wip: rewriting for clarity and accuracy

This commit is contained in:
Jeremy Wall 2025-05-09 09:16:26 -04:00
parent fe372b96fd
commit eec0dab6f0

View File

@ -4,49 +4,98 @@ use blake2::{Blake2b512, Digest};
use rand::Rng;
use serde::{Deserialize, Serialize};
/// A `Reference` represents a node in a content-addressable graph structure.
///
/// Each reference contains:
/// - An id that uniquely identifies the referenced object
/// - A content address hash that represents the content
/// - A path that provides a human-readable location for the reference
/// - A list of dependent references that this reference depends on
///
/// References form a directed acyclic graph where each node can have multiple dependents.
/// When the content of a reference changes, the object id of all parent references
/// are automatically recalculated to maintain content addressability throughout the graph.
#[derive(Serialize, Deserialize, Debug, Clone)]
pub struct Reference {
pub object_id: String,
pub id: String,
pub content_address: String,
pub path: String,
pub name: String,
#[serde(skip_serializing_if = "Vec::is_empty")]
pub dependents: Vec<Arc<Reference>>,
}
impl Reference {
pub fn new(object_id: String, content_address: String, path: String) -> Self {
/// Creates a new Reference with the specified object ID, content address, and path.
///
/// # Parameters
/// * `content_address` - The content-based hash address of the referenced object
/// * `name` - The name of the reference
///
/// # Returns
/// A new Reference instance with no dependents
pub fn new(content_address: String, name: String) -> Self {
// Calculate the reference_id from the content_address and path
let hasher = Self::initial_hash(&content_address, &name);
let calculated_id = format!("{:x}", hasher.finalize());
Self {
object_id,
id: calculated_id,
content_address,
path,
name,
dependents: Vec::new(),
}
}
/// Adds a dependent reference to this reference.
///
/// This establishes a parent-child relationship where this reference depends on
/// the provided reference. When a dependent reference changes, the parent's
/// id will be recalculated.
///
/// # Parameters
/// * `dep` - An Arc-wrapped Reference to add as a dependent
///
/// # Returns
/// The modified Reference with the new dependent added
pub fn add_dep(mut self, dep: Arc<Reference>) -> Self {
self.dependents.push(dep);
// Recalculate the ID based on dependents, content_address, and path
let mut hasher = Self::initial_hash(&self.content_address, &self.name);
for dependent in &self.dependents {
hasher.update(&dependent.id);
}
self.id = format!("{:x}", hasher.finalize());
self
}
/// Converts this Reference into an Arc-wrapped Reference for shared ownership.
///
/// This is useful when a Reference needs to be shared across multiple owners
/// in the graph structure.
///
/// # Returns
/// An Arc-wrapped version of this Reference
pub fn to_arc(self) -> Arc<Self> {
Arc::new(self)
}
/// Determines if this Reference is a leaf node (has no dependents).
///
/// Leaf nodes directly represent content, while non-leaf nodes derive their
/// content address from their dependents.
///
/// # Returns
/// `true` if this Reference has no dependents, `false` otherwise
pub fn is_leaf(&self) -> bool {
return self.dependents.is_empty();
}
/// Calculates a content address hash based on dependent references
pub fn calculate_content_address(&self) -> String {
if self.is_leaf() {
return self.content_address.clone();
}
fn initial_hash(content_address: &String, path: &String) -> Blake2b512 {
let mut hasher = Blake2b512::new();
for dep in &self.dependents {
hasher.update(&dep.content_address);
}
format!("{:x}", hasher.finalize())
hasher.update(content_address);
hasher.update(path);
hasher
}
}
@ -83,9 +132,12 @@ pub fn random_object() -> (String, String) {
}
impl Graph {
/// Updates a reference to point to a new object, recalculating content addresses
/// for all affected references in the graph
pub fn update_reference(&mut self, path: &String, new_object_id: String, new_content: String) -> Result<(), String> {
/// Updates a reference to point to a new object, recalculating content addresses and IDs
/// for all affected references in the graph.
///
/// The reference ID is calculated from the content address, name, and any dependents,
/// ensuring that it's truly content-addressable.
pub fn update_reference(&mut self, name: &String, new_content: String) -> Result<(), String> {
// Create a mutable copy of our maps
let mut refs = HashMap::new();
for (k, v) in self.refs.as_ref() {
@ -98,33 +150,40 @@ impl Graph {
}
// Find the reference to update
let ref_to_update = refs.get(path).ok_or_else(|| format!("Reference with path {} not found", path))?;
let ref_to_update = refs.get(name).ok_or_else(|| format!("Reference with name {} not found", name))?;
// Calculate hash for the new content
let mut hasher = Blake2b512::new();
hasher.update(&new_content);
let new_address = format!("{:x}", hasher.finalize());
// Create updated reference
let updated_ref = Arc::new(Reference {
object_id: new_object_id,
content_address: new_address.clone(),
path: path.to_string(),
dependents: ref_to_update.dependents.clone(),
});
// Create a new reference with the updated content address
// The ID will be calculated based on the content address, name, and dependents
let mut updated_ref = Reference::new(
new_address.clone(),
name.to_string()
);
// Add all dependents to the updated reference
for dep in &ref_to_update.dependents {
updated_ref = updated_ref.add_dep(dep.clone());
}
let updated_ref = updated_ref.to_arc();
// Update objects map with new content
objects.insert(new_address.clone(), new_content);
// Update references map with new reference
refs.insert(path.to_string(), updated_ref.clone());
refs.insert(name.to_string(), updated_ref.clone());
// Find and update all parent references that contain this reference
self.update_parent_references(&mut refs, path)?;
self.update_parent_references(&mut refs, name)?;
// Update the root reference if needed
if path == &self.root.path {
self.root = refs.get(path).unwrap().clone();
if name == &self.root.name {
self.root = refs.get(name).unwrap().clone();
}
// Update the Arc maps
@ -135,48 +194,41 @@ impl Graph {
}
/// Recursively updates parent references when a child reference changes
fn update_parent_references(&self, refs: &mut HashMap<String, Arc<Reference>>, updated_path: &str) -> Result<(), String> {
fn update_parent_references(&self, refs: &mut HashMap<String, Arc<Reference>>, updated_name: &str) -> Result<(), String> {
// Find all references that have the updated reference as a dependent
let parent_paths: Vec<String> = refs
let parent_names: Vec<String> = refs
.iter()
.filter(|(_, r)| r.dependents.iter().any(|dep| dep.path == updated_path))
.map(|(path, _)| path.clone())
.filter(|(_, r)| r.dependents.iter().any(|dep| dep.name == updated_name))
.map(|(name, _)| name.clone())
.collect();
for parent_path in parent_paths {
if let Some(parent_ref) = refs.get(&parent_path) {
// Create a new list of dependents with the updated reference
let mut new_dependents = Vec::new();
for parent_name in parent_names {
if let Some(parent_ref) = refs.get(&parent_name) {
// Create a new reference with the same content address and name
let mut updated_parent = Reference::new(
parent_ref.content_address.clone(),
parent_ref.name.clone()
);
// Add all dependents, replacing the updated one
for dep in &parent_ref.dependents {
if dep.path == updated_path {
if dep.name == updated_name {
// Use the updated reference
new_dependents.push(refs.get(updated_path).unwrap().clone());
updated_parent = updated_parent.add_dep(refs.get(updated_name).unwrap().clone());
} else {
// Keep the existing dependent
new_dependents.push(dep.clone());
updated_parent = updated_parent.add_dep(dep.clone());
}
}
// Calculate new content address based on updated dependents
let mut hasher = Blake2b512::new();
for dep in &new_dependents {
hasher.update(&dep.content_address);
}
let new_address = format!("{:x}", hasher.finalize());
// Create updated parent reference
let updated_parent = Arc::new(Reference {
object_id: parent_ref.object_id.clone(),
content_address: new_address,
path: parent_ref.path.clone(),
dependents: new_dependents,
});
// The ID is automatically calculated in the add_dep method
let updated_parent = Arc::new(updated_parent);
// Update the references map
refs.insert(parent_path.clone(), updated_parent);
refs.insert(parent_name.clone(), updated_parent);
// Recursively update parents of this parent
self.update_parent_references(refs, &parent_path)?;
self.update_parent_references(refs, &parent_name)?;
}
}
@ -184,49 +236,61 @@ impl Graph {
}
pub fn random_graph() -> Graph {
let path_root = String::from("ref/0");
let root_name = String::from("ref/0");
let mut objects = HashMap::new();
let mut refs = HashMap::new();
// Create the root reference
let mut root_ref = Reference::new(
"username:0".to_string(),
String::from("0"),
path_root.clone(),
String::from("root_content"),
root_name.clone(),
);
let mut root_hasher = Blake2b512::new();
// Create 10 item references
for i in 1..=10 {
let item_name = format!("/item/{}", i);
let mut item_ref = Reference::new(
format!("item:{}", i),
format!("0:{}", i),
format!("/item/{}", i),
format!("item_content_{}", i),
item_name.clone(),
);
let mut hasher = Blake2b512::new();
// Create 10 subitems for each item
for j in 1..=10 {
let (address, content) = random_object();
hasher.update(&content);
let subitem_name = format!("/item/{}/subitem/{}", i, j);
// Create a leaf reference
let leaf_ref = Reference::new(
format!("item:{}:subitem:{}", i, j),
format!("{}", address),
format!("/item/{}/subitem/{}", i, j),
)
.to_arc();
address.clone(),
subitem_name,
).to_arc();
// Add the leaf reference as a dependent to the item reference
item_ref = item_ref.add_dep(leaf_ref.clone());
// Store the content in the objects map
objects.insert(address.clone(), content);
hasher.update(&leaf_ref.content_address);
refs.insert(leaf_ref.path.clone(), leaf_ref);
// Store the leaf reference in the refs map
refs.insert(leaf_ref.name.clone(), leaf_ref);
}
let hash = format!("{:x}", hasher.finalize());
item_ref.content_address = hash;
root_hasher.update(&item_ref.content_address);
let rc_ref = item_ref.to_arc();
root_ref = root_ref.add_dep(rc_ref.clone());
refs.insert(rc_ref.path.clone(), rc_ref);
// Convert the item reference to Arc and add it to the root reference
let arc_item_ref = item_ref.to_arc();
root_ref = root_ref.add_dep(arc_item_ref.clone());
// Store the item reference in the refs map
refs.insert(arc_item_ref.name.clone(), arc_item_ref);
}
root_ref.content_address = format!("{:x}", root_hasher.finalize());
let rc_root = root_ref.to_arc();
refs.insert(rc_root.path.clone(), rc_root.clone());
dbg!(&objects);
// Convert the root reference to Arc
let arc_root_ref = root_ref.to_arc();
// Store the root reference in the refs map
refs.insert(arc_root_ref.name.clone(), arc_root_ref.clone());
Graph {
root: rc_root,
root: arc_root_ref,
refs: Arc::new(refs),
objects: Arc::new(objects),
}