sportdb-formats 1.1.2 → 1.1.3

data/lib/sportdb/formats/team/club_reader_history.rb

@@ -0,0 +1,203 @@
+# encoding: utf-8
+
+
+module SportDb
+module Import
+
+
+class ClubHistoryReader
+
+  def catalog() Import.catalog; end
+
+
+
+def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
+  txt = File.open( path, 'r:utf-8' ) { |f| f.read }
+  parse( txt )
+end
+
+def self.parse( txt )
+  new( txt ).parse
+end
+
+def initialize( txt )
+  @txt = txt
+end
+
+
+###
+## RENAME/RENAMED
+## MOVE/MOVED
+## BANKRUPT/BANKRUPTED
+## REFORM/REFORMED
+## MERGE/MERGED    - allow + or ++ or +++ or ; for "inline" - why? why not?
+
+
+KEYWORD_LINE_RE = %r{ ^(?<keyword>RENAMED?|
+                               MOVED?|
+                               BANKRUPT(?:ED)?|
+                               REFORM(?:ED)?|
+                               MERGED?
+                    )
+                      [ ]+
+                     (?<text>.*)    # rest of text
+                    $
+                  }x
+
+
+def parse
+  recs = []
+  last_rec  = nil
+
+  last_country = nil
+  last_season  = nil
+  last_keyword = nil
+  last_teams   = []
+
+  OutlineReader.parse( @txt ).each do |node|
+    if [:h1,:h2,:h3,:h4,:h5,:h6].include?( node[0] )
+      heading_level  = node[0][1].to_i
+      heading        = node[1]
+
+      puts "heading #{heading_level} >#{heading}<"
+
+
+        if heading_level == 1
+            ## assume country in heading; allow all "formats" supported by parse e.g.
+            ##   Österreich • Austria (at)
+            ##   Österreich • Austria
+            ##   Austria
+            ##   Deutschland (de) • Germany
+            country = catalog.countries.parse( heading )
+            ## check country code - MUST exist for now!!!!
+            if country.nil?
+              puts "!!! error [club history reader] - unknown country >#{heading}< - sorry - add country to config to fix"
+              exit 1
+            end
+            puts "  country >#{heading}< => #{country.name}, #{country.key}"
+            last_country = country
+            last_season  = nil  ## reset "lower levels" - season & keyword
+            last_keyword = nil
+         elsif heading_level == 2
+            ## assume season
+            season = Season.new( heading )
+            puts "  season >#{heading}< => #{season.key}"
+            last_season  = season  ## reset "lowwer levels" - keyword
+            last_keyword = nil
+         else
+            puts "!!! ERROR [club history reader] - for now only heading 1 & 2 supported; sorry"
+            exit 1
+         end
+
+    elsif node[0] == :p   ## paragraph with (text) lines
+      if last_country.nil?
+        puts "!!! ERROR [club history reader] - country heading 1 required, sorry"
+        exit 1
+      end
+      if last_season.nil?
+        puts "!!! ERROR [club history reader] - season heading 2 required, sorry"
+        exit 1
+      end
+
+      lines = node[1]
+      lines.each do |line|
+        if m=line.match(KEYWORD_LINE_RE)   ## extract keyword and continue
+          keyword = m[:keyword]
+          line    = m[:text].strip
+
+          puts "    keyword #{keyword}"
+          last_keyword = case keyword   ## "normalize" keywords
+                         when 'BANKRUPT', 'BANKRUPTED'
+                           'BANKRUPT'
+                         when 'RENAME', 'RENAMED'
+                           'RENAME'
+                         when 'REFORM', 'REFORMED'
+                           'REFORM'
+                         when 'MOVE',   'MOVED'
+                           'MOVE'
+                         when 'MERGE',  'MERGED'
+                           'MERGE'
+                         else
+                           puts "!!! ERROR [club history reader] - unexpected keyword >#{keyword}<; sorry - don't know how to normalize"
+                           exit 1
+                         end
+
+          last_teams   = []
+        end
+
+        if last_keyword.nil?
+          puts "!!! ERROR [club history reader] - line with keyword expected - got:"
+          puts line
+          exit 1
+        end
+
+        if    last_keyword == 'BANKRUPT'
+           ## requires / expects one team in one line
+           recs << [ last_keyword, last_season.key,
+                     [ squish(line), last_country.key ]
+                   ]
+        elsif last_keyword == 'RENAME' ||
+              last_keyword == 'REFORM' ||
+              last_keyword == 'MOVE'
+           ## requires / expects two teams in one line (separated by ⇒ or such)
+           teams = line.split( '⇒' )
+           if teams.size != 2
+            puts "!!! ERROR [club history reader] - expected two teams - got:"
+            pp teams
+            exit 1
+           end
+           teams = teams.map {|team| squish(team.strip) }  ## remove whitespaces
+           recs << [ last_keyword, last_season.key,
+                     [ teams[0], last_country.key ],
+                     [ teams[1], last_country.key ]
+                   ]
+        elsif last_keyword == 'MERGE'
+            ## check if line starts with separator
+            ## otherwise collect to be merged teams
+            if line.start_with?( '⇒' )
+              if last_teams.size < 2
+                puts "!!! ERROR [club history reader] - expected two or more teams for MERGE - got:"
+                pp last_teams
+                exit 1
+              end
+              ## auto-add country to all teams
+              teams = last_teams.map {|team| [team, last_country.key]}
+              recs << [ last_keyword, last_season.key,
+                        teams,
+                        [ squish(line.sub('⇒','').strip), last_country.key ]
+                      ]
+
+              last_teams = []
+            else
+              last_teams << squish(line)
+            end
+        else
+          puts "!!! ERROR [club history reader] - unknown keyword >#{last_keyword}<; cannot process; sorry"
+          exit 1
+        end
+      end  # each line (in paragraph)
+    else
+      puts "** !!! ERROR [club history reader] - unknown line type:"
+      pp node
+      exit 1
+    end
+  end
+
+  recs
+end  # method read
+
+
+###############
+## helper
+
+def squish( str )
+  ## colapse all whitespace to one
+  str.gsub( /[ ]+/,' ' )
+end
+
+
+end  # class ClubHistoryReader
+
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/formats/team/club_reader_props.rb

@@ -36,17 +36,16 @@ class ClubPropsReader
 
       ## find / match club by (canocial) name
       m = catalog.clubs.match( name )
-      if m && m.size > 1
+      if m.size > 1
         puts "** !!! WARN !!! ambigious (multiple) club matches (#{m.size}) for name >#{name}< in props row:"
         pp rec
         pp m
 
         ## todo/fix:  try filter by canonical name if more than one match
         m = m.select { |club| club.name == name }
-        m = nil    if m.empty?     ## note: reset to nil if no more matches
       end
 
-      if m.nil?
+      if m.empty?
         puts "** !!! ERROR !!! no club match for (canonical) name >#{name}< in props row:"
         pp rec
         exit 1

data/lib/sportdb/formats/version.rb

@@ -7,7 +7,7 @@ module Formats
 
   MAJOR = 1    ## todo: namespace inside version or something - why? why not??
   MINOR = 1
-  PATCH = 2
+  PATCH = 3
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/test/helper.rb

@@ -55,12 +55,56 @@ TXT
 = England
 
 Chelsea FC
-Arsenal FC
 Tottenham Hotspur
 West Ham United
 Crystal Palace
-Manchester United
-Manchester City
+
+### note add move entires for testing club name history
+Manchester United FC
+  | Manchester United
+  | Newton Heath FC
+
+Manchester City FC
+  | Manchester City
+  | Ardwick FC
+
+Arsenal FC
+  | The Arsenal FC
+  | Woolwich Arsenal FC
+  | Royal Arsenal FC
+
+Gateshead FC
+  | South Shields FC
+
+Sheffield Wednesday
+  | The Wednesday FC
+
+Port Vale FC
+  | Burslem Port Vale FC
+
+Chesterfield FC
+  | Chesterfield Town FC
+
+Birmingham FC
+  | Small Heath FC
+
+Burton Swifts FC
+Burton Wanderers FC
+Burton United FC
+
+Blackpool FC
+South Shore FC
+
+Glossop FC
+  | Glossop North End FC
+
+Walsall FC
+  | Walsall Town Swifts FC
+
+
+Newcastle West End FC
+Newcastle East End FC
+Newcastle United FC
 TXT
 
     index = ClubIndex.new
@@ -83,81 +127,3 @@ end  # module SportDb
 
 
 
-################
-## helper
-
-def parse_auto_conf( txt, lang: 'en', start: nil )
-  start = start ? start : Date.new( 2017, 7, 1 )
-
-  SportDb::Import.config.lang = lang
-
-  parser = SportDb::AutoConfParser.new( txt, start )
-  parser.parse
-end
-
-def parse_conf( txt )
-  parser = SportDb::ConfParser.new( txt )
-  parser.parse
-end
-
-
-## note: json always returns hash tables with string keys (not symbols),
-##        thus, always stringify keys before comparing!!!!
-class Object
-  def deep_stringify_keys
-    if self.is_a? Hash
-      self.reduce({}) {|memo,(k,v)| memo[k.to_s] = v.deep_stringify_keys; memo }
-    elsif self.is_a? Array
-      self.reduce([]) {|memo,v    | memo           << v.deep_stringify_keys; memo }
-    else
-      self
-    end
-  end
-end
-
-
-def read_blocks( path )
-  txt = File.open( path, 'r:utf-8' ).read
-
-  blocks = []
-  buf    = String.new('')
-  txt.each_line do |line|
-    if line =~ /^[ ]*
-              ([>]{3,} |
-               [<]{3,})
-              [ ]*
-             $/x   ## three or more markers
-      blocks << buf
-      buf = String.new('')
-    else
-      buf << line
-    end
-  end
-  blocks << buf
-  blocks
-end
-
-
-def parse_json( str )
-   ## note: allow empty string; fall back to empty hash
-  if str.strip.empty?
-    {}
-  else
-    JSON.parse( str )
-  end
-end
-
-def read_test( path )
-  blocks = read_blocks( "#{SportDb::Test.data_dir}/football.txt/#{path}" )
-
-  if blocks.size == 2
-    [blocks[0], parse_json( blocks[1] )]
-  elsif blocks.size == 3
-    ## note: returned in different order
-    ##         optional option block that comes first returned last!
-    [blocks[1], parse_json( blocks[2] ), blocks[0]]
-  else
-    puts "!! ERROR: expected two or three text blocks in >#{path}<; got #{blocks.size}"
-    exit 1
-  end
-end

data/test/test_club_index_history.rb

@@ -0,0 +1,107 @@
+# encoding: utf-8
+
+###
+#  to run use
+#     ruby -I ./lib -I ./test test/test_club_index_history.rb
+
+
+require 'helper'
+
+class TestClubHistoryIndex < MiniTest::Test
+
+  def test_eng
+    txt =<<TXT
+=  England
+
+##  note: use history log by season (instead of by year) - why? why not?
+##
+##  note/warn/remember !! a line starting with arrow (=>)
+##    will get turned into a heading 1!!!
+##    as an ascii-alternative to ⇒   use >> or -> or ??? - why? why not?
+
+== 1930/1
+MOVE        South Shields FC, South Shields  ⇒  Gateshead FC, Gateshead
+
+
+== 1929/30
+RENAME     The Wednesday FC, Sheffield   ⇒  Sheffield Wednesday
+
+== 1927/8
+RENAME     The Arsenal FC, London   ⇒  Arsenal FC
+
+
+== 1914/5
+RENAME     Woolwich Arsenal FC, London   ⇒  The Arsenal FC
+
+
+== 1911/2
+REFORM     Burslem Port Vale FC, Burslem  ⇒  Port Vale FC, Stoke-on-Trent
+  ## the towns of Burslem having been merged in 1910 with the towns of Fenton, Hanley,
+  ## Longton, Stoke-upon-Trent and Tunstall as the city of Stoke-on-Trent
+
+
+== 1909/10
+RENAME      Chesterfield Town FC, Chesterfield  ⇒  Chesterfield FC
+
+
+== 1905/6
+RENAME      Chesterfield FC, Chesterfield ⇒ Chesterfield Town FC
+            Small Heath FC, Birmingham    ⇒ Birmingham FC
+
+== 1902/3
+REFORM      Newton Heath FC, Manchester  ⇒  Manchester United
+
+== 1901/2
+MERGE       Burton Swifts FC,    Burton-upon-Trent
+            Burton Wanderers FC, Burton-upon-Trent
+              ⇒ Burton United FC
+
+BANKRUPT    Newton Heath FC, Manchester
+
+
+== 1899/00
+MERGE      Blackpool FC,   Blackpool
+           South Shore FC, Blackpool
+             ⇒ Blackpool FC
+
+== 1898/9
+RENAME     Glossop North End FC, Glossop  ⇒  Glossop FC
+
+
+== 1895/6
+RENAME      Walsall Town Swifts FC, Walsall  ⇒  Walsall FC
+
+
+== 1894/5
+REFORM      Ardwick FC, Manchester  ⇒  Manchester City FC
+
+== 1893/4
+BANKRUPT    Ardwick FC, Manchester
+
+MERGE      Newcastle West End FC, Newcastle-upon-Tyne
+           Newcastle East End FC, Newcastle-upon-Tyne
+            ⇒ Newcastle United FC
+
+== 1892/3
+RENAME     Royal Arsenal FC, London  ⇒  Woolwich Arsenal FC
+TXT
+
+    recs = SportDb::Import::ClubHistoryReader.parse( txt )
+
+    history = SportDb::Import::ClubHistoryIndex.new
+    history.add( recs )
+
+    pp history.errors
+    pp history.mappings
+
+    # [[1927/28, ["RENAME", [["The Arsenal FC, London", "eng"], ["Arsenal FC", "eng"]]]],
+    #  [1914/15, ["RENAME", [["Woolwich Arsenal FC, London", "eng"], ["The Arsenal FC", "eng"]]]],
+    #  [1892/93, ["RENAME", [["Royal Arsenal FC, London", "eng"], ["Woolwich Arsenal FC", "eng"]]]]],
+    assert_equal 'Arsenal FC',          history.find_name_by( name: 'Arsenal FC', season: '2000/1' )
+    assert_equal 'Arsenal FC',          history.find_name_by( name: 'Arsenal FC', season: '1927/8' )
+    assert_equal 'The Arsenal FC',      history.find_name_by( name: 'Arsenal FC', season: '1926/7' )
+    assert_equal 'Woolwich Arsenal FC', history.find_name_by( name: 'Arsenal FC', season: '1913/4' )
+    assert_equal 'Royal Arsenal FC',    history.find_name_by( name: 'Arsenal FC', season: '1891/2' )
+  end
+
+end # class TestClubHistoryIndex