Update upload_suggestions.py

upload_suggestions.py

@@ -7,11 +7,11 @@ from huggingface_hub import HfApi
 
 # Settings
 HF_TOKEN = os.environ.get("HF_TOKEN")
-REPO_ID = "axxam/libretranslate-suggestions"  # Dataset repo on HF
+REPO_ID = "axxam/libretranslate-suggestions"
 DEST_PATH_IN_REPO = "suggestions.json"
 REPO_TYPE = "dataset"
-JSON_OUTPUT_PATH = "/tmp/suggestions.json"
-CHECKSUM_FILE = "/tmp/.last_suggestions_checksum"
+JSON_OUTPUT_PATH = "/app/suggestions/suggestions.json"
+CHECKSUM_FILE = "/app/suggestions/.last_suggestions_checksum"
 
 # Known or expected locations for the DB
 possible_paths = [
@@ -22,12 +22,11 @@ possible_paths = [
 ]
 
 def find_db():
-    print(f"Running in CWD: {os.getcwd()}")
     for path in possible_paths:
         if os.path.exists(path):
             print(f"Found suggestions.db at {path}")
             return path
-    print("suggestions.db not found in any known path.")
+    print("suggestions.db not found.")
     return None
 
 def extract_suggestions(db_path):
@@ -61,15 +60,24 @@ def merge_with_existing(suggestions, json_path):
                 for item in json.load(f):
                     existing[item["id"]] = item
         except Exception as e:
-            print(f"Failed to read existing JSON: {e}")
+            print(f"Error loading existing JSON: {e}")
 
+    # Add only new suggestions
+    new_data_added = False
     for s in suggestions:
-        existing[s["id"]] = s
+        if s["id"] not in existing:
+            existing[s["id"]] = s
+            new_data_added = True
 
+    if not new_data_added:
+        print("No new suggestions — nothing to update.")
+        return False
+
+    # Only write if new data was added
     with open(json_path, "w", encoding="utf-8") as f:
         json.dump(list(existing.values()), f, indent=2, ensure_ascii=False)
-
-    return json_path
+    print(f"{len(suggestions)} suggestions processed. JSON updated.")
+    return True
 
 def get_checksum(filepath):
     if not os.path.exists(filepath):
@@ -86,7 +94,7 @@ def upload_if_updated(json_path):
             old_checksum = f.read().strip()
 
     if new_checksum != old_checksum:
-        print("Uploading updated suggestions.json to Hugging Face...")
+        print("New checksum detected — uploading to Hugging Face.")
         try:
             api = HfApi()
             api.upload_file(
@@ -98,11 +106,11 @@ def upload_if_updated(json_path):
             )
             with open(CHECKSUM_FILE, "w") as f:
                 f.write(new_checksum)
-            print(f"Upload successful at {datetime.now().isoformat()}")
+            print(f"Upload completed at {datetime.now().isoformat()}")
         except Exception as e:
             print("Upload failed:", e)
     else:
-        print("No changes in suggestions.json — skipping upload.")
+        print("Checksum unchanged — skipping upload.")
 
 def main():
     if not HF_TOKEN:
@@ -115,11 +123,12 @@ def main():
 
     suggestions = extract_suggestions(db_path)
     if not suggestions:
-        print("No suggestions found — skipping.")
+        print("No suggestions found.")
         return
 
-    json_path = merge_with_existing(suggestions, JSON_OUTPUT_PATH)
-    upload_if_updated(json_path)
+    updated = merge_with_existing(suggestions, JSON_OUTPUT_PATH)
+    if updated:
+        upload_if_updated(JSON_OUTPUT_PATH)
 
 if __name__ == "__main__":
     main()