Multiline terminal command isn't processed properly

[ivanmkc]

I've included two notebooks, which are identical except for the terminal command.

• one with a single-line terminal command

! bq --location=US load --autodetect --skip_leading_rows=1 --source_format=CSV {dataset_id}.us_states_local_file 'resources/us-states.csv'

• one with a multi-line terminal command

!bq \
--location=US load \
--autodetect \
--skip_leading_rows=1 \
--source_format=CSV \
{dataset_id}.us_states_local_file \
'resources/us-states.csv'

The one with multiline doesn't process properly, so when I run nbqa with black, I get:

Running black...
error: cannot format notebooks/official/bigquery_command_line_tool.ipynb: Cannot parse: cell_6:2:5: --lo 0x6AA55214

Archive.zip

[loferris]

I have had an almost identical error with this file and in addition
!bq load \ --location=US \
etc. is auto-revised by black to the second block @ivanmkc mentions above, which then throws the same error. This syntax should also be acceptable. https://cloud.google.com/bigquery/docs/bq-command-line-tool#loading_data

[MarcoGorelli]

Hey @loferris @ivanmkc , thanks for the report! I'll take a look

Does this also happen when you run black on the notebook directly? (As of version 21.8b0, it has support for notebooks)

[MarcoGorelli]

Can't reproduce either example with the latest version (1.1.0):

(venv) marcogorelli@OVMG025 mcve % cat Untitled.ipynb 
{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f013878e-4ee4-4d12-ba40-2afac49d4c29",
   "metadata": {},
   "outputs": [],
   "source": [
    "!bq \\\n",
    "--location=US load \\\n",
    "--autodetect \\\n",
    "--skip_leading_rows=1 \\\n",
    "--source_format=CSV \\\n",
    "{dataset_id}.us_states_local_file \\\n",
    "'resources/us-states.csv'\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "good-bayesian",
   "language": "python",
   "name": "good-bayesian"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}
(venv) marcogorelli@OVMG025 mcve % nbqa black Untitled.ipynb
All done! ✨ 🍰 ✨
1 file left unchanged.

From https://github.com/GoogleCloudPlatform/ai-platform-samples/blob/eed2ac7af3d552f16690c858439f22347a4d616e/.github/workflows/linter/requirements.txt it looks like you're on on an old version of nbQA

Please upgrade to the latest version (1.1.0) or use black directly (I contributed the PR psf/black#2357 , the logic is almost the same as nbQA's, it just works cell-by-cell)

If this still doesn't work for you with the latest version of nbqa, let me know and I'll reopen

[loferris]

@MarcoGorelli With the linter as currently defined, the same error is occurring with the updated version of nbqa. I am now seeing about using black directly to see if this solves the problem.

[MarcoGorelli]

Thanks @loferris - reopening for now then, I'll take another look

[loferris]

when I run the latest version of black directly python3 -m black "$notebook" I get a runtime error and a prompt to open a bug report with nbQA

[MarcoGorelli]

I just ran both nbqa and black on the files which @ivanmkc mentioned, and got no error. Please

1. show the output of:

nbqa --version
black --version

2. share a notebook which reproduces the issue
3. show the exact command you ran, along with the exact output

when I run the latest version of black directly python3 -m black "$notebook" I get a runtime error and a prompt to open a bug report with nbQA

There's no way that this could happen, black's codebase makes no mention of nbQA

[loferris]

1. nbqa --version was 0.6.0 on my machine (it's updated to 1.1.0 in the linter) but I upgraded it to 1.1.0 for the rest of this comment. black --version was 20.8b1 on my machine, upgraded to 21.8b0 for the rest of this comment. [This was because I didn't re-install the linter after changing the requirements .txt file.]
2. Currently all of the notebooks are failing in this repo, with the same error, captured below. The exact problem notebook black wasn't able to parse before is here: https://github.com/GoogleCloudPlatform/bigquery-notebooks/blob/update_linter/notebooks/official/template_notebooks/bigquery_command_line_tool.ipynb
3. Here's the code the linter runs:

if [ "$is_test" = true ] ; then
                echo "Running nbfmt..."
                python3 -m tensorflow_docs.tools.nbfmt --remove_outputs --test "$notebook"
                NBFMT_RTN=$?
                # echo "Running black..."
                # python3 -m nbqa black "$notebook" --check
                # BLACK_RTN=$?
                echo "Running pyupgrade..."
                python3 -m nbqa pyupgrade "$notebook"
                PYUPGRADE_RTN=$?
                echo "Running isort..."
                python3 -m nbqa isort "$notebook" --check
                ISORT_RTN=$?
                echo "Running flake8..."
                python3 -m nbqa flake8 "$notebook" --show-source --extend-ignore=W391,E501,F821,E402,F404,W503,W291,E203,E999,E111,E113 
                FLAKE8_RTN=$?
            else
                echo "Running black..."
                python3 -m black "$notebook"
                BLACK_RTN=$?            
                echo "Running pyupgrade..."
                python3 -m nbqa pyupgrade "$notebook"
                PYUPGRADE_RTN=$?
                echo "Running isort..."
                python3 -m nbqa isort "$notebook"
                ISORT_RTN=$?
                echo "Running nbfmt..."
                python3 -m tensorflow_docs.tools.nbfmt --remove_outputs "$notebook"
                NBFMT_RTN=$?
                echo "Running flake8..."
                python3 -m nbqa flake8 "$notebook" --show-source --extend-ignore=W391,E501,F821,E402,F404,W503,W291,E203,E999,E111,E113
                FLAKE8_RTN=$?                 
            fi

            NOTEBOOK_RTN="0"

(the file is here: https://github.com/GoogleCloudPlatform/bigquery-notebooks/blob/update_linter/.github/workflows/linter/run_linter.sh)
The output is captured here: black is failing the same way and the same error comes up for each subsequent invocation of nbqa.

Oh, and you're right, I misread the original error! black was running, but then all following nbqa commands were failing with the JSONDecodeError

[loferris]

When I go to that line in the .ipynb its questioning this:
'# Licensed under the Apache License, Version 2.0 (the "License");\n',
With the first character - ' - being l11,c17.
Manually changing around the quotes cleared that error, but there seems to be further JSON nits it gets bothered by.
Currently:

`nbQA failed to process bigquery_command_line_tool.ipynb with exception "JSONDecodeError('Expecting value: line 23 column 13 (char 1082)')"`
which is the closing bracket of `source`.
If you believe the notebook(s) to be valid, please report a bug at https://github.com/nbQA-dev/nbQA/issues

Another thing worth mentioning is that these odd changes seemed to have happened after I ran black (I think), as I haven't been manually messing around with the header code and previously it had double-quote formatting.

[MarcoGorelli]

OK, I see what's happened - you're missing the extra Jupyter requirements when running black on the notebooks, and so they're been formatted as if they were Python files. You probably ran black on the notebooks when the version was still 20.8b1, right?

If you want Jupyter support for black, you need to install it as

pip install -U black[jupyter]

I'd advise:

• checkout your notebooks to the original state
• pip install -U black[jupyter]
• run black on the notebooks
• run any other linting checks

---

The previous commit was fine btw (GoogleCloudPlatform/bigquery-notebooks@56712e9)