edifact_parser 0.9.0 → 0.9.1

data/lib/edifact_parser/parser.rb

@@ -30,38 +30,38 @@ module EdifactParser
 ##### State transition tables begin ###
 
 racc_action_table = [
-     7,    27,    26,     4,    22,    17,    18,    17,    18,    27,
-    26,     7,     7,     7,     8,    19 ]
+     7,    28,    27,     4,    20,    18,    19,    20,    18,    19,
+    28,    27,     7,     7,     7,     8,    21 ]
 
 racc_action_check = [
-     0,    15,    15,     0,    11,    11,    11,     6,     6,    16,
-    16,     3,    10,     2,     1,     8 ]
+     0,    16,    16,     0,    11,    11,    11,     6,     6,     6,
+    17,    17,     3,    10,     2,     1,     8 ]
 
 racc_action_pointer = [
-    -2,    14,    11,     9,   nil,   nil,     0,   nil,    15,   nil,
-    10,    -2,   nil,   nil,   nil,    -2,     6,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil ]
+    -2,    15,    12,    10,   nil,   nil,     1,   nil,    16,   nil,
+    11,    -2,   nil,   nil,   nil,   nil,    -2,     7,   nil,   nil,
+   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil ]
 
 racc_action_default = [
-   -23,   -23,    -1,   -23,    -3,    -5,   -23,   -13,   -23,    -4,
-    -2,   -23,    -8,    -9,   -10,   -11,   -12,   -20,   -21,    29,
-    -6,    -7,   -22,   -14,   -16,   -17,   -18,   -19,   -15 ]
+   -24,   -24,    -1,   -24,    -3,    -5,   -24,   -14,   -24,    -4,
+    -2,   -24,    -7,    -9,   -10,   -11,   -12,   -13,   -21,   -22,
+   -23,    30,    -6,    -8,   -15,   -17,   -18,   -19,   -20,   -16 ]
 
 racc_goto_table = [
-     9,    12,     2,    23,    28,    10,    21,    11,     9,    20,
-     3,     1 ]
+     9,    12,    13,    24,    29,    11,    22,    23,     9,     2,
+     3,     1,    10 ]
 
 racc_goto_check = [
-     4,     8,     2,    13,    13,     2,     8,     6,     4,     7,
-     3,     1 ]
+     4,     7,     8,    13,    13,     6,     7,     8,     4,     2,
+     3,     1,     2 ]
 
 racc_goto_pointer = [
-   nil,    11,     2,    10,    -2,   nil,     1,    -2,    -5,   nil,
-   nil,   nil,   nil,   -12,   nil,   nil ]
+   nil,    11,     9,    10,    -2,   nil,    -1,    -5,    -4,   nil,
+   nil,   nil,   nil,   -13,   nil,   nil ]
 
 racc_goto_default = [
-   nil,   nil,   nil,   nil,     5,     6,   nil,   nil,   nil,    13,
-    14,    15,    16,   nil,    24,    25 ]
+   nil,   nil,   nil,   nil,     5,     6,   nil,   nil,   nil,    14,
+    15,    16,    17,   nil,    25,    26 ]
 
 racc_reduce_table = [
   0, 0, :racc_error,
@@ -71,26 +71,27 @@ racc_reduce_table = [
   2, 11, :_reduce_none,
   1, 11, :_reduce_none,
   3, 13, :_reduce_none,
+  2, 13, :_reduce_none,
   2, 15, :_reduce_none,
   1, 15, :_reduce_none,
   1, 17, :_reduce_none,
   1, 17, :_reduce_none,
   1, 17, :_reduce_none,
   1, 17, :_reduce_none,
-  1, 14, :_reduce_13,
+  1, 14, :_reduce_14,
   2, 18, :_reduce_none,
   2, 19, :_reduce_none,
   1, 22, :_reduce_none,
   1, 22, :_reduce_none,
-  1, 24, :_reduce_18,
-  1, 23, :_reduce_19,
-  1, 20, :_reduce_20,
-  1, 21, :_reduce_21,
-  1, 16, :_reduce_22 ]
+  1, 24, :_reduce_19,
+  1, 23, :_reduce_20,
+  1, 20, :_reduce_21,
+  1, 21, :_reduce_22,
+  1, 16, :_reduce_23 ]
 
-racc_reduce_n = 23
+racc_reduce_n = 24
 
-racc_shift_n = 29
+racc_shift_n = 30
 
 racc_token_table = {
   false => 0,
@@ -180,40 +181,42 @@ Racc_debug_parser = true
 
 # reduce 12 omitted
 
-def _reduce_13(val, _values, result)
+# reduce 13 omitted
+
+def _reduce_14(val, _values, result)
  @handler.start_segment; @handler.qualifier val[0] 
     result
 end
 
-# reduce 14 omitted
-
 # reduce 15 omitted
 
 # reduce 16 omitted
 
 # reduce 17 omitted
 
-def _reduce_18(val, _values, result)
+# reduce 18 omitted
+
+def _reduce_19(val, _values, result)
  @handler.scalar val[0] 
     result
 end
 
-def _reduce_19(val, _values, result)
+def _reduce_20(val, _values, result)
  @handler.scalar val[0].gsub("?", "") 
     result
 end
 
-def _reduce_20(val, _values, result)
+def _reduce_21(val, _values, result)
  @handler.end_element; @handler.start_element 
     result
 end
 
-def _reduce_21(val, _values, result)
+def _reduce_22(val, _values, result)
  @handler.colon 
     result
 end
 
-def _reduce_22(val, _values, result)
+def _reduce_23(val, _values, result)
  @handler.end_element; @handler.end_segment 
     result
 end

data/lib/edifact_parser/tokenizer.rb

@@ -2,8 +2,23 @@ require 'strscan'
 
 module EdifactParser
   class Tokenizer
+    QUALIFIERS = [
+      'UNB', 'UNH', 'UNZ', 'UNT', 'UNS', 'IFI', 'IFT', 'ODI', 'TVL', 'APD',
+      'ADR', 'AGR', 'AJT', 'ALC', 'ALI', 'APP', 'APR', 'ARD', 'ARR', 'ASI', 'ATT', 'AUT',
+      'BAS', 'BGM', 'BII', 'BUS', 'CAV', 'CCD', 'CCI', 'CDI', 'CDS', 'CDV', 'CED', 'CIN',
+      'CLA', 'CLI', 'CMP', 'CNI', 'CNT', 'COD', 'COM', 'COT', 'CPI', 'CPS', 'CPT', 'CST',
+      'CTA', 'CUX', 'DAM', 'DFN', 'DGS', 'DII', 'DIM', 'DLI', 'DLM', 'DMS', 'DOC', 'DRD',
+      'DSG', 'DSI', 'DTM', 'EDT', 'EFI', 'ELM', 'ELU', 'ELV', 'EMP', 'EQA', 'EQD', 'EQN',
+      'ERC', 'ERP', 'EVE', 'FCA', 'FII', 'FNS', 'FNT', 'FOR', 'FSQ', 'FTX', 'GDS', 'GEI',
+      'GID', 'GIN', 'GIR', 'GOR', 'GRU', 'HAN', 'HYN', 'ICD', 'IDE', 'IFD', 'IHC', 'IMD',
+      'IND', 'INP', 'INV', 'IRQ', 'LAN', 'LIN', 'LOC', 'MEA', 'MEM', 'MKS', 'MOA', 'MSG',
+      'MTD', 'NAD', 'NAT', 'PAC', 'PAI', 'PAS', 'PCC', 'PCD', 'PCI', 'PDI', 'PER', 'PGI',
+      'PIA', 'PNA', 'POC', 'PRC', 'PRI', 'PRV', 'PSD', 'PTY', 'PYT', 'QRS', 'QTY', 'QUA',
+      'QVR', 'RCS', 'REL', 'RFF', 'RJL', 'RNG', 'ROD', 'RSL', 'RTE', 'SAL', 'SCC', 'SCD',
+      'SEG', 'SEL', 'SEQ', 'SFI', 'SGP', 'SGU', 'SPR', 'SPS', 'STA', 'STC', 'STG', 'STS',
+      'TAX', 'TCC', 'TDT', 'TEM', 'TMD', 'TMP', 'TOD', 'TPL', 'TRU', 'TSR', 'VLI' ]
+
     OPTIONAL_BEGIN = /^UNA:\+\.\?\s'/
-    QUALIFIER = /(^|(?<='))(UNB|UNH|BGM|DTM|PAI|ALI|IMD|FTX|LOC|GIS|DGS|GIR|RFF|MEA|QTY|MOA|RTE|NAD|DOC|CTA|COM|TAX|PCD|CUX|TDT|TSR|PII|TOD|PAC|EQD|SEL|ALC|RNG|INP|LIN|PRI|UNS|UNT|UNZ|CNT)/
     STRING = /[A-Za-z0-9\s\.]*(\?')*[A-Za-z0-9\s\.]*(\?\+)*[A-Za-z0-9\s\.]*(\?:)*[A-Za-z0-9\s\.]*/
     SPACE = /\s+/
     NUMBER = /[0-9]+(?=[\+:'])/
@@ -14,6 +29,7 @@ module EdifactParser
 
     def initialize(io)
       @ss = StringScanner.new io.read
+      @qualifiers = qualifier_regex
     end
 
     def next_token
@@ -27,7 +43,7 @@ module EdifactParser
 
       case
       when text = @ss.scan(OPTIONAL_BEGIN) then [:OPTIONAL_BEGIN, text]
-      when text = @ss.scan(QUALIFIER) then [:QUALIFIER, text]
+      when text = @ss.scan(@qualifiers) then [:QUALIFIER, text]
       when text = @ss.scan(SEGMENT_END) then [:SEGMENT_END, text]
       when text = @ss.scan(PLUS) then [:PLUS, text]
       when text = @ss.scan(COLON) then [:COLON, text]
@@ -38,5 +54,18 @@ module EdifactParser
         [x, x]
       end
     end
+
+    private
+
+    def qualifier_regex
+      reg_ex_string = ""
+
+      QUALIFIERS.each do |qual|
+        reg_ex_string += "#{qual}|"
+      end
+
+      /(^|(?<='))(#{reg_ex_string[0, reg_ex_string.length-1]})/
+    end
+
   end
-end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: edifact_parser
 version: !ruby/object:Gem::Version
-  version: 0.9.0
+  version: 0.9.1
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-01-26 00:00:00.000000000 Z
+date: 2013-01-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: racc