sportdb-readers 0.3.6 → 0.3.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: aa4225ef65ba9a0bddadba178cae48f142b2894d
-  data.tar.gz: '0966b042258895e134ce4236c1cf211588fd42d5'
+  metadata.gz: 0b82ac613158881351d512a4ffc9e7d15f34b133
+  data.tar.gz: 52251959f919b175feec37cadcc51e56c70e5037
 SHA512:
-  metadata.gz: 79eec4198136a2368667f49cf0b782551cca03f1d5cfd7e93ad6cd35f818589e8b3e6e35f9c3516cc441cd454da54516377fb1f04100b14b930aac2e275e441c
-  data.tar.gz: 3f599edf0c5aab6234c0bed6c8e73156b6ee9c844d944fb019aa906861abff067398dc2b0316e00aea1ed3545494aa9cfeb6cae0eea0465039503dc6476f5369
+  metadata.gz: 32c7921845dd7db5c2771a5941ab7ca0b27839eb88c80d71703354f6f533ca1c2eb99b47671e2462cbddcddb443fab3d0e115a4991f3d6155185a081d500796e
+  data.tar.gz: 072a747031829fb495773946c13ec6a43627c7b6de4b21b752c40eb2d1962c1fb3a047982fe83f1257be0b9be3f9115d0eed0d798de1b3dae2f93518e2ac02d3

data/Manifest.txt

@@ -5,6 +5,7 @@ Rakefile
 lib/sportdb/readers.rb
 lib/sportdb/readers/club_props_reader.rb
 lib/sportdb/readers/conf_linter.rb
+lib/sportdb/readers/conf_parser_auto.rb
 lib/sportdb/readers/conf_reader.rb
 lib/sportdb/readers/datafile.rb
 lib/sportdb/readers/league_outline_reader.rb
@@ -14,6 +15,7 @@ lib/sportdb/readers/match_reader.rb
 lib/sportdb/readers/package.rb
 lib/sportdb/readers/version.rb
 test/helper.rb
+test/test_conf_parser_auto.rb
 test/test_match_parser.rb
 test/test_package.rb
 test/test_props.rb

data/lib/sportdb/readers.rb

@@ -13,6 +13,7 @@ require 'sportdb/sync'
 # our own code
 require 'sportdb/readers/version' # let version always go first
 require 'sportdb/readers/league_outline_reader'
+require 'sportdb/readers/conf_parser_auto'
 require 'sportdb/readers/conf_reader'
 require 'sportdb/readers/conf_linter'
 require 'sportdb/readers/match_parser'

data/lib/sportdb/readers/conf_parser_auto.rb

@@ -0,0 +1,129 @@
+# encoding: utf-8
+
+module SportDb
+
+
+class AutoConfParser
+
+  include LogUtils::Logging
+
+  def initialize( lines, start )
+     @lines        = lines     ## todo/check: change to text instead of array of lines - why? why not?
+     @start        = start
+  end
+
+  def parse
+    ## try to  find all clubs in match schedule
+    @clubs = Hash.new(0)   ## keep track of usage counter
+
+    @lines.each do |line|
+      if is_round?( line )
+        logger.info "skipping matched round line: >#{line}<"
+      elsif try_parse_game( line )
+        # do nothing here
+      else
+        logger.info "skipping line (no match found): >#{line}<"
+      end
+    end # lines.each
+
+    @clubs
+  end
+
+  def is_round?( line )
+    ## note: =~ return nil if not match found, and 0,1, etc for match
+    (line =~ SportDb.lang.regex_round) != nil
+  end
+
+  def try_parse_game( line )
+    # note: clone line; for possible test do NOT modify in place for now
+    # note: returns true if parsed, false if no match
+    parse_game( line.dup )
+  end
+
+  def parse_game( line )
+    logger.debug "parsing game (fixture) line: >#{line}<"
+
+    ## remove all protected text runs e.g. []
+    ##   fix: add [ to end-of-line too
+
+    line = line.gsub( /\[
+                        [^\]]+?
+                       \]/x, '' ).strip
+    return true if line.empty?    ## note: return true (for valid line with no match/clubs)
+
+    ## split by geo (@) - remove for now
+    values = line.split( '@' )
+    line = values[0]
+
+
+    ## try find date
+    date = find_date!( line, start: @start )
+    if date   ## if found remove tagged run too; note using singular sub (NOT global gsub)
+      line = line.sub( /\[
+                          [^\]]+?
+                         \]/x, '' ).strip
+
+      return true if line.empty?    ## note: return true (for valid line with no match/clubs)
+    end
+
+
+    scores = find_scores!( line )
+
+    logger.debug "  line: >#{line}<"
+
+    line = line.sub( /\[
+                        [^\]]+?
+                       \]/x, '$$' )  # note: replace first score tag with $$
+    line = line.gsub( /\[
+                    [^\]]+?
+                   \]/x, '' )    # note: replace/remove all other score tags with nothing
+
+     ##  clean-up  remove all text run inside () or empty () too
+     line = line.gsub( /\(
+                     [^)]*?
+                    \)/x, '' )
+
+
+     ## check for more match separators e.g. - or vs for now
+     line = line.sub( / \s+
+                          (-|vs|v)
+                        \s+
+                       /ix, '$$' )
+
+     values = line.split( '$$' )
+     values = values.map { |value| value.strip }        ## strip spaces
+     values = values.select { |value| !value.empty? }   ## remove empty strings
+
+     return true    if values.size == 0  ## note: return true (for valid line with no match/clubs)
+     return false   if values.size != 2
+
+     puts "(auto config) try matching clubs:"
+     pp values
+     @clubs[ values[0] ] += 1    ## update usage counters
+     @clubs[ values[1] ] += 1
+
+     true
+  end
+
+
+
+  def find_scores!( line, opts={} )
+    # note: always call after find_dates !!!
+    #  scores match date-like patterns!!  e.g. 10-11  or 10:00 etc.
+    #   -- note: score might have two digits too
+
+    finder = ScoresFinder.new
+    finder.find!( line, opts )
+  end
+
+  def find_date!( line, start: )
+    ## NB: lets us pass in start_at/end_at date (for event)
+    #   for auto-complete year
+
+    # extract date from line
+    # and return it
+    # NB: side effect - removes date from line string
+    DateFormats.find!( line, start: start )
+  end
+end # class AutoConfParser
+end # module SportDb

data/lib/sportdb/readers/datafile.rb

@@ -22,7 +22,19 @@ module Datafile
 
 
 
-class DirPackage    ## todo/check: find a better name e.g. UnzippedPackage, FilesystemPackage, etc. - why? why not?
+  ## exclude pattern
+  ##  for now exclude all files in directories starting with a dot (e.g. .git/ or .github/ or .build/ etc.)
+  ##  todo/check: rename to EXCLUDE_DOT_DIRS_RE - why? why not?
+  EXCLUDE_RE = %r{  (?:^|/)               # beginning (^) or beginning of path (/)
+                        \.[a-zA-Z0-9_-]+  ## (almost) any name BUT must start with dot e.g.  .git, .build, etc.
+                        /
+                     }x
+  def self.match_exclude( path, pattern: EXCLUDE_RE ) pattern.match( path ); end
+
+
+class Package; end    ## use a shared base class for DirPackage, ZipPackage, etc.
+
+class DirPackage < Package    ## todo/check: find a better name e.g. UnzippedPackage, FilesystemPackage, etc. - why? why not?
 class Entry
   def initialize( pack, path )
     @pack = pack  ## parent package
@@ -38,6 +50,7 @@ end  # class DirPackage::Entry
   attr_reader :name, :path
 
   def initialize( path )
+    ## todo/fix:  expand_path ?! - why? why not? if you pass in ./ basename will be . and NOT directory name, for example!!!
     @path = path   ## rename to root_path or base_path or somehting - why? why not?
 
     basename = File.basename( path )   ## note: ALWAYS keeps "extension"-like name if present (e.g. ./austria.zip => austria.zip)
@@ -49,7 +62,9 @@ end  # class DirPackage::Entry
     ## note: incl. files starting with dot (.)) as candidates (normally excluded with just *)
     Dir.glob( "#{@path}/**/{*,.*}.#{extension}" ).each do |path|
       ## todo/fix: (auto) skip and check for directories
-      if pattern.match( path )
+      if EXCLUDE_RE.match( path )
+        ## note: skip dot dirs (e.g. .build/, .git/, etc.)
+      elsif pattern.match( path )
         yield( Entry.new( self, path ))
       else
         ## puts "  skipping >#{path}<"
@@ -64,7 +79,7 @@ end  # class DirPackage
 
 
 ## helper wrapper for datafiles in zips
-class ZipPackage
+class ZipPackage < Package
 class Entry
   def initialize( pack, entry )
     @pack  = pack
@@ -96,7 +111,9 @@ end # class ZipPackage::Entry
         if entry.directory?
           next ## skip
         elsif entry.file?
-          if pattern.match( entry.name )
+          if EXCLUDE_RE.match( entry.name )
+            ## note: skip dot dirs (e.g. .build/, .git/, etc.)
+          elsif pattern.match( entry.name )
             yield( Entry.new( self, entry ) )   # wrap entry in uniform access interface / api
           else
             ## puts "  skipping >#{entry.name}<"
@@ -137,8 +154,12 @@ private
         if entry.directory?
           next ## skip
         elsif entry.file?
-          if pattern.match( entry.name )
+          if EXCLUDE_RE.match( entry.name )
+            ## note: skip dot dirs (e.g. .build/, .git/, etc.)
+          elsif pattern.match( entry.name )
             entries << entry
+          else
+            ## no match; skip too
           end
         else
           puts "** !!! ERROR !!! #{entry.name} is unknown zip file type in >#{@path}<, sorry"

data/lib/sportdb/readers/package.rb

@@ -14,8 +14,7 @@ module SportDb
                 }x
 
     def initialize( path_or_pack )
-      if path_or_pack.is_a?( Datafile::DirPackage ) ||
-         path_or_pack.is_a?( Datafile::ZipPackage )
+      if path_or_pack.is_a?( Datafile::Package )
         @pack = path_or_pack
       else   ## assume it's a (string) path
         path = path_or_pack

data/lib/sportdb/readers/version.rb

@@ -6,7 +6,7 @@ module Readers
 
   MAJOR = 0    ## todo: namespace inside version or something - why? why not??
   MINOR = 3
-  PATCH = 6
+  PATCH = 7
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/test/test_conf_parser_auto.rb

@@ -0,0 +1,129 @@
+# encoding: utf-8
+
+###
+#  to run use
+#     ruby -I ./lib -I ./test test/test_conf_parser_auto.rb
+
+
+require 'helper'
+
+
+
+class TestAutoConfParser < MiniTest::Test
+
+  def test_eng
+    txt = <<TXT
+Matchday 1
+
+[Fri Aug/11]
+  Arsenal FC               4-3  Leicester City
+[Sat Aug/12]
+  Watford FC               3-3  Liverpool FC
+  Chelsea FC               2-3  Burnley FC
+  Crystal Palace           0-3  Huddersfield Town
+  Everton FC               1-0  Stoke City
+  Southampton FC           0-0  Swansea City
+  West Bromwich Albion     1-0  AFC Bournemouth
+  Brighton & Hove Albion   0-2  Manchester City
+[Sun Aug/13]
+  Newcastle United         0-2  Tottenham Hotspur
+  Manchester United        4-0  West Ham United
+
+
+Matchday 2
+
+[Sat Aug/19]
+  Swansea City             0-4  Manchester United
+  AFC Bournemouth          0-2  Watford FC
+  Burnley FC               0-1  West Bromwich Albion
+  Leicester City           2-0  Brighton & Hove Albion
+  Liverpool FC             1-0  Crystal Palace
+  Southampton FC           3-2  West Ham United
+  Stoke City               1-0  Arsenal FC
+[Sun Aug/20]
+  Huddersfield Town        1-0  Newcastle United
+  Tottenham Hotspur        1-2  Chelsea FC
+[Mon Aug/21]
+  Manchester City          1-1  Everton FC
+TXT
+
+    clubs = parse( txt )
+
+    assert_equal Hash(
+ 'Arsenal FC'     => 2,
+ 'Leicester City' => 2,
+ 'Watford FC'     => 2,
+ 'Liverpool FC'   => 2,
+ 'Chelsea FC'     => 2,
+ 'Burnley FC'     => 2,
+ 'Crystal Palace' => 2,
+ 'Huddersfield Town' => 2,
+ 'Everton FC'     => 2,
+ 'Stoke City'     => 2,
+ 'Southampton FC' => 2,
+ 'Swansea City'   => 2,
+ 'West Bromwich Albion' => 2,
+ 'AFC Bournemouth' => 2,
+ 'Brighton & Hove Albion' => 2,
+ 'Manchester City'   => 2,
+ 'Newcastle United'  => 2,
+ 'Tottenham Hotspur' => 2,
+ 'Manchester United' => 2,
+ 'West Ham United'   => 2 ), clubs
+  end   # method test_parse
+
+  def test_mauritius
+    txt = <<TXT
+Preliminary Round
+[Mon Jun/22]
+  Pointe-aux-Sables Mates      3-4  AS Port-Louis 2000              @ St. François Xavier Stadium, Port Louis
+
+Quarterfinals
+[Wed Jun/24]
+  Rivière du Rempart           3-1 pen (1-1) La Cure Sylvester      @ Auguste Vollaire Stadium, Central Flacq
+  Chamarel SC                  3-4           Petite Rivière Noire   @ Germain Comarmond Stadium, Bambous
+[Thu Jun/25]
+  Pamplemousses                2-0  AS Port-Louis 2000              @ Auguste Vollaire Stadium, Central Flacq
+[Sat Jun/27]
+  Savanne SC                   3-6  Entente Boulet Rouge            @ Anjalay Stadium, Mapou
+
+Semifinals
+[Wed Jul/15]
+  Rivière du Rempart           2-3  Petite Rivière Noire            @  New George V Stadium, Curepipe
+  Entente Boulet Rouge         0-2  Pamplemousses                   @  Germain Comarmond Stadium, Bambous
+
+Final
+[Sun Jul/19]
+  Petite Rivière Noire         2-0  Pamplemousses                   @ New George V Stadium, Curepipe
+TXT
+
+    clubs = parse( txt )
+
+    assert_equal Hash(
+     'Pointe-aux-Sables Mates' => 1,
+     'AS Port-Louis 2000'      => 2,
+     'Rivière du Rempart'      => 2,
+     'La Cure Sylvester'       => 1,
+     'Chamarel SC'             => 1,
+     'Petite Rivière Noire'    => 3,
+     'Pamplemousses'           => 3,
+     'Savanne SC'              => 1,
+     'Entente Boulet Rouge'    => 2), clubs
+  end
+
+
+################
+## helper
+def parse( txt, lang: 'en' )
+  lines = txt.split( /\n+/ )    # note: removes/strips empty lines
+  pp lines
+
+  start = Date.new( 2017, 7, 1 )
+
+  DateFormats.lang = lang  # e.g. 'en'
+  parser = SportDb::AutoConfParser.new( lines, start )
+  clubs = parser.parse
+  pp clubs
+  clubs
+end
+end   # class AutoConfParser

data/test/test_package.rb

@@ -10,6 +10,20 @@ require 'helper'
 
 class TestPackage < MiniTest::Test
 
+  def test_exclude
+    assert Datafile.match_exclude( '.build/' )
+    assert Datafile.match_exclude( '.git/' )
+
+    assert Datafile.match_exclude( '/.build/' )
+    assert Datafile.match_exclude( '/.git/' )
+
+    assert Datafile.match_exclude( '.build/leagues.txt' )
+    assert Datafile.match_exclude( '.git/leagues.txt' )
+
+    assert Datafile.match_exclude( '/.build/leagues.txt' )
+    assert Datafile.match_exclude( '/.git/leagues.txt' )
+  end
+
   def test_read
     [Datafile::DirPackage.new( '../../../openfootball/england' ),
      Datafile::ZipPackage.new( 'tmp/england-master.zip' )].each do |eng|

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-readers
 version: !ruby/object:Gem::Version
-  version: 0.3.6
+  version: 0.3.7
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-11-22 00:00:00.000000000 Z
+date: 2019-11-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: sportdb-config
@@ -97,6 +97,7 @@ files:
 - lib/sportdb/readers.rb
 - lib/sportdb/readers/club_props_reader.rb
 - lib/sportdb/readers/conf_linter.rb
+- lib/sportdb/readers/conf_parser_auto.rb
 - lib/sportdb/readers/conf_reader.rb
 - lib/sportdb/readers/datafile.rb
 - lib/sportdb/readers/league_outline_reader.rb
@@ -106,6 +107,7 @@ files:
 - lib/sportdb/readers/package.rb
 - lib/sportdb/readers/version.rb
 - test/helper.rb
+- test/test_conf_parser_auto.rb
 - test/test_match_parser.rb
 - test/test_package.rb
 - test/test_props.rb