Class: ScientificNameParser

Inherits:
Object
  • Object
show all
Defined in:
lib/biodiversity/parser.rb

Overview

we can use these expressions when we are ready to parse virus names class VirusParser

def initialize
  @order     = /^\s*[A-Z][a-z]\+virales/i
  @family    = /^\s*[A-Z][a-z]\+viridae|viroidae/i
  @subfamily = /^\s*[A-Z][a-z]\+virinae|viroinae/i
  @genus     = /^\s*[A-Z][a-z]\+virus|viroid/i
  @species   = /^\s*[A-z0-9u0391-u03C9\[\] ]\+virus|phage|viroid|satellite|prion[A-z0-9u0391-u03C9\[\] ]\+/i
  @parsed    = nil
end

end

Constant Summary collapse

VERSION = 
open(File.join(File.dirname(__FILE__), '..', '..', 'VERSION')).readline.strip

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize ⇒ ScientificNameParser

Returns a new instance of ScientificNameParser.



69
70
71
72
73
74
75
 
# File 'lib/biodiversity/parser.rb', line 69

def initialize
  @verbatim = ''
  @clean = ScientificNameCleanParser.new
  @dirty = ScientificNameDirtyParser.new
  @canonical = ScientificNameCanonicalParser.new
  @parsed = nil
end

Class Method Details

.all(verbatim = @verbatim) ⇒ Object 



105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
 
# File 'lib/biodiversity/parser.rb', line 105

def @parsed.all(verbatim = @verbatim)
  parsed = self.class != Hash
  res = { :parsed => parsed, :parser_version => ScientificNameParser::VERSION}
  if parsed
    hybrid = self.hybrid rescue false
    res.merge!({
      :verbatim => @verbatim,
      :normalized => self.value,
      :canonical => self.canonical,
      :hybrid => hybrid,
      :details => self.details,
      :parser_run => self.parser_run,
      :positions => self.pos
      })
  else
    res.merge!(self)
  end
  res = {:scientificName => res}
  res
end

.all_json ⇒ Object 



130
131
132
 
# File 'lib/biodiversity/parser.rb', line 130

def @parsed.all_json
  self.all.to_json rescue ''
end

.pos_json ⇒ Object 



126
127
128
 
# File 'lib/biodiversity/parser.rb', line 126

def @parsed.pos_json
  self.pos.to_json rescue ''
end

.verbatim=(a_string) ⇒ Object 



101
102
103
 
# File 'lib/biodiversity/parser.rb', line 101

def @parsed.verbatim=(a_string)
  @verbatim = a_string
end

Instance Method Details

#parse(a_string) ⇒ Object 



89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
 
# File 'lib/biodiversity/parser.rb', line 89

def parse(a_string)
  @verbatim = a_string
  a_string = PreProcessor::clean(a_string)
  
  if virus?(a_string)
    @parsed = { :verbatim => a_string, :virus => true }
  elsif unknown_placement?(a_string)
    @parsed = { :verbatim => a_string }
  else
    @parsed = @clean.parse(a_string) || @dirty.parse(a_string) || @canonical.parse(a_string) || { :verbatim => a_string }
  end

  def @parsed.verbatim=(a_string)
    @verbatim = a_string
  end

  def @parsed.all(verbatim = @verbatim)
    parsed = self.class != Hash
    res = { :parsed => parsed, :parser_version => ScientificNameParser::VERSION}
    if parsed
      hybrid = self.hybrid rescue false
      res.merge!({
        :verbatim => @verbatim,
        :normalized => self.value,
        :canonical => self.canonical,
        :hybrid => hybrid,
        :details => self.details,
        :parser_run => self.parser_run,
        :positions => self.pos
        })
    else
      res.merge!(self)
    end
    res = {:scientificName => res}
    res
  end
  
  def @parsed.pos_json
    self.pos.to_json rescue ''
  end
  
  def @parsed.all_json
    self.all.to_json rescue ''
  end

  @parsed.verbatim = @verbatim
  @parsed.all
end

#parsed ⇒ Object 



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

def parsed
  @parsed
end

#unknown_placement?(a_string) ⇒ Boolean

Returns:

  • (Boolean) 


81
82
83
 
# File 'lib/biodiversity/parser.rb', line 81

def unknown_placement?(a_string)
  !!(a_string.match(/incertae\s+sedis/i) || a_string.match(/inc\.\s*sed\./i))
end

#virus?(a_string) ⇒ Boolean

Returns:

  • (Boolean) 


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

def virus?(a_string)
  !!(a_string.match(/\sICTV\s*$/) || a_string.match(/\b(virus|viruses|phage|phages|viroid|viroids|satellite|satellites|prion|prions)\b/i))
end