structure

client-network/src/data.rs

@@ -0,0 +1,305 @@
+use std::collections::HashMap;
+use std::hash::{DefaultHasher, Hash, Hasher};
+use rand::{rng, Rng};
+
+// --- Constants ---
+const MAX_CHUNK_DATA_SIZE: usize = 1024;
+const MAX_DIRECTORY_ENTRIES: usize = 16;
+const MAX_BIG_CHILDREN: usize = 32;
+const MIN_BIG_CHILDREN: usize = 2;
+const FILENAME_HASH_SIZE: usize = 32;
+const DIRECTORY_ENTRY_SIZE: usize = FILENAME_HASH_SIZE * 2; // 64 bytes
+
+fn dummy_hash(data: &[u8]) -> NodeHash {
+    let mut hasher = DefaultHasher::new();
+    data.hash(&mut hasher);
+    let hash_u64 = hasher.finish();
+
+    let mut hash_array = [0u8; FILENAME_HASH_SIZE];
+    // Simple way to spread a 64-bit hash across 32 bytes for a unique-ish ID
+    for i in 0..8 {
+        hash_array[i] = (hash_u64 >> (i * 8)) as u8;
+    }
+    hash_array // The rest remains 0, satisfying the 32-byte requirement
+}
+
+fn generate_random_filename() -> [u8; FILENAME_HASH_SIZE] {
+    let mut rng = rand::rng();
+    let mut filename_bytes = [0; FILENAME_HASH_SIZE];
+
+    // Generate a random length for the base name
+    let name_len = rng.random_range(5..21);
+
+    // Generate random alphanumeric characters
+    for i in 0..name_len {
+        let char_code = rng.random_range(97..123); // 'a' through 'z'
+        if i < FILENAME_HASH_SIZE {
+            filename_bytes[i] = char_code as u8;
+        }
+    }
+
+    // Append a common extension
+    let ext = if rng.random_bool(0.5) { ".txt" } else { ".dat" };
+    let ext_bytes = ext.as_bytes();
+    let start_index = name_len.min(FILENAME_HASH_SIZE - ext_bytes.len());
+    if start_index < FILENAME_HASH_SIZE {
+        filename_bytes[start_index..(start_index + ext_bytes.len())].copy_from_slice(ext_bytes);
+    }
+
+    filename_bytes
+}
+
+pub type NodeHash = [u8; FILENAME_HASH_SIZE];
+
+pub fn node_hash_to_hex_string(hash: &NodeHash) -> String {
+    hash.iter()
+        .map(|b| format!("{:02x}", b))
+        .collect()
+}
+
+#[repr(u8)]
+#[derive(Debug, Clone)]
+pub enum MerkleNode {
+    // up to 1024 bytes of raw data.
+    Chunk(ChunkNode) = 0,
+    // 0 to 16 directory entries.
+    Directory(DirectoryNode) = 1,
+    // list of 2 to 32 hashes pointing to Chunk or Big nodes.
+    Big(BigNode) = 3,
+    // list of 2 to 32 hashes pointing to Directory or BigDirectory nodes.
+    BigDirectory(BigDirectoryNode) = 4,
+}
+
+fn generate_random_file_node(storage: &mut HashMap<NodeHash, MerkleNode>) -> Result<NodeHash, String> {
+    let mut rng = rng();
+    let is_big = rng.random_bool(0.2); // 20% chance of being a big file
+
+    if !is_big {
+        // Generate a simple Chunk Node
+        let node = MerkleNode::Chunk(ChunkNode::new_random());
+        let hash = dummy_hash(&node.serialize());
+        storage.insert(hash, node);
+        Ok(hash)
+    } else {
+        // Generate a Big Node (a file composed of chunks)
+        let num_children = rng.random_range(MIN_BIG_CHILDREN..=MAX_BIG_CHILDREN.min(8)); // Limit complexity
+        let mut children_hashes = Vec::with_capacity(num_children);
+
+        for _ in 0..num_children {
+            // Children must be Chunk or Big; for simplicity, we only generate Chunk children here.
+            let chunk_node = MerkleNode::Chunk(ChunkNode::new_random());
+            let chunk_hash = dummy_hash(&chunk_node.serialize());
+            storage.insert(chunk_hash, chunk_node);
+            children_hashes.push(chunk_hash);
+        }
+
+        let node = MerkleNode::Big(BigNode::new(children_hashes)?);
+        let hash = dummy_hash(&node.serialize());
+        storage.insert(hash, node);
+        Ok(hash)
+    }
+}
+
+fn generate_random_directory_node(
+    depth: u32,
+    max_depth: u32,
+    storage: &mut HashMap<NodeHash, MerkleNode>
+) -> Result<NodeHash, String> {
+    let mut rng = rng();
+    let current_depth = depth + 1;
+    let is_big_dir = rng.random_bool(0.3) && current_depth < max_depth;
+
+    if !is_big_dir || current_depth >= max_depth {
+        // Generate a simple Directory Node (leaf level directory)
+        let num_entries = rng.random_range(1..=MAX_DIRECTORY_ENTRIES.min(5)); // Limit directory size for testing
+        let mut entries = Vec::with_capacity(num_entries);
+
+        for _ in 0..num_entries {
+            if rng.random_bool(0.7) {
+                // 70% chance of creating a file (Chunk/Big)
+                let file_hash = generate_random_file_node(storage)?;
+                let entry = DirectoryEntry {
+                    filename: generate_random_filename(),
+                    content_hash: file_hash,
+                };
+                entries.push(entry);
+            } else if current_depth < max_depth {
+                // 30% chance of creating a subdirectory
+                let dir_hash = generate_random_directory_node(current_depth, max_depth, storage)?;
+
+                // Create a basic directory entry name
+                let mut filename_bytes = [0; 32];
+                let subdir_name = format!("dir_{}", current_depth);
+                filename_bytes[..subdir_name.len()].copy_from_slice(subdir_name.as_bytes());
+
+                let entry = DirectoryEntry {
+                    filename: filename_bytes,
+                    content_hash: dir_hash,
+                };
+                entries.push(entry);
+            }
+        }
+
+        let node = MerkleNode::Directory(DirectoryNode::new(entries)?);
+        let hash = dummy_hash(&node.serialize());
+        storage.insert(hash, node);
+        Ok(hash)
+
+    } else {
+        // Generate a BigDirectory Node (internal directory structure)
+        let num_children = rng.random_range(MIN_BIG_CHILDREN..=MAX_BIG_CHILDREN.min(4)); // Limit children count
+        let mut children = Vec::with_capacity(num_children);
+
+        for _ in 0..num_children {
+            // Children must be Directory or BigDirectory
+            let child_hash = generate_random_directory_node(current_depth, max_depth, storage)?;
+            children.push(child_hash);
+        }
+
+        let node = MerkleNode::BigDirectory(BigDirectoryNode::new(children)?);
+        let hash = dummy_hash(&node.serialize());
+        storage.insert(hash, node);
+        Ok(hash)
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct ChunkNode {
+    pub data: Vec<u8>,
+}
+impl ChunkNode {
+    pub fn new(data: Vec<u8>) -> Result<Self, String> {
+        if data.len() > MAX_CHUNK_DATA_SIZE {
+            return Err(format!("Chunk data exceeds {} bytes", data.len()));
+        }
+        Ok(ChunkNode { data })
+    }
+
+    pub fn new_random() -> Self {
+        let mut rng = rand::rng();
+
+        // Determine a random length between 1 and MAX_CHUNK_DATA_SIZE (inclusive).
+        // Using +1 ensures the range is up to 1024.
+        let random_len = rng.random_range(1..=MAX_CHUNK_DATA_SIZE);
+
+        // Initialize a vector with the random length
+        let mut data = vec![0u8; random_len];
+
+        // Fill the vector with random bytes
+        rng.fill(&mut data[..]);
+
+        // Since we generated the length based on MAX_CHUNK_DATA_SIZE,
+        // this is guaranteed to be valid and doesn't need to return a Result.
+        ChunkNode { data }
+    }
+}
+
+// Helper struct
+#[derive(Debug, Clone)]
+pub struct DirectoryEntry {
+    pub filename: [u8; FILENAME_HASH_SIZE],
+    pub content_hash: NodeHash,
+}
+
+#[derive(Debug, Clone)]
+pub struct DirectoryNode {
+    pub entries: Vec<DirectoryEntry>,
+}
+
+impl DirectoryNode {
+    pub fn new(entries: Vec<DirectoryEntry>) -> Result<Self, String> {
+        if entries.len() > MAX_DIRECTORY_ENTRIES {
+            return Err(format!("Directory exceeds {} bytes", entries.len()));
+        }
+        Ok(DirectoryNode { entries })
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct BigNode {
+    pub children_hashes: Vec<NodeHash>,
+}
+
+impl BigNode {
+    pub fn new(children_hashes: Vec<NodeHash>) -> Result<Self, String> {
+        let n = children_hashes.len();
+        if n < MIN_BIG_CHILDREN || n > MAX_BIG_CHILDREN {
+            return Err(format!(
+                "Big node must have between {} and {} children, found {}",
+                MIN_BIG_CHILDREN, MAX_BIG_CHILDREN, n
+            ));
+        }
+        Ok(BigNode { children_hashes })
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct BigDirectoryNode {
+    pub children_hashes: Vec<NodeHash>,
+}
+
+impl BigDirectoryNode {
+    pub fn new(children_hashes: Vec<NodeHash>) -> Result<Self, String> {
+        let n = children_hashes.len();
+        if n < MIN_BIG_CHILDREN || n > MAX_BIG_CHILDREN {
+            return Err(format!(
+                "BigDirectory node must have between {} and {} children, found {}",
+                MIN_BIG_CHILDREN, MAX_BIG_CHILDREN, n
+            ));
+        }
+        Ok(BigDirectoryNode { children_hashes })
+    }
+}
+
+impl MerkleNode {
+    pub fn get_type_byte(&self) -> u8 {
+        match self {
+            MerkleNode::Chunk(_) => 0,
+            MerkleNode::Directory(_) => 1,
+            MerkleNode::Big(_) => 3,
+            MerkleNode::BigDirectory(_) => 4,
+        }
+    }
+
+    pub fn serialize(&self) -> Vec<u8> {
+        let mut bytes = Vec::new();
+        // 1. Add the type byte
+        bytes.push(self.get_type_byte());
+
+        // 2. Add the node-specific data
+        match self {
+            MerkleNode::Chunk(node) => {
+                bytes.extend_from_slice(&node.data);
+            }
+            MerkleNode::Directory(node) => {
+                // The data is the sequence of directory entries
+                for entry in &node.entries {
+                    bytes.extend_from_slice(&entry.filename);
+                    bytes.extend_from_slice(&entry.content_hash);
+                }
+            }
+            MerkleNode::Big(node) => {
+                // The data is the list of child hashes
+                for hash in &node.children_hashes {
+                    bytes.extend_from_slice(hash);
+                }
+            }
+            MerkleNode::BigDirectory(node) => {
+                // The data is the list of child hashes
+                for hash in &node.children_hashes {
+                    bytes.extend_from_slice(hash);
+                }
+            }
+        }
+        bytes
+    }
+
+    pub fn generate_random_tree(max_depth: u32) -> Result<(NodeHash, HashMap<NodeHash, MerkleNode>), String> {
+        let mut storage = HashMap::new();
+
+        // Start tree generation from the root directory at depth 0
+        let root_hash = generate_random_directory_node(0, max_depth, &mut storage)?;
+
+        Ok((root_hash, storage))
+    }
+}