sem_core/parser/
differ.rs

1use rayon::prelude::*;
2use serde::Serialize;
3
4use crate::git::types::FileChange;
5use crate::model::change::{ChangeType, SemanticChange};
6use crate::model::identity::match_entities;
7use crate::parser::registry::ParserRegistry;
8use std::collections::HashSet;
9
10#[derive(Debug, Clone, Serialize)]
11#[serde(rename_all = "camelCase")]
12pub struct DiffResult {
13    pub changes: Vec<SemanticChange>,
14    pub file_count: usize,
15    pub added_count: usize,
16    pub modified_count: usize,
17    pub deleted_count: usize,
18    pub moved_count: usize,
19    pub renamed_count: usize,
20}
21
22pub fn compute_semantic_diff(
23    file_changes: &[FileChange],
24    registry: &ParserRegistry,
25    commit_sha: Option<&str>,
26    author: Option<&str>,
27) -> DiffResult {
28    // Process files in parallel: each file's entity extraction and matching is independent
29    let per_file_changes: Vec<(String, Vec<SemanticChange>)> = file_changes
30        .par_iter()
31        .filter_map(|file| {
32            let plugin = registry.get_plugin(&file.file_path)?;
33
34            let before_entities = if let Some(ref content) = file.before_content {
35                let before_path = file.old_file_path.as_deref().unwrap_or(&file.file_path);
36                match std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
37                    plugin.extract_entities(content, before_path)
38                })) {
39                    Ok(entities) => entities,
40                    Err(_) => Vec::new(),
41                }
42            } else {
43                Vec::new()
44            };
45
46            let after_entities = if let Some(ref content) = file.after_content {
47                match std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
48                    plugin.extract_entities(content, &file.file_path)
49                })) {
50                    Ok(entities) => entities,
51                    Err(_) => Vec::new(),
52                }
53            } else {
54                Vec::new()
55            };
56
57            let sim_fn = |a: &crate::model::entity::SemanticEntity,
58                          b: &crate::model::entity::SemanticEntity|
59             -> f64 { plugin.compute_similarity(a, b) };
60
61            let result = match_entities(
62                &before_entities,
63                &after_entities,
64                &file.file_path,
65                Some(&sim_fn),
66                commit_sha,
67                author,
68            );
69
70            if result.changes.is_empty() {
71                None
72            } else {
73                Some((file.file_path.clone(), result.changes))
74            }
75        })
76        .collect();
77
78    let mut all_changes: Vec<SemanticChange> = Vec::new();
79    let mut files_with_changes: HashSet<String> = HashSet::new();
80    for (file_path, changes) in per_file_changes {
81        files_with_changes.insert(file_path);
82        all_changes.extend(changes);
83    }
84
85    // Single-pass counting
86    let mut added_count = 0;
87    let mut modified_count = 0;
88    let mut deleted_count = 0;
89    let mut moved_count = 0;
90    let mut renamed_count = 0;
91
92    for c in &all_changes {
93        match c.change_type {
94            ChangeType::Added => added_count += 1,
95            ChangeType::Modified => modified_count += 1,
96            ChangeType::Deleted => deleted_count += 1,
97            ChangeType::Moved => moved_count += 1,
98            ChangeType::Renamed => renamed_count += 1,
99        }
100    }
101
102    DiffResult {
103        changes: all_changes,
104        file_count: files_with_changes.len(),
105        added_count,
106        modified_count,
107        deleted_count,
108        moved_count,
109        renamed_count,
110    }
111}
sem_core/parser/differ.rs

sem_core/parser/
differ.rs
