booksentences2anki.py

#!/usr/bin/env python3

from lxml import html
import requests
from bs4 import BeautifulSoup
import sys
import itertools
import genanki
import glob
import shutil
import os.path
import re
import hashlib
import struct

def writeCsvLine(f, t1, t2):
    f.write("\"{}\",\"{}\"\n".format(t1, t2))


origin_language = sys.argv[1].upper()
target_language = sys.argv[2].upper()

url = "https://www.goethe-verlag.com/book2"

target_url = f"{url}/{origin_language}/{origin_language}{target_language}/{origin_language}{target_language}"

def pad_number(n):
    if n < 10:
        return "00" + str(n)
    elif n < 100:
       return "0" + str(n)
    else:
        return str(n)

my_model = genanki.Model(
    1091735104,
    "50Languages_Import",
    fields=[
        {"name": "L2"},
        {"name": "L1"},
        {"name": "Audio"},  # ADD THIS
    ],
    templates=[
        {
            "name": "Card 1",
            "qfmt": "{{L2}}",  # AND THIS
            "afmt": '{{FrontSide}}<hr id="answer">{{L1}}<br>{{Audio}}',
        },
    ],
    css=""".card {
 font-family: arial;
 font-size: 20px;
 text-align: center;
 color: black;
 background-color: white;
}

.card1 { background-color: #FFFFFF; }
.card2 { background-color: #FFFFFF; }"""
)

# unique id for deck based on language pair and type -- probably overkill
h = hashlib.sha256()
h.update(origin_language.encode())
h.update(target_language.encode())
h.update(b'sentences')

my_deck = genanki.Deck(
    struct.unpack("<L", h.digest()[0:4])[0], f"Book2 {origin_language}-{target_language} (sentences)"
)

MIN_LESSON = 3 # 2 is the index page
MAX_LESSON = 102 # 103 is the youtube video

history = {} 
f = open(origin_language+target_language+".csv", "wt")
writeCsvLine(f, "Target language", "Native Language")

for i in range(MIN_LESSON, MAX_LESSON + 1):
    r = requests.get(f"{target_url}{pad_number(i)}.HTM") # no slash unlike vocab scraping
    soup = BeautifulSoup(r.content, "html.parser")

    # header
    header_l1_sentences = [t.text for t in soup.find_all("span", {"class": "Stil36"})]
    header_l2_sentences = [t.text for t in soup.find_all("span", {"class": "Stil46"})]
    l2_audio = [t.find_all("source")[0]["src"] for t in soup.find_all("audio")]

    body_l1_sentences = [t.text.strip() for t in soup.find_all("div", {"class": "Stil35"})][:18] # last element is some text about Alzheimer
    body_l2_sentences = [t.text.strip().split('\r\n\n')[1] for t in soup.find_all("div", {"class": "Stil45"})]

    l1_sentences = header_l1_sentences + body_l1_sentences
    l2_sentences = header_l2_sentences + body_l2_sentences

    for l1_s, l2_s, m in zip(l1_sentences, l2_sentences, l2_audio):

        l1_s = l1_s.lstrip()
        l1_s = l1_s.rstrip()
        l2_s = l2_s.lstrip()
        l2_s = l2_s.rstrip()

        # patch numbers -- hopefully this is sufficient
        if re.match(r"\d{1,} ?\[", l2_s):
            l2_s = l2_s.split('[', 1)[1].split(']')[0]
            l1_s = l1_s.split('[', 1)[1].split(']')[0]

        # avoid duplicates
        if l2_s in history:
            continue

        history[l2_s] = 1

        writeCsvLine(f, l2_s, l1_s)

        filename = f"sentence_{origin_language}{target_language}_" + m.split("/")[-1]

        if not os.path.isfile(filename):
            dl_file = requests.get(m, stream=True)
            print(m)
            with open(filename, "wb") as out_file:
                shutil.copyfileobj(dl_file.raw, out_file)

        my_note = genanki.Note(
            model=my_model, fields=[l2_s, l1_s, f"[sound:{filename}]"]
        )

        my_deck.add_note(my_note)

f.close()
my_package = genanki.Package(my_deck)
my_package.media_files = [m for m in glob.glob(f"sentence_{origin_language}{target_language}_*.mp3")]
my_package.write_to_file(f"book2_{origin_language}{target_language}_sentences.apkg")

If you need additional http parameters, you can enter them after the url in the requests.get call. You do need to provide the 2 character language codes for origin and target language which is generate a path (eg: https://www.goethe-verlag.com/book2/EM/EMJA/EMJA003.HTM). Admittedly, I haven't touched this since May and was mostly written because I don't like production decks or isolated vocabulary decks.