Skip to content

Processing Anki's .apkg files.md

Latest commit

 

History

History
277 lines (256 loc) · 10 KB

Processing Anki's .apkg files.md

File metadata and controls

277 lines (256 loc) · 10 KB

Processing Anki's .apkg files

What is this?

This document explains how the processing of the apkg archives should work.

Structure of .apkg files

The Anki .apkg files are just zip archives containing the following files:

  • collection.anki2: An sqlite database containing cards, notes, notes models, etc...
  • Files named 1,2,3,4...: The images that the cards use
  • media: A json file containing mappings between images and it's original names (eg: file 1 is actually called latex-96c15f8a1af25e7a2eec64f7c6fedafe12363352.png in the cards).

The structure of collection.anki2

This is the actual database that contains all the notes and other useful information that we'll use to generate the viewer. There is a very detailed guide about what exactly the database contains. In this document we'll just describe what we will need to pull off from the database in order to generate the viewer.

Fetching notes

The notes are stored in the notes table. The columns that we need from that table are:

  • id: The unique identifier of the note
  • mid: The ID of the cards model (described later how to get it)
  • tags: The tags associated to the note
  • fld: The fields of the note. This needs to be feed into the cards model to generate the cards. It contains all the fields, separated by the 0x1f (31) character. The order of the fields is exactly the one presented in the note's model.

Fetching cards models

In order to be able to generate the cards from the notes, we need to fetch the cards model. The model tells us how many cards have to be generated from a single note, what's the HTML surrounding the note, how the fields of notes.fld have to be presented.

The model can be found in the models column of the col table. Note that the col table always contains only one record. The model is stored in a JSON format; this is an example of what a model's JSON with two models looks like:

{
  "1471435193999": {
    "vers": [],
    "name": "Istruzioni Assembly",
    "tags": [],
    "did": 1493040141981,
    "usn": 625,
    "req": [
      [ 0, "any", [ 0, 2]],
      [ 1, "any", [ 1, 2]]
    ],
    "flds": [
      {
        "name": "Istruzione",
        "media": [],
        "sticky": false,
        "rtl": false,
        "ord": 0,
        "font": "Arial",
        "size": 20
      },
      {
        "name": "Descrizione",
        "media": [],
        "sticky": false,
        "rtl": false,
        "ord": 1,
        "font": "Arial",
        "size": 20
      },
      {
        "name": "Architettura",
        "media": [],
        "sticky": true,
        "rtl": false,
        "ord": 2,
        "font": "Arial",
        "size": 20
      }
    ],
    "sortf": 0,
    "tmpls": [
      {
        "name": "Carta 1",
        "qfmt": "<div></div>\n<div>Nell'architettura <strong>{{Architettura}}</strong> che cosa fa l'istruzione?</div>\n<div>\n<pre>{{Istruzione}}</pre>\n</div>",
        "did": null,
        "bafmt": "",
        "afmt": "{{FrontSide}}\n\n<hr id=answer>\n\n{{Descrizione}}",
        "ord": 0,
        "bqfmt": ""
      },
      {
        "name": "Carta 2",
        "qfmt": "<div>Quale istruzione <strong>{{Architettura}}</strong> corrisponde alla descrizione?</div>\n<br>\n<div><em>{{Descrizione}}</em></div>",
        "did": null,
        "bafmt": "",
        "afmt": "{{FrontSide}}\n\n<hr id=answer>\n\n<pre>{{Istruzione}}</pre>",
        "ord": 1,
        "bqfmt": ""
      }
    ],
    "mod": 1493150692,
    "latexPost": "\\end{document}",
    "type": 0,
    "id": "1471435193999",
    "css": ".card {\n font-family: arial;\n font-size: 20px;\n text-align: center;\n color: black;\n background-color: white;\n}\n",
    "latexPre": "\\documentclass[12pt]{article}\n\\special{papersize=3in,5in}\n\\usepackage[utf8]{inputenc}\n\\usepackage{amssymb,amsmath}\n\\pagestyle{empty}\n\\setlength{\\parindent}{0in}\n\\begin{document}\n"
  },
  "1471435194000": {
    "vers": [],
    "name": "Base (default)",
    "tags": [],
    "did": 1492955368330,
    "usn": 668,
    "req": [ [ 0, "all", [ 0 ] ] ],
    "flds": [
      {
        "name": "Fronte",
        "media": [],
        "sticky": false,
        "rtl": false,
        "ord": 0,
        "font": "Arial",
        "size": 20
      },
      {
        "name": "Retro",
        "media": [],
        "sticky": false,
        "rtl": false,
        "ord": 1,
        "font": "Arial",
        "size": 20
      }
    ],
    "sortf": 0,
    "tmpls": [
      {
        "name": "Carta 1",
        "qfmt": "{{Fronte}}",
        "did": null,
        "bafmt": "",
        "afmt": "{{FrontSide}}\n<hr id=answer>\n{{Retro}}",
        "ord": 0,
        "bqfmt": ""
      }
    ],
    "mod": 1494091839,
    "latexPost": "\\end{document}",
    "type": 0,
    "id": 1471435194000,
    "css": ".card {\n font-family: Arial;\n font-size: 20px;\n text-align: center;\n color: black;\n background-color: white;\n}\n\nimg {\n  vertical-align: middle;\n  padding-bottom: 7px;\n  padding-right: 6px;\n  padding-left: 6px\n}",
    "latexPre": "\\documentclass[12pt]{article}\n\\special{papersize=3in,5in}\n\\usepackage[utf8]{inputenc}\n\\usepackage{amssymb,amsmath}\n\\pagestyle{empty}\n\\setlength{\\parindent}{0in}\n\\begin{document}\n"
  }
}

In the same URL mentioned before there is a very nice description of what each field of the model does. Here will be described only the ones we need to create the cards browser.

In the root of the object, there is a "1471435193999" key. That key is the id of that model. This is mentioned in the mid column of the notes table.

Inside that object there is an array with key "flds". This array contains the fields that the cards specify when creating notes. Usually there are just "Front" and "Rear" fields, however, depending on the model, it's possible to have more different fields. As mentioned before, the notes store this information in the notes.fld column; the fields of the column are in the same order of the fields definitions in this array.

The "tmpls" object contains templates of the notes and provides information about the HTML structure of the cards. Models that define multiple templates generate multiple cards, one for each template. This object has two important fields:

  • "qfmt": The HTML of the front of the card
  • "afmt": The HTML of the rear of the card
  • "ord": The position of the template in the "tmpls" array. This is used by the cards table to identify the template that generated the card.

These fields can contain {{placeholders}} that have to be substituted with the values of the fields that the cards store in notes.fld. The association between field position and field name is the one of the objects in "flds". Note that "afmt" can contain a special placeholder called {{FrontSide}}; this is just a handy way of showing the content of the card without having to copy & paste the contents of "qfmt" while creating cards models.

The last field of our interest is "css"; it contains the CSS code that have to be applied to the card.

Fetching deck information

The information about the decks contained in the archive is stored in the col.decks column in JSON format. Here is a sample JSON:

{
  "1": {
    "desc": "",
    "name": "Predefinito",
    "extendRev": 50,
    "usn": 0,
    "collapsed": false,
    "newToday": [ 0, 0],
    "timeToday": [ 0, 0],
    "dyn": 0,
    "extendNew": 10,
    "conf": 1,
    "revToday": [ 0, 0],
    "lrnToday": [ 0, 0],
    "id": 1,
    "mod": 1494099120
  },
  "1493040141981": {
    "extendRev": 50,
    "collapsed": false,
    "newToday": [ 262, 0],
    "timeToday": [ 262, 158105],
    "dyn": 0,
    "extendNew": 10,
    "conf": 1,
    "revToday": [ 262, 11],
    "lrnToday": [ 262, 0],
    "id": 1493040141981,
    "mod": 1494073733,
    "name": "Universit\u00e0 - Calcolatori::Assembly",
    "usn": 661,
    "browserCollapsed": true,
    "mid": 1471435194000,
    "desc": ""
  },
  "1492955368330": {
    "extendRev": 50,
    "collapsed": false,
    "newToday": [ 262, 30],
    "timeToday": [ 262, 1384824],
    "dyn": 0,
    "extendNew": 10,
    "conf": 1,
    "revToday": [ 262, 25],
    "lrnToday": [ 262, 14],
    "id": 1492955368330,
    "mod": 1494095504,
    "name": "Universit\u00e0 - Calcolatori",
    "usn": 671,
    "browserCollapsed": true,
    "mid": 1471435194000,
    "desc": ""
  }
}

Some fields that should be noted:

  • id: Unique id of the deck. The id is used in the cards table, in the did column to specify the associations between cards and decks (explained later). Note that the default deck has always id 1.
  • name: The name of the deck. In case of child decks, the name will be in the format "Parent deck name::Child deck name"

Fetching cards

The information about cards are stored in the cards table. The columns that we need from that table are:

  • id: unique identifier of the card
  • nid: id of the note that generated this card
  • did: id identifier of the deck where this card is contained
  • ord: the position of the template of the model; it matches the template's ord field.

The structure of media file

media is a JSON file that associates the code-names of the images to its' original names. Note that any card's field can contain images, but the images are linked with their complete name. During the creation of the .apkg file anki compresses the names of the images (actually it just renames these with progressive numbers), however the cards still keep referring to the original name. The purpose of the media file it make us capable to translate numeric names in the images's original names.

The structure of media is pretty straight-forward:

{
    "2": "paste-61641370632193.jpg", 
    "3": "paste-81522774245377.jpg", 
    "1": "latex-0cc8b5131ccb25b20258394ebcf13773bb8b2d19.png", 
    "4": "paste-13327283519489.jpg"
}

Where the keys are the names of the compressed files and the string values are the original names.