Class: TextExtractor

Inherits:
Object
  • Object
show all
Defined in:
lib/text_extractor.rb,
lib/text_extractor/value.rb,
lib/text_extractor/record.rb,
lib/text_extractor/version.rb,
lib/text_extractor/filldown.rb,
lib/text_extractor/directives.rb,
lib/text_extractor/extraction.rb,
lib/text_extractor/inline_value.rb,
lib/text_extractor/directives/group.rb,
lib/text_extractor/directives/classes.rb

Overview

represents an extractor definition

Defined Under Namespace

Modules: Patterns Classes: Directives, Extraction, Filldown, InlineValue, Record, State, Value

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(&block) ⇒ TextExtractor

Returns a new instance of TextExtractor.



11
12
13
14
15
16
17
18
19
20
 
# File 'lib/text_extractor.rb', line 11

def initialize(&block)
  raise "#{self.class}.new requires a block" unless block
  @values = {}
  @fill = {}
  @values = {}
  @records = []
  @filldowns = []
  @current_record_values = []
  instance_exec(&block)
end

Instance Attribute Details

#records ⇒ Object (readonly)

Returns the value of attribute records.



9
10
11
 
# File 'lib/text_extractor.rb', line 9

def records
  @records
end

#values ⇒ Object (readonly)

Returns the value of attribute values.



9
10
11
 
# File 'lib/text_extractor.rb', line 9

def values
  @values
end

Class Method Details

.expand_directives(re) ⇒ Object 



7
8
9
 
# File 'lib/text_extractor/directives.rb', line 7

def self.expand_directives(re)
  Directives.new(re).expand
end

.version ⇒ Object 



2
3
4
 
# File 'lib/text_extractor/version.rb', line 2

def self.version
  '0.3.0'
end

Instance Method Details

#boolean(id, re = Patterns::BOOLEAN) ⇒ Object 



49
50
51
 
# File 'lib/text_extractor.rb', line 49

def boolean(id, re = Patterns::BOOLEAN)
  value(id, re) { |val| !val.match(Patterns::FALSE) }
end

#filldown(**kwargs, &block) ⇒ Object 



80
81
82
83
 
# File 'lib/text_extractor.rb', line 80

def filldown(**kwargs, &block)
  raise "#{self.class}.filldown requires a block" unless block
  record(Filldown, **kwargs, &block)
end

#find_record_for(match) ⇒ Object 



85
86
87
 
# File 'lib/text_extractor.rb', line 85

def find_record_for(match)
  records[records.length.times.find_index { |i| match["__#{i}"] }]
end

#float(id, re = Patterns::FLOAT) ⇒ Object 



57
58
59
 
# File 'lib/text_extractor.rb', line 57

def float(id, re = Patterns::FLOAT)
  value(id, re) { |val| Float(val) }
end

#inline(id, &block) ⇒ Object 



45
46
47
 
# File 'lib/text_extractor.rb', line 45

def inline(id, &block)
  @values[id] = InlineValue.new(id, &block)
end

#integer(id, re = Patterns::INTEGER) ⇒ Object 



53
54
55
 
# File 'lib/text_extractor.rb', line 53

def integer(id, re = Patterns::INTEGER)
  value(id, re) { |val| Integer(val) }
end

#ipaddr(id, re = Patterns::IPADDR) ⇒ Object 



65
66
67
 
# File 'lib/text_extractor.rb', line 65

def ipaddr(id, re = Patterns::IPADDR)
  value(id, re) { |val| IPAddr.new(val) }
end

#ipnetaddr(id, re = Patterns::IPNETADDR) ⇒ Object 



69
70
71
 
# File 'lib/text_extractor.rb', line 69

def ipnetaddr(id, re = Patterns::IPNETADDR)
  value(id, re) { |val| IPAddr.new(val) }
end

#rational(id, re = Patterns::RATIONAL) ⇒ Object 



61
62
63
 
# File 'lib/text_extractor.rb', line 61

def rational(id, re = Patterns::RATIONAL)
  value(id, re) { |val| Rational(val) }
end

#record(klass = Record, **kwargs, &block) ⇒ Object 



73
74
75
76
77
78
 
# File 'lib/text_extractor.rb', line 73

def record(klass = Record, **kwargs, &block)
  raise "#{self.class}.record requires a block" unless block
  kwargs[:extractor_values] = values
  kwargs[:values] = @current_record_values = []
  @records << klass.new(instance_exec(&block), **kwargs)
end

#regexps ⇒ Object 



93
94
95
96
97
 
# File 'lib/text_extractor.rb', line 93

def regexps
  @records.map.with_index do |record, i|
    Regexp.new("(?<__#{i}>#{record.source})", record.options)
  end
end

#scan(input) ⇒ Object 



89
90
91
 
# File 'lib/text_extractor.rb', line 89

def scan(input)
  Extraction.new(input, self).scan.extraction_matches
end

#to_re ⇒ Object 



99
100
101
 
# File 'lib/text_extractor.rb', line 99

def to_re
  Regexp.union(*regexps)
end

#value(id, re, &block) ⇒ Object 



37
38
39
40
41
42
43
 
# File 'lib/text_extractor.rb', line 37

def value(id, re, &block)
  val = @values[id] = Value.new(id, re, &block)
  define_singleton_method(id) do
    @current_record_values << val
    "(?<#{id}>#{re.source})"
  end
end