benchmark yaml parsers

read_yamls.py

import yaml
import ryaml
import pathlib

paths = pathlib.Path("yamls").glob("*.yaml")

def read_yaml():
    nodes = []
    for p in paths:
        with open(p, 'rb') as fid:
            node = yaml.safe_load(fid)
        nodes.append(node)
    return nodes

def read_ryaml():
    nodes = []
    for p in paths:
        with open(p, 'rb') as fid:
            node = ryaml.load(fid)
        nodes.append(node)
    return nodes


nodes = read_ryaml()
# nodes = read_yaml()

value = 0
for node in nodes:
    for edge in node['edges']:
        value += edge['value']

print(f"{value=}")

write_yamls.py

import string
import yaml
import uuid
import datetime
from dataclasses import dataclass, asdict
import random

@dataclass
class TreeEdge:
    value: int
    next_uuid: uuid.UUID | None
    name: str

@dataclass
class TreeNode:

    id: uuid.UUID | None
    name: str
    date_created: datetime.datetime
    edges: list[TreeEdge]

def random_uuid():
    return str(uuid.uuid4())

def random_date():
    return datetime.datetime.fromordinal(random.randint(1, 3650))

def random_str():
    return ''.join(random.choice(string.ascii_letters) for _ in range(10))


def random_tree(stop):
    """generate tree with randomised attributes"""
    parents = [None]
    n_nodes = 0
    while n_nodes < stop:
        next_parents = []
        for parent in parents:
            edges = [TreeEdge(value=random.randint(0, 100), next_uuid=random_uuid(), name=random_str()) for _ in range(10)]
            yield TreeNode(
                id=parent,
                name=random_str(),
                date_created=random_date(),
                edges=edges
            )
            n_nodes+=1
            next_parents.extend(edges)
        parents = next_parents

if __name__ == "__main__":
    for node in random_tree(1000):
        with open("yamls/node_{}.yaml".format(node.id), "w") as f:
            node_dict = asdict(node)
            yaml.safe_dump(node_dict, f)