Add support for fixing style as a task in another process

- Move body of style fix into new worker module
- Add Dockerfile/entrypoint defining worker container
- Allow log level to be specified in environment (SPACKBOTLOGLEVEL)
- Update gh workflow to build images for webservice and worker
- Fix branch name used to run pipelines (remove the 'github/' prefix)

Author: scottwittenburg

.env-dummy

@@ -1,6 +1,18 @@
 # Go to smee.io to generate a URL here
 SMEE_URL=https://smee.io/CHANGEME
 
+# Optionally customize redis host (local test server defined in docker-compose.yml)
+REDIS_HOST=rq-server
+
+# Optionally customize redis port
+REDIS_PORT=6379
+
+# Optionally configure time before jobs are killed and marked failed (in seconds, default 180s)
+WORKER_JOB_TIMEOUT=21600
+
+# Debug level (one of: "DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL")
+SPACKBOT_LOG_LEVEL=WARNING
+
 # You don't need to change this unless you change the docker-compose volumes
 GITHUB_PRIVATE_KEY=/app/spackbot/spack-bot-develop.private-key.pem
 

.github/workflows/build-deploy.yaml

@@ -12,19 +12,24 @@ on:
 jobs:
   deploy-test-containers:
     runs-on: ubuntu-latest
-    name: Build Spackbot Container
+    strategy:
+      fail-fast: false
+      # matrix: [tag, path to Dockerfile, label]
+      matrix:
+        dockerfile: [[spack-bot, ./Dockerfile, Spackbot],
+                     [spackbot-workers, ./workers/Dockerfile, "Spackbot Workers"]]
+    name: Build ${{matrix.dockerfile[2]}} Container
     steps:
       - name: Checkout
         uses: actions/checkout@v2        
 
       - name: Build and Run Test Container
         run: |
-            docker build -t ghcr.io/spack/spack-bot:latest .
-            docker tag ghcr.io/spack/spack-bot:latest ghcr.io/spack/spack-bot:${GITHUB_SHA::8} 
-
+            docker build -f ${{matrix.dockerfile[1]}} -t ghcr.io/spack/${{matrix.dockerfile[0]}}:latest .
+            docker tag ghcr.io/spack/${{matrix.dockerfile[0]}}:latest ghcr.io/spack/${{matrix.dockerfile[0]}}:${GITHUB_SHA::8}
       - name: Login and Deploy Test Container
         if: (github.event_name != 'pull_request')
         run: |
             docker images
             echo ${{ secrets.GITHUB_TOKEN }} | docker login ghcr.io -u ${{ secrets.GHCR_USERNAME }} --password-stdin
-            docker push --all-tags ghcr.io/spack/spack-bot
+            docker push --all-tags ghcr.io/spack/${{matrix.dockerfile[0]}}

Dockerfile

@@ -4,28 +4,8 @@ EXPOSE 8080
 
 # dependencies first since they're the slowest
 COPY requirements.txt .
-RUN pip3 install -r requirements.txt
-
-# make the bot trust GitHub's host key (and verify it)
-# If GitHub's fingerprint changes, update the code below with a new one:
-# https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/githubs-ssh-key-fingerprints
-# and rebuild this container.
-RUN ssh-keyscan -t rsa github.com 2>/dev/null > github.key \
-  && fingerprint=$(ssh-keygen -lf ./github.key | grep -o 'SHA256:[^ ]*') \
-  && if [ "$fingerprint" != "SHA256:nThbg6kXUpJWGl7E1IGOCspRomTxdCARLviKw6E5SY8" ]; \
-     then echo "GitHub key is invalid!" && exit 1; \
-     fi \
-  && mkdir -p /root/.ssh \
-  && cat ./github.key >> /root/.ssh/known_hosts
 
-# use identity file in /git_rsa (the mount point for our public/private keys)
-RUN \
-     echo "Host github.com"              >> /root/.ssh/config \
-  && echo "IdentityFile /git_rsa/id_rsa" >> /root/.ssh/config
-
-# ensure /root/.ssh has correct permissions
-RUN chmod -R go-rwx /root/.ssh
-RUN mkdir -p /git_rsa && chmod -R go-rwx /git_rsa
+RUN pip3 install -r requirements.txt
 
 # copy app in last so that everything above can be cached
 COPY spackbot /app/spackbot

docker-compose.yml

@@ -8,6 +8,25 @@ services:
       context: .
       dockerfile: smee/Dockerfile
 
+  rq-worker:
+    build:
+      context: .
+      dockerfile: workers/Dockerfile
+    env_file:
+      - ./.env
+    deploy:
+      replicas: 1
+
+  rq-server:
+    env_file:
+      - ./.env
+    image: redis:alpine
+    expose:
+      - ${REDIS_PORT}
+    volumes:
+      - redis-data:/data
+      - redis-conf:/usr/local/etc/redis/redis.conf
+
   spackbot:
     build:
       context: .
@@ -29,3 +48,7 @@ services:
       - ./.env
     links:
       - smee
+
+volumes:
+  redis-data:
+  redis-conf:

requirements.txt

@@ -1,9 +1,5 @@
 aiohttp
 gidgethub
 python_dotenv
+rq
 sh
-
-# Add these so we don't wait for install
-mypy
-flake8
-isort

spackbot/__main__.py

@@ -3,7 +3,6 @@
 #
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
-import logging
 import os
 
 import aiohttp
@@ -13,12 +12,12 @@
 from gidgethub import aiohttp as gh_aiohttp
 from .routes import router
 from .auth import authenticate_installation
+from .helpers import get_logger
 
 # take environment variables from .env file (if present)
 load_dotenv()
 
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger("spackbot")
+logger = get_logger(__name__)
 
 #: Location for authenticatd app to get a token for one of its installations
 INSTALLATION_TOKEN_URL = "app/installations/{installation_id}/access_tokens"
@@ -41,13 +40,19 @@ async def main(request):
     logger.info(f"Received event {event}")
 
     # get an installation token to make a GitHubAPI for API calls
-    token = await authenticate_installation(event.data)
+    installation_id = event.data["installation"]["id"]
+    token = await authenticate_installation(installation_id)
+
+    dispatch_kwargs = {
+        "installation_id": installation_id,
+        "token": token,
+    }
 
     async with aiohttp.ClientSession() as session:
         gh = gh_aiohttp.GitHubAPI(session, REQUESTER, oauth_token=token)
 
         # call the appropriate callback for the event
-        await router.dispatch(event, gh, session=session)
+        await router.dispatch(event, gh, session=session, **dispatch_kwargs)
 
     # return a "Success"
     return web.Response(status=200)

spackbot/auth.py

@@ -3,7 +3,6 @@
 #
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
-import logging
 import os
 import re
 import time
@@ -16,9 +15,6 @@
 
 load_dotenv()
 
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger("spackbot")
-
 #: Location for authenticatd app to get a token for one of its installations
 INSTALLATION_TOKEN_URL = "app/installations/{installation_id}/access_tokens"
 
@@ -79,14 +75,13 @@ async def renew_jwt():
     return await _tokens.get_token("JWT", renew_jwt)
 
 
-async def authenticate_installation(payload):
+async def authenticate_installation(installation_id):
     """Get an installation access token for the application.
 
     Renew the JWT if necessary, then use it to get an installation access
     token from github, if necessary.
 
     """
-    installation_id = payload["installation"]["id"]
 
     async def renew_installation_token():
         async with aiohttp.ClientSession() as session:

spackbot/comments.py

@@ -3,17 +3,23 @@
 #
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
+import io
 import random
+import traceback
 import spackbot.helpers as helpers
 
 
 async def tell_joke(gh):
     """
     Tell a joke to ease the PR tension!
     """
-    joke = await gh.getitem(
-        "https://official-joke-api.appspot.com/jokes/programming/random"
-    )
+    try:
+        joke = await gh.getitem(
+            "https://official-joke-api.appspot.com/jokes/programming/random"
+        )
+    except Exception:
+        return "To be honest, I haven't heard any good jokes lately."
+
     joke = joke[0]
     return f"> {joke['setup']}\n *{joke['punchline']}*\n😄️"
 
@@ -57,6 +63,31 @@ def get_style_message(output):
 """
 
 
+def get_style_error_message(e_type, e_value, tb):
+    """
+    Given job failure details, format an error message to post.  The
+    parameters e_type, e_value, and tb (for traceback) should be the same as
+    returned by sys.exc_info().
+    """
+    buffer = io.StringIO()
+    traceback.print_tb(tb, file=buffer)
+    tb_contents = buffer.getvalue()
+    buffer.close()
+
+    return f"""
+I encountered an error attempting to format style.
+<details>
+<summary><b>Details</b></summary>
+
+```bash
+Error: {e_type}, {e_value}
+Stack trace:
+{tb_contents}
+```
+</details>
+"""
+
+
 commands_message = f"""
 You can interact with me in many ways!
 
@@ -70,8 +101,14 @@ def get_style_message(output):
 If you need help or see there might be an issue with me, open an issue [here](https://github.com/spack/spack-bot/issues)
 """
 
-style_message = """
-It looks like you had an issue with style checks! To fix this, you can run:
+style_message = f"""
+It looks like you had an issue with style checks! I can help with that if you ask me!  Just say:
+
+`{helpers.botname} fix style`
+
+... and I'll try to fix style and push a commit to your fork with the fix.
+
+Alternatively, you can run:
 
 ```bash
 $ spack style --fix

spackbot/handlers/labels.py

@@ -4,10 +4,10 @@
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
 import re
-import logging
 
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger("spackbot")
+import spackbot.helpers as helpers
+
+logger = helpers.get_logger(__name__)
 
 
 #: ``label_patterns`` maps labels to patterns that tell us to apply the labels.
@@ -163,7 +163,18 @@ async def add_labels(event, gh):
             attr_matches = []
             # Pattern matches for for each attribute are or'd together
             for attr, patterns in pattern_dict.items():
-                attr_matches.append(any(p.search(file[attr]) for p in patterns))
+                # 'patch' is an example of an attribute that is not required to
+                # appear in response when listing pull request files.  See here:
+                #
+                #    https://docs.github.com/en/rest/pulls/pulls#list-pull-requests-files
+                #
+                # If we don't get some attribute in the response, no labels that
+                # depend on finding a match in that attribute should be added.
+                attr_matches.append(
+                    any(p.search(file[attr]) for p in patterns)
+                    if attr in file
+                    else False
+                )
             # If all attributes have at least one pattern match, we add the label
             if all(attr_matches):
                 labels.append(label)

spackbot/handlers/pipelines.py

@@ -3,14 +3,13 @@
 #
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
-import logging
 import os
 import urllib.parse
 import spackbot.helpers as helpers
 
 import aiohttp
 
-logger = logging.getLogger(__name__)
+logger = helpers.get_logger(__name__)
 
 # We can only make the request with a GITLAB TOKEN
 GITLAB_TOKEN = os.environ.get("GITLAB_TOKEN")
@@ -50,7 +49,7 @@ async def run_pipeline(event, gh):
 
     # We need the branch name plus number to assemble the GitLab CI
     branch = pr["head"]["ref"]
-    branch = f"github/pr{number}_{branch}"
+    branch = f"pr{number}_{branch}"
     branch = urllib.parse.quote_plus(branch)
 
     url = f"{helpers.gitlab_spack_project_url}/pipeline?ref={branch}"