21 lines
1.0 KiB
Smalltalk
21 lines
1.0 KiB
Smalltalk
|
accessing
|
||
|
|
wordcloudDataLanguage: language
|
||
|
|
|
||
|
|
| stopwords stopwordsCapitalized occurrencesWords wordAndValue |
|
||
|
|
stopwords := (TweetsCollection dataStore / 'commons' / ('stopwords-', language, '.txt')) contents splitOn: Character lf.
|
||
|
|
stopwordsCapitalized := stopwords collect: [:each | each first asString asUppercase, each allButFirst asLowercase ].
|
||
|
|
occurrencesWords := ((((self writeWordsFile contents) splitOn: ' ') asBag asDictionary)
|
||
|
|
associations asSortedCollection: [:x :y | x value > y value]) asOrderedDictionary.
|
||
|
|
occurrencesWords removeKeys: stopwords.
|
||
|
|
occurrencesWords removeKeys: stopwordsCapitalized.
|
||
|
|
occurrencesWords removeKey: ''.
|
||
|
|
|
||
|
|
occurrencesWords size > 50 ifTrue: [
|
||
|
|
occurrencesWords := (occurrencesWords associations copyFrom: 1 to: 50) asOrderedDictionary ].
|
||
|
|
|
||
|
|
wordAndValue := OrderedCollection new.
|
||
|
|
occurrencesWords keysAndValuesDo: [ :k :v |
|
||
|
|
wordAndValue add: ('{name:', ($' asString), k, ($' asString), ',value:', v asString, '}')
|
||
|
|
].
|
||
|
|
^ {'[', ((',' join: wordAndValue) copyWithout: Character lf), ']'.
|
||
|
|
occurrencesWords}
|