Skip to main content

provenant/copyright/
mod.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4//! Copyright detection module.
5//!
6//! Detects copyright statements, holder names, and author information
7//! from source code files using a four-stage pipeline:
8//! 1. Text preparation (normalization)
9//! 2. Candidate line selection
10//! 3. Lexing (POS tagging) and parsing (grammar rules)
11//! 4. Refinement and junk filtering
12
13use std::time::Duration;
14
15mod candidates;
16mod credits;
17mod detector;
18mod detector_input_normalization;
19pub mod golden_utils;
20mod grammar;
21mod hints;
22mod lexer;
23mod line_tracking;
24mod parser;
25mod patterns;
26mod prepare;
27mod refiner;
28mod types;
29
30pub use credits::{detect_credits_authors, is_credits_file};
31pub(crate) use prepare::prepare_text_line;
32pub(crate) use refiner::has_copyright_year;
33pub(crate) use refiner::looks_like_source_code;
34pub(crate) use refiner::refine_author;
35pub use refiner::refine_copyright;
36pub use types::{AuthorDetection, CopyrightDetection, HolderDetection};
37
38pub fn detect_copyrights(
39    content: &str,
40    max_runtime: Option<Duration>,
41) -> (
42    Vec<CopyrightDetection>,
43    Vec<HolderDetection>,
44    Vec<AuthorDetection>,
45) {
46    if let Some(max_runtime) = max_runtime {
47        detector::detect_copyrights_from_text_with_deadline(content, Some(max_runtime))
48    } else {
49        detector::detect_copyrights_from_text(content)
50    }
51}