food_fish_parser 0.3.7 → 0.3.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4ec812d752c0ebb045d364f95e1343c902d84e430e14c2c817b04213b91f7603
-  data.tar.gz: 7d18f20a23b718d4d27e8bf2f67c879fbaccd6b81af5cf2657995e08becb665a
+  metadata.gz: 1276f2546a66d95c826ec9abe326186077010c0186a8bcdeec5a6fa6595091b6
+  data.tar.gz: 3702ffd5cc2f17f75e936c4938b61a33e4688f15e30677dd6e9130af21a51f3f
 SHA512:
-  metadata.gz: cbb1d4177bd203662006584e871bb5fe1262ced8885c4705039e25e50700116716357cfb0db9e0c3819ed98769f5881109cc8bec622f714d339a7028a0039fcf
-  data.tar.gz: 64e44abbf0aa1a87eec8c6d4671b5f2f2d3ec5d247b2cd5fc9996488f4f1e9ef0985be7d24bda0d6fe7b3f6655cb7460e540b04e307dad6f70e87de9b3a0a2fc
+  metadata.gz: 141366cd9da78e28a6cfaa026e3a66c349e9ba09a06692cc01ea6df19c0e846b09ce10c9f92c83143777737cd5ae4dcb5f67db8c48d4e8ea359cea5925b08266
+  data.tar.gz: 417463cfc6470b951ece7fadf0d50bae63e209933775d9bce974019674acddd3855eeb5fda26024851d17ab39b4d0f3f1ef3deef8ebbe82489e2c8d8a3a1b0bc

data/README.md

@@ -121,7 +121,7 @@ FAO regions and fish names found.
 require 'food_fish_parser'
 
 parser = FoodFishParser::Flat::Parser.new
-s = "Foobar zalm (salmo salar) *&! gevangen pangasius spp FAO 61 ?or ?FAO 67 what more.")
+s = "Foobar zalm (salmo salar) *&! gevangen met lijnen pangasius spp FAO 61 ?or ?FAO 67 what more.")
 puts parser.parse(s).to_a.inspect
 ```
 
@@ -136,7 +136,7 @@ puts parser.parse(s).to_a.inspect
       { :name=>nil, :fao_codes=>["61"] },
       { :name=>nil, :fao_codes=>["67"] }
     ],
-    :catch_methods       => [],
+    :catch_methods       => [{ :text=>"lijnen" }],
     :aquaculture_areas   => [],
     :aquaculture_methods => []
   }

data/lib/food_fish_parser/flat/aquac_method.rb

@@ -0,0 +1,26 @@
+module FoodFishParser
+  module Flat
+    module AquacMethod
+
+      REGEX = /
+        (
+          kwekerij(?:en)? |
+          \bkweekvijvers? |
+          \bvijvers?(?:\s*\/\s*bassins?)? |
+          \b(?:open\s+)? kooien (?: \s+in\s+zee | \s+in\s+rivier )? |
+          \bdoorstroomsystemen |
+          \b(?:open\s+)? hangcultuur |
+          \bgekweekt\s+in\s+netten
+        )
+        \b
+      /ix
+
+      def self.find_all(text)
+        text
+          .scan(REGEX)
+          .flatten
+          .map {|s| { text: s } }
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/{fao_region.rb → area_fao.rb}

@@ -1,6 +1,6 @@
 module FoodFishParser
   module Flat
-    module FaoRegion
+    module AreaFao
 
       REGEX_CODE = / 0?\d{2} (?: (?: -0?\d{2} | (?: \/ | - | \s+ | ) [lIV]+ ) [a-d]? )? /ix
       REGEX = /
@@ -28,7 +28,7 @@ module FoodFishParser
 
       def self.find_all(text)
         find_all_code_groups(text)
-          .map {|a| { name: nil, fao_codes: a } }
+          .map {|a| { text: nil, fao_codes: a } }
       end
     end
   end

data/lib/food_fish_parser/flat/area_name.rb

@@ -0,0 +1,64 @@
+module FoodFishParser
+  module Flat
+    module AreaName
+
+      REGEX_AREA_NAMES = /
+        \b
+        (?:
+          Cornwall |
+          Noorwegen |
+          India |
+          Ierland |
+          Vi[eë]tnam |
+          Indonesi[eë] |
+          Griekenland |
+          Ecuador |
+          Limfjord |
+          Denemarken |
+          Chili |
+          Faroër eilanden |
+          Groenland |
+          Californi[eë] |
+          Malediven |
+          Spitsbergen |
+          Bereneiland |
+          Maleisi[eë] |
+          (?:
+            atlantische | atlantic | pacifische | pacific | stille | noorse |
+            middelandse | indische
+          ) \s+ (?: oceaan | ocean | zee) |
+          Noordzee |
+          Barentszzee |
+          Ijsland
+        )
+        \b
+      /ix
+
+      REGEX_AREA = /
+        \b
+        (
+          (?: (?: rond | bij | aan | voor ) \s+ )?
+          (?: de\s+ (?: kust | fjorden ) \s+van\s+ )?
+          (?: (?:\(sub\))?tropische\s+wateren\s+in\s+ )?
+          (?: (?: de | het ) \s+ )?
+          (?: (?: \s* (?:noord|zuid|oost|west)-? )+(?:elijke?\s+(?: deel | gedeelte )\s+van)? \s+ )?
+          (?: (?: de | het ) \s+ )?
+          #{REGEX_AREA_NAMES}
+        )
+        \b
+      /ix
+
+      REGEX = /
+        \b#{REGEX_AREA} (?: (?: \s+(?:en|of|en\/of)\s+ | \s*,\s* ) #{REGEX_AREA} )* \b
+      /ix
+
+      def self.find_all(text)
+        text
+          .scan(REGEX)
+          .flatten
+          .compact
+          .map {|s| { text: s.strip.sub(/^(de|het|een)\s*/, ''), fao_codes: [] } }
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/catch_method.rb

@@ -0,0 +1,42 @@
+module FoodFishParser
+  module Flat
+    module CatchMethod
+
+      REGEX = /
+        \b
+        (
+          trawlnet(?:ten)? |
+          trawler |
+          haken\s+en\s+lijnen |
+          (machinale\s+)? handlijnen |
+          ring-\s+en\s+kruisnet(?:ten)? |
+          ringnet(?:ten)? |
+          kruisnet(?:ten)? |
+          sleeplijn(?:en)? |
+          kieuwnet(?:ten)? |
+          (?:soortgelijke|vergelijkbare)\s+net(?:ten)? |
+          (?:zwevende\s+)? (?:bodem)?(?:otter)?trawls? |
+          (?:deense\s+)? zegens? |
+          ringzegens? |
+          hengels?\s+en\s+lijn(?:en)? |
+          met\s+de\s+hengel |
+          met\s+lijnen |
+          vallen |
+          purse\s*seiner
+        ) (?:visserij)?
+        \b
+      /ix
+
+      def self.find_all(text)
+        text
+          .scan(REGEX)
+          .flatten
+          .compact
+          .map do |s|
+            s.gsub!(/^met\b(\s+de\b)?\s*/, '')
+            { text: s }
+          end
+      end
+    end
+  end
+end

data/lib/food_fish_parser/flat/fish_name.rb

@@ -4,14 +4,15 @@ 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_ALLERGEN_NAMES = /\b(?:visvlees|vis|weekdieren|schaaldieren)\b/i
+      REGEX_ALLERGEN = /\s*(?: \(\s*#{REGEX_ALLERGEN_NAMES}\s*\) | \b,?\s*#{REGEX_ALLERGEN_NAMES}\s*,?\s*\b )/ix
 
       REGEX = /
         (?:
-          #{FishNameNL::REGEX} (?: #{REGEX_VIS} )? \s* \( \s* #{FishNameLatin::REGEX} \s* \) |
-          #{FishNameNL::REGEX} \s*,?\s* \b #{FishNameLatin::REGEX} |
+          (?!#{REGEX_ALLERGEN}) #{FishNameNL::REGEX} (?: #{REGEX_ALLERGEN} )? \s* \( \s* #{FishNameLatin::REGEX} \s* \) |
+          (?!#{REGEX_ALLERGEN}) #{FishNameNL::REGEX} \s*,?\s* \b #{FishNameLatin::REGEX} |
           #{FishNameLatin::REGEX} |
-          #{FishNameNL::REGEX}
+          (?!#{REGEX_ALLERGEN}) #{FishNameNL::REGEX}
         )
       /ix
 

data/lib/food_fish_parser/flat/fish_name_latin.rb

@@ -1,4 +1,4 @@
-# autogenerated by species-treetop-gen-latin.rb on 2020-03-25
+# autogenerated by species-treetop-gen-latin.rb on 2020-03-31
 module FoodFishParser
   module Flat
     module FishNameLatin

data/lib/food_fish_parser/flat/fish_name_nl.rb

@@ -1,11 +1,11 @@
-# autogenerated by species-treetop-gen-nl.rb on 2020-03-25
+# autogenerated by species-treetop-gen-nl.rb on 2020-03-31
 module FoodFishParser
   module Flat
     module FishNameNL
-      REGEX_AREA = /pacifische|noorse|japanse|indische|ijsselmeer|groenlandse|atlantische|argentijnse|alaska/i
+      REGEX_AREA = /pacifische|pacific|noorse|japanse|indische|ijsselmeer|groenlandse|atlantische|argentijnse|alaska/i
       REGEX_ATTR = /zwarte|zwart|witte|witpoot|wit|roze|rood|rode|rivier|pijl|kleine|klein|grote|groot|groene|groen|grise|doorn|coho|chum|blonde|blond|blauwe|blauw/i
-      REGEX_NAME = /zwaardvis|zonnevis|zeewolf|zeesnoek|zeekreeft|zeeforel|zeebaars|zalm|wijting|weekdieren|weekdier|vliegende\ vis|vis|vintonijn|vin\ tonijn|tonijn\ albacore|tonijn|tong|tilapia|thon\ albacore|tarbot|tapijtschelp|surimi|steur|st\.\ pierre|sprot|spie|snotolf|snoekbaars|snoek|snapper|sliptong|skrei|skipjack\ tuna|skipjack\ tonijn|skipjack|seabob\ garnalen|schol|schelvis|schelpen|schelp|schar|sardines|sardinen|sardinelle|sardine|sardienen|sardien|saibling|rouget|rogvleugel|riddervis|regenboogforel|red\ snapper|red\ grouper|raat|poon|pollock|pollak|pilchard|pieterman|pangasius|paling|pagre|oogtonijn|nijlbaars|neustong|mul|mosselen|mossel|meerval|marlijn|mantelschelp|makreel|mahi\-mahi|mahi\ mahi|maatjesha|lom|leng|kreeft|krab|koolvis|kokkel|king\ clip|karper|kabeljauw|inktvis|hondstong|hoki|heilbot|heekrug|heek|haring|harder\ diklip|hamvis|hake|haai|gruis|grouper|griet|geep|geelvintonijn|geelvin\ tonijn|garnalen|garnaal|forel|fint|escolar|dorade|diklipharder|diklip\ harder|daurade|corvina|coquilles|conger\ aal|cocquilles|cobia|claresse|caviaar|braadha|botervis|bot|blood\ snapper|beekridder|barracumundi|barracuda|bacalao|baars|arkschelp|ansovis|ansjovis|alfonsino|albacore\ tonijn/i
-      REGEX_SUFX = /wang|vlees|steur|ringen|ring|moten|kaviaar|filets|filetes|filet|eiwit/i
+      REGEX_NAME = /zwaardvis|zonnevis|zeewolf|zeesnoek|zeekreeft|zeeforel|zeebaars|zalm|wijting|weekdieren|weekdier|vliegende\ vis|vis|vintonijn|vin\ tonijn|tonijn\ albacore|tonijn|tong|tilapia|thon\ albacore|tarbot|tapijtschelp|surimi|steur|st\.\ pierre|sprot|spie|sockeye\ zalm|snotolf|snoekbaars|snoek|snapper|sliptong|skrei|skipjack\ tuna|skipjack\ tonijn|skipjack|seabob\ garnalen|schol|schelvis|schelpen|schelp|schar|sardines|sardinen|sardinelle|sardine|sardienen|sardien|saibling|rouget|rogvleugel|riddervis|regenboogforel|red\ snapper|red\ grouper|raat|poon|pollock|pollak|pilchard|pieterman|pangasius|paling|pagre|oogtonijn|nijlbaars|neustong|mul|mosselen|mossel|meerval|marlijn|mantelschelp|makreel|mahi\-mahi|mahi\ mahi|maatjesha|lom|leng|kreeft|krab|koolvis|kokkel|king\ clip|karper|kabeljauw|inktvis|hondstong|hoki|heilbot|heekrug|heek|haring|harder\ diklip|hamvis|hake|haai|gruis|grouper|griet|geep|geelvintonijn|geelvin\ tonijn|garnalen|garnaal|forel|fint|escolar|dorade|diklipharder|diklip\ harder|daurade|corvina|coquilles|conger\ aal|cocquilles|cobia|claresse|caviaar|braadha|botervis|bot|blood\ snapper|beekridder|barracumundi|barracuda|bacalao|baars|arkschelp|ansovis|ansjovis|alfonsino|albacore\ tonijn/i
+      REGEX_SUFX = /wang|vlees|steur|ringen|ring|moten|lever|kaviaar|filets|filetes|filet|eiwit/i
 
       REGEX = /
         (?: \b verse \s+ | \b kaviaar \s+ | \b kaviaar \s+ van \s+ | \b gevilde \s+ )?

data/lib/food_fish_parser/flat/kind.rb

@@ -9,10 +9,11 @@ module FoodFishParser
         (?:
           gevangen |
           visgebied |
-          vangstgebied |
-          vangsgebied |
+          vangst?gebied |
+          visvangst |
           betrapt \s+ bij |
-          wilde? \s+ #{FishNameNL::REGEX}
+          wilde? \s+ #{FishNameNL::REGEX} |
+          MSC # certificate is for wild fish only
         )
         \b
       /ix
@@ -24,7 +25,8 @@ module FoodFishParser
           aquacultuurproduct |
           aquacultuur \s+ product |
           kweekmethode |
-          kweekmethoden
+          kweekmethoden |
+          ASC # certificate is for aquaculture fish only
         )
         \b
       /ix

data/lib/food_fish_parser/flat/parser.rb

@@ -1,5 +1,8 @@
 require_relative 'fish_name'
-require_relative 'fao_region'
+require_relative 'area_fao'
+require_relative 'area_name'
+require_relative 'catch_method'
+require_relative 'aquac_method'
 require_relative 'kind'
 
 module FoodFishParser
@@ -17,31 +20,32 @@ module FoodFishParser
       # @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
+        areas = AreaName.find_all(s) + AreaFao.find_all(s)
+        catch_methods = CatchMethod.find_all(s)
+        aquac_methods = AquacMethod.find_all(s)
 
-        is_wild = Kind.is_wild?(s)
-        is_aquac = Kind.is_aquac?(s)
+        is_wild = catch_methods.any? || Kind.is_wild?(s)
+        is_aquac = aquac_methods.any? || Kind.is_aquac?(s)
 
-        return [] unless names.any? || areas.any?
+        return [] unless names.any? || aquac_methods.any? || catch_methods.any? || areas.any?
 
         attrs = {
           names: names,
           catch_areas: [],
-          catch_methods: [],
+          catch_methods: catch_methods,
           aquaculture_areas: [],
-          aquaculture_methods: []
+          aquaculture_methods: aquac_methods
         }
 
         if is_wild && !is_aquac
-          [attrs.merge(catch_areas: areas, catch_methods: methods)]
+          [attrs.merge(catch_areas: areas)]
         elsif !is_wild && is_aquac
-          [attrs.merge(aquaculture_areas: areas, aquaculture_methods: methods)]
-        elsif areas.any? || methods.any?
+          [attrs.merge(aquaculture_areas: areas)]
+        elsif areas.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)]
+          [attrs.merge(areas: areas)]
         else
           # just names
           [attrs]

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

@@ -7,7 +7,7 @@ module FoodFishParser::Strict::Grammar
     rule catch_area_indicator
       ( ( comma ws* )? ( 'wildfang'i / 'wild'i ) ws* ( comma ws* )? )?
       (
-        'gevangen'i ws+ ( 'in'i / 'op'i ) /
+        'gevangen'i ws+ catch_area_indicator_preposition /
         'visgebied'i / 'vangstgebied'i / 'vangsgebied'i / 'fanggebiet'i /
         'betrapt bij'i
       )
@@ -17,7 +17,11 @@ module FoodFishParser::Strict::Grammar
 
     rule catch_area_indicator_short
       catch_area_indicator /
-      ( 'in'i / 'op'i ) !char ( ws* ':' )?
+      catch_area_indicator_preposition !char ( ws* ':' )?
+    end
+
+    rule catch_area_indicator_preposition
+      'in'i / 'op'i / 'voor'i
     end
 
     rule catch_area_content

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

@@ -9,7 +9,7 @@ module FoodFishParser::Strict::Grammar
     end
 
     rule fish_allergen_text
-      'vis'i
+      'visvlees'i / 'vis'i / 'weekdieren'i / 'schaaldieren'i
     end
   end
 end

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

@@ -6,7 +6,7 @@ module FoodFishParser::Strict::Grammar
     include FishNameNL
 
     rule fish_name_both
-      ( fish_name_nl ws* '(' ( ws* fish_allergen )? ws* fish_name_latin ( ws* ')' / comma )? )
+      ( !fish_allergen fish_name_nl ws* '(' ( ws* fish_allergen )? ws* fish_name_latin ( ws* ')' / comma )? )
     end
 
     rule fish_name_both_list
@@ -20,8 +20,8 @@ module FoodFishParser::Strict::Grammar
     end
 
     rule fish_name_nl_list
-      ( fish_name_nl <FishNameNode> )
-      ( ws+ and_or ws+ fish_name_nl <FishNameNode> )*
+      ( !fish_allergen fish_name_nl <FishNameNode> )
+      ( ws+ and_or ws+ !fish_allergen fish_name_nl <FishNameNode> )*
     end
 
     rule fish_name_any_list

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

@@ -1,4 +1,4 @@
-# autogenerated by species-treetop-gen-latin.rb on 2020-03-25
+# autogenerated by species-treetop-gen-latin.rb on 2020-03-31
 module FoodFishParser::Strict::Grammar
   grammar FishNameLatin
     include Common

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

@@ -1,4 +1,4 @@
-# autogenerated by species-treetop-gen-nl.rb on 2020-03-25
+# autogenerated by species-treetop-gen-nl.rb on 2020-03-31
 module FoodFishParser::Strict::Grammar
   grammar FishNameNL
     include Common
@@ -7,13 +7,19 @@ module FoodFishParser::Strict::Grammar
     rule fish_name_nl
       '(' ws* fish_name_nl ws* ')' /
       ( 'verse'i ws+ / 'kaviaar'i ws+ 'van'i ws+ / 'kaviaar'i ws+ / 'gevilde'i ws+ )?
-      ( fish_name_nl_area ws+ )? ( fish_name_nl_attr ws* )? fish_name_nl_name fish_name_nl_suffix?
+      (
+        fish_name_nl_area ws+ fish_name_nl_attr ws+ /
+        fish_name_nl_attr ws+ fish_name_nl_area ws+ /
+        fish_name_nl_area ws+ /
+        fish_name_nl_attr ws+
+      )?
+      fish_name_nl_name fish_name_nl_suffix?
       ( ws* fish_allergen )?
       <FishNameCommonNode>
     end
 
     rule fish_name_nl_area
-      'pacifische'i / 'noorse'i / 'japanse'i / 'indische'i / 'ijsselmeer'i / 'groenlandse'i / 'atlantische'i / 'argentijnse'i / 'alaska'i
+      'pacifische'i / 'pacific'i / 'noorse'i / 'japanse'i / 'indische'i / 'ijsselmeer'i / 'groenlandse'i / 'atlantische'i / 'argentijnse'i / 'alaska'i
     end
 
     rule fish_name_nl_attr
@@ -21,11 +27,11 @@ module FoodFishParser::Strict::Grammar
     end
 
     rule fish_name_nl_name
-      'zwaardvis'i / 'zonnevis'i / 'zeewolf'i / 'zeesnoek'i / 'zeekreeft'i / 'zeeforel'i / 'zeebaars'i / 'zalm'i / 'wijting'i / 'weekdieren'i / 'weekdier'i / 'vliegende vis'i / 'vis'i / 'vintonijn'i / 'vin tonijn'i / 'tonijn albacore'i / 'tonijn'i / 'tong'i / 'tilapia'i / 'thon albacore'i / 'tarbot'i / 'tapijtschelp'i / 'surimi'i / 'steur'i / 'st. pierre'i / 'sprot'i / 'spie'i / 'snotolf'i / 'snoekbaars'i / 'snoek'i / 'snapper'i / 'sliptong'i / 'skrei'i / 'skipjack tuna'i / 'skipjack tonijn'i / 'skipjack'i / 'seabob garnalen'i / 'schol'i / 'schelvis'i / 'schelpen'i / 'schelp'i / 'schar'i / 'sardines'i / 'sardinen'i / 'sardinelle'i / 'sardine'i / 'sardienen'i / 'sardien'i / 'saibling'i / 'rouget'i / 'rogvleugel'i / 'riddervis'i / 'regenboogforel'i / 'red snapper'i / 'red grouper'i / 'raat'i / 'poon'i / 'pollock'i / 'pollak'i / 'pilchard'i / 'pieterman'i / 'pangasius'i / 'paling'i / 'pagre'i / 'oogtonijn'i / 'nijlbaars'i / 'neustong'i / 'mul'i / 'mosselen'i / 'mossel'i / 'meerval'i / 'marlijn'i / 'mantelschelp'i / 'makreel'i / 'mahi-mahi'i / 'mahi mahi'i / 'maatjesha'i / 'lom'i / 'leng'i / 'kreeft'i / 'krab'i / 'koolvis'i / 'kokkel'i / 'king clip'i / 'karper'i / 'kabeljauw'i / 'inktvis'i / 'hondstong'i / 'hoki'i / 'heilbot'i / 'heekrug'i / 'heek'i / 'haring'i / 'harder diklip'i / 'hamvis'i / 'hake'i / 'haai'i / 'gruis'i / 'grouper'i / 'griet'i / 'geep'i / 'geelvintonijn'i / 'geelvin tonijn'i / 'garnalen'i / 'garnaal'i / 'forel'i / 'fint'i / 'escolar'i / 'dorade'i / 'diklipharder'i / 'diklip harder'i / 'daurade'i / 'corvina'i / 'coquilles'i / 'conger aal'i / 'cocquilles'i / 'cobia'i / 'claresse'i / 'caviaar'i / 'braadha'i / 'botervis'i / 'bot'i / 'blood snapper'i / 'beekridder'i / 'barracumundi'i / 'barracuda'i / 'bacalao'i / 'baars'i / 'arkschelp'i / 'ansovis'i / 'ansjovis'i / 'alfonsino'i / 'albacore tonijn'i
+      'zwaardvis'i / 'zonnevis'i / 'zeewolf'i / 'zeesnoek'i / 'zeekreeft'i / 'zeeforel'i / 'zeebaars'i / 'zalm'i / 'wijting'i / 'weekdieren'i / 'weekdier'i / 'vliegende vis'i / 'vis'i / 'vintonijn'i / 'vin tonijn'i / 'tonijn albacore'i / 'tonijn'i / 'tong'i / 'tilapia'i / 'thon albacore'i / 'tarbot'i / 'tapijtschelp'i / 'surimi'i / 'steur'i / 'st. pierre'i / 'sprot'i / 'spie'i / 'sockeye zalm'i / 'snotolf'i / 'snoekbaars'i / 'snoek'i / 'snapper'i / 'sliptong'i / 'skrei'i / 'skipjack tuna'i / 'skipjack tonijn'i / 'skipjack'i / 'seabob garnalen'i / 'schol'i / 'schelvis'i / 'schelpen'i / 'schelp'i / 'schar'i / 'sardines'i / 'sardinen'i / 'sardinelle'i / 'sardine'i / 'sardienen'i / 'sardien'i / 'saibling'i / 'rouget'i / 'rogvleugel'i / 'riddervis'i / 'regenboogforel'i / 'red snapper'i / 'red grouper'i / 'raat'i / 'poon'i / 'pollock'i / 'pollak'i / 'pilchard'i / 'pieterman'i / 'pangasius'i / 'paling'i / 'pagre'i / 'oogtonijn'i / 'nijlbaars'i / 'neustong'i / 'mul'i / 'mosselen'i / 'mossel'i / 'meerval'i / 'marlijn'i / 'mantelschelp'i / 'makreel'i / 'mahi-mahi'i / 'mahi mahi'i / 'maatjesha'i / 'lom'i / 'leng'i / 'kreeft'i / 'krab'i / 'koolvis'i / 'kokkel'i / 'king clip'i / 'karper'i / 'kabeljauw'i / 'inktvis'i / 'hondstong'i / 'hoki'i / 'heilbot'i / 'heekrug'i / 'heek'i / 'haring'i / 'harder diklip'i / 'hamvis'i / 'hake'i / 'haai'i / 'gruis'i / 'grouper'i / 'griet'i / 'geep'i / 'geelvintonijn'i / 'geelvin tonijn'i / 'garnalen'i / 'garnaal'i / 'forel'i / 'fint'i / 'escolar'i / 'dorade'i / 'diklipharder'i / 'diklip harder'i / 'daurade'i / 'corvina'i / 'coquilles'i / 'conger aal'i / 'cocquilles'i / 'cobia'i / 'claresse'i / 'caviaar'i / 'braadha'i / 'botervis'i / 'bot'i / 'blood snapper'i / 'beekridder'i / 'barracumundi'i / 'barracuda'i / 'bacalao'i / 'baars'i / 'arkschelp'i / 'ansovis'i / 'ansjovis'i / 'alfonsino'i / 'albacore tonijn'i
     end
 
     rule fish_name_nl_suffix
-      'wang'i / 'vlees'i / 'steur'i / 'ringen'i / 'ring'i / 'moten'i / 'kaviaar'i / 'filets'i / 'filetes'i / 'filet'i / 'eiwit'i
+      'wang'i / 'vlees'i / 'steur'i / 'ringen'i / 'ring'i / 'moten'i / 'lever'i / 'kaviaar'i / 'filets'i / 'filetes'i / 'filet'i / 'eiwit'i
     end
   end
 end

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

@@ -54,8 +54,9 @@ module FoodFishParser::Strict::Grammar
     # fish with catch or aquaculture info
     rule fish_with_info
       (
-        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_catch_info ) /
-        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_aquac_info )
+        # @todo move optional '(' after common fish name to root and properly match start and end brackets
+        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_catch_info ( ws* ')' )? ) /
+        ( fish_name_any_list ( ws* ( comma / ':' ) )? ws+ fish_aquac_info ( ws* ')' )? )
       )
       <FishNode>
     end

data/lib/food_fish_parser/version.rb

@@ -1,4 +1,4 @@
 module FoodFishParser
-  VERSION      = '0.3.7'
-  VERSION_DATE = '2020-03-25'
+  VERSION      = '0.3.8'
+  VERSION_DATE = '2020-03-31'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: food_fish_parser
 version: !ruby/object:Gem::Version
-  version: 0.3.7
+  version: 0.3.8
 platform: ruby
 authors:
 - wvengen
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-03-25 00:00:00.000000000 Z
+date: 2020-03-31 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: treetop
@@ -43,7 +43,10 @@ files:
 - bin/food_fish_parser
 - food_fish_parser.gemspec
 - lib/food_fish_parser.rb
-- lib/food_fish_parser/flat/fao_region.rb
+- lib/food_fish_parser/flat/aquac_method.rb
+- lib/food_fish_parser/flat/area_fao.rb
+- lib/food_fish_parser/flat/area_name.rb
+- lib/food_fish_parser/flat/catch_method.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