feat: add option to exclude specific files from language stats totals

Co-authored-by: aider (o3) <aider@aider.chat>

scripts/language_stats.py

@@ -30,6 +30,12 @@ if not BASE_URL or not TOKEN or not USERNAME:
 session = requests.Session()
 session.headers["Authorization"] = f"token {TOKEN}"
 
+# List of files whose byte-counts should be excluded from the totals.
+# Each tuple: (repository_name, file_path, language_name)
+EXCLUDE_FILES: list[tuple[str, str, str]] = [
+    ("ar-scavenger-hunt", "static/aframe.js", "JavaScript"),
+]
+
 # ---------- collect repositories ----------
 page = 1
 repos = []
@@ -57,6 +63,20 @@ for repo in repos:
     for lang, bytes_ in langs.items():
         totals[lang] += bytes_
 
+# ---------- subtract excluded files ----------
+for repo_name, file_path, language in EXCLUDE_FILES:
+    try:
+        meta = session.get(
+            f"{BASE_URL}/api/v1/repos/{USERNAME}/{repo_name}/contents/{file_path}",
+            timeout=30,
+        )
+        if meta.ok:
+            size = meta.json().get("size", 0)
+            totals[language] = max(0, totals[language] - size)
+    except requests.RequestException:
+        # Ignore failures – continue with best-effort stats
+        pass
+
 total_bytes = sum(totals.values())
 if total_bytes == 0:
     sys.exit("No language data returned from API.")