Skip to content

Latest commit

 

History

History
171 lines (120 loc) · 5.34 KB

TRANSLATE.md

File metadata and controls

171 lines (120 loc) · 5.34 KB

How to translate Anki deck

This how-to describes the process of translating Anki deck.

What we gonna do is:

  • export deck into JSON;
  • extract needed information into a CSV using jq;
  • edit (e.g. translate) this information in Google Spreadsheets;
  • export the updated information into JSON array using Export Sheet Data Google Spreadsheets add-on;
  • apply it as a patch for the original JSON.

Prerequisites

  • Install Anki for Desktop application if it's not installed yet.

  • Install CrownAnki add-on for Anki. It provides additional Export/Import options to save/load decks using its own JSON format.

  • Install jq - a command-line JSON processor.

Guide

Export deck into JSON

Here you have two variants:

  • If the deck is distributed in CrowdAnki JSON, then just get its sources.

  • If it's a shared deck then install it into Anki for Desktop and export it to CrowdAnki JSON.

In both case you end up with a directory with a json file representing the deck's data and media/ directory with deck's media files.

Extract information into CSV

First you need to know the ordinal number of the field you want to update. Open the JSON-file you got on the previous step and find the notes section in there. Here goes a collection of deck's notes. Check out the first note and locate the field of your interest.

For example, let's pick the JSON-file from Ultimate Geography deck:

  "notes": [
    {
      "__type__": "Note", 
      "data": "", 
      "fields": [
        "England", 
        "Constituent country of the United Kingdom.",
        ...

Let's imagine we're translating country names only. Notice that they go in the 0'th position of the fields array. We'll be referencing it in our examples below.

Here is how you can export a field's data into CSV using jq:

$ { echo "data"; jq -r '. | .notes[] | .fields[OFFSET] | [.] | @csv' < JSON; } > CSV

where:

  • OFFSET is the index of the field we're exporting;
  • JSON is the original JSON-file;
  • CSV is a CSV-file to export data into.

This is how we export our country names:

$ { echo "name"; jq -r '. | .notes[] | .fields[0] | [.] | @csv' < Ultimate_Geography.json; } > CountryNames.csv

The resulting CountryNames.csv is a single-column plain CSV file with a list of countries:

$ cat CountryNames.csv | head -n 6
name
"England"
"Scotland"
"United Kingdom"
"Northern Ireland"
"France"

Now let's translate them.

Editing CSV data

You can use any tool to edit your CSV data but since Google Spreadsheets can export data back into JSON it's reasonable to use it.

Continuing our example, check out this shared spreadsheet with Country names. The first CountryNames sheet contains the original data imported from the CSV file. The second RU sheet contains translated country names. Translation is performed automatically using =GoogleTranslate() function.

Export new data into JSON

To export data into JSON we will use Export Sheet Data Google Spreadsheets add-on.

Switch to the appropriate sheet and open Export Sheet Data's sidebar:

Add-ons > Export Sheet Data > Open Sidebar

Ensure that Select Format field is set to "JSON" and tick the following checkboxes:

  • [x] Export value arrays in the JSON group
  • [x] Export contents as array in the Advanced JSON group

Click Export button to download the JSON data or Visualize button to preview it first.

The resulting JSON-file should be just a plain array:

$ cat RU.json | head -n 6
[
  "Англия",
  "Шотландия",
  "Великобритания",
  "Северная Ирландия",
  "Франция",

Patching the original JSON-file

To apply the patch we will use jq one more time. In a generic form our command look like:

$ jq -s '
         .[1] as $array
           |
         .[0]
           |
         .notes 
           |= reduce range(0;length) as $i (.;
               .[$i].fields[OFFSET] = $array[$i])' \
    JSON PATCH > PATCHED

where:

  • OFFSET is the index of the field we're updating;
  • JSON is the original JSON-file;
  • PATCH is our patch file;
  • PATCHED is a patched version of the original JSON-file.

Back to our example, this is real command which applies countries translation to Ultimate Geography deck:

$ jq --indent 4 -s '
         .[1] as $array
           |
         .[0]
           |
         .notes 
           |= reduce range(0;length) as $i (.;
               .[$i].fields[0] = $array[$i])' \
    Ultimate_Geography.json RU.json | sed 's/,$/, /' > Ultimate_Geography_ru.json

Here -- indent 4 and sed 's/,$/, /' augmentations make the final JSON look almost the same as the original one which simplifies real patching.