vtcode_commons/
workspace_snapshot.rs1use std::collections::BTreeMap;
14use std::io;
15use std::path::{Path, PathBuf};
16
17use serde::{Deserialize, Serialize};
18
19const SKIP_DIRS: &[&str] = &[
21 ".git",
22 ".svn",
23 "target",
24 "node_modules",
25 "dist",
26 "build",
27 ".vtcode",
28 ".next",
29 "vendor",
30];
31
32#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
34pub struct FileStat {
35 pub size: u64,
37 pub mtime_ns: i64,
39 pub head_hash: u64,
41}
42
43const HASH_SAMPLE_BYTES: usize = 4096;
45
46#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
48pub struct WorkspaceSnapshot {
49 pub files: BTreeMap<String, FileStat>,
51 pub captured_at: String,
53}
54
55#[derive(Debug, Clone, Default, PartialEq, Eq)]
57pub struct SnapshotDelta {
58 pub added: Vec<String>,
60 pub changed: Vec<String>,
62 pub removed: Vec<String>,
64}
65
66pub fn capture(workspace: &Path, max_file_bytes: u64) -> io::Result<WorkspaceSnapshot> {
71 let mut files = BTreeMap::new();
72 collect(workspace, workspace, max_file_bytes, &mut files)?;
73 Ok(WorkspaceSnapshot { files, captured_at: now_rfc3339() })
74}
75
76fn collect(
77 root: &Path,
78 dir: &Path,
79 max_file_bytes: u64,
80 out: &mut BTreeMap<String, FileStat>,
81) -> io::Result<()> {
82 for entry in std::fs::read_dir(dir)? {
83 let entry = entry?;
84 let path = entry.path();
85 let ft = entry.file_type()?;
86 if ft.is_dir() {
87 if let Some(name) = path.file_name().and_then(|n| n.to_str())
88 && SKIP_DIRS.contains(&name)
89 {
90 continue;
91 }
92 collect(root, &path, max_file_bytes, out)?;
93 } else if ft.is_file() {
94 let rel = match path.strip_prefix(root).ok().and_then(|p| p.to_str()) {
95 Some(r) => r.to_string(),
96 None => continue,
97 };
98 match entry.metadata() {
99 Ok(meta) => {
100 let size = meta.len();
101 let mtime_ns = meta
102 .modified()
103 .ok()
104 .and_then(|t| t.duration_since(std::time::UNIX_EPOCH).ok())
105 .map(|d| d.as_nanos() as i64)
106 .unwrap_or(0);
107 let head_hash = if size <= max_file_bytes {
108 hash_head(&path)
109 } else {
110 0
111 };
112 out.insert(rel, FileStat { size, mtime_ns, head_hash });
113 }
114 Err(_) => continue,
115 }
116 }
117 }
118 Ok(())
119}
120
121fn hash_head(path: &Path) -> u64 {
123 const SEED: u64 = 0xcbf29ce484222325;
124 const PRIME: u64 = 0x100000001b3;
125 let mut hash = SEED;
126 let mut buf = [0u8; HASH_SAMPLE_BYTES];
127 if let Ok(mut f) = std::fs::File::open(path) {
128 use std::io::Read;
129 if let Ok(n) = f.read(&mut buf) {
130 for &b in &buf[..n] {
131 hash ^= u64::from(b);
132 hash = hash.wrapping_mul(PRIME);
133 }
134 }
135 }
136 hash
137}
138
139#[must_use]
141pub fn diff(old: &WorkspaceSnapshot, new: &WorkspaceSnapshot) -> SnapshotDelta {
142 let mut delta = SnapshotDelta::default();
143 for (path, new_stat) in &new.files {
144 match old.files.get(path) {
145 None => delta.added.push(path.clone()),
146 Some(old_stat) if old_stat != new_stat => delta.changed.push(path.clone()),
147 Some(_) => {}
148 }
149 }
150 for path in old.files.keys() {
151 if !new.files.contains_key(path) {
152 delta.removed.push(path.clone());
153 }
154 }
155 delta
156}
157
158#[must_use]
160pub fn is_drift(delta: &SnapshotDelta) -> bool {
161 !delta.added.is_empty() || !delta.changed.is_empty() || !delta.removed.is_empty()
162}
163
164pub fn save_json(snapshot: &WorkspaceSnapshot, path: &Path) -> io::Result<()> {
166 if let Some(parent) = path.parent() {
167 std::fs::create_dir_all(parent)?;
168 }
169 let bytes = serde_json::to_vec_pretty(snapshot)
170 .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
171 std::fs::write(path, bytes)
172}
173
174pub fn load_json(path: &Path) -> io::Result<WorkspaceSnapshot> {
176 let bytes = std::fs::read(path)?;
177 serde_json::from_slice(&bytes).map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
178}
179
180fn now_rfc3339() -> String {
181 chrono::Utc::now().to_rfc3339()
182}
183
184#[must_use]
186pub fn snapshot_path(session_dir: &Path) -> PathBuf {
187 session_dir.join("derived").join("workspace_snapshot.json")
188}
189
190#[cfg(test)]
191mod tests {
192 use super::*;
193
194 #[test]
195 fn capture_and_diff_detects_changes() {
196 let tmp = std::env::temp_dir().join(format!("vtcode-snap-{}", std::process::id()));
197 let ws = tmp.join("ws");
198 std::fs::create_dir_all(&ws).unwrap();
199 std::fs::write(ws.join("a.txt"), b"hello").unwrap();
200 std::fs::write(ws.join("b.txt"), b"world").unwrap();
201
202 let snap1 = capture(&ws, 1_000_000).unwrap();
203 assert_eq!(snap1.files.len(), 2);
204 assert!(snap1.files.contains_key("a.txt"));
205 assert!(snap1.files.contains_key("b.txt"));
206
207 std::fs::write(ws.join("a.txt"), b"changed").unwrap();
209 std::fs::write(ws.join("c.txt"), b"new").unwrap();
210 std::fs::remove_file(ws.join("b.txt")).unwrap();
211
212 let snap2 = capture(&ws, 1_000_000).unwrap();
213 let delta = diff(&snap1, &snap2);
214 assert_eq!(delta.added, vec!["c.txt".to_string()]);
215 assert_eq!(delta.changed, vec!["a.txt".to_string()]);
216 assert_eq!(delta.removed, vec!["b.txt".to_string()]);
217 assert!(is_drift(&delta));
218 let _ = std::fs::remove_dir_all(&tmp);
219 }
220
221 #[test]
222 fn json_round_trips() {
223 let tmp = std::env::temp_dir().join(format!("vtcode-snap-json-{}", std::process::id()));
224 let path = tmp.join("snap.json");
225 let snap = WorkspaceSnapshot {
226 files: {
227 let mut m = BTreeMap::new();
228 m.insert("x.rs".to_string(), FileStat { size: 3, mtime_ns: 42, head_hash: 7 });
229 m
230 },
231 captured_at: "2026-01-01T00:00:00Z".to_string(),
232 };
233 save_json(&snap, &path).unwrap();
234 let loaded = load_json(&path).unwrap();
235 assert_eq!(loaded, snap);
236 let _ = std::fs::remove_dir_all(&tmp);
237 }
238}