Import Omnivore reading list into Wallabag

omnivore-to-wallabag.py

"""Convert exported Omnivore reading list to Wallabag v2 JSON format.

This is by no means well checked or perfect, but it should serve as a start for further efforts.
It got my list of articles imported into Wallabag, which is all I wanted.
"""

import json
import csv
from datetime import datetime
import glob


def convert_to_wallabag_json(omnivore_data):
    wallabag_entries = []
    for item in omnivore_data:
        entry = {
            "is_archived": 1 if item["state"] == "Archived" else 0,
            "is_starred": 0,
            "tags": item["labels"],
            "is_public": False,
            "id": None,  # Wallabag will assign a new ID
            "title": item["title"],
            "url": item["url"],
            "given_url": item["url"],
            "content": item.get("description", ""),  # Use description as content if available
            "created_at": datetime.fromisoformat(item["savedAt"].replace('Z', '+00:00')).strftime("%Y-%m-%d %H:%M:%S"),
            "updated_at": datetime.fromisoformat(item["updatedAt"].replace('Z', '+00:00')).strftime("%Y-%m-%d %H:%M:%S"),
            "published_by": [],
            "annotations": [],
            "domain_name": item["url"].split("//")[-1].split("/")[0],
            "preview_picture": item.get("thumbnail", "")
        }
        if item["publishedAt"] is not None:
            entry["published_at"] = datetime.fromisoformat(item["publishedAt"].replace('Z', '+00:00')).strftime("%Y-%m-%d %H:%M:%S")
        wallabag_entries.append(entry)
    return wallabag_entries


# Read Omnivore JSON data
omnivore_data = []
for f in glob.glob('metadata_*.json'):
    with open(f, 'r') as f:
        omnivore_data.extend(json.load(f))

# Convert to Wallabag format
wallabag_data = convert_to_wallabag_json(omnivore_data)

# Write to JSON file
with open('wallabag_import.json', 'w', encoding='utf-8') as f:
    json.dump(wallabag_data, f, ensure_ascii=False, indent=2)

print("Conversion complete. Output saved to 'wallabag_import.json'")