@anysphere/file-service 0.0.0-bff0125b → 0.0.0-c8c4f8fc

package/Cargo.toml

@@ -20,6 +20,8 @@ prost = "0.11.9"
 tracing = "0.1.37"
 tracing-subscriber = "0.3.17"
 tracing-appender = "0.2.2"
+binaryornot = "1.0.0"
+dunce = "1.0.1"
 
 [build-dependencies]
 napi-build = "2.0.1"

package/index.d.ts

@@ -4,7 +4,7 @@
 /* auto-generated by NAPI-RS */
 
 export class MerkleClient {
-  constructor(rootDirectory: string)
+  constructor(absoluteRootDirectory: string)
   init(): Promise<void>
   computeMerkleTree(): Promise<void>
   updateFile(filePath: string): Promise<void>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@anysphere/file-service",
-  "version": "0.0.0-bff0125b",
+  "version": "0.0.0-c8c4f8fc",
   "main": "index.js",
   "types": "index.d.ts",
   "napi": {
@@ -9,7 +9,8 @@
       "additional": [
         "aarch64-apple-darwin",
         "aarch64-pc-windows-msvc",
-        "universal-apple-darwin"
+        "universal-apple-darwin",
+        "aarch64-unknown-linux-gnu"
       ]
     }
   },
@@ -35,11 +36,12 @@
     "version": "napi version"
   },
   "optionalDependencies": {
-    "@anysphere/file-service-win32-x64-msvc": "0.0.0-bff0125b",
-    "@anysphere/file-service-darwin-x64": "0.0.0-bff0125b",
-    "@anysphere/file-service-linux-x64-gnu": "0.0.0-bff0125b",
-    "@anysphere/file-service-darwin-arm64": "0.0.0-bff0125b",
-    "@anysphere/file-service-win32-arm64-msvc": "0.0.0-bff0125b",
-    "@anysphere/file-service-darwin-universal": "0.0.0-bff0125b"
+    "@anysphere/file-service-win32-x64-msvc": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-darwin-x64": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-linux-x64-gnu": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-darwin-arm64": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-win32-arm64-msvc": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-darwin-universal": "0.0.0-c8c4f8fc",
+    "@anysphere/file-service-linux-arm64-gnu": "0.0.0-c8c4f8fc"
   }
 }

package/src/file_utils.rs

@@ -62,7 +62,7 @@ pub fn is_good_file(file_path: &Path) -> Result<(), Error> {
     Some(extension) => match extension.to_str() {
       Some(ext_str) => {
         if bad_extensions.contains(&ext_str) {
-          return Err(anyhow::anyhow!("File is not a valid UTF-8 string"));
+          return Err(anyhow::anyhow!("Binary file excluded from indexing."));
         }
       }
       None => {
@@ -88,10 +88,12 @@ pub fn is_good_file(file_path: &Path) -> Result<(), Error> {
   Ok(())
 }
 
+// use binaryornot::is_binary;
+// use anyhow::Context;
 // implement the buffer above:
 pub async fn is_good_file_runtime_check(
   file_path: &Path,
-  buffer: &[u8],
+  _buffer: &[u8],
 ) -> Result<(), Error> {
   match get_file_size(file_path).await {
     Ok(size) if size > 2 * 1024 * 1024 => {
@@ -101,13 +103,10 @@ pub async fn is_good_file_runtime_check(
     _ => {}
   }
 
-  for &byte in buffer.iter().take(2048) {
-    if byte.is_ascii() {
-      continue;
-    } else {
-      return Err(anyhow::anyhow!("File is not a valid UTF-8 string"));
-    }
-  }
+	// if is_binary(file_path).context("Failed to check if file is binary")? {
+	// 	return Err(anyhow::anyhow!("File is binary"));
+	// }
+
   Ok(())
 }
 

package/src/lib.rs

@@ -17,7 +17,7 @@ extern crate napi_derive;
 #[napi]
 pub struct MerkleClient {
   tree: MerkleTree,
-  root_directory: String,
+  absolute_root_directory: String,
   _guard: tracing_appender::non_blocking::WorkerGuard,
 }
 
@@ -40,12 +40,22 @@ pub fn init_logger() -> tracing_appender::non_blocking::WorkerGuard {
 #[napi]
 impl MerkleClient {
   #[napi(constructor)]
-  pub fn new(root_directory: String) -> MerkleClient {
+  pub fn new(absolute_root_directory: String) -> MerkleClient {
     let _guard = init_logger();
 
+    let canonical_root_directory = std::path::Path::new(&absolute_root_directory);
+    // use dunce::canonicalize;
+    let canonical_root_directory = match dunce::canonicalize(&canonical_root_directory) {
+      Ok(path) => path.to_str().unwrap_or(&absolute_root_directory).to_string(),
+      Err(e) => {
+        info!("Error in canonicalizing path: path: {:?}, error {:?}", canonical_root_directory, e);
+        absolute_root_directory
+      }
+    };
+
     MerkleClient {
       tree: MerkleTree::empty_tree(),
-      root_directory,
+      absolute_root_directory: canonical_root_directory,
       _guard,
     }
   }
@@ -55,6 +65,7 @@ impl MerkleClient {
     // 1. compute the merkle tree
     // 2. update the backend
     // 3. sync with the remote
+    info!("Merkle tree compute started!");
     unsafe {
       self.compute_merkle_tree().await?;
     }
@@ -71,18 +82,7 @@ impl MerkleClient {
     &mut self,
   ) -> Result<(), napi::Error> {
     let t =
-      MerkleTree::construct_merkle_tree(self.root_directory.clone()).await;
-
-    let files = self.tree.get_all_files().await;
-
-    match files {
-      Ok(files) => {
-        info!("files: {:?}", files);
-      }
-      Err(e) => {
-        info!("Error in get_all_files: {:?}", e);
-      }
-    }
+      MerkleTree::construct_merkle_tree(self.absolute_root_directory.clone()).await;
 
     match t {
       Ok(tree) => {
@@ -111,19 +111,36 @@ impl MerkleClient {
     &self,
     relative_path: String,
   ) -> Result<String, napi::Error> {
-    info!("relative_path: {:?}", relative_path);
     let absolute_path =
-      std::path::Path::new(&self.root_directory).join(relative_path);
-    let canonical_path = absolute_path.canonicalize().unwrap();
+      std::path::Path::new(&self.absolute_root_directory).join(&relative_path);
+
+    let canonical_path = match dunce::canonicalize(&absolute_path) {
+      Ok(path) => path,
+      Err(e) => {
+        return Err(napi::Error::new(
+          napi::Status::Unknown,
+          format!("Error in canonicalizing path: {:?}", e),
+        ))
+      }
+    };
+
+    let cononical_str = match canonical_path.to_str() {
+      Some(s) => s,
+      None => {
+        return Err(napi::Error::new(
+          napi::Status::Unknown,
+          format!("Error in converting canonical path to string"),
+        ))
+      }
+    };
 
-    info!("canonical_path: {:?}", canonical_path);
-    let hash = self.tree.get_subtree_hash(canonical_path).await;
+    let hash = self.tree.get_subtree_hash(cononical_str).await;
 
     match hash {
       Ok(hash) => Ok(hash),
       Err(e) => Err(napi::Error::new(
         napi::Status::Unknown,
-        format!("Error in get_subtree_hash: {:?}", e),
+        format!("Error in get_subtree_hash. \nRelative path: {:?}, \nAbsolute path: {:?}, \nCanonical path: {:?}, \nRoot directory: {:?}\nError: {:?}", &relative_path, absolute_path, canonical_path, self.absolute_root_directory, e),
       )),
     }
   }
@@ -145,7 +162,7 @@ impl MerkleClient {
     &self,
     relative_path: String,
   ) -> Result<i32, napi::Error> {
-    let absolute_path = std::path::Path::new(&self.root_directory)
+    let absolute_path = std::path::Path::new(&self.absolute_root_directory)
       .join(relative_path)
       .canonicalize()?;
 
@@ -209,11 +226,7 @@ impl MerkleClient {
         // TODO(sualeh): we should assert that the path is ascending up to the path.
 
         let ret = vec![file];
-        info!("file: {:?}", ret);
-
         let ret = ret.into_iter().chain(path.into_iter()).collect::<Vec<_>>();
-        info!("ret to js: {:?}", ret);
-
         Ok(ret)
       }
       Err(e) => Err(napi::Error::new(
@@ -259,6 +272,6 @@ impl MerkleClient {
 
   #[napi]
   pub fn update_root_directory(&mut self, root_directory: String) {
-    self.root_directory = root_directory;
+    self.absolute_root_directory = root_directory;
   }
 }

package/src/merkle_tree/local_construction.rs

@@ -46,8 +46,6 @@ impl LocalConstruction for MerkleTree {
       Err(_e) => HashSet::new(),
     };
 
-    tracing::info!("git_ignored_files: {:?}", git_ignored_files);
-
     let root_node = MerkleNode::new(
       path,
       None,
@@ -73,6 +71,7 @@ impl LocalConstruction for MerkleTree {
         let node_reader = node.read().await;
         match &node_reader.node_type {
           NodeType::Branch(n) => {
+            tracing::info!("Branch: {:?}", n.0);
             let children = &n.1;
             files.insert(n.0.clone(), File { node: node.clone() });
             for child in children {
@@ -81,7 +80,12 @@ impl LocalConstruction for MerkleTree {
           }
           NodeType::File(file_name) => {
             let f = File { node: node.clone() };
-            files.insert(file_name.clone(), f);
+            let canonical_file_name = match dunce::canonicalize(file_name) {
+              Ok(path) => path.to_str().unwrap_or(file_name).to_string(),
+              Err(_) => file_name.clone(),
+            };
+
+            files.insert(canonical_file_name, f);
           }
           NodeType::ErrorNode(_) => {
             // do nothing
@@ -92,6 +96,9 @@ impl LocalConstruction for MerkleTree {
 
     add_nodes_to_hashmap(&mt.root, &mut mt.files).await;
 
+    tracing::info!("Merkle tree compute finished!");
+    tracing::info!("Merkle tree: {}", mt);
+
     Ok(mt)
   }
 

package/src/merkle_tree/mod.rs

@@ -101,21 +101,14 @@ impl MerkleTree {
 
   pub async fn get_subtree_hash(
     &self,
-    absolute_path: PathBuf,
+    absolute_path: &str,
   ) -> Result<String, anyhow::Error> {
-    let abs_string = match absolute_path.to_str() {
-      Some(s) => s.to_string(),
-      None => {
-        return Err(anyhow::anyhow!(
-          "get_subtree_hash: Failed to convert path to string"
-        ))
-      }
-    };
 
-    let node = match self.files.get(&abs_string) {
+    let node = match self.files.get(absolute_path) {
       Some(file) => file.node.clone(),
       None => {
-        return Err(anyhow::anyhow!("Could not find file in tree!"));
+        let all_files: Vec<String> = self.files.keys().cloned().collect();
+        return Err(anyhow::anyhow!("Could not find file in tree! Looking for: {}. All files: {:?}", absolute_path, all_files));
       }
     };
 
@@ -806,6 +799,10 @@ impl MerkleNode {
 
       if is_git_ignored && !bypass_git {
         // println!("skipping directory: {}", path_str);
+        tracing::info!(
+          "skipping directory because its git ignored: {}",
+          path_str
+        );
         return Arc::new(RwLock::new(MerkleNode::empty_node(
           Some(absolute_file_or_directory),
           Some("Directory is git ignored!".to_string()),