marxbook/misc/ff_import.py

import sqlite3
from marxbook import Store
from pathlib import Path
import sys
import logging

logger = logging.getLogger()


def is_nongeneric_url(url):
    """Returns True for URLs which are non-http and non-generic.

    Parameters
    ----------
    url : str
        URL to scan.

    Returns
    -------
    bool
        True if URL is a non-generic URL, False otherwise.
    """

    ignored_prefix = ["about:", "apt:", "chrome://", "file://", "place:"]

    for prefix in ignored_prefix:
        if url.startswith(prefix):
            return True

    return False


def load_firefox_database(store: Store, path):
    """Connect to Firefox sqlite db and import bookmarks into BukuDb.

    Parameters
    ----------
    path : str
        Path to Firefox bookmarks sqlite database.
    """

    path = Path(path).expanduser()
    # Connect to input DB
    if sys.version_info >= (3, 4, 4):
        # Python 3.4.4 and above
        conn = sqlite3.connect("file:%s?mode=ro" % path, uri=True)
    else:
        conn = sqlite3.connect(path)

    cur = conn.cursor()
    res = cur.execute(
        "SELECT DISTINCT fk, parent, title FROM moz_bookmarks WHERE type=1"
    )
    # get id's and remove duplicates
    for fk, parent_id, bm_title in res.fetchall():
        # get the url
        res = cur.execute("SELECT url FROM moz_places where id={}".format(fk))
        url = res.fetchone()[0]
        if is_nongeneric_url(url):
            continue

        # get tags
        res = cur.execute(
            "SELECT parent FROM moz_bookmarks WHERE "
            "fk={} AND title IS NULL".format(fk)
        )
        bm_tag_ids = [tid for item in res.fetchall() for tid in item]

        bookmark_tags = []
        for bm_tag_id in bm_tag_ids:
            res = cur.execute(
                "SELECT title FROM moz_bookmarks WHERE id={}".format(bm_tag_id)
            )
            bookmark_tags.append(res.fetchone()[0])

        # add folder name
        folder: list = []
        while parent_id:
            res = cur.execute(
                "SELECT title,parent FROM moz_bookmarks "
                "WHERE id={}".format(parent_id)
            )
            parent = res.fetchone()
            if parent:
                title, parent_id = parent
                if title:
                    folder.insert(0, title)
        folder_name = "/".join(folder).lstrip("/")

        # get the title
        if not bm_title:
            bm_title = ""
        print(f'store.add({folder_name}, url={url}, title={bm_title}, tags={bookmark_tags})')
        store.add(folder_name, url=url, title=bm_title, tags=bookmark_tags)
    try:
        cur.close()
        conn.close()
    except Exception:
        logger.exception("Couldnt close FF db")


if __name__ == "__main__":
    s = Store("~/.local/share/marxbook/bookmarks/")
    load_firefox_database(s, sys.argv[1])