Drupal testing
Practical Drupal testing patterns using curl smoke tests, drush eval, and test scripts. Use when verifying Drupal implementations actually work - module enabled, service exists, page renders, config correct.From its SKILL.md
npx -y skills add gkastanis/drupal-workflow --skill drupal-testingAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 18 stars18 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
7.9 KB, as published. Nobody here has run it
Drupal Testing Skill
Practical verification patterns for Drupal implementations.
Environment Detection
The testing patterns default to DDEV but support other environments. Detect and adapt:
# Auto-detect environment and set the execution prefix.
if command -v ddev >/dev/null 2>&1 && ddev describe >/dev/null 2>&1; then
DRUSH="ddev drush"
EXEC="ddev exec"
elif command -v lando >/dev/null 2>&1; then
DRUSH="lando drush"
EXEC="lando ssh -c"
else
DRUSH="drush"
EXEC="bash -c"
fi
All patterns below use ddev drush and ddev exec. Substitute with $DRUSH and $EXEC for non-DDEV environments.
Curl Smoke Test Patterns
Pattern 1: Status Code + Response Size (Same-Shell Auth)
The login URL and curl MUST run in the same shell so the session cookie is valid. Write a script file and run it inside ddev.
#!/bin/bash
# scripts/tests/verify-page-access.sh
# Run with: ddev exec bash scripts/tests/verify-page-access.sh
LOGIN_URL=$(drush uli --uid=1 --no-browser --uri=http://localhost 2>/dev/null)
COOKIE=$(curl -s -D - -o /dev/null -L "$LOGIN_URL" 2>/dev/null \
| grep -i 'set-cookie' | head -1 \
| sed 's/.*set-cookie: *//i' | cut -d';' -f1)
STATUS=$(curl -s -o /dev/null -w "%{http_code}" -b "$COOKIE" "http://localhost/TARGET_PATH")
echo "TARGET_PATH: $STATUS"
Why --uri=http://localhost: Without it, drush uli may generate a URL with a different domain (e.g., https://mysite.ddev.site), causing a cookie domain mismatch when curling http://localhost.
Pattern 2: Download Full HTML
ddev exec curl -s -b "$COOKIE_FILE" "http://localhost/TARGET_PATH" \
> /tmp/claude/page-output.html
# Then read /tmp/claude/page-output.html to inspect content
Pattern 3: Extract Specific Content
# Check for specific text/element on page
ddev exec curl -s -b "$COOKIE_FILE" "http://localhost/TARGET_PATH" \
| grep -o '<title>[^<]*</title>'
Reusable Template
#!/bin/bash
# scripts/tests/verify-PROJECT_NAME.sh
# Verify: DESCRIPTION
set -e
COOKIE_FILE="/tmp/cookies-$$.txt"
LOGIN_URL=$(ddev drush uli --uid=1 --no-browser 2>/dev/null)
ddev exec curl -s -c "$COOKIE_FILE" -L "$LOGIN_URL" -o /dev/null
STATUS=$(ddev exec curl -s -o /dev/null -w "%{http_code}" \
-b "$COOKIE_FILE" "http://localhost/TARGET_PATH")
if [ "$STATUS" = "200" ]; then
echo "PASS: TARGET_PATH returns 200"
else
echo "FAIL: TARGET_PATH returns $STATUS (expected 200)"
exit 1
fi
rm -f "$COOKIE_FILE"
DDEV Shell Gotchas
ddev exec passes commands through multiple shell layers, destroying pipes, variable expansion, nested quotes, and grep -P.
# BAD — variables get expanded/mangled by ddev exec shell layers
ddev exec "drush eval 'foreach ($groups as $g) { echo $g->id(); }'"
# GOOD — write a script file, run the file
cat > scripts/tests/my-check.php << 'EOF'
<?php
$groups = \Drupal::entityTypeManager()->getStorage('group')->loadMultiple();
foreach ($groups as $g) { echo $g->id() . "\n"; }
EOF
ddev exec drush scr scripts/tests/my-check.php
Rule: If the command has pipes (|), variables ($var), nested quotes, or multi-line PHP, write it to a file first.
Drush Eval Patterns
CRITICAL escaping rules:
- Use
Drupal::not\Drupal::(backslash gets eaten by shell) - Use
Exceptionnot\Exception - No
usestatements (not supported in eval context) - Keep PHP on one line
- Complex logic →
drush scr script.phpinstead - Redirect stderr:
2>/dev/null - Use single quotes around PHP code
Service Verification
ddev drush eval 'print json_encode(["exists" => Drupal::hasService("my_module.my_service")]);' 2>/dev/null
Module Enabled Check
ddev drush eval 'print json_encode(["enabled" => Drupal::moduleHandler()->moduleExists("my_module")]);' 2>/dev/null
Entity Field Check
ddev drush eval '$defs = Drupal::service("entity_field.manager")->getFieldDefinitions("node", "article"); print json_encode(["has_field" => isset($defs["field_custom"])]);' 2>/dev/null
Config Value Check
ddev drush eval 'print json_encode(Drupal::config("my_module.settings")->get());' 2>/dev/null
Permission Exists
ddev drush eval '$perms = Drupal::service("user.permissions")->getPermissions(); print json_encode(["exists" => isset($perms["administer my_module"])]);' 2>/dev/null
Route Exists
ddev drush eval 'try { $url = Drupal::service("url_generator")->generateFromRoute("my_module.page"); print json_encode(["route" => "exists", "url" => $url]); } catch (Exception $e) { print json_encode(["route" => "missing"]); }' 2>/dev/null
Test Script Creation
Directory Convention
scripts/tests/
index.md # Index of all test scripts
verify-*.sh # Feature verification scripts
check-*.sh # Quick check scripts
smoke-*.sh # Page smoke test scripts
Creating a Script
- Write to
scripts/tests/verify-<feature>.sh - Make executable:
chmod +x scripts/tests/verify-<feature>.sh - Update
scripts/tests/index.mdwith entry
Script Template
#!/bin/bash
# scripts/tests/verify-<feature>.sh
# Verifies: <what this tests>
# Created: <date>
set -e
echo "=== Verifying <feature> ==="
# Test 1: Check module enabled
RESULT=$(ddev drush eval 'print Drupal::moduleHandler()->moduleExists("my_module") ? "yes" : "no";' 2>/dev/null)
if [ "$RESULT" = "yes" ]; then
echo "PASS: Module enabled"
else
echo "FAIL: Module not enabled"
exit 1
fi
# Test 2: Check service exists
RESULT=$(ddev drush eval 'print Drupal::hasService("my_module.service") ? "yes" : "no";' 2>/dev/null)
if [ "$RESULT" = "yes" ]; then
echo "PASS: Service registered"
else
echo "FAIL: Service not found"
exit 1
fi
echo "=== All checks passed ==="
Common Verification Scenarios
| What to Verify | Method | Command |
|---|---|---|
| Module enabled | drush eval | Drupal::moduleHandler()->moduleExists("x") |
| Service exists | drush eval | Drupal::hasService("x.y") |
| Field on bundle | drush eval | getFieldDefinitions("node", "article") |
| Route accessible | curl | Status code 200 check |
| Config value | drush eval | Drupal::config("x.settings")->get("key") |
| Permission defined | drush eval | getPermissions() check |
| Cache clear works | drush | ddev drush cr exits 0 |
| Config imports | drush | ddev drush cim -y exits 0 |
Finding the Real File (Vendor vs Contrib)
Patched modules may live in both vendor/drupal/ (original) and web/modules/contrib/ (patched). Editing the wrong one has no effect.
// Find which file is actually loaded at runtime
$ref = new \ReflectionMethod($service, 'methodName');
echo $ref->getFileName();
Rule: When a patched module exists in both locations, use ReflectionMethod to find which file PHP actually loads. Edit that file, not the one you assume.
Role-Based Access Testing
Create dedicated test users rather than relying on existing users with unknown role combinations.
# Create test users
ddev drush user:create testauth --password=testauth123
ddev drush user:create testadmin --password=testadmin123
ddev drush user:role:add my_admin_role testadmin
Test matrix -- verify each user type against each resource state:
| User type | Published resource | Unpublished resource |
|---|---|---|
| Anonymous | view only | 403 |
| Authenticated | view only | 403 |
| Role-based admin | view + edit | view + edit |
| uid 1 | view + edit | view + edit (bypass) |
Check content, not just status codes: Download the HTML and grep for Drupal form IDs (edit-*) to verify what users actually see -- presence of edit forms, workflow fields, and local task tabs.