food_fish_parser 0.2.1 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 94e64c5de393e971e5bbf4da925f19b4437142b23707fa40d85ddb7b7a1c5c9b
-  data.tar.gz: 2e546183d0fbcf49a8365b36ce93085f33ce6720d2ec630a09e0eb621bb9ddc0
+  metadata.gz: e92312ff6eec700a2c57b1c48cad77fb9710dd3086b2fa4ded31e352d0bef9c5
+  data.tar.gz: 404c5692f4a731e1614c9c4f8628ddf01f599192740e205c4e9f064411d6e18a
 SHA512:
-  metadata.gz: f2b425d44262ed2e86a6b3b4eb394c80a407d14ea7f37751cedcfd6f11e6c53c7c16c50e2290c12202638daa7995c2896674dac64b0d91e18f48137d14afd044
-  data.tar.gz: 31532dfd05906b070154b23f0c4bb55e2cdb6f8022a163df20525148b6513eff0a1f05804d5b356b542a78300dfc301cc8e189790909ad64868e8c96b46f9b15
+  metadata.gz: 2de059ea4900f0aea03194559b275f1c7b307c1c06dc6a0b02a1bf17c21bccdc5a8fce98ad6fa178ad99181ad0ab36d5a609c82edd960280c46efbc2f1e71cb7
+  data.tar.gz: 7fdc6287c1285c7868d784d771b0f983d0c4fb975e3fc82eeff52a2f80768b12455caebad5d487b4b1b7816126f33ec6c51e04e91d498bb128e500aa34e183d2

data/README.md

@@ -18,6 +18,8 @@ gem install food_fish_parser
 
 ## Example
 
+### Strict parser
+
 ```ruby
 require 'food_fish_parser'
 
@@ -28,7 +30,7 @@ s = <<EOT.gsub(/\n/, '').strip
   kabeljauw (gadus macrocephalus), vangstgebied stille oceaan fao 67, garnaal
   (litopenaeus vannamei), gekweekt in ecuador, kweekmethode: vijver.
 EOT
-parser = FoodFishParser::Parser.new
+parser = FoodFishParser::Strict::Parser.new
 puts parser.parse(s).to_a.inspect
 ```
 
@@ -74,13 +76,15 @@ Results in a list of detected fishes
 ]
 ```
 
+### Anywhere
+
 When you have a piece of text and don't know where (or if) any fish details are
 present, you can use the `anywhere` option.
 
 ```ruby
 require 'food_fish_parser'
 
-parser = FoodFishParser::Parser.new
+parser = FoodFishParser::Strict::Parser.new
 s = "tomaat, vis (zalm (salmo salar) gevangen in Noorwegen), zout"
 puts parser.parse(s, anywhere: true).to_a.inspect
 ```
@@ -104,6 +108,43 @@ While the parser would normally return nothing, with `anywhere` it returns:
 
 Please note that the `anywhere` option can make the parser much slower.
 
+### Flat parser
+
+While the strict parser can recognize the structure of multiple fishes, it is really
+strict about what it expects. Many cases are not recognized, or sometimes incomplete.
+
+The flat parser does basic named entity recognition anywhere in the text. Any structure
+is lost, so it always returns an array with one or zero items - but you get all the
+FAO regions and fish names found.
+
+```ruby
+require 'food_fish_parser'
+
+parser = FoodFishParser::Flat::Parser.new
+s = "Foobar zalm (salmo salar) *&! gevangen pangasius spp FAO 61 ?or ?FAO 67 what more.")
+puts parser.parse(s).to_a.inspect
+```
+
+```
+[
+  {
+    :names => [
+      { :common=>"zalm", :latin=>"salmo salar" },
+      { :common=>nil, :latin=>"pangasius spp" }
+    ],
+    :catch_areas => [
+      { :name=>nil, :fao_codes=>["61"] },
+      { :name=>nil, :fao_codes=>["67"] }
+    ],
+    :catch_methods       => [],
+    :aquaculture_areas   => [],
+    :aquaculture_methods => []
+  }
+]
+```
+
+This might be expanded to more information at some point.
+
 
 ## Test tool
 
@@ -120,6 +161,7 @@ Usage: bin/food_fish_parser [options] --file|-f <filename>
     -q, --[no-]quiet                 Only show summary.
     -p, --parsed                     Only show lines that were successfully parsed.
     -n, --noresult                   Only show lines that had no result.
+    -r, --parser PARSER              Use specific parser (strict, flat).
     -a, --[no-]anywhere              Search for fish details anywhere in the text.
     -e, --[no-]escape                Escape newlines
     -c, --[no-]color                 Use color

data/bin/food_fish_parser

@@ -72,9 +72,14 @@ verbosity = 1
 files = []
 strings = []
 print = nil
+parser_name = :strict
 anywhere = false
 escape = false
 color = true
+PARSERS = {
+  strict: FoodFishParser::Strict::Parser,
+  flat:   FoodFishParser::Flat::Parser
+}
 OptionParser.new do |opts|
   opts.banner = <<-EOF.gsub(/^    /, '')
     Usage: #{$0} [options] --file|-f <filename>
@@ -88,7 +93,8 @@ OptionParser.new do |opts|
   opts.on("-q", "--[no-]quiet", "Only show summary.") {|q| verbosity = q ? 0 : 1 }
   opts.on("-p", "--parsed", "Only show lines that were successfully parsed.") {|p| print = :parsed }
   opts.on("-n", "--noresult", "Only show lines that had no result.") {|p| print = :noresult }
-  opts.on("-a", "--[no-]anywhere", "Search for fish details anywhere in the text.") {|a| anywhere = !!a }
+  opts.on("-r", "--parser PARSER", "Use specific parser (#{PARSERS.keys.join(", ")}).") {|p| parser_name = p&.downcase&.to_sym }
+  opts.on("-a", "--[no-]anywhere", "Search for fish details anywhere in the text (only strict parser).") {|a| anywhere = !!a }
   opts.on("-e", "--[no-]escape", "Escape newlines") {|e| escape = !!e }
   opts.on("-c", "--[no-]color", "Use color") {|e| color = !!e }
   opts.on("-v", "--[no-]verbose", "Show more data (parsed tree).") {|v| verbosity = v ? 2 : 1 }
@@ -103,7 +109,10 @@ OptionParser.new do |opts|
 end.parse!
 
 if strings.any? || files.any?
-  parser = FoodFishParser::Parser.new
+  unless parser = PARSERS[parser_name]&.new
+    STDERR.puts("Please specify one of the known parsers: #{PARSERS.keys.join(", ")}.")
+    exit(1)
+  end
   success = true
   strings.each {|s| success &= parse_single(s, parser: parser, verbosity: verbosity, print: print, escape: escape, color: color, anywhere: anywhere) }
   files.each   {|f| success &= parse_file(f, parser: parser, verbosity: verbosity, print: print, escape: escape, color: color, anywhere: anywhere) }

data/lib/food_fish_parser/flat/fao_region.rb

@@ -0,0 +1,35 @@
+module FoodFishParser
+  module Flat
+    module FaoRegion
+
+      REGEX_CODE = / 0?\d{2} (?: (?: -0?\d{2} | (?: \/ | - | \s+ | ) [lIV]+ ) [a-d]? )? /ix
+      REGEX = /
+        \b FA[O0]
+        \s* (?: -? (?:zone|gebied) )? \s* (?:nummer|nr\.?|,)?
+        \s*
+        (
+          #{REGEX_CODE}
+          (?:
+            (?: ,?\s*en\s* #{REGEX_CODE} ) |
+            (?: \/ #{REGEX_CODE} )+ |
+            (?: (?: ,\s* #{REGEX_CODE} )+ (?: ,?\s*en\s* #{REGEX_CODE} )? )
+          )?
+        )
+        \b
+      /ix
+
+      def self.find_all_code_groups(text)
+        text
+          .scan(REGEX)
+          .flatten(1)
+          .map {|s| s.split(/\s*(?: en | ,\s*en | , | \/ )\s*/xi) }
+          .map {|a| a.map {|s| s.gsub(/\A0(\d{2})\z/, '\1') } } # normalize trailing zeroes
+      end
+
+      def self.find_all(text)
+        find_all_code_groups(text)
+          .map {|a| { name: nil, fao_codes: a } }
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/fish_name.rb

@@ -0,0 +1,34 @@
+require_relative 'fish_name_latin'
+require_relative 'fish_name_nl'
+
+module FoodFishParser
+  module Flat
+    module FishName
+      REGEX_VIS = /\s*(?: \(\s*vis\s*\) | \b,?\s*vis\s*,?\s*\b )/ix
+
+      REGEX = /
+        (?:
+          #{FishNameNL::REGEX} (?: #{REGEX_VIS} )? \s* \( \s* #{FishNameLatin::REGEX} \s* \) |
+          #{FishNameNL::REGEX} \s*,?\s* \b #{FishNameLatin::REGEX} |
+          #{FishNameLatin::REGEX} |
+          #{FishNameNL::REGEX}
+        )
+      /ix
+
+      def self.find_all(text)
+        # Because scan doesn't support named captures, we have to use numbered capture groups.
+        # Make sure to keep all groups you don't want to reference below as non-capturing groups.
+        # Each name regex has a capture group (so as to avoid noise), so you don't see them here.
+        # The order of the captures corresponds to the order of the fish names in the regex above.
+        text.scan(REGEX).map do |m|
+          case
+          when m[0] && m[1] then { common: m[0], latin: m[1] }
+          when m[2] && m[3] then { common: m[2], latin: m[3] }
+          when m[4]         then { common: nil,  latin: m[4] }
+          when m[5]         then { common: m[5], latin: nil  }
+          end
+        end.compact
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/fish_name_latin.rb

@@ -0,0 +1,17 @@
+# autogenerated by species-treetop-gen-latin.rb on 2020-03-19
+module FoodFishParser
+  module Flat
+    module FishNameLatin
+      REGEX_FIRST = /zygochlamys|zeus|xiphopenaeus|xiphias|undaria|ulva|trichiurus|trachurus|todarodes|thunnus|theragra|stolephorus|sprattus|spirulina|sparus|solea|sepiella|sepia|sebastes|scomber|sardinella|sardina|salmo|saccharina|reinhardtius|psetta|procambarus|portunus|porphyra|pollachius|pleuronectes|pleoticus|placopecten|phymatolithon|perna|penaeus|penaeidae|pelvetia|pecten|patinopecten|parapenaeopsis|paralomis|paphia|pangasius|pandalus|palmaria|pagellus|ovalipes|ostrea|oreochromis|oncorhynchus|octopus|nephrops|nemipterus|nelumbo|mytilus|mulinia|micromesistius|metapenaeus|merluccius|merlangius|melanogrammus|macruronus|macrocystis|lophius|loligo|litopenaeus|lithodes|limanda|lethrinus|lepidotrigla|lepidopsetta|lates|laminaria|katwonus|katwomus|katsuwonus|katsuwomus|illex|homarus|himanthalia|haematococcus|gracilaria|gelidium|gadus|fucus|euthynnus|ensis|engraulis|dunaliella|dosidicus|dicentrarchus|crassostrea|crangon|clupea|clarias|chondrus|chlorella|cerastoderma|caulerpa|ascophyllum|anguilla|anadara|alle|alaria|acipenser/i
+      REGEX_SECND = /yessoensis|vulgaris|virens|vesiculosus|verrucosa|vannamei|undulata|umbilicalis|tenera|stylifera|sprattus|spp\.|spp|sombrus|solea|solar|scombrus|santolla|salina|salar|ringens|pyrifera|pyrenoidosa|punctatus|productus|pluvialis|platessa|platensis|piscatorius|pinnatifida|pilchardus|pelatis|pelanis|pelamis|pelagicus|patagonica|pangasius|palmata|pacificus|officinalis|ocellatus|nucifera|novaezelandiae|novaezealandiae|norvegicus|nodosum|niloticus|nerka|mykiss|mykis|myki\?s|murphyi|muelleri|morhue|morhua|monodon|monoceros|microptera|merluccius|merlangus|merguiensis|maximus|maxima|marinus|magellanicus|macrocephalus|limanda|lepturus|lentillifera|latissima|lactuca|labrax|kroyeri|kisutch|keta|jordani|japonicus|japonica|hyptophthalmus|hypophtalmus|hippoglossoides|hexodon|harengus|gueldenstaedtii|granulosa|gorbusche|gorbuscha|gladius|gigas|gibbosa|gariepinus|galloprovincialis|faber|esculenta|encrasicolus|elongata|edulis|edule|directus|digitata|crispus|crangon|clarkii|chilensis|chalcogramma|capensis|cannamei|canaliculus|canaliculata|calcareum|borealis|bogaraveo|bilineata|australis|aurata|argentinus|antiquata|anquilla|anguilla|anchoita|americanus|alle|albacares|alalunga|aeglefinus/i
+
+      REGEX = /
+        \b
+        (
+          #{REGEX_FIRST} (?: \s+ #{REGEX_SECND} )?
+        )
+        \b
+      /ix
+    end
+  end
+end

data/lib/food_fish_parser/flat/fish_name_nl.rb

@@ -0,0 +1,20 @@
+# autogenerated by species-treetop-gen-nl.rb on 2020-03-19
+module FoodFishParser
+  module Flat
+    module FishNameNL
+      REGEX_AREA = /pacifische|noorse|indische|groenlandse|atlantische|argentijnse|alaska/i
+      REGEX_ATTR = /zwarte|zwart|witte|witpoot|wit|roze|rood|rode|rivier|pijl|kleine|klein|grote|groot|groene|groen|doorn|coho|chum|blauwe|blauw/i
+      REGEX_NAME = /zonnevis|zeewolf|zeesnoek|zeekreeft|zeeforel|zeebaars|zalm|wijting|weekdieren|weekdier|vintonijn|tonijn\ albacore|tonijn|tong|tilapia|thon\ albacore|tarbot|tapijtschelp|sprot|spie|snotolf|snoekbaars|snoek|skipjack\ tonijn|schol|schelvis|schelpen|schelp|schar|sardines|regenboogforel|raat|poon|pollock|pollak|pangasius|paling|oogtonijn|mul|mosselen|mossel|meerval|mantelschelp|makreel|lom|leng|kreeft|koolvis|kokkel|karper|kabeljauw|hondstong|hoki|heilbot|heek|hake|haai|ha|gruis|griet|geep|geelvintonijn|garnalen|garnaal|fint|coquilles|cocquilles|botervis|bot|beekridder|barracuda|baars|arkschelp|ansjovis|albacore\ tonijn/i
+      REGEX_SUFX = /vlees|ringen|ring|filets|filet/i
+
+      REGEX = /
+        (?: \b verse \s+ )?
+        \b
+        (
+          (?: #{REGEX_AREA} \s+ )? (?: #{REGEX_ATTR} \s* )? #{REGEX_NAME} (?: #{REGEX_SUFX} )?
+        )
+        \b
+      /ix
+    end
+  end
+end

data/lib/food_fish_parser/flat/kind.rb

@@ -0,0 +1,41 @@
+require_relative 'fish_name_nl'
+
+module FoodFishParser
+  module Flat
+    module Kind
+
+      REGEX_WILD = /
+        \b
+        (?:
+          gevangen |
+          visgebied |
+          vangstgebied |
+          vangsgebied |
+          betrapt \s+ bij |
+          wilde? \s+ #{FishNameNL::REGEX}
+        )
+        \b
+      /ix
+
+      REGEX_AQUAC = /
+        \b
+        (?:
+          gekweekt |
+          aquacultuurproduct |
+          aquacultuur \s+ product |
+          kweekmethode |
+          kweekmethoden
+        )
+        \b
+      /ix
+
+      def self.is_wild?(text)
+        !!REGEX_WILD.match(text)
+      end
+
+      def self.is_aquac?(text)
+        !!REGEX_AQUAC.match(text)
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/parser.rb

@@ -0,0 +1,52 @@
+require_relative 'fish_name'
+require_relative 'fao_region'
+require_relative 'kind'
+
+module FoodFishParser
+  module Flat
+    class Parser
+
+      # Create a new fish detail parser
+      # @return [FoodFishParser::Flat::Parser]
+      def initialize
+      end
+
+      # Parse food fish text into a structured representation.
+      #
+      # @param s [String] text to parse
+      # @return [Array<Hash>] structured representation of fish details (maximum one item)
+      def parse(s, **options)
+        names = FishName.find_all(s)
+        areas = FaoRegion.find_all(s)
+        methods = [] # @todo
+
+        is_wild = Kind.is_wild?(s)
+        is_aquac = Kind.is_aquac?(s)
+
+        return [] unless names.any? || areas.any?
+
+        attrs = {
+          names: names,
+          catch_areas: nil,
+          catch_methods: nil,
+          aquaculture_areas: nil,
+          aquaculture_methods: nil
+        }
+
+        if is_wild && !is_aquac
+          [attrs.merge(catch_areas: areas, catch_methods: methods)]
+        elsif !is_wild && is_aquac
+          [attrs.merge(aquaculture_areas: areas, aquaculture_methods: methods)]
+        elsif areas.any? || methods.any?
+          # We have a problem: either there are multiple fish and they're a mix of
+          # wild and aquaculture fish, or there is no such indication at all.
+          # For now, we return it in a non-standard way (this needs to be tackled).
+          [attrs.merge(areas: areas, methods: methods)]
+        else
+          # just names
+          [attrs]
+        end
+      end
+    end
+  end
+end

data/lib/food_fish_parser/{grammar → strict/grammar}/aquac_area.treetop

@@ -1,5 +1,5 @@
 
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar AquacArea
     include Common
     include Words

data/lib/food_fish_parser/{grammar → strict/grammar}/aquac_method.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar AquacMethod
     include Common
     include Words

data/lib/food_fish_parser/{grammar → strict/grammar}/catch_area.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar CatchArea
     include Common
     include Words

data/lib/food_fish_parser/{grammar → strict/grammar}/catch_method.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar CatchMethod
     include Common
     include Words

data/lib/food_fish_parser/{grammar → strict/grammar}/common.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar Common
 
     # whitespace

data/lib/food_fish_parser/{grammar → strict/grammar}/fao_area.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar FaoArea
     include Common
 
@@ -45,8 +45,9 @@ module FoodFishParser::Grammar
 
     rule fao_area_sub_code
       (
-        ( ( dash / '/' / ws* )? [ivxIVX]+ ) /
-        ( dash [0-9] [0-9] )
+        # IVX for roman numals, but l for incorrectly OCR-ed text
+        ( ( dash / '/' / ws* )? [ivxIVXl]+ ) /
+        ( dash [0-9]+ )
       )
       fao_area_suffix?
       (

data/lib/food_fish_parser/strict/grammar/fish_name.treetop

@@ -0,0 +1,30 @@
+module FoodFishParser::Strict::Grammar
+  grammar FishName
+    include Common
+    include FishNameLatin
+    include FishNameNL
+
+    rule fish_name_both
+      ( fish_name_nl ws* '(' ws* fish_name_latin ( ws* ')' / comma )? )
+    end
+
+    rule fish_name_both_list
+      ( fish_name_both <FishNameNode> )
+      ( ws+ and_or ws+ fish_name_both <FishNameNode> )*
+    end
+
+    rule fish_name_latin_list
+      ( fish_name_latin <FishNameNode> )
+      ( ws+ and_or ws+ fish_name_latin <FishNameNode> )*
+    end
+
+    rule fish_name_nl_list
+      ( fish_name_nl <FishNameNode> )
+      ( ws+ and_or ws+ fish_name_nl <FishNameNode> )*
+    end
+
+    rule fish_name_any_list
+      fish_name_both_list / fish_name_latin_list / fish_name_nl_list
+    end
+  end
+end

data/lib/food_fish_parser/{grammar → strict/grammar}/fish_name_latin.treetop

@@ -1,5 +1,5 @@
-# autogenerated by species-treetop-gen-latin.rb on 2020-03-18
-module FoodFishParser::Grammar
+# autogenerated by species-treetop-gen-latin.rb on 2020-03-19
+module FoodFishParser::Strict::Grammar
   grammar FishNameLatin
     include Common
 
@@ -9,11 +9,11 @@ module FoodFishParser::Grammar
     end
 
     rule fish_name_latin_first
-      'zygochlamys'i / 'zeus'i / 'xiphopenaeus'i / 'xiphias'i / 'undaria'i / 'ulva'i / 'trichiurus'i / 'trachurus'i / 'todarodes'i / 'thunnus'i / 'theragra'i / 'stolephorus'i / 'sprattus'i / 'spirulina'i / 'sparus'i / 'solea'i / 'sepiella'i / 'sepia'i / 'sebastes'i / 'scomber'i / 'sardinella'i / 'sardina'i / 'salmo'i / 'saccharina'i / 'reinhardtius'i / 'psetta'i / 'procambarus'i / 'portunus'i / 'porphyra'i / 'pollachius'i / 'pleuronectes'i / 'pleoticus'i / 'placopecten'i / 'phymatolithon'i / 'perna'i / 'penaeus'i / 'penaeidae'i / 'pelvetia'i / 'pecten'i / 'patinopecten'i / 'parapenaeopsis'i / 'paralomis'i / 'paphia'i / 'pangasius'i / 'pandalus'i / 'palmaria'i / 'pagellus'i / 'pacifische'i / 'ovalipes'i / 'ostrea'i / 'oreochromis'i / 'oncorhynchus'i / 'octopus'i / 'nephrops'i / 'nemipterus'i / 'nelumbo'i / 'mytilus'i / 'mulinia'i / 'micromesistius'i / 'metapenaeus'i / 'merluccius'i / 'merlangius'i / 'melanogrammus'i / 'macruronus'i / 'macrocystis'i / 'lophius'i / 'loligo'i / 'litopenaeus'i / 'lithodes'i / 'limanda'i / 'lethrinus'i / 'lepidotrigla'i / 'lepidopsetta'i / 'lates'i / 'laminaria'i / 'katwonus'i / 'katwomus'i / 'katsuwonus'i / 'katsuwomus'i / 'illex'i / 'homarus'i / 'himanthalia'i / 'haematococcus'i / 'gracilaria'i / 'gelidium'i / 'gadus'i / 'fucus'i / 'euthynnus'i / 'ensis'i / 'engraulis'i / 'dunaliella'i / 'dosidicus'i / 'dicentrarchus'i / 'crassostrea'i / 'crangon'i / 'clupea'i / 'clarias'i / 'chondrus'i / 'chlorella'i / 'cerastoderma'i / 'caulerpa'i / 'ascophyllum'i / 'anguilla'i / 'anadara'i / 'alle'i / 'alaria'i / 'acipenser'i
+      'zygochlamys'i / 'zeus'i / 'xiphopenaeus'i / 'xiphias'i / 'undaria'i / 'ulva'i / 'trichiurus'i / 'trachurus'i / 'todarodes'i / 'thunnus'i / 'theragra'i / 'stolephorus'i / 'sprattus'i / 'spirulina'i / 'sparus'i / 'solea'i / 'sepiella'i / 'sepia'i / 'sebastes'i / 'scomber'i / 'sardinella'i / 'sardina'i / 'salmo'i / 'saccharina'i / 'reinhardtius'i / 'psetta'i / 'procambarus'i / 'portunus'i / 'porphyra'i / 'pollachius'i / 'pleuronectes'i / 'pleoticus'i / 'placopecten'i / 'phymatolithon'i / 'perna'i / 'penaeus'i / 'penaeidae'i / 'pelvetia'i / 'pecten'i / 'patinopecten'i / 'parapenaeopsis'i / 'paralomis'i / 'paphia'i / 'pangasius'i / 'pandalus'i / 'palmaria'i / 'pagellus'i / 'ovalipes'i / 'ostrea'i / 'oreochromis'i / 'oncorhynchus'i / 'octopus'i / 'nephrops'i / 'nemipterus'i / 'nelumbo'i / 'mytilus'i / 'mulinia'i / 'micromesistius'i / 'metapenaeus'i / 'merluccius'i / 'merlangius'i / 'melanogrammus'i / 'macruronus'i / 'macrocystis'i / 'lophius'i / 'loligo'i / 'litopenaeus'i / 'lithodes'i / 'limanda'i / 'lethrinus'i / 'lepidotrigla'i / 'lepidopsetta'i / 'lates'i / 'laminaria'i / 'katwonus'i / 'katwomus'i / 'katsuwonus'i / 'katsuwomus'i / 'illex'i / 'homarus'i / 'himanthalia'i / 'haematococcus'i / 'gracilaria'i / 'gelidium'i / 'gadus'i / 'fucus'i / 'euthynnus'i / 'ensis'i / 'engraulis'i / 'dunaliella'i / 'dosidicus'i / 'dicentrarchus'i / 'crassostrea'i / 'crangon'i / 'clupea'i / 'clarias'i / 'chondrus'i / 'chlorella'i / 'cerastoderma'i / 'caulerpa'i / 'ascophyllum'i / 'anguilla'i / 'anadara'i / 'alle'i / 'alaria'i / 'acipenser'i
     end
 
     rule fish_name_latin_second
-      'yessoensis'i / 'vulgaris'i / 'virens'i / 'vesiculosus'i / 'verrucosa'i / 'vannamei'i / 'undulata'i / 'umbilicalis'i / 'tenera'i / 'stylifera'i / 'sprattus'i / 'spp.'i / 'spp'i / 'sombrus'i / 'solea'i / 'solar'i / 'scombrus'i / 'santolla'i / 'salina'i / 'salar'i / 'ringens'i / 'pyrifera'i / 'pyrenoidosa'i / 'punctatus'i / 'productus'i / 'pluvialis'i / 'platessa'i / 'platensis'i / 'piscatorius'i / 'pinnatifida'i / 'pilchardus'i / 'pelatis'i / 'pelanis'i / 'pelamis'i / 'pelagicus'i / 'patagonica'i / 'pangasius'i / 'palmata'i / 'pacificus'i / 'officinalis'i / 'ocellatus'i / 'nucifera'i / 'novaezelandiae'i / 'novaezealandiae'i / 'norvegicus'i / 'nodosum'i / 'niloticus'i / 'nerka'i / 'mykiss'i / 'mykis'i / 'myki?s'i / 'murphyi'i / 'muelleri'i / 'morhue'i / 'morhua'i / 'monodon'i / 'monoceros'i / 'microptera'i / 'merluccius'i / 'merlangus'i / 'merguiensis'i / 'maximus'i / 'maxima'i / 'marinus'i / 'magellanicus'i / 'macrocephalus'i / 'limanda'i / 'lepturus'i / 'lentillifera'i / 'latissima'i / 'lactuca'i / 'labrax'i / 'kroyeri'i / 'kisutch'i / 'keta'i / 'kabeljauw'i / 'jordani'i / 'japonicus'i / 'japonica'i / 'hyptophthalmus'i / 'hypophtalmus'i / 'hippoglossoides'i / 'hexodon'i / 'harengus'i / 'gueldenstaedtii'i / 'granulosa'i / 'gorbusche'i / 'gorbuscha'i / 'gladius'i / 'gigas'i / 'gibbosa'i / 'gariepinus'i / 'galloprovincialis'i / 'faber'i / 'esculenta'i / 'encrasicolus'i / 'elongata'i / 'edulis'i / 'edule'i / 'directus'i / 'digitata'i / 'crispus'i / 'crangon'i / 'clarkii'i / 'chilensis'i / 'chalcogramma'i / 'capensis'i / 'cannamei'i / 'canaliculus'i / 'canaliculata'i / 'calcareum'i / 'borealis'i / 'bogaraveo'i / 'bilineata'i / 'australis'i / 'aurata'i / 'argentinus'i / 'antiquata'i / 'anquilla'i / 'anguilla'i / 'anchoita'i / 'americanus'i / 'alle'i / 'albacares'i / 'alalunga'i / 'aeglefinus'i
+      'yessoensis'i / 'vulgaris'i / 'virens'i / 'vesiculosus'i / 'verrucosa'i / 'vannamei'i / 'undulata'i / 'umbilicalis'i / 'tenera'i / 'stylifera'i / 'sprattus'i / 'spp.'i / 'spp'i / 'sombrus'i / 'solea'i / 'solar'i / 'scombrus'i / 'santolla'i / 'salina'i / 'salar'i / 'ringens'i / 'pyrifera'i / 'pyrenoidosa'i / 'punctatus'i / 'productus'i / 'pluvialis'i / 'platessa'i / 'platensis'i / 'piscatorius'i / 'pinnatifida'i / 'pilchardus'i / 'pelatis'i / 'pelanis'i / 'pelamis'i / 'pelagicus'i / 'patagonica'i / 'pangasius'i / 'palmata'i / 'pacificus'i / 'officinalis'i / 'ocellatus'i / 'nucifera'i / 'novaezelandiae'i / 'novaezealandiae'i / 'norvegicus'i / 'nodosum'i / 'niloticus'i / 'nerka'i / 'mykiss'i / 'mykis'i / 'myki?s'i / 'murphyi'i / 'muelleri'i / 'morhue'i / 'morhua'i / 'monodon'i / 'monoceros'i / 'microptera'i / 'merluccius'i / 'merlangus'i / 'merguiensis'i / 'maximus'i / 'maxima'i / 'marinus'i / 'magellanicus'i / 'macrocephalus'i / 'limanda'i / 'lepturus'i / 'lentillifera'i / 'latissima'i / 'lactuca'i / 'labrax'i / 'kroyeri'i / 'kisutch'i / 'keta'i / 'jordani'i / 'japonicus'i / 'japonica'i / 'hyptophthalmus'i / 'hypophtalmus'i / 'hippoglossoides'i / 'hexodon'i / 'harengus'i / 'gueldenstaedtii'i / 'granulosa'i / 'gorbusche'i / 'gorbuscha'i / 'gladius'i / 'gigas'i / 'gibbosa'i / 'gariepinus'i / 'galloprovincialis'i / 'faber'i / 'esculenta'i / 'encrasicolus'i / 'elongata'i / 'edulis'i / 'edule'i / 'directus'i / 'digitata'i / 'crispus'i / 'crangon'i / 'clarkii'i / 'chilensis'i / 'chalcogramma'i / 'capensis'i / 'cannamei'i / 'canaliculus'i / 'canaliculata'i / 'calcareum'i / 'borealis'i / 'bogaraveo'i / 'bilineata'i / 'australis'i / 'aurata'i / 'argentinus'i / 'antiquata'i / 'anquilla'i / 'anguilla'i / 'anchoita'i / 'americanus'i / 'alle'i / 'albacares'i / 'alalunga'i / 'aeglefinus'i
     end
   end
 end

data/lib/food_fish_parser/{grammar → strict/grammar}/fish_name_nl.treetop

@@ -1,5 +1,5 @@
-# autogenerated by species-treetop-gen-nl.rb on 2020-03-18
-module FoodFishParser::Grammar
+# autogenerated by species-treetop-gen-nl.rb on 2020-03-19
+module FoodFishParser::Strict::Grammar
   grammar FishNameNL
     include Common
 

data/lib/food_fish_parser/{grammar → strict/grammar}/root.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar Root
     include Common
     include FishName
@@ -18,8 +18,10 @@ module FoodFishParser::Grammar
     rule root
       (
         ( fish_with_info ( ws* fish_sep ws* fish_with_info )* ) /
-        ( fish_only_names ( ws* fish_sep ws* fish_only_names )* ) /
-        ( fish_only_info ( ws* fish_sep ws* fish_only_info )* )
+        ( fish_names_both ( ws* fish_sep ws* fish_names_both )* ) /
+        ( fish_names_latin ( ws* fish_sep ws* fish_names_latin )* ) /
+        ( fish_only_info ( ws* fish_sep ws* fish_only_info )* ) /
+        ( fish_names_nl ( ws* fish_sep ws* fish_names_nl )* )
       )
       ws* '.'? ws*
       <RootNode>
@@ -31,8 +33,10 @@ module FoodFishParser::Grammar
     rule root_anywhere
       (
         ( !fish_with_info . )* ( fish_with_info ( ( !fish_with_info . )* fish_with_info )* ) ( !fish_with_info . )* /
-        ( !fish_only_names . )* ( fish_only_names ( ( !fish_only_names . )+ fish_only_names )* ) ( !fish_only_names . )* /
-        ( !fish_only_info . )* ( fish_only_info ( ( !fish_only_info . )+ fish_only_info )* ) ( !fish_only_info . )*
+        ( !fish_names_both . )* ( fish_names_both ( ( !fish_names_both . )+ fish_names_both )* ) ( !fish_names_both . )* /
+        ( !fish_names_latin . )* ( fish_names_latin ( ( !fish_names_latin . )+ fish_names_latin )* ) ( !fish_names_latin . )* /
+        ( !fish_only_info . )* ( fish_only_info ( ( !fish_only_info . )+ fish_only_info )* ) ( !fish_only_info . )* /
+        ( !fish_names_nl . )* ( fish_names_nl ( ( !fish_names_nl . )+ fish_names_nl )* ) ( !fish_names_nl . )*
       )
       <RootNode>
     end
@@ -45,15 +49,25 @@ module FoodFishParser::Grammar
     # fish with catch or aquaculture info
     rule fish_with_info
       (
-        ( fish_name_list ( ws* ( comma / ':' ) )? ws+ fish_catch_info ) /
-        ( fish_name_list ( ws* ( comma / ':' ) )? ws+ fish_aquac_info )
+        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_catch_info ) /
+        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_aquac_info )
       )
       <FishNode>
     end
 
-    # fish names only
-    rule fish_only_names
-      fish_name_list <FishNode>
+    # fish names common and latin
+    rule fish_names_both
+      fish_name_both_list <FishNode>
+    end
+
+    # fish names Latin only
+    rule fish_names_latin
+      fish_name_latin_list <FishNode>
+    end
+
+    # fish names NL only
+    rule fish_names_nl
+      fish_name_nl_list <FishNode>
     end
 
     # catch or aquaculture info only (no names)

data/lib/food_fish_parser/{grammar → strict/grammar}/words.treetop

@@ -1,4 +1,4 @@
-module FoodFishParser::Grammar
+module FoodFishParser::Strict::Grammar
   grammar Words
     include Common
     include FishNameLatin

data/lib/food_fish_parser/{nodes.rb → strict/nodes.rb}

@@ -1,7 +1,7 @@
 require 'treetop/runtime'
 
 # Needs to be in grammar namespace so Treetop can find the nodes.
-module FoodFishParser
+module FoodFishParser::Strict
   module Grammar
 
     # Additions for Treetop nodes, include this in other nodes where needed.

data/lib/food_fish_parser/strict/parser.rb

@@ -0,0 +1,34 @@
+require_relative 'grammar'
+
+module FoodFishParser
+  module Strict
+    class Parser
+
+      # @!attribute [r] parser
+      #   @return [Treetop::Runtime::CompiledParser] low-level parser object
+      #   @note This attribute is there for convenience, but may change in the future. Take care.
+      attr_reader :parser
+
+      # Create a new fish detail parser
+      # @return [FoodFishParser::Strict::Parser]
+      def initialize
+        @parser = Grammar::RootParser.new
+      end
+
+      # Parse food fish text into a structured representation.
+      #
+      # @param s [String] text to parse
+      # @param anywhere [Bool] +false+ assume the text is only fish details, +true+ to search for fish details in the text
+      # @return [FoodFishParser::Strict::Grammar::RootNode] structured representation of fish details
+      # @note Unrecognized options are passed to Treetop, but this is not guarenteed to remain so forever.
+      def parse(s, anywhere: false, **options)
+        if anywhere
+          options = options.merge(root: :root_anywhere, consume_all_input: false)
+        end
+
+        @parser.parse(s, **options)
+      end
+
+    end
+  end
+end

data/lib/food_fish_parser/version.rb

@@ -1,4 +1,4 @@
 module FoodFishParser
-  VERSION      = '0.2.1'
-  VERSION_DATE = '2020-03-18'
+  VERSION      = '0.3.0'
+  VERSION_DATE = '2020-03-19'
 end

data/lib/food_fish_parser.rb

@@ -1,2 +1,3 @@
 require_relative 'food_fish_parser/version'
-require_relative 'food_fish_parser/parser'
+require_relative 'food_fish_parser/flat/parser'
+require_relative 'food_fish_parser/strict/parser'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: food_fish_parser
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.3.0
 platform: ruby
 authors:
 - wvengen
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-03-18 00:00:00.000000000 Z
+date: 2020-03-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: treetop
@@ -43,20 +43,26 @@ files:
 - bin/food_fish_parser
 - food_fish_parser.gemspec
 - lib/food_fish_parser.rb
-- lib/food_fish_parser/grammar.rb
-- lib/food_fish_parser/grammar/aquac_area.treetop
-- lib/food_fish_parser/grammar/aquac_method.treetop
-- lib/food_fish_parser/grammar/catch_area.treetop
-- lib/food_fish_parser/grammar/catch_method.treetop
-- lib/food_fish_parser/grammar/common.treetop
-- lib/food_fish_parser/grammar/fao_area.treetop
-- lib/food_fish_parser/grammar/fish_name.treetop
-- lib/food_fish_parser/grammar/fish_name_latin.treetop
-- lib/food_fish_parser/grammar/fish_name_nl.treetop
-- lib/food_fish_parser/grammar/root.treetop
-- lib/food_fish_parser/grammar/words.treetop
-- lib/food_fish_parser/nodes.rb
-- lib/food_fish_parser/parser.rb
+- lib/food_fish_parser/flat/fao_region.rb
+- lib/food_fish_parser/flat/fish_name.rb
+- lib/food_fish_parser/flat/fish_name_latin.rb
+- lib/food_fish_parser/flat/fish_name_nl.rb
+- lib/food_fish_parser/flat/kind.rb
+- lib/food_fish_parser/flat/parser.rb
+- lib/food_fish_parser/strict/grammar.rb
+- lib/food_fish_parser/strict/grammar/aquac_area.treetop
+- lib/food_fish_parser/strict/grammar/aquac_method.treetop
+- lib/food_fish_parser/strict/grammar/catch_area.treetop
+- lib/food_fish_parser/strict/grammar/catch_method.treetop
+- lib/food_fish_parser/strict/grammar/common.treetop
+- lib/food_fish_parser/strict/grammar/fao_area.treetop
+- lib/food_fish_parser/strict/grammar/fish_name.treetop
+- lib/food_fish_parser/strict/grammar/fish_name_latin.treetop
+- lib/food_fish_parser/strict/grammar/fish_name_nl.treetop
+- lib/food_fish_parser/strict/grammar/root.treetop
+- lib/food_fish_parser/strict/grammar/words.treetop
+- lib/food_fish_parser/strict/nodes.rb
+- lib/food_fish_parser/strict/parser.rb
 - lib/food_fish_parser/version.rb
 homepage: https://github.com/q-m/food-fish-parser-ruby
 licenses:

data/lib/food_fish_parser/grammar/fish_name.treetop

@@ -1,21 +0,0 @@
-module FoodFishParser::Grammar
-  grammar FishName
-    include Common
-    include FishNameLatin
-    include FishNameNL
-
-    rule fish_name
-      (
-        fish_name_nl ws* '(' ws* fish_name_latin ws* ')' /
-        fish_name_nl /
-        fish_name_latin
-      )
-      <FishNameNode>
-    end
-
-    rule fish_name_list
-      fish_name ( ws+ and_or ws+ fish_name )*
-    end
-
-  end
-end

data/lib/food_fish_parser/parser.rb

@@ -1,32 +0,0 @@
-require_relative 'grammar'
-
-module FoodFishParser
-  class Parser
-
-    # @!attribute [r] parser
-    #   @return [Treetop::Runtime::CompiledParser] low-level parser object
-    #   @note This attribute is there for convenience, but may change in the future. Take care.
-    attr_reader :parser
-
-    # Create a new fish detail parser
-    # @return [FoodFishParser::Parser]
-    def initialize
-      @parser = Grammar::RootParser.new
-    end
-
-    # Parse food fish text into a structured representation.
-    #
-    # @param s [String] text to parse
-    # @param anywhere [Bool] +false+ assume the text is only fish details, +true+ to search for fish details in the text
-    # @return [FoodFishParser::Grammar::RootNode] structured representation of fish details
-    # @note Unrecognized options are passed to Treetop, but this is not guarenteed to remain so forever.
-    def parse(s, anywhere: false, **options)
-      if anywhere
-        options = options.merge(root: :root_anywhere, consume_all_input: false)
-      end
-
-      @parser.parse(s, **options)
-    end
-
-  end
-end