Script to migrate reading history and highlights from Readera (reads from an unzipped backup file) to Librera (writes to a profile directory), a pair of Android reading apps.

readera-to-librera.py

#!/usr/bin/env python3

import hashlib
import json
import logging
import os.path
from pathlib import Path


LOG = logging.getLogger(__name__)


PROGRESS = {
    "cp": False,  # crop pages
    "d": 0,  # delta
    "dc": False,  # double pages cover
    "dp": False,  # double pages normal
    "lk": 1,  # lock (0 = None, 1 = Yes, 2 = None)
    # "p": 1,  # progress ratio
    "s": 120,  # scroll speed
    "sp": False,  # split pages
    # "t": 1682461584539,  # time
    "x": 0,  # x-axis offset
    "y": 0,  # y-axis offset
    "z": 100  # zoom percent
}

BOOKMARK = {
    "isF": False,
    # "p": 0.54811203,  # position ratio
    # "path": "$aroot/path/to/file.epub",
    # "t": 1682461584539,  # time
    # "text": "quoted text",
}


def sha1(target: Path):
    with open(target, "rb") as f:
        return hashlib.sha1(f.read()).hexdigest()


def main(sroot: Path, aroot: Path, rdir: Path, ldir: Path, hfile: Path):
    LOG.info("Loading Readera library")
    with open(rdir / "library.json") as f:
        readera = json.load(f)
    LOG.info("Loading Librera profile")
    with open(ldir / "app-Progress.json") as f:
        librera_progress = json.load(f)
    with open(ldir / "app-Recent.json") as f:
        librera_recent = json.load(f)
    with open(ldir / "app-Bookmarks.json") as f:
        librera_bookmarks = json.load(f)
    hashes: dict[str, Path]
    try:
        LOG.info("Loading hashes")
        with open(hfile) as f:
            hashes = {k: Path(v) for k, v in json.load(f).items()}
    except:
        LOG.exception("Regenerating hashes")
        hashes = {}
        for path in sroot.rglob("*"):
            if path.is_file():
                relpath = path.relative_to(sroot)
                LOG.debug(relpath)
                hashes[sha1(path)] = relpath
        LOG.info("Saving hashes")
        with open(hfile, "w") as f:
            json.dump(hashes, f, default=lambda o: str(o))
    LOG.info("Processing Readera library")
    for doc in readera["docs"]:
        try:
            path = hashes[doc["data"]["doc_sha1"]]
        except KeyError:
            continue
        if not doc["data"]["doc_have_read_time"]:
            continue
        position = json.loads(doc["data"]["doc_position"])
        if position["ratio"] <= 0:
            continue
        lpath = os.path.join(aroot, path)
        LOG.debug(lpath)
        time = max(
            doc["data"]["doc_have_read_time"],
            doc["data"]["doc_last_read_time"],
        )
        progress = dict(PROGRESS)
        progress["p"] = position["ratio"]
        progress["t"] = time
        librera_progress[path.name] = progress
        recent = {
            "path": lpath,
            "time": time,
        }
        librera_recent.append(recent)
        for citation in doc["citations"]:
            bookmark = dict(BOOKMARK)
            bookmark["p"] = citation["note_index"]
            bookmark["path"] = lpath
            bookmark["t"] = citation["note_insert_time"]
            bookmark["text"] = citation["note_body"]
            librera_bookmarks[citation["note_insert_time"]] = bookmark
    LOG.info("Saving Librera profile")
    with open(ldir / "app-Progress.json", "w") as f:
        json.dump(librera_progress, f)
    with open(ldir / "app-Recent.json", "w") as f:
        json.dump(librera_recent, f)
    with open(ldir / "app-Bookmarks.json", "w") as f:
        json.dump(librera_bookmarks, f)


if __name__ == "__main__":

    import sys

    logging.basicConfig(level=logging.DEBUG)

    main(*(Path(arg) for arg in sys.argv[1:]))