@anysphere/file-service 0.0.0-a48d29cb → 0.0.0-a4fe55c1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/Cargo.toml CHANGED
@@ -17,9 +17,6 @@ tempfile = "3.8.0"
17
17
  anyhow = "1.0.75"
18
18
  tonic = "0.9.2"
19
19
  prost = "0.11.9"
20
- tracing = "0.1.37"
21
- tracing-subscriber = "0.3.17"
22
- tracing-appender = "0.2.2"
23
20
 
24
21
  [build-dependencies]
25
22
  napi-build = "2.0.1"
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@anysphere/file-service",
3
- "version": "0.0.0-a48d29cb",
3
+ "version": "0.0.0-a4fe55c1",
4
4
  "main": "index.js",
5
5
  "types": "index.d.ts",
6
6
  "napi": {
@@ -35,11 +35,11 @@
35
35
  "version": "napi version"
36
36
  },
37
37
  "optionalDependencies": {
38
- "@anysphere/file-service-win32-x64-msvc": "0.0.0-a48d29cb",
39
- "@anysphere/file-service-darwin-x64": "0.0.0-a48d29cb",
40
- "@anysphere/file-service-linux-x64-gnu": "0.0.0-a48d29cb",
41
- "@anysphere/file-service-darwin-arm64": "0.0.0-a48d29cb",
42
- "@anysphere/file-service-win32-arm64-msvc": "0.0.0-a48d29cb",
43
- "@anysphere/file-service-darwin-universal": "0.0.0-a48d29cb"
38
+ "@anysphere/file-service-win32-x64-msvc": "0.0.0-a4fe55c1",
39
+ "@anysphere/file-service-darwin-x64": "0.0.0-a4fe55c1",
40
+ "@anysphere/file-service-linux-x64-gnu": "0.0.0-a4fe55c1",
41
+ "@anysphere/file-service-darwin-arm64": "0.0.0-a4fe55c1",
42
+ "@anysphere/file-service-win32-arm64-msvc": "0.0.0-a4fe55c1",
43
+ "@anysphere/file-service-darwin-universal": "0.0.0-a4fe55c1"
44
44
  }
45
45
  }
package/src/file_utils.rs CHANGED
@@ -12,8 +12,9 @@ pub fn is_in_bad_dir(file_path: &Path) -> Result<bool, Error> {
12
12
  let item_path = file_path
13
13
  .to_str()
14
14
  .ok_or(anyhow::anyhow!("Failed to convert path to string"))?;
15
- let is_bad_dir =
16
- item_path.contains("node_modules") || item_path.contains(".git");
15
+ let is_bad_dir = (item_path.contains("node_modules")
16
+ || item_path.contains(".git"))
17
+ && !(item_path.ends_with(".git") || item_path.ends_with("node_modules"));
17
18
  Ok(is_bad_dir)
18
19
  }
19
20
 
@@ -37,14 +38,14 @@ pub fn is_good_file(file_path: &Path) -> Result<(), Error> {
37
38
 
38
39
  match file_name {
39
40
  "package-lock.json" | "pnpm-lock.yaml" | "yarn.lock" | "composer.lock"
40
- | "Gemfile.lock" | "bun.lockb" => {
41
+ | "Gemfile.lock" => {
41
42
  return Err(anyhow::anyhow!("File is just a lock file"));
42
43
  }
43
44
  _ => {}
44
45
  }
45
46
 
46
47
  match extension {
47
- "lock" | "bak" | "tmp" | "bin" | "exe" | "dll" | "so" | "lockb" => {
48
+ "lock" | "bak" | "tmp" | "bin" | "exe" | "dll" | "so" => {
48
49
  return Err(anyhow::anyhow!("File is just a lock file"));
49
50
  }
50
51
  _ => {}
package/src/git_utils.rs CHANGED
@@ -1,10 +1,10 @@
1
1
  use std::collections::HashSet;
2
+ use std::path::PathBuf;
2
3
  use std::process::Command;
3
4
 
4
5
  pub fn list_ignored_files(
5
6
  workspace_root_path: &str,
6
- should_return_absolute_paths: bool,
7
- ) -> Result<HashSet<String>, Box<dyn std::error::Error>> {
7
+ ) -> Result<HashSet<PathBuf>, Box<dyn std::error::Error>> {
8
8
  let mut gitignored_files = HashSet::new();
9
9
 
10
10
  let commands = vec![
@@ -15,66 +15,12 @@ pub fn list_ignored_files(
15
15
  "--ignored",
16
16
  "--exclude-standard",
17
17
  ],
18
- // FIXME(sualeh): this is super sketchy and might totally break in like a bazillion ways. i dont like it.
19
- vec![
20
- "sh",
21
- "-c",
22
- "git submodule foreach --quiet 'git ls-files --others --ignored --exclude-standard | sed \"s|^|$path/|\"'",
23
- ],
24
- ];
25
-
26
- for command in commands {
27
- let output = Command::new(&command[0])
28
- .args(&command[1..])
29
- .current_dir(workspace_root_path)
30
- .output()?;
31
-
32
- if output.status.success() {
33
- let files = String::from_utf8(output.stdout)?
34
- .lines()
35
- .filter(|line| !line.is_empty())
36
- .map(|line| {
37
- if should_return_absolute_paths {
38
- let mut path = std::path::PathBuf::from(workspace_root_path);
39
- path.push(line);
40
-
41
- match path.canonicalize() {
42
- Ok(canonical_path) => {
43
- canonical_path.to_string_lossy().into_owned()
44
- }
45
- Err(_) => String::from(line),
46
- }
47
- } else {
48
- String::from(line)
49
- }
50
- })
51
- .collect::<HashSet<_>>();
52
-
53
- gitignored_files.extend(files);
54
- }
55
- }
56
-
57
- Ok(gitignored_files)
58
- }
59
-
60
- pub fn list_ignored_files_with_absolute_paths(
61
- workspace_root_path: &str,
62
- ) -> Result<HashSet<String>, Box<dyn std::error::Error>> {
63
- let mut gitignored_files = HashSet::new();
64
-
65
- let commands = vec![
66
18
  vec![
67
19
  "git",
68
- "ls-files",
69
- "--others",
70
- "--ignored",
71
- "--exclude-standard",
72
- ],
73
- // FIXME(sualeh): this is super sketchy and might totally break in like a bazillion ways. i dont like it.
74
- vec![
75
- "sh",
76
- "-c",
77
- "git submodule foreach --quiet 'git ls-files --others --ignored --exclude-standard | sed \"s|^|$path/|\"'",
20
+ "submodule",
21
+ "foreach",
22
+ "--quiet",
23
+ "git ls-files --others --ignored --exclude-standard | sed 's|^|$path/|'",
78
24
  ],
79
25
  ];
80
26
 
@@ -88,7 +34,7 @@ pub fn list_ignored_files_with_absolute_paths(
88
34
  let files = String::from_utf8(output.stdout)?
89
35
  .lines()
90
36
  .filter(|line| !line.is_empty())
91
- .map(|line| String::from(line))
37
+ .map(|line| PathBuf::from(line))
92
38
  .collect::<HashSet<_>>();
93
39
 
94
40
  gitignored_files.extend(files);
@@ -111,18 +57,6 @@ pub async fn is_git_ignored(
111
57
  Ok(output.status.success())
112
58
  }
113
59
 
114
- pub async fn is_git_directory(
115
- workspace_root_path: &str,
116
- ) -> Result<bool, anyhow::Error> {
117
- let output = tokio::process::Command::new("git")
118
- .args(&["rev-parse", "--is-inside-work-tree"])
119
- .current_dir(workspace_root_path)
120
- .output()
121
- .await?;
122
-
123
- Ok(output.status.success())
124
- }
125
-
126
60
  #[cfg(test)]
127
61
  mod tests {
128
62
  use super::*;
@@ -132,8 +66,7 @@ mod tests {
132
66
  #[test]
133
67
  fn test_no_ignored_files() {
134
68
  let dir = tempfile::tempdir().unwrap();
135
- let gitignored_files =
136
- list_ignored_files(dir.path().to_str().unwrap(), false).unwrap();
69
+ let gitignored_files = list_ignored_files(dir.path().to_str().unwrap()).unwrap();
137
70
  Command::new("git")
138
71
  .args(&["init"])
139
72
  .current_dir(dir.path())
@@ -159,14 +92,13 @@ mod tests {
159
92
  .current_dir(dir.path())
160
93
  .output()
161
94
  .unwrap();
162
- let gitignored_files =
163
- list_ignored_files(dir.path().to_str().unwrap(), false).unwrap();
95
+ let gitignored_files = list_ignored_files(dir.path().to_str().unwrap()).unwrap();
164
96
  println!(
165
97
  "ignored files for test_one_ignored_file: {:?}",
166
98
  gitignored_files
167
99
  );
168
100
  // assert_eq!(gitignored_files.len(), 1);
169
- assert!(gitignored_files.contains(&String::from("ignored.txt")));
101
+ assert!(gitignored_files.contains(&PathBuf::from("ignored.txt")));
170
102
  }
171
103
 
172
104
  #[test]
@@ -189,83 +121,19 @@ mod tests {
189
121
  .current_dir(dir.path())
190
122
  .output()
191
123
  .unwrap();
192
- let gitignored_files =
193
- list_ignored_files(dir.path().to_str().unwrap(), false).unwrap();
124
+ let gitignored_files = list_ignored_files(dir.path().to_str().unwrap()).unwrap();
194
125
  println!(
195
126
  "ignored files for test_multiple_ignored_files: {:?}",
196
127
  gitignored_files
197
128
  );
198
129
  // assert_eq!(gitignored_files.len(), 2);
199
- assert!(gitignored_files.contains(&String::from("ignored1.txt")));
200
- assert!(gitignored_files.contains(&String::from("ignored2.txt")));
201
- }
202
-
203
- #[test]
204
- fn test_git_submodule_ignored_files() {
205
- let dir = tempfile::tempdir().unwrap();
206
- let submodule_path = dir.path().join("submodule");
207
- std::fs::create_dir(&submodule_path).unwrap();
208
-
209
- let o = Command::new("git")
210
- .args(&["init"])
211
- .current_dir(&submodule_path)
212
- .output()
213
- .unwrap();
214
- println!("git init output: {:?}", o);
215
-
216
- let file_path = submodule_path.join("ignored.txt");
217
- let mut file = File::create(&file_path).unwrap();
218
- writeln!(file, "This is an ignored file.").unwrap();
219
-
220
- let file2 = submodule_path.join("ignored2.txt");
221
- let mut file = File::create(&file2).unwrap();
222
- writeln!(file, "This is another ignored file.").unwrap();
223
-
224
- let gitignore_path = submodule_path.join(".gitignore");
225
- let mut gitignore = File::create(&gitignore_path).unwrap();
226
- writeln!(gitignore, "*.txt").unwrap();
227
-
228
- let o = Command::new("git")
229
- .args(&["init"])
230
- .current_dir(dir.path())
231
- .output()
232
- .unwrap();
233
- println!("git init output: {:?}", o);
234
-
235
- // make a commit in the submodule
236
- let o = Command::new("git")
237
- .args(&["add", "."])
238
- .current_dir(&submodule_path)
239
- .output()
240
- .unwrap();
241
- println!("git add output: {:?}", o);
242
- let o = Command::new("git")
243
- .args(&["commit", "-m", "initial commit"])
244
- .current_dir(&submodule_path)
245
- .output()
246
- .unwrap();
247
- println!("git commit output: {:?}", o);
248
-
249
- let o = Command::new("git")
250
- .args(&["submodule", "add", "./submodule"])
251
- .current_dir(dir.path())
252
- .output()
253
- .unwrap();
254
- println!("git submodule add output: {:?}", o);
255
-
256
- let gitignored_files =
257
- list_ignored_files(dir.path().to_str().unwrap(), false).unwrap();
258
- println!(
259
- "ignored files for test_git_submodule_ignored_files: {:?}",
260
- gitignored_files
261
- );
262
- assert!(gitignored_files.contains(&String::from("submodule/ignored.txt")));
263
- assert!(gitignored_files.contains(&String::from("submodule/ignored2.txt")));
130
+ assert!(gitignored_files.contains(&PathBuf::from("ignored1.txt")));
131
+ assert!(gitignored_files.contains(&PathBuf::from("ignored2.txt")));
264
132
  }
265
133
 
266
134
  #[test]
267
135
  fn test_multiple_ignored_files_in_current_dir() {
268
- let gitignored_files = list_ignored_files(".", false).unwrap();
136
+ let gitignored_files = list_ignored_files(".").unwrap();
269
137
  assert!(gitignored_files.len() > 1);
270
138
 
271
139
  // print a sample of the ignored files
@@ -279,6 +147,7 @@ mod tests {
279
147
  }
280
148
  }
281
149
 
150
+
282
151
  #[tokio::test]
283
152
  async fn test_file_not_ignored() {
284
153
  let dir = tempfile::tempdir().unwrap();
@@ -291,10 +160,7 @@ mod tests {
291
160
  .current_dir(dir.path())
292
161
  .output()
293
162
  .unwrap();
294
- let is_ignored =
295
- is_git_ignored(dir.path().to_str().unwrap(), "not_ignored.txt")
296
- .await
297
- .unwrap();
163
+ let is_ignored = is_git_ignored(dir.path().to_str().unwrap(), "not_ignored.txt").await.unwrap();
298
164
  assert_eq!(is_ignored, false);
299
165
  }
300
166
 
@@ -314,10 +180,7 @@ mod tests {
314
180
  .current_dir(dir.path())
315
181
  .output()
316
182
  .unwrap();
317
- let is_ignored =
318
- is_git_ignored(dir.path().to_str().unwrap(), "ignored.txt")
319
- .await
320
- .unwrap();
183
+ let is_ignored = is_git_ignored(dir.path().to_str().unwrap(), "ignored.txt").await.unwrap();
321
184
  assert_eq!(is_ignored, true);
322
185
  }
323
186
 
@@ -337,10 +200,8 @@ mod tests {
337
200
  .current_dir(dir.path())
338
201
  .output()
339
202
  .unwrap();
340
- let is_ignored =
341
- is_git_ignored(dir.path().to_str().unwrap(), "ignored.txt")
342
- .await
343
- .unwrap();
203
+ let is_ignored = is_git_ignored(dir.path().to_str().unwrap(), "ignored.txt").await.unwrap();
344
204
  assert_eq!(is_ignored, true);
345
205
  }
206
+
346
207
  }
package/src/lib.rs CHANGED
@@ -1,5 +1,4 @@
1
1
  #![deny(clippy::all)]
2
- #![deny(unsafe_op_in_unsafe_fn)]
3
2
  pub mod file_utils;
4
3
  pub mod git_utils;
5
4
  pub mod merkle_tree;
@@ -7,9 +6,6 @@ pub mod merkle_tree;
7
6
  use std::vec;
8
7
 
9
8
  use merkle_tree::{LocalConstruction, MerkleTree};
10
- use tracing::{info, Level};
11
- use tracing_appender::rolling::{RollingFileAppender, Rotation};
12
- use tracing_subscriber::fmt;
13
9
 
14
10
  #[macro_use]
15
11
  extern crate napi_derive;
@@ -18,35 +14,15 @@ extern crate napi_derive;
18
14
  pub struct MerkleClient {
19
15
  tree: MerkleTree,
20
16
  root_directory: String,
21
- _guard: tracing_appender::non_blocking::WorkerGuard,
22
- }
23
-
24
- pub fn init_logger() -> tracing_appender::non_blocking::WorkerGuard {
25
- let file_appender =
26
- RollingFileAppender::new(Rotation::NEVER, "./", "rust_log.txt");
27
- let (non_blocking, _guard) = tracing_appender::non_blocking(file_appender);
28
- let subscriber = fmt::Subscriber::builder()
29
- .with_max_level(Level::TRACE)
30
- .with_writer(non_blocking)
31
- .with_ansi(false)
32
- .with_line_number(true)
33
- .finish();
34
-
35
- let _ = tracing::subscriber::set_global_default(subscriber);
36
-
37
- _guard
38
17
  }
39
18
 
40
19
  #[napi]
41
20
  impl MerkleClient {
42
21
  #[napi(constructor)]
43
22
  pub fn new(root_directory: String) -> MerkleClient {
44
- let _guard = init_logger();
45
-
46
23
  MerkleClient {
47
24
  tree: MerkleTree::empty_tree(),
48
25
  root_directory,
49
- _guard,
50
26
  }
51
27
  }
52
28
 
@@ -55,10 +31,7 @@ impl MerkleClient {
55
31
  // 1. compute the merkle tree
56
32
  // 2. update the backend
57
33
  // 3. sync with the remote
58
- info!("Merkle tree compute started!");
59
- unsafe {
60
- self.compute_merkle_tree().await?;
61
- }
34
+ self.compute_merkle_tree().await?;
62
35
 
63
36
  Ok(())
64
37
  }
@@ -74,17 +47,6 @@ impl MerkleClient {
74
47
  let t =
75
48
  MerkleTree::construct_merkle_tree(self.root_directory.clone()).await;
76
49
 
77
- let files = self.tree.get_all_files().await;
78
-
79
- match files {
80
- Ok(files) => {
81
- info!("files: {:?}", files);
82
- }
83
- Err(e) => {
84
- info!("Error in get_all_files: {:?}", e);
85
- }
86
- }
87
-
88
50
  match t {
89
51
  Ok(tree) => {
90
52
  self.tree = tree;
@@ -114,9 +76,7 @@ impl MerkleClient {
114
76
  ) -> Result<String, napi::Error> {
115
77
  let absolute_path =
116
78
  std::path::Path::new(&self.root_directory).join(relative_path);
117
- let canonical_path = absolute_path.canonicalize().unwrap();
118
-
119
- let hash = self.tree.get_subtree_hash(canonical_path).await;
79
+ let hash = self.tree.get_subtree_hash(absolute_path).await;
120
80
 
121
81
  match hash {
122
82
  Ok(hash) => Ok(hash),
@@ -144,10 +104,8 @@ impl MerkleClient {
144
104
  &self,
145
105
  relative_path: String,
146
106
  ) -> Result<i32, napi::Error> {
147
- let absolute_path = std::path::Path::new(&self.root_directory)
148
- .join(relative_path)
149
- .canonicalize()?;
150
-
107
+ let absolute_path =
108
+ std::path::Path::new(&self.root_directory).join(relative_path);
151
109
  let num = self
152
110
  .tree
153
111
  .get_num_embeddable_files_in_subtree(absolute_path)
@@ -209,6 +167,7 @@ impl MerkleClient {
209
167
 
210
168
  let ret = vec![file];
211
169
  let ret = ret.into_iter().chain(path.into_iter()).collect::<Vec<_>>();
170
+
212
171
  Ok(ret)
213
172
  }
214
173
  Err(e) => Err(napi::Error::new(
@@ -224,16 +183,17 @@ impl MerkleClient {
224
183
  &self,
225
184
  absolute_file_path: String,
226
185
  ) -> Result<Vec<String>, napi::Error> {
227
- let absolute_path_str = absolute_file_path.as_str();
228
- let spline = self.tree.get_spline(absolute_path_str).await;
186
+ // let spline = self.tree.get_spline(absolute_file_path).await;
229
187
 
230
- match spline {
231
- Ok(spline) => Ok(spline),
232
- Err(e) => Err(napi::Error::new(
233
- napi::Status::Unknown,
234
- format!("Error in get_spline: {:?}", e),
235
- )),
236
- }
188
+ return Ok(vec![]);
189
+
190
+ // match spline {
191
+ // Ok(spline) => Ok(spline),
192
+ // Err(e) => Err(napi::Error::new(
193
+ // napi::Status::Unknown,
194
+ // format!("Error in get_spline: {:?}", e),
195
+ // )),
196
+ // }
237
197
  }
238
198
 
239
199
  #[napi]
@@ -1,11 +1,11 @@
1
- use crate::git_utils;
2
1
  use crate::merkle_tree::{
3
2
  File, MerkleNode, MerkleNodePtr, NodeType, PinnedFuture,
4
3
  };
5
4
 
6
5
  use super::{LocalConstruction, MerkleTree};
7
- use std::collections::{BTreeMap, HashSet};
8
- use std::path::{Path, PathBuf};
6
+ use std::collections::BTreeMap;
7
+ use std::path::PathBuf;
8
+ use std::{collections::HashMap, path::Path, sync::Arc};
9
9
  use tonic::async_trait;
10
10
 
11
11
  #[async_trait]
@@ -29,36 +29,20 @@ impl LocalConstruction for MerkleTree {
29
29
  /// 3. construct merkle tree
30
30
  /// 4. return merkle tree
31
31
  async fn construct_merkle_tree(
32
- absolute_path_to_root_directory: String,
32
+ root_directory: String,
33
33
  ) -> Result<MerkleTree, anyhow::Error> {
34
- let path = PathBuf::from(absolute_path_to_root_directory.clone());
34
+ let path = PathBuf::from(root_directory.clone());
35
35
  if !path.exists() {
36
36
  // FIXME: we should report this via a good logger.
37
37
  panic!("Root directory does not exist!");
38
38
  }
39
39
 
40
- // 1. get all the gitignored files
41
- let git_ignored_files = match git_utils::list_ignored_files(
42
- absolute_path_to_root_directory.as_str(),
43
- true,
44
- ) {
45
- Ok(git_ignored) => git_ignored,
46
- Err(_e) => HashSet::new(),
47
- };
48
-
49
- let root_node = MerkleNode::new(
50
- path,
51
- None,
52
- &git_ignored_files,
53
- absolute_path_to_root_directory.as_str(),
54
- )
55
- .await;
40
+ let root_node = MerkleNode::new(path, None).await;
56
41
  let mut mt = MerkleTree {
57
42
  root: root_node,
58
43
  files: BTreeMap::new(),
59
- root_path: absolute_path_to_root_directory,
44
+ root_path: root_directory,
60
45
  cursor: None,
61
- git_ignored_files,
62
46
  };
63
47
 
64
48
  // we now iterate over all the nodes and add them to the hashmap
@@ -71,7 +55,6 @@ impl LocalConstruction for MerkleTree {
71
55
  let node_reader = node.read().await;
72
56
  match &node_reader.node_type {
73
57
  NodeType::Branch(n) => {
74
- tracing::info!("Branch: {:?}", n.0);
75
58
  let children = &n.1;
76
59
  files.insert(n.0.clone(), File { node: node.clone() });
77
60
  for child in children {
@@ -80,7 +63,6 @@ impl LocalConstruction for MerkleTree {
80
63
  }
81
64
  NodeType::File(file_name) => {
82
65
  let f = File { node: node.clone() };
83
- tracing::info!("File: {:?}", file_name);
84
66
  files.insert(file_name.clone(), f);
85
67
  }
86
68
  NodeType::ErrorNode(_) => {
@@ -92,9 +74,6 @@ impl LocalConstruction for MerkleTree {
92
74
 
93
75
  add_nodes_to_hashmap(&mt.root, &mut mt.files).await;
94
76
 
95
- tracing::info!("Merkle tree compute finished!");
96
- // tracing::info!("Merkle tree: {:?}", mt.files);
97
-
98
77
  Ok(mt)
99
78
  }
100
79
 
@@ -1,13 +1,10 @@
1
- use crate::git_utils;
2
-
3
1
  use super::file_utils;
4
2
  use sha2::Digest;
5
- use std::collections::{BTreeMap, HashSet};
3
+ use std::collections::BTreeMap;
6
4
  use std::path::PathBuf;
7
5
  use std::{fs, path::Path, sync::Arc};
8
6
  use tokio::sync::RwLock;
9
7
  use tonic::async_trait;
10
- use tracing::info;
11
8
  pub mod local_construction;
12
9
  pub mod test;
13
10
 
@@ -17,8 +14,7 @@ pub struct MerkleTree {
17
14
  root_path: String,
18
15
  root: MerkleNodePtr,
19
16
  files: BTreeMap<String, File>,
20
- cursor: Option<usize>,
21
- git_ignored_files: HashSet<String>,
17
+ cursor: Option<MerkleNodePtr>,
22
18
  }
23
19
 
24
20
  #[derive(Debug)]
@@ -95,7 +91,6 @@ impl MerkleTree {
95
91
  files: BTreeMap::new(),
96
92
  root_path: "".to_string(),
97
93
  cursor: None,
98
- git_ignored_files: HashSet::new(),
99
94
  }
100
95
  }
101
96
 
@@ -115,8 +110,7 @@ impl MerkleTree {
115
110
  let node = match self.files.get(&abs_string) {
116
111
  Some(file) => file.node.clone(),
117
112
  None => {
118
- let all_files: Vec<String> = self.files.keys().cloned().collect();
119
- return Err(anyhow::anyhow!("Could not find file in tree! Looking for: {}. All files: {:?}", abs_string, all_files));
113
+ return Err(anyhow::anyhow!("Could not find file in tree!"));
120
114
  }
121
115
  };
122
116
 
@@ -162,7 +156,7 @@ impl MerkleTree {
162
156
  }
163
157
  };
164
158
 
165
- // TODO(sualeh): worth keeping this list sorted. its now a btree
159
+ // TODO(sualeh): worth keeping this list sorted.
166
160
 
167
161
  for (_, file) in &self.files {
168
162
  let file_reader = file.node.read().await;
@@ -240,44 +234,83 @@ impl MerkleTree {
240
234
  pub async fn get_next_file_to_embed(
241
235
  &mut self,
242
236
  ) -> Result<(String, Vec<String>), anyhow::Error> {
243
- // if the cursor is none, set it to 0
244
- let cursor = match self.cursor {
245
- Some(cursor) => cursor,
246
- None => {
247
- self.cursor = Some(0);
248
- 0
237
+ // the plan is to do an in-order traversal of the tree.
238
+
239
+ // first the edge case to deal with:
240
+ // cursor == None
241
+ if self.cursor.is_none() {
242
+ // If the root is a file, return its name.
243
+ if let NodeType::File(file_path) = &self.root.read().await.node_type {
244
+ return Ok((file_path.clone(), vec![]));
249
245
  }
250
- };
251
246
 
252
- // get the thing at the cursor. while we dont find a file, we keep incrementing the cursor.
253
- let mut cursor = cursor;
254
- loop {
255
- // O(log n)
256
- let file = match self.files.values().nth(cursor) {
257
- Some(file) => file,
258
- None => {
259
- return Err(anyhow::anyhow!("Could not find file to embed!"));
260
- }
261
- };
247
+ // if the path is not empty, we can iterate till we find the first child.
248
+ let mut potential_first_child = self.root.clone();
249
+ let mut is_branch = true;
250
+ let mut path = Vec::new();
251
+
252
+ while is_branch {
253
+ let node = {
254
+ let potential_first_child_reader = potential_first_child.read().await;
255
+ match &potential_first_child_reader.node_type {
256
+ NodeType::Branch(branch) => branch.clone(),
257
+ NodeType::File(_) => {
258
+ return Err(anyhow::anyhow!(
259
+ "get_next_file_to_embed: This should not happen! the branch happened to be file."
260
+ ));
261
+ }
262
+ NodeType::ErrorNode(_) => {
263
+ return Err(anyhow::anyhow!("Cursor is an error node!"));
264
+ }
265
+ }
266
+ };
262
267
 
263
- let file_reader = file.node.read().await;
264
- match &file_reader.node_type {
265
- NodeType::File(f) => {
266
- // update the cursor.
267
- self.cursor = Some(cursor + 1);
268
- let spline = self.get_spline(f).await?;
269
- return Ok((f.clone(), spline));
270
- }
271
- NodeType::Branch(_) => {
272
- cursor += 1;
273
- continue;
274
- }
275
- NodeType::ErrorNode(_) => {
276
- cursor += 1;
277
- continue;
268
+ let current_node_name = &node.0;
269
+ let child_list = &node.1;
270
+
271
+ if let Some(c) = child_list.first() {
272
+ let c_reader = c.read().await;
273
+
274
+ match &c_reader.node_type {
275
+ NodeType::File(file_path) => {
276
+ // must set the cursor!
277
+ self.cursor = Some(c.clone());
278
+
279
+ return Ok((file_path.clone(), path));
280
+ }
281
+ NodeType::Branch(_) => {
282
+ potential_first_child = c.clone();
283
+ is_branch = true;
284
+
285
+ // add the path to the current node.
286
+ path.push(current_node_name.clone());
287
+ }
288
+ NodeType::ErrorNode(_) => {
289
+ return Err(anyhow::anyhow!("Cursor is an error node!"));
290
+ }
291
+ }
292
+ } else {
293
+ // If the root has no children, return an error.
294
+ return Err(anyhow::anyhow!("Root has no children!"));
278
295
  }
279
296
  }
280
297
  }
298
+
299
+ // THE DEFAULT CASE:
300
+ // we already have a cursor at a file.
301
+
302
+ // UNWRAP checked and fine. see the none case above.
303
+ let cursor_name = self.cursor.as_ref().unwrap();
304
+ let cursor_reader = cursor_name.read().await;
305
+
306
+ // invariant: you must be a file!!
307
+
308
+ // everytime we get to a child list, we will add all the children to a fifo, and then pull from it as long as we need it.
309
+
310
+ // algorithm:
311
+ // 1.
312
+
313
+ Err(anyhow::anyhow!("Could not find file to embed!"))
281
314
  }
282
315
 
283
316
  pub async fn get_all_dir_files_to_embed(
@@ -307,54 +340,6 @@ impl MerkleTree {
307
340
  Ok(files)
308
341
  }
309
342
 
310
- // TODO(sualeh): i need tests for this!!
311
- pub async fn get_spline(
312
- &self,
313
- absolute_path: &str,
314
- ) -> Result<Vec<String>, anyhow::Error> {
315
- info!("get_spline called with absolute_path: {}", absolute_path);
316
- let mut files = Vec::new();
317
-
318
- let current_node = match self.files.get(absolute_path) {
319
- Some(node) => {
320
- info!("Found node for absolute_path: {}", absolute_path);
321
- node.node.clone()
322
- }
323
- None => {
324
- info!("File not found for absolute_path: {}", absolute_path);
325
- return Err(anyhow::anyhow!("File not found: {}", absolute_path));
326
- }
327
- };
328
-
329
- let mut stack = Vec::new();
330
- stack.push(current_node);
331
-
332
- while let Some(node) = stack.pop() {
333
- let parent = node.read().await.parent.clone();
334
- if let Some(parent) = parent {
335
- info!("Adding parent hash to files vector");
336
- {
337
- let parent_node = parent.read().await;
338
- match &parent_node.node_type {
339
- NodeType::File(file_name) => {
340
- files.push(file_name.clone());
341
- }
342
- NodeType::Branch((branch_name, _)) => {
343
- files.push(branch_name.clone());
344
- }
345
- _ => {
346
- continue;
347
- }
348
- }
349
- }
350
-
351
- stack.push(parent);
352
- }
353
- }
354
- info!("Returning files vector with {} elements", files.len());
355
- Ok(files)
356
- }
357
-
358
343
  /// creates a new node and attaches it to the current tree.
359
344
  /// SPEC:
360
345
  /// - you are allowed to create a file with a node such that the
@@ -390,18 +375,12 @@ impl MerkleTree {
390
375
  // 1. the path is empty. this means that the ancestor is the root.
391
376
  // 2. the path is non-empty. that means there exist a non-empty element btwn till the root.
392
377
 
393
- let absolute_root_path = self.root_path.clone();
394
378
  let new_node = match path.len() {
395
379
  0 => {
396
380
  // this means that the ancestor is the root.
397
381
  // we need to create a new node and attach it to the ancestor.
398
- let new_node = MerkleNode::new(
399
- file_path.clone(),
400
- Some(ancestor.clone()),
401
- &self.git_ignored_files,
402
- &absolute_root_path.as_str(),
403
- )
404
- .await;
382
+ let new_node =
383
+ MerkleNode::new(file_path.clone(), Some(ancestor.clone())).await;
405
384
  ancestor.write().await.attach_child(new_node.clone()).await;
406
385
  new_node
407
386
  }
@@ -412,13 +391,9 @@ impl MerkleTree {
412
391
  // UNSURE: not sure this is the correct thing to do but it is the fastest.
413
392
  // get the last thing that is not in the tree.
414
393
  let first_child_path = path.last().unwrap();
415
- let first_child = MerkleNode::new(
416
- first_child_path.clone(),
417
- Some(ancestor.clone()),
418
- &self.git_ignored_files,
419
- &absolute_root_path.as_str(),
420
- )
421
- .await;
394
+ let first_child =
395
+ MerkleNode::new(first_child_path.clone(), Some(ancestor.clone()))
396
+ .await;
422
397
 
423
398
  // TODO(sualeh): we should do an assertion check that the entire vec is contained here.
424
399
 
@@ -695,62 +670,18 @@ use std::future::Future;
695
670
  use std::pin::Pin;
696
671
 
697
672
  type PinnedFuture<'a, T> = Pin<Box<dyn Future<Output = T> + Send + 'a>>;
698
- type IgnoredFiles = HashSet<String>;
699
673
 
700
674
  impl MerkleNode {
701
675
  /// please be careful using this.
702
676
  async fn __new_unchecked(
703
677
  file_or_directory: String,
704
678
  parent: ParentPtr,
705
- ignored_files: &IgnoredFiles,
706
- absolute_root_path: &str,
707
679
  ) -> MerkleNodePtr {
708
- // check if the root is a git directory.
709
- let is_git_repo =
710
- match git_utils::is_git_directory(absolute_root_path).await {
711
- Ok(is_git_repo) => is_git_repo,
712
- Err(e) => false,
713
- };
714
- let bypass_git = !is_git_repo;
715
-
716
- MerkleNode::construct_node(
717
- Path::new(&file_or_directory),
718
- parent,
719
- ignored_files,
720
- absolute_root_path,
721
- bypass_git,
722
- )
723
- .await
680
+ MerkleNode::construct_node(Path::new(&file_or_directory), parent).await
724
681
  }
725
682
 
726
- async fn new(
727
- absolute_file_or_directory: PathBuf,
728
- parent: ParentPtr,
729
- ignored_files: &IgnoredFiles,
730
- absolute_root_path: &str,
731
- ) -> MerkleNodePtr {
732
- // check if the root is a git directory.
733
- let is_git_repo =
734
- match git_utils::is_git_directory(absolute_root_path).await {
735
- Ok(is_git_repo) => is_git_repo,
736
- Err(_e) => false,
737
- };
738
- let bypass_git = !is_git_repo;
739
-
740
- info!(
741
- "constructing node for absolute_file_or_directory: {:?}",
742
- absolute_file_or_directory
743
- );
744
- info!("bypass_git: {}, is_git_repo: {}", bypass_git, is_git_repo);
745
-
746
- MerkleNode::construct_node(
747
- Path::new(&absolute_file_or_directory),
748
- parent,
749
- ignored_files,
750
- absolute_root_path,
751
- bypass_git,
752
- )
753
- .await
683
+ async fn new(file_or_directory: PathBuf, parent: ParentPtr) -> MerkleNodePtr {
684
+ MerkleNode::construct_node(Path::new(&file_or_directory), parent).await
754
685
  }
755
686
 
756
687
  /// NOT added to the tree by default.
@@ -761,68 +692,38 @@ impl MerkleNode {
761
692
  // let file_hash = self.files.get_mut(&file_path).unwrap();
762
693
 
763
694
  fn construct_node<'a>(
764
- absolute_file_or_directory: &'a Path,
695
+ file_or_directory: &'a Path,
765
696
  parent: ParentPtr,
766
- ignored_files: &'a IgnoredFiles,
767
- absolute_root_path: &'a str,
768
- bypass_git: bool,
769
697
  ) -> PinnedFuture<'a, MerkleNodePtr> {
770
698
  Box::pin(async move {
771
699
  // check if it is a file
772
- let path_str = absolute_file_or_directory.to_str().unwrap().to_string();
773
- if absolute_file_or_directory.is_file() {
700
+ let path_str = file_or_directory.to_str().unwrap().to_string();
701
+ if file_or_directory.is_file() {
774
702
  return Arc::new(RwLock::new(
775
703
  MerkleNode::construct_file_node_or_error_node(
776
- absolute_file_or_directory,
704
+ file_or_directory,
777
705
  parent,
778
- ignored_files,
779
706
  )
780
707
  .await,
781
708
  ));
782
709
  }
783
710
 
784
711
  // check if the directory fails the bad dir test.
785
- let is_bad_dir = file_utils::is_in_bad_dir(absolute_file_or_directory);
712
+ let is_bad_dir = file_utils::is_in_bad_dir(file_or_directory);
786
713
  if is_bad_dir.is_err() || is_bad_dir.unwrap_or(false) {
787
714
  // println!("skipping directory: {}", path_str);
788
715
  return Arc::new(RwLock::new(MerkleNode::empty_node(
789
- Some(absolute_file_or_directory),
716
+ Some(file_or_directory),
790
717
  Some("Directory is in bad dir!".to_string()),
791
718
  )));
792
719
  }
793
720
 
794
- // check if the directory is git ignored
795
- let is_git_ignored =
796
- match git_utils::is_git_ignored(absolute_root_path, path_str.as_str())
797
- .await
798
- {
799
- Ok(is_git_ignored) => is_git_ignored,
800
- Err(e) => {
801
- return Arc::new(RwLock::new(MerkleNode::empty_node(
802
- Some(absolute_file_or_directory),
803
- Some(e.to_string()),
804
- )));
805
- }
806
- };
807
-
808
- if is_git_ignored && !bypass_git {
809
- // println!("skipping directory: {}", path_str);
810
- tracing::info!(
811
- "skipping directory because its git ignored: {}",
812
- path_str
813
- );
814
- return Arc::new(RwLock::new(MerkleNode::empty_node(
815
- Some(absolute_file_or_directory),
816
- Some("Directory is git ignored!".to_string()),
817
- )));
818
- }
819
-
820
- let entries = fs::read_dir(absolute_file_or_directory);
721
+ let entries = fs::read_dir(file_or_directory);
821
722
  match entries {
822
723
  Ok(_) => (),
823
724
  Err(e) => {
824
725
  return Arc::new(RwLock::new(MerkleNode::empty_node(
825
- Some(absolute_file_or_directory),
726
+ Some(file_or_directory),
826
727
  Some(e.to_string()),
827
728
  )));
828
729
  }
@@ -842,19 +743,13 @@ impl MerkleNode {
842
743
  match entry {
843
744
  Ok(entry) => {
844
745
  children.push(
845
- MerkleNode::construct_node(
846
- &entry.path(),
847
- Some(node.clone()),
848
- ignored_files,
849
- absolute_root_path,
850
- bypass_git,
851
- )
852
- .await,
746
+ MerkleNode::construct_node(&entry.path(), Some(node.clone()))
747
+ .await,
853
748
  );
854
749
  }
855
750
  Err(e) => {
856
751
  children.push(Arc::new(RwLock::new(MerkleNode::empty_node(
857
- Some(absolute_file_or_directory),
752
+ Some(file_or_directory),
858
753
  Some(e.to_string()),
859
754
  ))));
860
755
  }
@@ -874,33 +769,23 @@ impl MerkleNode {
874
769
  }
875
770
 
876
771
  async fn construct_file_node(
877
- absolute_file_path: &Path,
772
+ file_path: &Path,
878
773
  parent: ParentPtr,
879
- ignored_files: &IgnoredFiles,
880
774
  ) -> Result<MerkleNode, String> {
881
- let file_str = absolute_file_path
775
+ let file_str = file_path
882
776
  .to_str()
883
777
  .ok_or("Could not convert file path to string!")?
884
778
  .to_string();
885
779
  // first see if it passes the
886
- match file_utils::is_good_file(absolute_file_path) {
780
+ match file_utils::is_good_file(file_path) {
887
781
  Ok(_) => {}
888
782
  Err(e) => {
889
783
  return Err(format!("File failed runtime checks! {}", e.to_string()));
890
784
  }
891
785
  }
892
786
 
893
- // check if the file is in the git ignore buffer.
894
- // this is a bug right because we are not checking absoluteness here.
895
- match ignored_files.contains(&file_str) {
896
- true => {
897
- return Err(format!("File is in git ignore buffer!"));
898
- }
899
- false => {}
900
- }
901
-
902
787
  // read the file_content to a buffer
903
- let file_content = match tokio::fs::read(absolute_file_path).await {
788
+ let file_content = match tokio::fs::read(file_path).await {
904
789
  Ok(content) => content,
905
790
  Err(e) => {
906
791
  return Err(format!("Could not read file! {}", e.to_string()));
@@ -908,11 +793,7 @@ impl MerkleNode {
908
793
  };
909
794
 
910
795
  // check if the file passes runtime checks.
911
- match file_utils::is_good_file_runtime_check(
912
- absolute_file_path,
913
- &file_content,
914
- )
915
- .await
796
+ match file_utils::is_good_file_runtime_check(file_path, &file_content).await
916
797
  {
917
798
  Ok(_) => {}
918
799
  Err(e) => {
@@ -943,22 +824,15 @@ impl MerkleNode {
943
824
  }
944
825
 
945
826
  async fn construct_file_node_or_error_node(
946
- absolute_file_path: &Path,
827
+ file_path: &Path,
947
828
  parent: ParentPtr,
948
- ignored_files: &IgnoredFiles,
949
829
  ) -> MerkleNode {
950
- let node = match MerkleNode::construct_file_node(
951
- absolute_file_path,
952
- parent,
953
- ignored_files,
954
- )
955
- .await
956
- {
830
+ let node = match MerkleNode::construct_file_node(file_path, parent).await {
957
831
  Ok(node) => node,
958
832
  Err(e) => {
959
833
  // println!("constructing error node. error: {}", e);
960
834
  // println!("file_path: {:?}", file_path);
961
- MerkleNode::empty_node(Some(absolute_file_path), Some(e))
835
+ MerkleNode::empty_node(Some(file_path), Some(e))
962
836
  }
963
837
  };
964
838
 
package/src/test.rs DELETED
@@ -1,5 +0,0 @@
1
- #[cfg(test)]
2
- mod tests {
3
- use super::super::*;
4
- use std::path::PathBuf;
5
- }