From dc52158bf02697c0b0f3022829f0df78665cfcc6 Mon Sep 17 00:00:00 2001
From: Matthew Harrigan <mpharrigan@google.com>
Date: Wed, 15 Dec 2021 14:28:13 -0800
Subject: [PATCH] Add dev_tools/check_notebooks.py (#244)

* Add dev_tools/check_notebooks.py

Change-Id: I578c7a45d1946d89b8de1ca15d3353ccc8578f76

* docs

Change-Id: I07c189f27b2c7c2adcf184e56437abab2e8eafe8

* nbfmt

Change-Id: I4308d149b9f088f565c68aad304ad6ee420e976b
---
 dev_tools/.gitignore                  |   1 +
 dev_tools/check_notebooks.py          | 109 ++++++++++++++++++++++++++
 docs/qaoa/optimization_analysis.ipynb |  25 ++++++
 3 files changed, 135 insertions(+)
 create mode 100644 dev_tools/.gitignore
 create mode 100644 dev_tools/check_notebooks.py

diff --git a/dev_tools/.gitignore b/dev_tools/.gitignore
new file mode 100644
index 00000000..fd02abca
--- /dev/null
+++ b/dev_tools/.gitignore
@@ -0,0 +1 @@
+notebook_envs
\ No newline at end of file
diff --git a/dev_tools/check_notebooks.py b/dev_tools/check_notebooks.py
new file mode 100644
index 00000000..04f15c18
--- /dev/null
+++ b/dev_tools/check_notebooks.py
@@ -0,0 +1,109 @@
+import os
+import shutil
+import time
+from multiprocessing import Pool
+from subprocess import run as _run, CalledProcessError
+
+BASE_PACKAGES = [
+    # for running the notebooks
+    "jupyter",
+    # assumed to be part of colab
+    "seaborn~=0.11.1",
+]
+
+
+def _check_notebook(notebook_fn: str, notebook_id: str, stdout, stderr):
+    """Helper function to actually do the work in `check_notebook`.
+
+    `check_notebook` has all the context managers and exception handling,
+    which would otherwise result in highly indented code.
+
+    Args:
+        notebook_fn: The notebook filename
+        notebook_id: A unique string id for the notebook that does not include `/`
+        stdout: A file-like object to redirect stdout
+        stderr: A file-like object to redirect stderr
+    """
+    print(f'Starting {notebook_id}')
+
+    def run(*args, **kwargs):
+        return _run(*args, check=True, stdout=stdout, stderr=stderr, **kwargs)
+
+    # 1. create venv
+    venv_dir = os.path.abspath(f'./notebook_envs/{notebook_id}')
+    run(['python', '-m', 'venv', '--clear', venv_dir])
+
+    # 2. basic colab-like environment
+    pip = f'{venv_dir}/bin/pip'
+    run([pip, 'install'] + BASE_PACKAGES)
+
+    # 3. execute
+    jupyter = f'{venv_dir}/bin/jupyter'
+    env = os.environ.copy()
+    env['PATH'] = f'{venv_dir}/bin:{env["PATH"]}'
+    run([jupyter, 'nbconvert', '--to', 'html', '--execute', notebook_fn], cwd='../', env=env)
+
+    # 4. clean up
+    shutil.rmtree(venv_dir)
+
+
+def check_notebook(notebook_fn: str):
+    """Check a notebook.
+
+     1. Create a venv just for that notebook
+     2. Verify the notebook executes without error (and that it installs its own dependencies!)
+     3. Clean up venv dir
+
+    A scratch directory dev_tools/notebook_envs will be created containing tvenv as well as
+    stdout and stderr logs for each notebook. Each of these files and directories will be
+    named according to the "notebook_id", which is `notebook_fn.replace('/', '-')`.
+
+    The executed notebook will be rendered to html alongside its original .ipynb file for
+    spot-checking.
+
+    Args:
+        notebook_fn: The filename of the notebook relative to the repo root.
+    """
+    notebook_id = notebook_fn.replace('/', '-')
+    start = time.perf_counter()
+    with open(f'./notebook_envs/{notebook_id}.stdout', 'w') as stdout, \
+            open(f'./notebook_envs/{notebook_id}.stderr', 'w') as stderr:
+        try:
+            _check_notebook(notebook_fn, notebook_id, stdout, stderr)
+        except CalledProcessError:
+            print('ERROR!', notebook_id)
+    end = time.perf_counter()
+    print(f'{notebook_id} {end - start:.1f}s')
+
+
+NOTEBOOKS = [
+    'docs/otoc/otoc_example.ipynb',
+    'docs/guide/data_analysis.ipynb',
+    'docs/guide/data_collection.ipynb',
+    'docs/qaoa/example_problems.ipynb',
+    'docs/qaoa/precomputed_analysis.ipynb',
+    'docs/qaoa/hardware_grid_circuits.ipynb',
+    'docs/qaoa/optimization_analysis.ipynb',
+    'docs/qaoa/tasks.ipynb',
+    'docs/qaoa/landscape_analysis.ipynb',
+    'docs/qaoa/routing_with_tket.ipynb',
+    'docs/quantum_chess/concepts.ipynb',
+    # 'docs/quantum_chess/quantum_chess_rest_api.ipynb', # runs a server, never finishes.
+    # 'docs/quantum_chess/quantum_chess_client.ipynb',   # uses the server, requires modification.
+    'docs/hfvqe/molecular_data.ipynb',
+    'docs/hfvqe/quickstart.ipynb',
+    'docs/fermi_hubbard/publication_results.ipynb',
+    'docs/fermi_hubbard/experiment_example.ipynb',
+]
+
+
+def main():
+    os.chdir(os.path.dirname(__file__))
+    os.makedirs('./notebook_envs', exist_ok=True)
+    with Pool(4) as pool:
+        results = pool.map(check_notebook, NOTEBOOKS)
+    print(results)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/docs/qaoa/optimization_analysis.ipynb b/docs/qaoa/optimization_analysis.ipynb
index 6fb6e173..1241317c 100644
--- a/docs/qaoa/optimization_analysis.ipynb
+++ b/docs/qaoa/optimization_analysis.ipynb
@@ -62,6 +62,31 @@
     "</table>"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "314b4703b85b"
+   },
+   "source": [
+    "## Setup\n",
+    "\n",
+    "Install the ReCirq package:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "id": "ca9285a3a162"
+   },
+   "outputs": [],
+   "source": [
+    "try:\n",
+    "    import recirq\n",
+    "except ImportError:\n",
+    "    !pip install -q git+https://github.com/quantumlib/ReCirq sympy~=1.6"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {