs",
]
[dependencies.thiserror]
version = "2"
[dependencies.tiktoken-rs]
version = "0.12"
[dependencies.tree-sitter]
version = "0.26"
[dependencies.tree-sitter-c]
version = "0.24.2"
[dependencies.tree-sitter-cpp]
version = "0.23.4"
[dependencies.tree-sitter-go]
version = "0.25"
[dependencies.tree-sitter-javascript]
version = "0.25"
[dependencies.tree-sitter-python]
version = "0.25"
[dependencies.tree-sitter-rust]
version = "0.24.2"
[dependencies.tree-sitter-typescript]
version = "0.23.2"
[dependencies.url]
version = "2.5"