Class: WordDataSource

Inherits:
BaseDataSource show all
Defined in:
lib/data/word_data_source.rb

Direct Known Subclasses

DelimitedWordDataSource

Instance Attribute Summary collapse

Attributes inherited from BaseDataSource

#startOffset

Instance Method Summary collapse

Methods inherited from BaseDataSource

#each_with_index, #extendWith, #has_terminator?, #nextDataSourceValueAt, #valueSequence

Constructor Details

#initialize(filePath, regex = "/[^a-z0-9\-\s]/i") ⇒ WordDataSource

Returns a new instance of WordDataSource.



6
7
8
9
10
11
12
13
14
15
16
17
18
19
 
# File 'lib/data/word_data_source.rb', line 6

def initialize(filePath, regex = "/[^a-z0-9\-\s]/i")
  @filePath = filePath
  @words = []
  @regex = regex
  File.open(filePath, "r") do |file|
    file.each_line do |line|
      line.chomp!
      if (self.process(line)) then
        break
      end
    end
  end
  @numberWordsInFile = @words.length
end

Instance Attribute Details

#numberWordsInFile ⇒ Object (readonly)

Returns the value of attribute numberWordsInFile.



4
5
6
 
# File 'lib/data/word_data_source.rb', line 4

def numberWordsInFile
  @numberWordsInFile
end

#words ⇒ Object (readonly)

Returns the value of attribute words.



4
5
6
 
# File 'lib/data/word_data_source.rb', line 4

def words
  @words
end

Instance Method Details

#numberValues ⇒ Object 



21
22
23
 
# File 'lib/data/word_data_source.rb', line 21

def numberValues
  return @words.length
end

#preprocessLine(line) ⇒ Object 



38
39
40
 
# File 'lib/data/word_data_source.rb', line 38

def preprocessLine(line)
  line.downcase.gsub(@regex, ' ')
end

#process(line) ⇒ Object 



25
26
27
28
 
# File 'lib/data/word_data_source.rb', line 25

def process(line)
  line = self.preprocessLine(line)
  return self.processData(line.split)
end

#processData(data) ⇒ Object 



30
31
32
33
34
35
36
 
# File 'lib/data/word_data_source.rb', line 30

def processData(data)
  data.each do |word|
    word = word.chomp(",")
    @words << word
  end
  return false
end

#toString(startOffset, endOffset) ⇒ Object 



47
48
49
50
51
52
53
54
55
56
57
 
# File 'lib/data/word_data_source.rb', line 47

def toString(startOffset, endOffset)
  if (endOffset == -1) then
    result = "#{@words[startOffset]} ..*"
  else
    result = ""
    (startOffset..endOffset).each do |offset|
      result += "#{@words[offset]} "
    end
  end
  result
end

#valueAt(offset) ⇒ Object 



42
43
44
45
 
# File 'lib/data/word_data_source.rb', line 42

def valueAt(offset)
  return @words[offset] if (offset < @numberWordsInFile)
  return nil
end