food_ingredient_parser 1.1.10 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a56d22b7e67a3a913b051bcbda8da885ddd467dc53f5a0df0faa5b40759a1f35
-  data.tar.gz: 427dd79c9f9203dc7901ead6264e08c05183d02aec266ac1d3bff930a5ba1dcd
+  metadata.gz: 7c478a080e36c8f48ee3dbd6e9978eadec3758a4b0ab6fab571e18f103ed6bf0
+  data.tar.gz: aa078366f72ab03d038d497c908a3ad92f5816f37d3f0308fa64e81680905dea
 SHA512:
-  metadata.gz: 0b07032ade3a55ce208bcb0c069223b41aee21f185a2b6a9bb91332881dfef8e1d829ae966097e48ffdba9984517be43b10bd027099f9bdce04e3a4c6fc41ca8
-  data.tar.gz: ebdf452a09d54b151ce8cfa9bb65b4477dd1afc81bfc5cd1d94055f726d387f522dd04a3e11b73d4b26222a18bc1068912a81c8e2e3cd8439b0cee1c1ec290d7
+  metadata.gz: d8acbd71e431958a72350e6fd1d3e5e8d21db8ee53525c53a08bbe2c564734fca9601ac0fdc33d9737695f292bc7cd6da898721f02f68ca8f87175c5b276c709
+  data.tar.gz: f261a1537a6e903d55b36dc91c0a1c302893d7a092a83afb8598730c87142d041fdaf86540918697f297f5024461082e5e56594df64f603f56f4b22148c7c9fd

data/README.md

@@ -104,7 +104,7 @@ RootNode+Root3 offset=0, "tomato" (contains,notes):
   SyntaxNode offset=6, ""
 {:contains=>[{:name=>"tomato"}]}
 
-$ bin/food_ingredient_parser --html -s "tomato"
+$ food_ingredient_parser --html -s "tomato"
 <div class="root"><span class='depth0'><span class='name'>tomato</span></span></div>
 
 $ food_ingredient_parser -v -r loose -s "tomato"
@@ -197,7 +197,7 @@ plus a bit of English and German. Support for other languages is already good, b
 areas: improvements are welcome (starting with a corpus in [data/](data/)).
 
 Many ingredient lists from the USA are structured a bit differently than those from Europe, they
-parse less well (that that's a matter of tine-tuning).
+parse less well (that is probably a matter of tine-tuning).
 
 ## Test data
 

data/lib/food_ingredient_parser/loose/scanner.rb

@@ -4,8 +4,9 @@ module FoodIngredientParser::Loose
   class Scanner
 
     SEP_CHARS  = "|;,.".freeze
+    AND_SEP_RE = /\A\s*(and|en|und)\s+/i.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
+    PREFIX_RE  = /\A\s*(ingredients(\s*list)?|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.
     ABBREV_RE  = Regexp.union(
@@ -23,8 +24,8 @@ module FoodIngredientParser::Loose
         www\.[-_\/:%.A-Za-z0-9]+
       )/xi,
       *%w[
-        a.o.p b.g.a b.o.b c.a c.i d.e d.m.v d.o.c d.o.p d.s e.a e.g e.u f.i.l f.o.s i.a
-        i.d i.e i.g.m.e i.g.p i.m.v i.o i.v.m l.s.l n.a n.b n.o n.v.t o.a o.b.v p.d.o
+        a.o.p b.g.a b.o.b c.a c.i d.e d.m.v d.o.c d.o.p d.s e.a e.g e.u f.i.l f.o.s h.o.h
+        i.a i.d i.e i.g.m.e i.g.p i.m.v i.o i.v.m l.s.l n.a n.b n.o n.v.t o.a o.b.v p.d.o
         p.g.i q.s s.l s.s t.o.v u.h.t v.g v.s w.a w.o w.v vit denat alc vol conc subsp
         min max ca
       ].map {|s| /\A#{Regexp.escape(s)}\b\.?/i}
@@ -75,6 +76,11 @@ module FoodIngredientParser::Loose
       elsif ")]".include?(c)      # close nesting
         add_child
         close_parent
+        # after bracket check for 'and' to not lose text
+        if is_and_sep?(@i+1)
+          @i += and_sep_len(@i+1)
+          add_child
+        end
       elsif is_notes_start?       # usually a dot marks the start of notes
         close_all_ancestors
         @iterator = :notes
@@ -148,6 +154,15 @@ module FoodIngredientParser::Loose
       chars.include?(c) && @s[@i-1..@i+1] !~ /\A\d.\d\z/
     end
 
+    def is_and_sep?(i = @i)
+      and_sep_len(i) > 0
+    end
+
+    def and_sep_len(i = @i)
+      m = @s[i..-1].match(AND_SEP_RE)
+      m ? m.offset(0).last : 0
+    end
+
     def is_mark?(i = @i)
       mark_len(i) > 0 && @s[i..i+1] !~ /\A°[CF]/
     end

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

@@ -102,6 +102,7 @@ module FoodIngredientParser::Strict::Grammar
         'e.u'i /
         'f.i.l'i /
         'f.o.s'i /
+        'h.o.h'i /
         'i.a'i /
         'i.d'i /
         'i.e'i /

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

@@ -5,7 +5,12 @@ module FoodIngredientParser::Strict::Grammar
     include IngredientColoned
 
     rule ingredient
-      ws* ( ingredient_nested / ingredient_coloned / ingredient_simple_with_amount )
+      ws*
+      (
+        ingredient_nested ( ws* and ws+ ingredient )? /
+        ingredient_coloned /
+        ingredient_simple_with_amount
+      )
     end
 
   end

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

@@ -19,9 +19,10 @@ module FoodIngredientParser::Strict::Grammar
 
     rule root_prefix
       (
-        'ingredients'i / 'contains'i /
+        'ingredients'i ( ws+ 'list'i )? / 'contains'i /
         ('ingred'i [IÏiï] [EËeë] 'n'i ( 't'i 'en'i? 'declaratie'i? )? ) / 'bevat'i / 'dit zit er in'i / 'samenstelling'i /
-        'zutaten'i
+        'zutaten'i /
+        'ingredienser'i
       )
       ( ws* [:;.] ( ws* newline )? / ws* newline / ws ) ws*  # optional colon or other separator
       "'"? ws*                                               # stray quote occurs sometimes

data/lib/food_ingredient_parser/version.rb

@@ -1,4 +1,4 @@
 module FoodIngredientParser
-  VERSION      = '1.1.10'
-  VERSION_DATE = '2021-03-23'
+  VERSION      = '1.2.0'
+  VERSION_DATE = '2024-01-19'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: food_ingredient_parser
 version: !ruby/object:Gem::Version
-  version: 1.1.10
+  version: 1.2.0
 platform: ruby
 authors:
 - wvengen
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-03-23 00:00:00.000000000 Z
+date: 2024-01-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: treetop
@@ -87,7 +87,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.3
+rubygems_version: 3.1.6
 signing_key: 
 specification_version: 4
 summary: Parser for ingredient lists found on food products.