parse_dictionaries

The following blog post contains some backgroud about this repo: Reverse-Engineering Apple Dictionary.

Parsing with reverse_data.py

Parses the great Apple Dictionary (for now tested with the New Oxfor American Dictionary).

Here is what the built-in Dictionary app gives for "handle":

And here is what this script gives (on a Mac), with

#
# NOTE: might be at a different location for you!
# If so, you can find it using a glob, e.g.,
# ls /System/Library/AssetsV2/ \
#   com_apple_MobileAsset_DictionaryServices_dictionaryOSX/*/*
# then take the one that has `New Oxford American Dictionary`!
#
# New Oxford American Dictionary
NOAD='/System/Library/AssetsV2/ \
       com_apple_MobileAsset_DictionaryServices_dictionaryOSX/ \
       4094df88727a054b658681dfb74f23702d3c985e.asset/ \
       AssetData/ \
       New Oxford American Dictionary.dictionary/ \
       Contents/Resources/Body.data'

python reverse_data.py \ 
        --dictionary_path $NOAD --lookup handle --output_path lookup/lookup.html

Extracting words and definitions from a book with extract.py

If you want to split a book into all its words and look them all up, you can use extract.py. This relies on nltk to properly get definitions, e.g., to turn "he builds houses" into ["he", "build", "house"].

python extract.py PATH_TO_BOOK.txt PATH_TO_OUTPUT.zip

The resulting zip file contains a single file master.json, which contains three keys. Example:

{
 "definitions": {
   "cozen": "<d:entry xmlns:d=...",
   "house": "<d:entry xmlns:d=...",
   "related": "<d:entry xmlns:d=...",
   "rod": "<d:entry xmlns:d=...",
   "...": "..."
 },
 "links": {
   "vitals": "vital",
   "...": "..."
 },
 "scores": {
   "cozen": 0.5,
   "house": 1.0,
   "related": 10.0,
   "rod": 20.0,
   "...": "..."
 }
}

• definitions are just definitions of all words.
• links contains links to definitions, if a word in the book does not have it's own definition.
• scores is a crude estimate for how likely it is that the reader knows a word, see the _get_scores in extract.py.