Idea: Cleaning up unused translations by amazingphilippe · Pull Request #2635 · cds-snc/notification-admin

amazingphilippe · 2026-03-30T15:24:28Z

Summary | Résumé

We have a lot of translated content strings that we don't use anymore. Mostly from the days where content was hardcoded. Do we want to clean this up? This script can do it automatically, and we can always run make test-translations as a safety measure.

Test instructions | Instructions pour tester la modification

run make cleanup-translations
It might find unused translations, and prompt you to confirm deletion
run make test-translations
It should not find missing translations

+                            unused -= to_remove
+                            if not unused:
+                                return unused
+                    except Exception:


General fix: keep the non-fatal behavior (continue scanning) but replace the empty handler with explicit handling that records what failed and why.

Best fix without changing functionality: in find_unused_translations, change the empty except Exception: pass to print a warning to stderr containing the file path and exception message, then continue. This preserves current control flow (no re-raise, still skips bad files) while removing silent failure.

Specific edit needed:

File: scripts/cleanup_translations.py

Region: inside find_unused_translations, around lines 87–100

Change only the except Exception: block to emit a warning and continue.

No new imports are needed (sys is already imported).

Copilot

Pull request overview

Adds an interactive utility to identify and remove unused French translation strings by combining Babel extraction results with a fallback text search, and wires it into the Makefile.

Changes:

Introduces scripts/cleanup_translations.py to extract used strings via Babel and detect unused keys.
Adds an interactive deletion flow to remove unused rows from app/translations/csv/fr.csv.
Adds make cleanup-translations target to run the script.

Reviewed changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 8 comments.

File	Description
scripts/cleanup_translations.py	New script to detect and optionally delete unused translation strings using Babel + source search.
Makefile	Adds a new `cleanup-translations` target to run the cleanup script.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-30T15:27:46Z

+        return set()
+
+
+def find_unused_translations(keys, search_dirs):
+    # First, get keys that are used in code via Babel (covers Python/HTML)
+    babel_keys = get_keys_from_babel()


On any Babel extraction error, get_keys_from_babel() returns an empty set, which makes unused = set(keys) - babel_keys - extra_keys treat almost everything as unused and can lead to mass deletion. Safer behavior is to fail fast (raise/exit non-zero) or return a sentinel (e.g., None) and have find_unused_translations abort without offering deletion when extraction fails.

Suggested change

return set()

def find_unused_translations(keys, search_dirs):

# First, get keys that are used in code via Babel (covers Python/HTML)

babel_keys = get_keys_from_babel()

return None

def find_unused_translations(keys, search_dirs):

# First, get keys that are used in code via Babel (covers Python/HTML)

babel_keys = get_keys_from_babel()

if babel_keys is None:

print("Aborting: unable to extract keys via Babel; not proceeding with cleanup.")

sys.exit(1)

Copilot · 2026-03-30T15:27:47Z

+        subprocess.run(
+            ["poetry", "run", "pybabel", "extract", "-F", "babel.cfg", "-k", "_l", "-o", "/tmp/cleanup_messages.po", "."],
+            check=True,
+            capture_output=True,
+        )
+
+        # 2. Run po2csv
+        subprocess.run(
+            ["poetry", "run", "po2csv", "/tmp/cleanup_messages.po", "/tmp/cleanup_messages.csv"], check=True, capture_output=True
+        )


The script uses fixed /tmp/cleanup_messages.* paths and only cleans them up on the success path. This can clash across concurrent runs (e.g., parallel CI/jobs) and can leak temp files when any subprocess step fails. Use tempfile to create unique temp paths and ensure cleanup in a finally block (also consider specifying encoding='utf-8' when reading the temp CSV for consistent behavior).

Copilot · 2026-03-30T15:27:47Z

+
+        # 3. Read extracted keys
+        extracted_keys = set()
+        with open("/tmp/cleanup_messages.csv", newline="") as csvfile:


The script uses fixed /tmp/cleanup_messages.* paths and only cleans them up on the success path. This can clash across concurrent runs (e.g., parallel CI/jobs) and can leak temp files when any subprocess step fails. Use tempfile to create unique temp paths and ensure cleanup in a finally block (also consider specifying encoding='utf-8' when reading the temp CSV for consistent behavior).

Copilot · 2026-03-30T15:27:47Z

+        os.remove("/tmp/cleanup_messages.po")
+        os.remove("/tmp/cleanup_messages.csv")


The script uses fixed /tmp/cleanup_messages.* paths and only cleans them up on the success path. This can clash across concurrent runs (e.g., parallel CI/jobs) and can leak temp files when any subprocess step fails. Use tempfile to create unique temp paths and ensure cleanup in a finally block (also consider specifying encoding='utf-8' when reading the temp CSV for consistent behavior).

Copilot · 2026-03-30T15:27:48Z

+def get_translation_keys(csv_path):
+    keys = []
+    if not os.path.exists(csv_path):
+        return keys
+
+    with open(csv_path, mode="r", encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            if row["source"]:
+                if row["source"].startswith("!/!/"):
+                    continue
+                keys.append(row["source"])
+    return keys


The log message claims the script found 'unique' source strings, but get_translation_keys returns a list and can include duplicates. Either deduplicate (e.g., return a set or de-dupe before printing) or adjust the message to not claim uniqueness.

Copilot · 2026-03-30T15:27:48Z

+
+    print(f"Loading translations from {csv_path}...")
+    keys = get_translation_keys(csv_path)
+    print(f"Found {len(keys)} unique translation source strings.")


The log message claims the script found 'unique' source strings, but get_translation_keys returns a list and can include duplicates. Either deduplicate (e.g., return a set or de-dupe before printing) or adjust the message to not claim uniqueness.

Suggested change

print(f"Found {len(keys)} unique translation source strings.")

print(f"Found {len(keys)} translation source strings.")

Copilot · 2026-03-30T15:27:48Z

+    # Second pass: search remaining unused keys in JS and CSS files
+    for directory in search_dirs:
+        for root, dirs, files in os.walk(directory):
+            for file in files:
+                if file.endswith((".js", ".css", ".scss", ".txt", ".md")):
+                    file_path = os.path.join(root, file)
+                    if "translations/csv" in file_path or "node_modules" in file_path:
+                        continue
+                    try:
+                        with open(file_path, "r", encoding="utf-8") as f:
+                            content = f.read()
+
+                            to_remove = set()
+                            for key in unused:
+                                if key in content:
+                                    to_remove.add(key)
+
+                            unused -= to_remove
+                            if not unused:
+                                return unused


This does an O(number_of_files × number_of_candidate_keys) substring scan, which can get expensive if unused is large. A more efficient approach is to aggregate searchable content once (or stream files) and check each key once, or use a multi-pattern search approach (e.g., building an index, or a compiled matcher) to avoid re-scanning the same content for every key.

Copilot · 2026-03-30T15:27:48Z

+    # Also need to consider hardcoded constant keys used in test-translations.py
+    extra_keys = set(
+        [
+            "English Government of Canada signature",
+            "French Government of Canada signature",
+            "Empty",
+            "1 template",
+            "Number must have 10 digits",
+            "bad invitation link",
+            "invitation expired",
+            "password",
+            "Your service already uses ",
+            "Try again. Something’s wrong with this code",
+            "Code already sent, wait 10 seconds",
+            "You cannot delete a default email reply to address if other reply to addresses exist",
+            "Code has expired",
+            "Code already sent",
+            "Code has already been used",
+            "Code not found",
+            "as an email reply-to address.",
+            "You cannot remove the only user for a service",
+            "Cannot send to international mobile numbers",
+        ]
+    )


Hardcoding extra_keys creates a drift risk against the real source of truth (the constants used by test-translations). Consider loading these from a shared module/config (or importing the list from the test-translations implementation if it’s in Python), so future translation key additions don’t require updating multiple places.

github-actions · 2026-03-30T15:30:27Z

🧪 Review environment

https://bw5fulsda4vf5476xpfzrmclwy0yhdcl.lambda-url.ca-central-1.on.aws/

+            if temp_file and os.path.exists(temp_file):
+                try:
+                    os.remove(temp_file)
+                except Exception:


Use non-empty exception handling in the temp-file cleanup block by catching OSError (the expected deletion failure type) and logging a warning to stderr. This keeps functionality unchanged (script continues even if cleanup fails) while preserving diagnostic information.

Best fix in this snippet:

In scripts/cleanup_translations.py, in get_keys_from_babel() finally block (lines ~62–65), replace:

except Exception: pass

With:

except OSError as cleanup_error: print(..., file=sys.stderr)

No new imports are required because sys is already imported.

github-actions · 2026-07-08T14:22:27Z

🧪 Review environment

https://taylefytzpreqhynysprh63jvy0eqazj.lambda-url.ca-central-1.on.aws/

Cleaning up unused translations

6bc2864

amazingphilippe requested a review from Copilot March 30, 2026 15:24

Copilot started reviewing on behalf of amazingphilippe March 30, 2026 15:25 View session

github-advanced-security AI found potential problems Mar 30, 2026

View reviewed changes

Copilot AI reviewed Mar 30, 2026

View reviewed changes

amazingphilippe added 2 commits April 1, 2026 13:07

address copilot review

0c6d9a6

format

90595e4

github-advanced-security AI found potential problems Apr 1, 2026

View reviewed changes

format

2bbb58b

jzbahrai approved these changes Jul 8, 2026

View reviewed changes

amazingphilippe added 2 commits July 8, 2026 14:31

remove auto-generated files created by accident

3364c71

Merge branch 'main' into debt/unused-translation-script

b6193cd

amazingphilippe enabled auto-merge (squash) July 15, 2026 19:17

Merge branch 'main' into debt/unused-translation-script

92044d7

Navigation Menu

Uh oh!

Idea: Cleaning up unused translations - #2635

Idea: Cleaning up unused translations#2635
amazingphilippe wants to merge 7 commits into
mainfrom
debt/unused-translation-script

amazingphilippe commented Mar 30, 2026

Uh oh!

Check notice

Copilot Autofix

Copilot AI left a comment

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

Copilot AI Mar 30, 2026

Uh oh!

github-actions Bot commented Mar 30, 2026

Uh oh!

Check notice

Copilot Autofix

github-actions Bot commented Jul 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		os.remove("/tmp/cleanup_messages.po")
		os.remove("/tmp/cleanup_messages.csv")

	print(f"Found {len(keys)} unique translation source strings.")
	print(f"Found {len(keys)} translation source strings.")

Uh oh!

Conversation

amazingphilippe commented Mar 30, 2026

Summary | Résumé

Test instructions | Instructions pour tester la modification

Uh oh!

Check notice

Copilot Autofix

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 30, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Mar 30, 2026

🧪 Review environment

Uh oh!

Check notice

Copilot Autofix

github-actions Bot commented Jul 8, 2026

🧪 Review environment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants