codesynapse_core/ts_extract/
astro.rs

1use super::make_file_node;
2use crate::error::Result;
3use crate::extract::{make_id, ImportNode, LanguageExtractor};
4use crate::types::{Edge, ExtractionFragment, Node};
5use std::collections::{HashMap, HashSet};
6use std::path::Path;
7
8pub struct AstroExtractor;
9
10impl AstroExtractor {
11    pub fn extract(source: &[u8], path: &Path) -> Result<ExtractionFragment> {
12        let (file_id, _, file_node) = make_file_node(path);
13        let mut fragment = ExtractionFragment {
14            nodes: vec![file_node],
15            edges: vec![],
16        };
17
18        let content = std::str::from_utf8(source).unwrap_or("");
19        let str_path = path.to_string_lossy().to_string();
20
21        let mut regions: Vec<String> = Vec::new();
22
23        // Extract frontmatter block: ---\n...\n---
24        let trimmed = content.trim_start_matches(['\r', '\n']);
25        if let Some(rest) = trimmed.strip_prefix("---") {
26            if let Some(end) = rest.find("\n---") {
27                regions.push(rest[..end].to_string());
28            }
29        }
30
31        // Extract <script> blocks
32        let content_lower = content.to_ascii_lowercase();
33        let mut search_start = 0usize;
34        while let Some(rel_open) = content_lower[search_start..].find("<script") {
35            let abs_open = search_start + rel_open;
36            if let Some(rel_tag_end) = content_lower[abs_open..].find('>') {
37                let content_start = abs_open + rel_tag_end + 1;
38                if let Some(rel_close) = content_lower[content_start..].find("</script") {
39                    let abs_close = content_start + rel_close;
40                    regions.push(content[content_start..abs_close].to_string());
41                    search_start = abs_close + 8;
42                    continue;
43                }
44            }
45            break;
46        }
47
48        // Regex patterns (compiled once per call — acceptable for extraction workloads)
49        let static_re = regex::Regex::new(r#"import\s+(?:[^'"`;\n]+?\s+from\s+)?['"]([^'"]+)['"]"#)
50            .expect("static import regex");
51        let dynamic_re = regex::Regex::new(r#"import\(\s*['"]([^'"]+)['"]\s*\)?"#)
52            .expect("dynamic import regex");
53
54        let mut seen: HashSet<String> = HashSet::new();
55
56        let add_import = |fragment: &mut ExtractionFragment,
57                          seen: &mut HashSet<String>,
58                          raw: &str,
59                          relation: &str| {
60            if raw.is_empty() {
61                return;
62            }
63            let module = raw.rsplit('/').next().unwrap_or(raw);
64            if module.is_empty() {
65                return;
66            }
67            let node_id = make_id(&[module]);
68            let is_new = seen.insert(node_id.clone());
69            if is_new {
70                fragment.nodes.push(Node {
71                    id: node_id.clone(),
72                    label: raw.to_string(),
73                    file_type: "module".to_string(),
74                    source_file: str_path.clone(),
75                    source_location: None,
76                    community: None,
77                    rationale: None,
78                    docstring: None,
79                    metadata: HashMap::new(),
80                });
81            }
82            fragment.edges.push(Edge {
83                source: file_id.clone(),
84                target: node_id,
85                relation: relation.to_string(),
86                confidence: "EXTRACTED".to_string(),
87                source_file: Some(str_path.clone()),
88                weight: 1.0,
89                context: None,
90            });
91        };
92
93        for region in &regions {
94            for cap in static_re.captures_iter(region) {
95                if let Some(m) = cap.get(1) {
96                    add_import(&mut fragment, &mut seen, m.as_str(), "imports_from");
97                }
98            }
99            for cap in dynamic_re.captures_iter(region) {
100                if let Some(m) = cap.get(1) {
101                    add_import(&mut fragment, &mut seen, m.as_str(), "dynamic_import");
102                }
103            }
104        }
105
106        Ok(fragment)
107    }
108}
109
110impl LanguageExtractor for AstroExtractor {
111    fn file_extensions(&self) -> Vec<&'static str> {
112        vec!["astro"]
113    }
114    fn extract(&self, source: &[u8], path: &Path) -> Result<ExtractionFragment> {
115        Self::extract(source, path)
116    }
117    fn resolve_imports(&self, _imports: &[ImportNode]) -> Vec<Edge> {
118        vec![]
119    }
120    fn collect_type_refs(&self, _fragment: &mut ExtractionFragment) {}
121}
codesynapse_core/ts_extract/astro.rs

codesynapse_core/ts_extract/
astro.rs
