historical_dating 1.4.0 → 1.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 47c1a58ce160b759e2787853d78b7aa96428fe9e2f7076fdf96158a38664902f
-  data.tar.gz: 236c5fadd1663f7516fa09f5504d2fc5227c0290a8739814da2ebc7e0cf1dd36
+  metadata.gz: 8d4c906bc72a0b96c769243f6b025b0f15b6d0e96293f0a2b715706f80ff00e7
+  data.tar.gz: db559bafccbefa59ed6b24ff8a3d5797e197f29a323860a405fab056fb959d94
 SHA512:
-  metadata.gz: c9f810f66539567fa1cdc3f83e22106d533e94c6bb334d28ca884d1dcbf60e078835cd2a8b47f2490149cc7e4063cb76ff6f7d2078afdf13a4432f447a2300fb
-  data.tar.gz: 6b5ce82213a10cb258cad1062eccc431aeed3444679062b3584b08771351f467c12d65209344a4677c9380365eda9dca2ef8898c9b49742a045893ac99507a0a
+  metadata.gz: 61d0cbd348f836dd2bfb9e1b07ea7c5042534bbdfdfb7d0333e97f8332548d53d7668ec2606582e37acaece60fc8558257c89b21dd6504b7d544828d5aa544a0
+  data.tar.gz: 7b72d6a0d078730008dede42c8df257c20fdd43221d80eab32c605abb2d44b5632a3ba901eeea72624b21bc0582ae536c0f0a4840a9649b13c32f41cea290a9b

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    historical_dating (1.4.0)
+    historical_dating (1.5.0)
       activesupport (>= 3.0.0)
       parslet (~> 2.0.0)
 

data/lib/historical_dating/api.rb

@@ -15,4 +15,4 @@ module HistoricalDating::Api
     result = parser.transform(string)
     HistoricalDating::Range.new(result[:from], result[:to])
   end
-end
+end

data/lib/historical_dating/parser.rb

@@ -3,29 +3,68 @@ class HistoricalDating::Parser < Parslet::Parser
 
   rule(:zero){ str '0' }
   rule(:natural_number){ match['1-9'] >> match['0-9'].repeat }
+  rule(:two_digit_natural_number){ match['1-9'] >> match['0-9'].repeat(0,1) }
+  rule(:more_than_two_digit_natural_number){ match['1-9'] >> match['0-9'].repeat(2,nil) }
   rule(:positive_number){ zero | natural_number }
+  rule(:two_digit_positive_number){ zero | two_digit_natural_number }
+  rule(:more_than_two_digit_positive_number){ zero | more_than_two_digit_natural_number }
   rule(:minus){ match '-' }
   rule(:whole_number_without_zero){ natural_number | minus >> natural_number }
+  rule(:whole_number_with_zero){ positive_number | minus >> natural_number }
+  rule(:two_digit_whole_number_with_zero){ two_digit_positive_number | minus >> two_digit_natural_number }
+  rule(:more_than_two_digit_whole_number_with_zero){ more_than_two_digit_positive_number | minus >> more_than_two_digit_natural_number }
   rule(:whole_number){ positive_number | minus >> natural_number }
-  rule(:day){ match['1-2'] >> match['0-9'] | str('3') >> match['0-1'] | match['1-9'] | str('0') >> match['1-9'] }
-  rule(:month){ str('1') >> match['0-2'] | match['1-9'] | str('0') >> match['1-9'] }
 
   # Utility
 
   rule(:space){ str(' ').repeat(1, nil) }
+  rule(:prefix){ (str('von') | str('zwischen') | str('Zwischen')) >> space }
   rule(:christ){ str('Chr.') | str('Christus') }
-  rule(:age){ str('v.') | str('vor') | str('n.') | str('nach') }
-  rule(:acbc){ age >> space >> christ | str('BC') | str('AC') }
-  rule(:century_string){ str('Jahrhundert') | str('Jh.') | str('Jhd') }
-  rule(:approx){ str('ca.') | str('um') | str('circa') }
+  rule(:age_before){ str('v.') | str('vor') }
+  rule(:age_after){ str('n.') | str('nach') }
+  rule(:ac){
+    age_after >> space >> christ |
+    str('AC') |
+    str('Ac') |
+    str('Anno Domini') |
+    str('A. D.') |
+    str('AD')
+  }
+  rule(:bc){
+    age_before >> space >> christ |
+    str('BC') |
+    str('Bc')
+  }
+  rule(:acbc){
+    ac | bc
+  }
+  rule(:century_string){
+    str('Jahrhundert') |
+    str('Jhd.') |
+    str('jhd.') |
+    str('Jhd') |
+    str('jhd') |
+    str('Jh.') |
+    str('jh.') |
+    str('Jh') |
+    str('jh')
+  }
+  rule(:approx){ str('ca.') | str('Ca.') | str('ca') | str('um') | str('Um') | str('circa') }
   rule(:unknown){ str('?') }
+  rule(:to_characters){
+    str('bis') | str('-') | str('/') | str('und')
+  }
+  rule(:to_two_digit_year){
+    (space >> to_characters >> space) |
+    (to_characters >> space) |
+    (space >> to_characters) |
+    str('/')
+  }
   rule(:to){
-    ((space >> (str('bis') | str('-') | str('/')) >> space) |
-    str('-') |
-    str('/'))
+    to_two_digit_year | str('-')
   }
-  rule(:before){ str('vor') >> space }
-  rule(:after){ str('nach') >> space }
+  rule(:before){ (str('vor') | str('Vor') | str('before')) >> space }
+  rule(:after){ (str('Nach') | str('nach') | str('ab') | str('after')) >> space }
   rule(:negate){ str('nicht') >> space }
   rule(:part){
     str('Anfang') |
@@ -44,41 +83,69 @@ class HistoricalDating::Parser < Parslet::Parser
 
   # Dating
 
+  rule(:day){ match['1-2'] >> match['0-9'] | str('3') >> match['0-1'] | match['1-9'] | str('0') >> match['1-9'] }
+  rule(:month){ (str('0') >> match['1-9']) | (str('1') >> match['0-2']) | match['1-9'] }
+  rule(:two_digit_year){ (approx >> space).maybe.as(:approx) >> two_digit_whole_number_with_zero.as(:num) >> (space >> acbc).maybe.as(:acbc) }
+  rule(:more_than_two_digit_year){ (approx >> space).maybe.as(:approx) >> more_than_two_digit_whole_number_with_zero.as(:num) >> (space >> acbc).maybe.as(:acbc) }
+  rule(:year){ (approx >> space).maybe.as(:approx) >> whole_number_with_zero.as(:num) >> (space >> acbc).maybe.as(:acbc) }
   rule(:century){
     (approx >> space).maybe.as(:approx) >>
-    positive_number.as(:num) >> str('.').as(:cd) >> 
-    (space >> century_string.as(:cs) >> (space >> acbc).maybe.as(:acbc)).maybe
+    natural_number.as(:num) >>
+    ((str('.').as(:cd) >> (space >> century_string.as(:cs) >> (space >> acbc).maybe.as(:acbc)).maybe) |
+    (space >> century_string.as(:cs) >> (space >> acbc).maybe.as(:acbc)))
+  }
+  rule(:century_number){
+    (approx >> space).maybe.as(:approx) >>
+    natural_number.as(:num) >>
+    (space >> century_string).maybe.as(:cs) >>
+    (space >> acbc).maybe.as(:acbc)
   }
-  rule(:year){ (approx >> space).maybe.as(:approx) >> whole_number_without_zero.as(:num) >> (space >> acbc).maybe.as(:acbc) }
   rule(:century_part){ part.as(:part) >> space >> positive_number.as(:num) >> str('.') >> space >> century_string.as(:cs) >> (space >> acbc).maybe.as(:acbc) }
-  rule(:european_date){ day.as(:day) >> str('.') >> month.as(:month) >> str('.') >> whole_number.as(:yearnum) }
-  rule(:machine_date){ whole_number.as(:yearnum) >> (str('.') | str('-')) >> month.as(:month) >> (str('.') | str('-')) >> day.as(:day) }
+  rule(:european_date){
+    (day.as(:day) >> (str('.') | str('-')) >>
+    month.as(:month) >>
+    (str('.') | str('-')) >>
+    whole_number.as(:yearnum)) |
+    (month.as(:month) >>
+    (str('.') | str('-')) >>
+    whole_number.as(:yearnum))
+  }
+  rule(:machine_date){ whole_number.as(:yearnum) >> (str('.') | str('-')) >> month.as(:month) >> ((str('.') | str('-')) >> day.as(:day)).maybe }
   rule(:date){ european_date | machine_date }
   rule(:date_interval){ date.as(:from) >> to >> date.as(:to) }
-  rule(:century_interval){ century.as(:from) >> to >> century.as(:to) }
+  rule(:century_interval){
+    prefix.maybe >>
+    (century.as(:from) | century_number.as(:from)) >>
+    to >>
+    century.as(:to)
+  }
   rule(:before_year){ negate.maybe.as(:not) >> before >> year.as(:date) }
   rule(:after_year){ negate.maybe.as(:not) >> after >> year.as(:date) }
   rule(:year_interval){
-    year.as(:from) >> to >> (year | unknown).as(:to) |
-    (year | unknown).as(:from) >> to >> year.as(:to)
+    prefix.maybe >>
+    (year | unknown).as(:from) >>
+    ((to >>
+    (more_than_two_digit_year | unknown).as(:to)) |
+    (to_two_digit_year >>
+    two_digit_year.as(:to)))
   }
   rule(:before_century){ before >> century.as(:century) }
 
   rule(:interval){
     before_year.as(:before_year) |
-      after_year.as(:after_year) |
-      date_interval.as(:date_interval) |
-      century_interval.as(:century_interval) |
-      year_interval.as(:year_interval) |
-      before_century.as(:before_century)
+    after_year.as(:after_year) |
+    date_interval.as(:date_interval) |
+    century_interval.as(:century_interval) |
+    year_interval.as(:year_interval) |
+    before_century.as(:before_century)
   }
 
   rule(:dating){
     interval.as(:interval) |
-      century_part.as(:century_part) |
-      century.as(:century) |
-      date.as(:date) |
-      year.as(:year)
+    century_part.as(:century_part) |
+    century.as(:century) |
+    date.as(:date) |
+    year.as(:year)
   }
 
   root(:dating)

data/lib/historical_dating/pre_transform.rb

@@ -1,11 +1,19 @@
 class HistoricalDating::PreTransform < Parslet::Transform
   rule(from: subtree(:from), to: subtree(:to)) do
-    if to[:acbc] && (to[:acbc].match(/(vor|v.) (Chr.|Christus)/) || to[:acbc].match(/BC/))
+    bc = to[:acbc] && (to[:acbc].match(/(vor|v.) (Chr.|Christus)/) || to[:acbc].match(/B(C|c)/))
+
+    if bc
       from[:acbc] = to[:acbc]
     end
 
-    if to[:num].to_s.size == 2 && to[:cs].nil?
-      # e.g.: 1890/91
+    # To detect the first range of e.g.: 2. - 3. Jhd, :acbc key is needed.
+    # TODO: Find a better solution.
+    if from.is_a?(Hash) && to.is_a?(Hash) && !from.key?(:acbc) && to.key?(:acbc) && !to[:acbc]
+      from[:acbc] = nil
+    end
+
+    if !bc && to[:num].to_s.size == 2 && to[:cs].nil?
+      # for e.g.: 1890/91 but not for e.g.: 150 - 60 v. Chr.
       to[:num] = (from[:num].to_i / 100).to_s + to[:num]
     end
 
@@ -14,18 +22,28 @@ class HistoricalDating::PreTransform < Parslet::Transform
 
   # detect century by :cd and add :cs
   rule(num: simple(:num), approx: simple(:approx), acbc: simple(:acbc), cd: simple(:cd)) do |data|
-    data.delete :cd
-    data.merge(
-      cs: 'Jahrhundert'
-    )
+    if data[:cd]
+      data.delete :cd
+      data.merge(
+        cs: 'Jahrhundert'
+      )
+    else
+      data.delete :cd
+      data
+    end
   end
 
   # detect century by :cd and add :cs
   rule(num: simple(:num), approx: simple(:approx), cd: simple(:cd)) do |data|
-    data.delete :cd
-    data.merge(
-      cs: 'Jahrhundert'
-    )
+    if data[:cd]
+      data.delete :cd
+      data.merge(
+        cs: 'Jahrhundert'
+      )
+    else
+      data.delete :cd
+      data
+    end
   end
 
   # remove :cd if :cs already present (with acbc)
@@ -39,4 +57,10 @@ class HistoricalDating::PreTransform < Parslet::Transform
     data.delete :cd
     data
   end
+
+  # remove :prefix
+  rule(prefix: simple(:prefix)) do |data|
+    data.delete :prefix
+    data
+  end
 end

data/lib/historical_dating/transform.rb

@@ -15,7 +15,7 @@ class HistoricalDating::Transform < Parslet::Transform
     when "4. Viertel" then [75, 0]
     end
 
-    if acbc.nil? || acbc.match(/(nach|n.) (Chr.|Christus)/) || acbc.match(/AC/)
+    if HistoricalDating::Transform.ac?(acbc)
       {
         :from => Date.new((num.to_i - 1) * 100 + modifier.first, 1, 1),
         :to => Date.new((num.to_i - 1) * 100 + 99 - modifier.last, 12, 31)
@@ -29,7 +29,7 @@ class HistoricalDating::Transform < Parslet::Transform
   end
 
   rule(:num => simple(:num), :approx => simple(:approx), :acbc => simple(:acbc), :cs => simple(:cs)) do
-    result = if acbc.nil? || acbc.match(/(nach|n.) (Chr.|Christus)/) || acbc.match(/AC/)
+    result = if HistoricalDating::Transform.ac?(acbc)
       {
         :from => Date.new((num.to_i - 1) * 100, 1, 1),
         :to => Date.new((num.to_i - 1) * 100 + 99, 12, 31)
@@ -52,7 +52,7 @@ class HistoricalDating::Transform < Parslet::Transform
   rule(:num => simple(:num), :approx => simple(:approx), :acbc => simple(:acbc)) do
     modifier = (approx ? 5 : 0)
 
-    if acbc.nil? || acbc.match(/(nach|n.) (Chr.|Christus)/) || acbc.match(/AC/)
+    if HistoricalDating::Transform.ac?(acbc)
       {
         :from => Date.new(num.to_i - modifier, 1, 1),
         :to => Date.new(num.to_i + modifier, 12, 31)
@@ -76,6 +76,13 @@ class HistoricalDating::Transform < Parslet::Transform
     }
   end
 
+  rule(:month => simple(:month), :yearnum => simple(:yearnum)) do
+    {
+      :from => Date.new(yearnum.to_i, month.to_i, 1),
+      :to => Date.new(yearnum.to_i, month.to_i, -1),
+    }
+  end
+
   rule(:date => {:from => simple(:from), :to => simple(:to)}) do
     {:from => from, :to => to}
   end
@@ -136,6 +143,10 @@ class HistoricalDating::Transform < Parslet::Transform
     end
   end
 
+  def self.ac?(acbc)
+    acbc.nil? || acbc.match(/(nach|n.) (Chr.|Christus)/) || acbc.match(/AC|Ac|Anno Domini|A. D.|AD/)
+  end
+
   def self.open_start(year)
     return {
       from: Date.new(year - distance(year), 1, 1),

data/lib/historical_dating/version.rb

@@ -1,3 +1,3 @@
 module HistoricalDating
-  VERSION = "1.4.0"
+  VERSION = "1.5.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: historical_dating
 version: !ruby/object:Gem::Version
-  version: 1.4.0
+  version: 1.5.0
 platform: ruby
 authors:
 - Moritz Schepp
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-11-23 00:00:00.000000000 Z
+date: 2020-11-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: parslet