From 3d8a54d3841a69f16dec9e061c3a8b1b19a9e0c1 Mon Sep 17 00:00:00 2001
From: Ethiraric <ethiraric@gmail.com>
Date: Wed, 14 Feb 2024 00:35:41 +0100
Subject: [PATCH] Add a generator for nested objects.

---
 saphyr/examples/gen_large_yaml/main.rs   |   6 +-
 saphyr/examples/gen_large_yaml/nested.rs | 108 +++++++++++++++++++++++
 2 files changed, 112 insertions(+), 2 deletions(-)
 create mode 100644 saphyr/examples/gen_large_yaml/nested.rs

diff --git a/saphyr/examples/gen_large_yaml/main.rs b/saphyr/examples/gen_large_yaml/main.rs
index 4ba7ad9..cfc8d70 100644
--- a/saphyr/examples/gen_large_yaml/main.rs
+++ b/saphyr/examples/gen_large_yaml/main.rs
@@ -1,6 +1,7 @@
 #![allow(dead_code)]
 
 mod gen;
+mod nested;
 
 use std::collections::HashMap;
 
@@ -8,8 +9,9 @@ use rand::{rngs::ThreadRng, Rng};
 
 fn main() -> std::fmt::Result {
     let mut s = String::new();
-    let mut g = Generator::new();
-    g.gen_strings_array(&mut s, 1_300_000, 1_300_001, 10, 40)?;
+    // let mut g = Generator::new();
+    // g.gen_strings_array(&mut s, 1_300_000, 1_300_001, 10, 40)?;
+    nested::create_deep_object(&mut s, 5_000_000)?;
     println!("{s}");
     Ok(())
 }
diff --git a/saphyr/examples/gen_large_yaml/nested.rs b/saphyr/examples/gen_large_yaml/nested.rs
new file mode 100644
index 0000000..3977901
--- /dev/null
+++ b/saphyr/examples/gen_large_yaml/nested.rs
@@ -0,0 +1,108 @@
+use std::{cell::RefCell, rc::Rc};
+
+use rand::{rngs::ThreadRng, Rng};
+
+/// Create a deep object with the given amount of nodes.
+pub fn create_deep_object<W: std::fmt::Write>(writer: &mut W, n_nodes: usize) -> std::fmt::Result {
+    let mut tree = Tree::new();
+    for _ in 0..n_nodes {
+        tree.push_node();
+    }
+    tree.write_to(writer)
+}
+
+/// An n-tree.
+///
+/// The algorithm used to generate a potentially deep object is to create a tree, one node at a
+/// time, where each node is put as a child of a random existing node in the tree.
+struct Tree {
+    /// The tree-view of the tree.
+    root: Rc<RefCell<Node>>,
+    /// Array of all the nodes in the tree, including the root node.
+    nodes: Vec<Rc<RefCell<Node>>>,
+    /// The RNG state.
+    rng: ThreadRng,
+}
+
+/// A node in a tree.
+struct Node {
+    /// All the children of the node.
+    children: Vec<Rc<RefCell<Node>>>,
+}
+
+impl Tree {
+    /// Create a new tree.
+    fn new() -> Self {
+        let root = Node::new_rc_refcell();
+        Tree {
+            root: root.clone(),
+            nodes: vec![root],
+            rng: rand::thread_rng(),
+        }
+    }
+
+    /// Add a new node as a child of a random node in the tree.
+    fn push_node(&mut self) {
+        let new_node = Node::new_rc_refcell();
+        let n_nodes = self.nodes.len();
+        let parent = &mut self.nodes[self.rng.gen_range(0..n_nodes)];
+        (**parent).borrow_mut().push_child(new_node.clone());
+        self.nodes.push(new_node);
+    }
+
+    /// Write the YAML representation of the tree to `writer`.
+    fn write_to<W: std::fmt::Write>(&self, writer: &mut W) -> std::fmt::Result {
+        (*self.root).borrow().write_to(writer, 0)
+    }
+}
+
+impl Node {
+    /// Create a new node.
+    fn new() -> Self {
+        Node { children: vec![] }
+    }
+
+    fn new_rc_refcell() -> Rc<RefCell<Self>> {
+        Rc::new(RefCell::new(Self::new()))
+    }
+
+    /// Append a child to the node.
+    fn push_child(&mut self, child: Rc<RefCell<Self>>) {
+        self.children.push(child);
+    }
+
+    /// Write the YAML representation of the node to `writer`.
+    fn write_to<W: std::fmt::Write>(&self, writer: &mut W, indent: usize) -> std::fmt::Result {
+        if self.children.is_empty() {
+            write_n(writer, ' ', indent)?;
+            writer.write_str("a: 1\n")?;
+        } else {
+            for (n, child) in self.children.iter().enumerate() {
+                write_n(writer, ' ', indent)?;
+                write_id_for_number(writer, n)?;
+                writer.write_str(":\n")?;
+                (**child).borrow().write_to(writer, indent + 2)?;
+            }
+        }
+        Ok(())
+    }
+}
+
+/// Write `n` times `c` to `out`.
+fn write_n<W: std::fmt::Write>(out: &mut W, c: char, n: usize) -> std::fmt::Result {
+    for _ in 0..n {
+        out.write_char(c)?;
+    }
+    Ok(())
+}
+
+/// Create a valid identifier for the given number.
+fn write_id_for_number<W: std::fmt::Write>(out: &mut W, mut n: usize) -> std::fmt::Result {
+    const DIGITS: &[u8] = b"_abcdefghijklmnopqrstuvwxyz";
+    n += 1;
+    while n > 0 {
+        out.write_char(DIGITS[n % DIGITS.len()] as char)?;
+        n /= DIGITS.len();
+    }
+    Ok(())
+}