provenant/license_detection/
hash_match.rs1use sha1::{Digest, Sha1};
10
11use crate::license_detection::index::LicenseIndex;
12use crate::license_detection::index::dictionary::{TokenId, TokenKind};
13use crate::license_detection::models::position_span::PositionSpan;
14use crate::license_detection::models::{LicenseMatch, MatchCoordinates, MatcherKind};
15use crate::license_detection::query::QueryRun;
16use crate::models::LineNumber;
17use crate::models::MatchScore;
18
19pub const MATCH_HASH: MatcherKind = MatcherKind::Hash;
20
21pub fn compute_hash(tokens: &[TokenId]) -> [u8; 20] {
34 let mut hasher = Sha1::new();
35
36 for token in tokens {
37 let signed = token.raw() as i16;
38 hasher.update(signed.to_le_bytes());
39 }
40
41 hasher.finalize().into()
42}
43
44pub fn hash_match(index: &LicenseIndex, query_run: &QueryRun) -> Vec<LicenseMatch> {
58 let mut matches = Vec::new();
59 let query_hash = compute_hash(query_run.tokens());
60
61 if let Some(&rid) = index.rid_by_hash.get(&query_hash) {
62 let rule = &index.rules_by_rid[rid];
63 let itokens = &index.tids_by_rid[rid];
64
65 let rule_length = rule.tokens.len();
66
67 let matched_length = query_run.tokens().len();
68 let match_coverage = 100.0;
69
70 let start_line = query_run
71 .line_for_pos(query_run.start)
72 .and_then(LineNumber::new)
73 .unwrap_or(LineNumber::ONE);
74 let end_line = if let Some(end) = query_run.end {
75 query_run
76 .line_for_pos(end)
77 .and_then(LineNumber::new)
78 .unwrap_or(start_line)
79 } else {
80 start_line
81 };
82
83 let end = query_run.end.unwrap_or(query_run.start);
84 let qspan = PositionSpan::range(query_run.start, end + 1);
85 let ispan = PositionSpan::range(0, rule_length);
86 let hispan = PositionSpan::from_positions(
87 (0..rule_length)
88 .filter(|&p| index.dictionary.token_kind(itokens[p]) == TokenKind::Legalese),
89 );
90
91 let license_match = LicenseMatch {
92 license_expression: rule.license_expression.clone(),
93 license_expression_spdx: index
94 .rule_metadata_by_identifier
95 .get(&rule.identifier)
96 .and_then(|metadata| metadata.license_expression_spdx.clone()),
97 from_file: None,
98 start_line,
99 end_line,
100 start_token: query_run.start,
101 end_token: query_run.end.map_or(query_run.start, |e| e + 1),
102 matcher: MATCH_HASH,
103 score: MatchScore::MAX,
104 matched_length,
105 rule_length,
106 match_coverage,
107 rule_relevance: rule.relevance,
108 rid,
109 rule_identifier: rule.identifier.clone(),
110 rule_url: rule.rule_url().unwrap_or_default(),
111 matched_text: None,
112 referenced_filenames: rule.referenced_filenames.clone(),
113 rule_kind: rule.kind(),
114 is_from_license: rule.is_from_license,
115 rule_start_token: 0,
116 coordinates: MatchCoordinates::rule_aligned(qspan, ispan, hispan),
117 candidate_resemblance: 0.0,
118 candidate_containment: 0.0,
119 };
120
121 matches.push(license_match);
122 }
123
124 matches
125}
126
127#[cfg(test)]
128#[path = "hash_match_test.rs"]
129mod tests;