@anysphere/file-service 0.0.0-e3fdf62d → 0.0.0-e6124fba

package/src/merkle_tree/mod.rs

@@ -1,9 +1,13 @@
 use super::file_utils;
 use sha2::Digest;
+use std::collections::{BTreeMap, HashSet};
 use std::path::PathBuf;
-use std::{collections::HashMap, fs, path::Path, sync::Arc};
+use std::vec;
+use std::{fs, path::Path, sync::Arc};
 use tokio::sync::RwLock;
 use tonic::async_trait;
+use tracing::info;
+
 pub mod local_construction;
 pub mod test;
 
@@ -12,8 +16,10 @@ pub type MerkleNodePtr = Arc<RwLock<MerkleNode>>;
 pub struct MerkleTree {
   root_path: String,
   root: MerkleNodePtr,
-  files: HashMap<String, File>,
-  cursor: Option<MerkleNodePtr>,
+  files: BTreeMap<String, File>,
+  cursor: Option<usize>,
+  git_ignored_files_and_dirs: HashSet<String>,
+  is_git_repo: bool,
 }
 
 #[derive(Debug)]
@@ -57,6 +63,8 @@ pub trait LocalConstruction {
 
   async fn construct_merkle_tree(
     root_directory: String,
+    git_ignored_files_and_dirs: HashSet<String>,
+    is_git_repo: bool
   ) -> Result<MerkleTree, anyhow::Error>;
 
   async fn update_file(
@@ -87,27 +95,38 @@ impl MerkleTree {
   pub fn empty_tree() -> MerkleTree {
     MerkleTree {
       root: Arc::new(RwLock::new(MerkleNode::empty_node(None, None))),
-      files: HashMap::new(),
+      files: BTreeMap::new(),
       root_path: "".to_string(),
       cursor: None,
+      git_ignored_files_and_dirs: HashSet::new(),
+      is_git_repo: false
     }
   }
 
   pub async fn get_subtree_hash(
     &self,
-    path: String,
+    absolute_path: &str,
   ) -> Result<String, anyhow::Error> {
-    let path = PathBuf::from(path);
-    let node = match self.files.get(path.to_str().unwrap()) {
+    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
+        ));
       }
     };
 
     let node_reader = node.read().await;
     let node_hash = node_reader.hash.clone();
 
+    info!(
+      "get_subtree_hash for path: {}, node_hash: {}",
+      absolute_path, node_hash
+    );
+
     Ok(node_hash)
   }
 
@@ -132,6 +151,43 @@ impl MerkleTree {
     Ok(count)
   }
 
+  pub async fn get_num_embeddable_files_in_subtree(
+    &self,
+    absolute_path: PathBuf,
+  ) -> Result<i32, anyhow::Error> {
+    let mut count = 0;
+
+    let absolute_path = match absolute_path.to_str() {
+      Some(s) => s.to_string(),
+      None => {
+        return Err(anyhow::anyhow!(
+        "get_num_embeddable_files_in_subtree: Failed to convert path to string"
+      ))
+      }
+    };
+
+    // TODO(sualeh): worth keeping this list sorted. its now a btree
+
+    for (_, file) in &self.files {
+      let file_reader = file.node.read().await;
+      match &file_reader.node_type {
+        NodeType::File(file_name) => {
+          if file_name.contains(&absolute_path) {
+            count += 1;
+          }
+        }
+        NodeType::Branch(_) => {
+          continue;
+        }
+        NodeType::ErrorNode(_) => {
+          continue;
+        }
+      }
+    }
+
+    Ok(count)
+  }
+
   pub async fn get_all_files(&self) -> Result<Vec<String>, anyhow::Error> {
     let mut files = Vec::new();
 
@@ -188,83 +244,120 @@ impl MerkleTree {
   pub async fn get_next_file_to_embed(
     &mut self,
   ) -> Result<(String, Vec<String>), anyhow::Error> {
-    // the plan is to do an in-order traversal of the tree.
-
-    // first the edge case to deal with:
-    // cursor == None
-    if self.cursor.is_none() {
-      // If the root is a file, return its name.
-      if let NodeType::File(file_path) = &self.root.read().await.node_type {
-        return Ok((file_path.clone(), vec![]));
-      }
-
-      // if the path is not empty, we can iterate till we find the first child.
-      let mut potential_first_child = self.root.clone();
-      let mut is_branch = true;
-      let mut path = Vec::new();
-
-      while is_branch {
-        let node = {
-          let potential_first_child_reader = potential_first_child.read().await;
-          match &potential_first_child_reader.node_type {
-            NodeType::Branch(branch) => branch.clone(),
-            NodeType::File(_) => {
-              return Err(anyhow::anyhow!(
-                "get_next_file_to_embed: This should not happen! the branch happened to be file."
-              ));
-            }
-            NodeType::ErrorNode(_) => {
-              return Err(anyhow::anyhow!("Cursor is an error node!"));
-            }
-          }
-        };
+    // if the cursor is none, set it to 0
+    let cursor = match self.cursor {
+      Some(cursor) => cursor,
+      None => {
+        self.cursor = Some(0);
+        0
+      }
+    };
+
+    // get the thing at the cursor. while we dont find a file, we keep incrementing the cursor.
+    let mut cursor = cursor;
+    loop {
+      // O(log n)
+      let file = match self.files.values().nth(cursor) {
+        Some(file) => file,
+        None => {
+          return Err(anyhow::anyhow!("Could not find file to embed!"));
+        }
+      };
 
-        let current_node_name = &node.0;
-        let child_list = &node.1;
+      let file_reader = file.node.read().await;
+      match &file_reader.node_type {
+        NodeType::File(f) => {
+          // update the cursor.
+          self.cursor = Some(cursor + 1);
+          let spline = self.get_spline(f).await?;
+          return Ok((f.clone(), spline));
+        }
+        NodeType::Branch(_) => {
+          cursor += 1;
+          continue;
+        }
+        NodeType::ErrorNode(_) => {
+          cursor += 1;
+          continue;
+        }
+      }
+    }
+  }
 
-        if let Some(c) = child_list.first() {
-          let c_reader = c.read().await;
+  pub async fn get_all_dir_files_to_embed(
+    &self,
+    absolute_path: &str,
+  ) -> Result<Vec<String>, anyhow::Error> {
+    let mut files = Vec::new();
 
-          match &c_reader.node_type {
-            NodeType::File(file_path) => {
-              // must set the cursor!
-              self.cursor = Some(c.clone());
+    // 1. should check that this absolute path is actually a directory.
+    let file_node = self.files.get(absolute_path);
+    if file_node.is_none() {
+      return Err(anyhow::anyhow!("Could not find directory the in tree!"));
+    }
 
-              return Ok((file_path.clone(), path));
-            }
-            NodeType::Branch(_) => {
-              potential_first_child = c.clone();
-              is_branch = true;
+    for (file_path, f) in &self.files {
+      if !file_path.contains(absolute_path) {
+        continue;
+      }
 
-              // add the path to the current node.
-              path.push(current_node_name.clone());
-            }
-            NodeType::ErrorNode(_) => {
-              return Err(anyhow::anyhow!("Cursor is an error node!"));
-            }
-          }
-        } else {
-          // If the root has no children, return an error.
-          return Err(anyhow::anyhow!("Root has no children!"));
+      match f.node.read().await.node_type {
+        NodeType::File(_) => {
+          files.push(file_path.clone());
+        }
+        NodeType::Branch(_) => {
+          continue;
+        }
+        NodeType::ErrorNode(_) => {
+          continue;
         }
       }
     }
 
-    // THE DEFAULT CASE:
-    // we already have a cursor at a file.
-
-    // UNWRAP checked and fine. see the none case above.
-    let cursor_name = self.cursor.as_ref().unwrap();
-    let cursor_reader = cursor_name.read().await;
+    Ok(files)
+  }
 
-    // invariant: you must be a file!!
+  // TODO(sualeh): i need tests for this!!
+  pub async fn get_spline(
+    &self,
+    absolute_path: &str,
+  ) -> Result<Vec<String>, anyhow::Error> {
+    let mut files = Vec::new();
 
-    // 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.
+    let current_node = match self.files.get(absolute_path) {
+      Some(node) => {
+        node.node.clone()
+      }
+      None => {
+        return Err(anyhow::anyhow!("File not found: {}", absolute_path));
+      }
+    };
 
-    // algorithm:
-    // 1.
+    let mut stack = Vec::new();
+    stack.push(current_node);
+
+    while let Some(node) = stack.pop() {
+      let parent = node.read().await.parent.clone();
+      if let Some(parent) = parent {
+        {
+          let parent_node = parent.read().await;
+          match &parent_node.node_type {
+            NodeType::File(file_name) => {
+              files.push(file_name.clone());
+            }
+            NodeType::Branch((branch_name, _)) => {
+              files.push(branch_name.clone());
+            }
+            _ => {
+              continue;
+            }
+          }
+        }
 
-    Err(anyhow::anyhow!("Could not find file to embed!"))
+        stack.push(parent);
+      }
+    }
+    Ok(files)
   }
 
   /// creates a new node and attaches it to the current tree.
@@ -302,12 +395,19 @@ impl MerkleTree {
     // 1. the path is empty. this means that the ancestor is the root.
     // 2. the path is non-empty. that means there exist a non-empty element btwn till the root.
 
+    let absolute_root_path = self.root_path.clone();
     let new_node = match path.len() {
       0 => {
         // this means that the ancestor is the root.
         // we need to create a new node and attach it to the ancestor.
-        let new_node =
-          MerkleNode::new(file_path.clone(), Some(ancestor.clone())).await;
+        let new_node = MerkleNode::new(
+          file_path.clone(),
+          Some(ancestor.clone()),
+          &self.git_ignored_files_and_dirs,
+          &absolute_root_path.as_str(),
+          self.is_git_repo
+        )
+        .await;
         ancestor.write().await.attach_child(new_node.clone()).await;
         new_node
       }
@@ -318,9 +418,14 @@ impl MerkleTree {
         // UNSURE: not sure this is the correct thing to do but it is the fastest.
         // get the last thing that is not in the tree.
         let first_child_path = path.last().unwrap();
-        let first_child =
-          MerkleNode::new(first_child_path.clone(), Some(ancestor.clone()))
-            .await;
+        let first_child = MerkleNode::new(
+          first_child_path.clone(),
+          Some(ancestor.clone()),
+          &self.git_ignored_files_and_dirs,
+          &absolute_root_path.as_str(),
+          self.is_git_repo
+        )
+        .await;
 
         // TODO(sualeh): we should do an assertion check that the entire vec is contained here.
 
@@ -597,18 +702,57 @@ use std::future::Future;
 use std::pin::Pin;
 
 type PinnedFuture<'a, T> = Pin<Box<dyn Future<Output = T> + Send + 'a>>;
+type IgnoredFiles = HashSet<String>;
 
 impl MerkleNode {
   /// please be careful using this.
   async fn __new_unchecked(
     file_or_directory: String,
     parent: ParentPtr,
+    ignored_files: &IgnoredFiles,
+    absolute_root_path: &str,
+    is_git_repo: bool,
   ) -> MerkleNodePtr {
-    MerkleNode::construct_node(Path::new(&file_or_directory), parent).await
+    // // check if the root is a git directory.
+    // let is_git_repo =
+    //   match git_utils::is_git_directory(absolute_root_path).await {
+    //     Ok(is_git_repo) => is_git_repo,
+    //     Err(_e) => false,
+    //   };
+    let bypass_git = !is_git_repo;
+
+    MerkleNode::construct_node(
+      Path::new(&file_or_directory),
+      parent,
+      ignored_files,
+      absolute_root_path,
+      bypass_git,
+    )
+    .await
   }
 
-  async fn new(file_or_directory: PathBuf, parent: ParentPtr) -> MerkleNodePtr {
-    MerkleNode::construct_node(Path::new(&file_or_directory), parent).await
+  async fn new(
+    absolute_file_or_directory: PathBuf,
+    parent: ParentPtr,
+    ignored_files: &IgnoredFiles,
+    absolute_root_path: &str,
+    is_git_repo: bool,
+  ) -> MerkleNodePtr {
+    let bypass_git = !is_git_repo;
+
+    info!(
+      "constructing node for absolute_file_or_directory: {:?}",
+      absolute_file_or_directory
+    );
+
+    MerkleNode::construct_node(
+      Path::new(&absolute_file_or_directory),
+      parent,
+      ignored_files,
+      absolute_root_path,
+      bypass_git,
+    )
+    .await
   }
 
   /// NOT added to the tree by default.
@@ -619,38 +763,51 @@ impl MerkleNode {
   // 		let file_hash = self.files.get_mut(&file_path).unwrap();
 
   fn construct_node<'a>(
-    file_or_directory: &'a Path,
+    absolute_file_or_directory: &'a Path,
     parent: ParentPtr,
+    ignored_files: &'a IgnoredFiles,
+    absolute_root_path: &'a str,
+    bypass_git: bool,
   ) -> PinnedFuture<'a, MerkleNodePtr> {
     Box::pin(async move {
       // check if it is a file
-      let path_str = file_or_directory.to_str().unwrap().to_string();
-      if file_or_directory.is_file() {
+      let path_str = absolute_file_or_directory.to_str().unwrap().to_string();
+      if absolute_file_or_directory.is_file() {
         return Arc::new(RwLock::new(
           MerkleNode::construct_file_node_or_error_node(
-            file_or_directory,
+            absolute_file_or_directory,
             parent,
+            ignored_files,
           )
           .await,
         ));
       }
 
       // check if the directory fails the bad dir test.
-      let is_bad_dir = file_utils::is_in_bad_dir(file_or_directory);
+      let is_bad_dir = file_utils::is_in_bad_dir(absolute_file_or_directory);
       if is_bad_dir.is_err() || is_bad_dir.unwrap_or(false) {
         // println!("skipping directory: {}", path_str);
         return Arc::new(RwLock::new(MerkleNode::empty_node(
-          Some(file_or_directory),
+          Some(absolute_file_or_directory),
           Some("Directory is in bad dir!".to_string()),
         )));
       }
 
-      let entries = fs::read_dir(file_or_directory);
+      let is_git_ignored_dir = ignored_files.contains(&path_str);
+
+      if is_git_ignored_dir && !bypass_git {
+        return Arc::new(RwLock::new(MerkleNode::empty_node(
+          Some(absolute_file_or_directory),
+          Some("Directory is git ignored!".to_string()),
+        )));
+      }
+
+      let entries = fs::read_dir(absolute_file_or_directory);
       match entries {
         Ok(_) => (),
         Err(e) => {
           return Arc::new(RwLock::new(MerkleNode::empty_node(
-            Some(file_or_directory),
+            Some(absolute_file_or_directory),
             Some(e.to_string()),
           )));
         }
@@ -670,13 +827,19 @@ impl MerkleNode {
         match entry {
           Ok(entry) => {
             children.push(
-              MerkleNode::construct_node(&entry.path(), Some(node.clone()))
-                .await,
+              MerkleNode::construct_node(
+                &entry.path(),
+                Some(node.clone()),
+                ignored_files,
+                absolute_root_path,
+                bypass_git,
+              )
+              .await,
             );
           }
           Err(e) => {
             children.push(Arc::new(RwLock::new(MerkleNode::empty_node(
-              Some(file_or_directory),
+              Some(absolute_file_or_directory),
               Some(e.to_string()),
             ))));
           }
@@ -696,31 +859,37 @@ impl MerkleNode {
   }
 
   async fn construct_file_node(
-    file_path: &Path,
+    absolute_file_path: &Path,
     parent: ParentPtr,
+    ignored_files: &IgnoredFiles,
   ) -> Result<MerkleNode, String> {
-    let file_str = file_path
+    let file_str = absolute_file_path
       .to_str()
       .ok_or("Could not convert file path to string!")?
       .to_string();
     // first see if it passes the
-    match file_utils::is_good_file(file_path) {
+    match file_utils::is_good_file(absolute_file_path) {
       Ok(_) => {}
       Err(e) => {
         return Err(format!("File failed runtime checks! {}", e.to_string()));
       }
     }
 
-    // read the file_content to a buffer
-    let file_content = match tokio::fs::read(file_path).await {
-      Ok(content) => content,
-      Err(e) => {
-        return Err(format!("Could not read file! {}", e.to_string()));
+    // check if the file is in the git ignore buffer.
+    // this is a bug right because we are not checking absoluteness here.
+    match ignored_files.contains(&file_str) {
+      true => {
+        return Err(format!("File is in git ignore buffer!"));
       }
-    };
+      false => {}
+    }
 
     // check if the file passes runtime checks.
-    match file_utils::is_good_file_runtime_check(file_path, &file_content).await
+    match file_utils::is_good_file_runtime_check(
+      absolute_file_path,
+      // &file_content,
+    )
+    .await
     {
       Ok(_) => {}
       Err(e) => {
@@ -728,15 +897,14 @@ impl MerkleNode {
       }
     }
 
-    let file_content = match std::str::from_utf8(&file_content) {
-      Ok(content) => content,
-      Err(e) => {
-        return Err(format!(
-          "UTF8 Failure. Could not convert file content to string! {}",
-          e.to_string()
-        ));
-      }
-    };
+    // read the file_content to a buffer
+    let file_content =
+      match file_utils::read_string_without_bom(absolute_file_path).await {
+        Ok(content) => content,
+        Err(e) => {
+          return Err(format!("Could not read file! {}", e.to_string()));
+        }
+      };
 
     let file_hash = compute_hash(&file_content);
     let node = MerkleNode {
@@ -751,15 +919,22 @@ impl MerkleNode {
   }
 
   async fn construct_file_node_or_error_node(
-    file_path: &Path,
+    absolute_file_path: &Path,
     parent: ParentPtr,
+    ignored_files: &IgnoredFiles,
   ) -> MerkleNode {
-    let node = match MerkleNode::construct_file_node(file_path, parent).await {
+    let node = match MerkleNode::construct_file_node(
+      absolute_file_path,
+      parent,
+      ignored_files,
+    )
+    .await
+    {
       Ok(node) => node,
       Err(e) => {
         // println!("constructing error node. error: {}", e);
         // println!("file_path: {:?}", file_path);
-        MerkleNode::empty_node(Some(file_path), Some(e))
+        MerkleNode::empty_node(Some(absolute_file_path), Some(e))
       }
     };
 
@@ -785,15 +960,51 @@ impl MerkleNode {
 
   async fn compute_branch_hash(children: &[MerkleNodePtr]) -> String {
     let mut hasher = sha2::Sha256::new();
+    let mut names_and_hashes = vec![];
+    let mut non_zero_children = 0;
+
     for child in children {
       // check if it is an error node
       let child_reader = child.read().await;
-      if let NodeType::ErrorNode(_) = &child_reader.node_type {
+
+      match &child_reader.node_type {
+        NodeType::File(file_name) => {
+          non_zero_children += 1;
+          names_and_hashes.push((file_name.clone(), child_reader.hash.clone()));
+        }
+        NodeType::Branch((file_name, _)) => {
+          let hash = child_reader.hash.clone();
+          if hash == "" {
+            continue;
+          }
+
+          non_zero_children += 1;
+          names_and_hashes.push((file_name.clone(), hash));
+        }
+        NodeType::ErrorNode(_) => {
+          continue;
+        }
+      }
+    }
+
+    // sort the list of names and hashes by the hashes!!
+    names_and_hashes
+      .sort_by(|a, b| a.1.to_lowercase().cmp(&b.1.to_lowercase()));
+
+    for (name, hash) in names_and_hashes {
+      if hash == "" {
         continue;
       }
+      info!("name: {}, hash: {}", name, hash);
+      hasher.update(hash);
+    }
 
-      hasher.update(child_reader.hash.as_bytes());
+    if non_zero_children == 0 {
+      // this means that the branch is empty.
+      // we should return an empty string.
+      return "".to_string();
     }
+
     let result = hasher.finalize();
     format!("{:x}", result)
   }

package/src/merkle_tree/test.rs

@@ -43,8 +43,9 @@ mod tests {
     // let path = Path::new(&temp_dir_path);
 
     // Test construct_merkle_tree() function
+    let new_set = std::collections::HashSet::<String>::new();
     let tree =
-      MerkleTree::construct_merkle_tree(temp_dir_path.clone()).await;
+      MerkleTree::construct_merkle_tree(temp_dir_path.clone(), new_set, false).await;
     let mut tree = match tree {
       Ok(tree) => {
         assert_eq!(tree.files.len(), 2);

package/src/test.rs

@@ -0,0 +1,5 @@
+#[cfg(test)]
+mod tests {
+  use super::super::*;
+  use std::path::PathBuf;
+}