Class: TextExtractor

Inherits:
Object
  • Object
show all
Defined in:
lib/text_extractor.rb,
lib/text_extractor/skip.rb,
lib/text_extractor/guard.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, Guard, GuardError, InlineValue, Record, Skip, State, Value

Constant Summary collapse

INDENTED = 
{
  description: 'indented line',
  block: proc {
    /
    ^[^\n\S]+[^\n]*$
    /
  }
}.freeze
UNINDENTED = 
{
  description: 'unindented line',
  block: proc {
    /
    ^\S+[^\n]*$
    /
  }
}.freeze
DEFAULT = 
[
  INDENTED,
  UNINDENTED
].freeze

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(&block) ⇒ TextExtractor

rubocop: disable Metrics/MethodLength



14
15
16
17
18
19
20
21
22
23
24
25
26
27
 
# File 'lib/text_extractor.rb', line 14

def initialize(&block)
  raise "#{self.class}.new requires a block" unless block
  @values = {}
  @fill = {}
  @values = {}
  @records = []
  @filldowns = []
  @current_record_values = []
  @section_delimiter = nil
  @section_terminator = nil
  @append_guards = []
  instance_exec(&block)
  @append_guards.each { |g| guard(**g, &g[:block]) }
end

Instance Attribute Details

#records ⇒ Object (readonly)

Returns the value of attribute records.



11
12
13
 
# File 'lib/text_extractor.rb', line 11

def records
  @records
end

#values ⇒ Object (readonly)

Returns the value of attribute values.



11
12
13
 
# File 'lib/text_extractor.rb', line 11

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
  '0.5.0'
end

Instance Method Details

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



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

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

#factory(object = nil) ⇒ Object 



94
95
96
97
98
99
100
 
# File 'lib/text_extractor.rb', line 94

def factory(object = nil)
  if object
    @factory = object
  else
    @factory
  end
end

#filldown(**kwargs, &block) ⇒ Object 



102
103
104
105
 
# File 'lib/text_extractor.rb', line 102

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

#find_record_for(match) ⇒ Object 



107
108
109
 
# File 'lib/text_extractor.rb', line 107

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

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



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

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

#guard(**kwargs, &block) ⇒ Object 



111
112
113
114
 
# File 'lib/text_extractor.rb', line 111

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

#guards(*guard_args) ⇒ Object 



116
117
118
119
 
# File 'lib/text_extractor.rb', line 116

def guards(*guard_args)
  guard_args = Guards::DEFAULT if guard_args.empty?
  @append_guards = guard_args
end

#inline(id, &block) ⇒ Object 



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

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

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



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

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

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



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

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

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



77
78
79
 
# File 'lib/text_extractor.rb', line 77

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

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



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

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

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



81
82
83
84
85
86
87
 
# File 'lib/text_extractor.rb', line 81

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

#regexps ⇒ Object 



142
143
144
145
146
 
# File 'lib/text_extractor.rb', line 142

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

#scan(input) ⇒ Object 



121
122
123
124
125
126
 
# File 'lib/text_extractor.rb', line 121

def scan(input)
  prefill = {}
  sections(input).flat_map { |section|
    Extraction.new(section, self, prefill).scan.extraction_matches
  }
end

#section(delimiter, terminator = nil) ⇒ Object 



89
90
91
92
 
# File 'lib/text_extractor.rb', line 89

def section(delimiter, terminator = nil)
  @section_delimiter = delimiter
  @section_terminator = terminator
end

#sections(input) ⇒ Object 



128
129
130
131
132
133
134
135
 
# File 'lib/text_extractor.rb', line 128

def sections(input)
  return [input] unless @section_delimiter

  texts = input.split(@section_delimiter)
  return texts unless @section_terminator

  texts.map { |section| section + @section_terminator }
end

#skip(**kwargs, &block) ⇒ Object 



137
138
139
140
 
# File 'lib/text_extractor.rb', line 137

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

#to_re ⇒ Object 



148
149
150
 
# File 'lib/text_extractor.rb', line 148

def to_re
  Regexp.union(*regexps)
end

#value(id, re, &block) ⇒ Object 



45
46
47
48
49
50
51
 
# File 'lib/text_extractor.rb', line 45

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