food_ingredient_parser 1.1.6 → 1.1.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e65fb234a122c893fcb4de453e4cac07851d5000b848fa99d2b356002e187a60
-  data.tar.gz: c72d35c6fc978a6ddc139e3adf2059c24104221627f04857ba73aad159b29a0f
+  metadata.gz: 59825ee90990b2c4f52c9e59fae2e34e5b4558bc63a57fc59946db7f71335351
+  data.tar.gz: b8201945554a11fddbac8eb6676c4cbff8c1f5d3523780ec54125e3172fd05ac
 SHA512:
-  metadata.gz: 295728820e893f1277bb06cfdad572df1fde9d496df230caf55f41b23d124677131b004a5ca0b2a2e4a25b2273f98feb18840c47c6361ec474082d66d4fde7fb
-  data.tar.gz: 0ca48100273055b6c8f6e1d89bd1b44976e325cc9633080d6570e52cb6ef5d07dd419013d9b9aff67f95b6e4fac6ae12396c9248123daf4902d3eb97d84d1b28
+  metadata.gz: 3b97c863f9da5b26162883a3627809857fa9277e60e76f2c805312ceea34ad40317bcdf6ca2f1c56b2eafc4a5c637bbe0b5ac3c3c24ef8f874dd008bb5cc3bd7
+  data.tar.gz: bfcc88aea38c3db84670e84dd1166cbcdf8ce60edff297a2ed7324616ea53a4625ab137423f2f6701981a27a51b398ab652b207abe435328e4fa67974bad3f01

data/lib/food_ingredient_parser/cleaner.rb

@@ -2,14 +2,14 @@ module FoodIngredientParser
   module Cleaner
 
     def self.clean(s)
-      s.gsub!("\u00ad", "")             # strip soft hyphen
-      s.gsub!("\u0092", "'")            # windows-1252 apostrophe - https://stackoverflow.com/a/15564279/2866660
-      s.gsub!("‚", ",")                 # normalize unicode comma
-      s.gsub!("aÄs", "aïs")             # encoding issue for maïs
-      s.gsub!("ï", "ï")                # encoding issue
-      s.gsub!("ë", "ë")                # encoding issue
-      s.gsub!(/\A\s*"(.*)"\s*\z/, '\1') # enclosing double quotation marks
-      s.gsub!(/\A\s*'(.*)'\s*\z/, '\1') # enclosing single quotation marks
+      s.gsub!(/(_x005f_|_)x000d_/i, "\n") # fix sometimes encoding for newline
+      s.gsub!("\u00ad", "")               # strip soft hyphen
+      s.gsub!("\u0092", "'")              # windows-1252 apostrophe - https://stackoverflow.com/a/15564279/2866660
+      s.gsub!("‚", ",")                   # normalize unicode comma
+      s.gsub!("aÄs", "aïs")               # encoding issue for maïs
+      s.gsub!("ï", "ï")                  # encoding issue
+      s.gsub!("ë", "ë")                  # encoding issue
+      s.gsub!(/\A\s*(["']+)(.*)\1\s*\z/, '\2') # enclosing quotation marks
       s
     end
 

data/lib/food_ingredient_parser/loose/scanner.rb

@@ -4,7 +4,7 @@ module FoodIngredientParser::Loose
   class Scanner
 
     SEP_CHARS  = "|;,.".freeze
-    MARK_CHARS = "¹²³⁴⁵ᵃᵇᶜᵈᵉᶠᵍªº⁽⁾†‡•°▪◊#^˄*~".freeze
+    MARK_CHARS = "¹²³⁴⁵ᵃᵇᶜᵈᵉᶠᵍªº⁽⁾†‡⁺•°▪◊#^˄*~".freeze
     PREFIX_RE  = /\A\s*(ingredients|contains|ingred[iï][eë]nt(en)?(declaratie)?|bevat|dit zit er\s?in|samenstelling|zutaten)\b\s*[:;.]?\s*/i.freeze
     NOTE_RE    = /\A\b(dit product kan\b|deze verpakking kan\b|kan sporen\b.*?\bbevatten\b|voor allergenen\b|allergenen\b|allergie[- ]informatie(\s*:|\b)|E\s*=|gemaakt in\b|geproduceerd in\b|bevat mogelijk\b|kijk voor meer\b|allergie-info|in de fabriek\b|in dit bedrijf\b|voor [0-9,.]+ (g\.?|gr\.?|ram|ml).*\bis [0-9,.]+ (g\.?|gr\.?|ram|ml).*\bgebruikt\b)/i.freeze
     # Keep in sync with +abbrev+ in the +Common+ grammar, plus relevant ones from the +Amount+ grammar.

data/lib/food_ingredient_parser/strict/grammar/amount.treetop

@@ -21,22 +21,28 @@ module FoodIngredientParser::Strict::Grammar
       ) ws* )?
       amount_simple_quantity
       ( ws+ (
-        'of'i / 'or less of'i / 'or more of'i /
-        'minimaal'i / 'minimum'i / 'van het uitlekgewicht'i / 'van het geheel'i /
+        'of a'i / 'of'i / 'or less of'i / 'or more of'i /
+        'van een'i / 'minimaal'i / 'minimum'i / 'van het uitlekgewicht'i / 'van het geheel'i /
         'min.'i / 'min'i / 'max.'i / 'max'i
       ) )?
     end
 
     rule amount_simple_quantity
-      amount_simple_number ( ws* amount_simple_unit? ws* dash ws* amount_simple_number )? ws* amount_simple_unit
+      amount_simple_number ( ws* amount_simple_unit? ws* dash ws* amount_simple_number )? ( ws* amount_simple_unit )?
     end
 
     rule amount_simple_number
-      ( [±∓~∼∽≂≃≈≲≤<>≥≳] ws* )? number
+      ( amount_simple_comparator ws* )? number
+    end
+
+    rule amount_simple_comparator
+      '=' ws* [<>] /
+      [<>] ws* ( '=' / 'of gelijk aan'i !char / 'or equal to'i !char ) /
+      [±∓~∼∽≂≃≈≲≤<>≥≳] / '+/-' / '-/+'
     end
 
     rule amount_simple_unit
-      ( percent / ( ( 'procent' / 'percent' / 'gram'i / 'ml'i / 'mg'i / 'gr'i / 'g'i ) !char ) )
+      ( percent / ( ( 'procent' / 'percent' / 'gram'i / 'ml'i / 'mg'i / 'gr'i / 'g'i / 'ppm'i ) !char ) )
       ( ws 'vol'i ( !char / '.' ) )?
       ( ws* '℮' )?
     end

data/lib/food_ingredient_parser/strict/grammar/common.treetop

@@ -23,9 +23,9 @@ module FoodIngredientParser::Strict::Grammar
       [¹²³⁴⁵ᵃᵇᶜᵈᵉᶠᵍªº] '⁾'? /
       '⁽' [¹²³⁴⁵ᵃᵇᶜᵈᵉᶠᵍªº] '⁾' /
       [˄^] digit /
-      [†‡•°▪◊#˄^~˛] /
+      [†‡⁺•°▪◊#˄^~˛] /
       '*'+ /
-      '(' ws* ( [†‡•°▪◊#˄^~˛] / '*'+ ) ws* ')'
+      '(' ws* ( [†‡⁺•°▪◊#˄^~˛] / '*'+ ) ws* ')'
     end
 
     rule digit
@@ -33,7 +33,8 @@ module FoodIngredientParser::Strict::Grammar
     end
 
     rule fraction
-      [½⅓⅔¼¾⅕⅖⅗⅘⅙⅚⅐⅛⅜⅝⅞⅑⅒]
+      [½⅓⅔¼¾⅕⅖⅗⅘⅙⅚⅐⅛⅜⅝⅞⅑⅒] /
+      digit+ '/' digit+
     end
 
     rule percent
@@ -61,6 +62,20 @@ module FoodIngredientParser::Strict::Grammar
       ![[:alnum:]] / ( ws* '(' 'i'i+ ')' ) # e.g. "E450 (iii)"
     end
 
+    rule chem_systematic_name
+      ( chem_systematic_name_num dash ) ( chem_systematic_name_word dash chem_systematic_name_num dash ws? )* chem_systematic_name_word /
+      ( chem_systematic_name_word dash chem_systematic_name_num dash ws? )+ chem_systematic_name_word
+    end
+
+    rule chem_systematic_name_word
+      [A-Za-z]+ ( dash [A-Za-z]+ dash [A-Za-z]+ )*
+    end
+
+    rule chem_systematic_name_num
+      digit+ [RH] /
+      digit+ ( ',' digit+ )* '\''?
+    end
+
     rule abbrev
       # These are listed explicitely to avoid incorrect interpretations, and allow missing trailing dots.
       # To get an idea of what occurs (second one omits trailing dots):
@@ -115,6 +130,7 @@ module FoodIngredientParser::Strict::Grammar
         'w.o'i /
         'w.v'i /
         # not auto-generated additions
+        'nr.'i /
         'vit'i /   # vitamin
         'denat'i / # denaturated
         'alc'i /   # alcohol
@@ -138,7 +154,8 @@ module FoodIngredientParser::Strict::Grammar
         'T.' ws* 'aestivum'i (ws+ 'vitt.'i)? /
         'nucifera' ws+ 'L.'i /
         'type'i ws+ '"' [0-9]+ '"' /
-        e_number
+        e_number /
+        chem_systematic_name
       ) ![[:alpha:]]
     end
   end

data/lib/food_ingredient_parser/version.rb

@@ -1,4 +1,4 @@
 module FoodIngredientParser
-  VERSION      = '1.1.6'
-  VERSION_DATE = '2020-11-19'
+  VERSION      = '1.1.7'
+  VERSION_DATE = '2020-12-28'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: food_ingredient_parser
 version: !ruby/object:Gem::Version
-  version: 1.1.6
+  version: 1.1.7
 platform: ruby
 authors:
 - wvengen
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-11-19 00:00:00.000000000 Z
+date: 2020-12-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: treetop