words-to-readlang 1.3.0

Convert vocabulary exports from Pod101, Language Reactor, and more to Readlang CSV format

words-to-readlang

A Python library and CLI tool for converting vocabulary exports from language learning platforms (Pod101, Language Reactor, etc.) into the CSV format accepted by Readlang's word import.

It also comes with a web interface for uploading, previewing, editing, and exporting vocabulary — with automatic example sentence fetching from Tatoeba and the Leipzig Corpora Collection.

Background

Readlang is a reading-focused vocabulary tool with spaced-repetition flashcards. It supports importing words via CSV, but the format it expects is specific, and the exports from other learning platforms don't match it out of the box. This library bridges that gap.

Installation

pip install words-to-readlang

Or install from source:

git clone https://codeberg.org/psy-q/words-to-readlang
cd words-to-readlang
pip install -e .

To use the web interface, install with the web extras:

pip install -e ".[web]"

Supported Input Formats

Pod101 (FinnishPod101, SpanishPod101, etc.)

Pod101 language learning sites let Premium subscribers save words to a Word Bank, which can be exported as a CSV file. The export is a simple two-column file (Word, English) encoded in UTF-16, which is automatically detected.

Language Reactor

Language Reactor is a browser extension for learning languages through Netflix and YouTube. Pro subscribers can save words and export them as a tab-separated file from the Saved Items panel. This format includes the base/dictionary form, translations, and subtitle sentences.

CLI Usage

Commands

words-to-readlang convert      Convert a vocabulary file to Readlang CSV
words-to-readlang serve        Start the web interface
words-to-readlang list-formats List all available input formats

Basic Conversion

# From Pod101
words-to-readlang convert input.csv output.csv --format pod101

# From Language Reactor
words-to-readlang convert saved.csv output.csv --format languagereactor

# Short alias
words-to-readlang convert saved.csv output.csv --format lr

If your input has more than 200 words (Readlang's import limit), the output is automatically split into multiple files: output (part 1).csv, output (part 2).csv, etc.

Example Sentence Lookup

Language Reactor saves words in their inflected form as found in subtitles. Readlang requires the example sentence to contain the exact word, so mismatches are detected and the context is cleared.

The --fetch flag fetches replacement example sentences automatically. It tries Tatoeba first, then falls back to the Leipzig Corpora Collection if Tatoeba has no match.

words-to-readlang convert input.csv output.csv --format languagereactor --fetch

This is also useful for Pod101 exports, which never include example sentences.

By default the source language is Finnish (--lang fin). Pass the ISO 639-3 code for other languages:

words-to-readlang convert words.csv output.csv --format pod101 --fetch --lang swe

convert options

Option	Short	Default	Description
--format	-f	(required)	Input format: pod101, languagereactor, or lr
--fetch		off	Fetch missing examples from Tatoeba / Leipzig
--lang		fin	ISO 639-3 source language code
--version			Print version and exit

List Available Formats

words-to-readlang list-formats

Web Interface

# Development server (localhost only)
words-to-readlang serve

# Custom host/port
words-to-readlang serve --port 8080

# Bind to all interfaces with auto-reload
words-to-readlang serve --host 0.0.0.0 --port 8080 --debug

serve options:

Option	Short	Default	Description
--host	-h	127.0.0.1	Address to bind to
--port	-p	5000	Port to listen on
--debug		off	Enable Flask debug mode with auto-reload
--metrics-port		0 (off)	Expose Prometheus metrics on this port
--metrics-host		127.0.0.1	Address for the metrics endpoint

See README_WEB.md for full web interface and Docker deployment documentation.

Programmatic API

from pathlib import Path
from words_to_readlang import Converter

# Basic usage
converter = Converter()
output_files = converter.convert(
    input_path=Path("words.csv"),
    output_path=Path("readlang.csv"),
    parser_name="pod101"
)

# With example sentence lookup
converter = Converter(fetch_examples=True, src_lang="fin")
output_files = converter.convert(
    input_path=Path("saved_items.csv"),
    output_path=Path("readlang.csv"),
    parser_name="languagereactor"
)

print(f"Created {len(output_files)} file(s)")

Advanced: Custom Processing

from pathlib import Path
from words_to_readlang import get_parser, ReadlangWriter, ExampleFetcher

# Parse input
parser = get_parser("languagereactor")
entries = parser.parse(Path("saved.csv"))

# Filter or modify entries
entries = [e for e in entries if len(e.word) > 3]

# Write output with example lookup
fetcher = ExampleFetcher(delay=1.0, verbose=True)
writer = ReadlangWriter(example_fetcher=fetcher, src_lang="fin")
output_files = writer.write(entries, Path("output.csv"), fetch_examples=True)

Adding Custom Parsers

from pathlib import Path
from typing import List
from words_to_readlang import Entry, register_parser

@register_parser("myformat")
class MyFormatParser:
    @property
    def name(self) -> str:
        return "myformat"

    @property
    def description(self) -> str:
        return "My custom vocabulary format"

    def parse(self, file_path: Path) -> List[Entry]:
        entries = []
        # ... parse file ...
        return entries

words-to-readlang convert input.txt output.csv --format myformat

Readlang CSV Format Reference

Column	Content	Notes
1	Word or phrase	/ separates synonyms
2	Translation	/ separates alternatives
3	Context sentence (optional)	Must contain the exact word from column 1
4	Practice interval in days (optional)
5	Next practice date (optional)	YYYY-MM-DD

• No header row
• No newlines within cells
• Maximum 200 words per file

Development

git clone https://codeberg.org/psy-q/words-to-readlang
cd words-to-readlang
pip install -e ".[dev]"

# Run tests
pytest

# Type check
ty check

# Format / lint
black src tests
ruff check src tests

License

MIT