sportdb-readers 0.3.2 → 0.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: fbf4efcbeb2e1af833862fa9654f7832552082d3
-  data.tar.gz: 108ddbf3e0951307fc5ad1c3c36a9b11e00c16cb
+  metadata.gz: 2b15f6d3dbaef2a2d05eca174b86731092ba54b6
+  data.tar.gz: 265e5151c0e81d1beaf5b7cad751f4abfef484e5
 SHA512:
-  metadata.gz: 720dcb7703b7687eeae81485ee95a702c9ac433d468ba9bba8ab04a3dc0417e97d9a72fc6b862fca4f5e9a29b228ab2310cc1f2a8cbddb215e168e06f6c1a818
-  data.tar.gz: 193d88f88d0404499f5a2f18c5034fabec41d977a1864936edc46f26b3260552fa44028cd8134940acb4dc912a12a2f47aba8f84642c730ce2ebdd790cd04583
+  metadata.gz: b943c21d84d36738e562e68a73f2a5de58e7ec693361ad74a96aa04d9ed0f9b5201c876cea62fae0d8daa532f090a6556ade2f78abfac3d12282c987f6710ef5
+  data.tar.gz: 66ee60adbd4ed8bfc34718d3520b24b5840e11193ffeb9f0fab83bc1fec9e7947febc0bc09bfa06123df0b310d1435d74310f5e9a422d3b0a4d0c25f689a5da2

data/Manifest.txt

@@ -11,6 +11,7 @@ lib/sportdb/readers/league_outline_reader.rb
 lib/sportdb/readers/match_linter.rb
 lib/sportdb/readers/match_parser.rb
 lib/sportdb/readers/match_reader.rb
+lib/sportdb/readers/package.rb
 lib/sportdb/readers/version.rb
 test/helper.rb
 test/test_match_parser.rb

data/lib/sportdb/readers/datafile.rb

@@ -8,15 +8,13 @@ module Datafile
   CONF_RE  = %r{ /\.conf\.txt$
                }x
 
-  MATCH_RE = %r{ /\d{4}-\d{2}        ## season folder e.g. /2019-20
-                 /[a-z0-9_-]+\.txt$  ## txt e.g /1-premierleague.txt
-              }x
 
-  CLUBS_PROPS_RE = %r{  (?:^|/)               # beginning (^) or beginning of path (/)
+  CLUB_PROPS_RE = %r{  (?:^|/)               # beginning (^) or beginning of path (/)
                          (?:[a-z]{1,4}\.)?   # optional country code/key e.g. eng.clubs.props.txt
                         clubs\.props\.txt$
                      }x
-  def self.match_clubs_props( path, pattern: CLUBS_PROPS_RE ) pattern.match( path ); end
+  def self.match_club_props( path, pattern: CLUB_PROPS_RE ) pattern.match( path ); end
+
 
   ZIP_RE = %r{ \.zip$
             }x
@@ -24,124 +22,73 @@ module Datafile
 
 
 
-class PackageBase
-
-  ## note: "abstract" methods - each and read required in derived class !!!!
-
-  def each_conf( &blk )   each( pattern: CONF_RE, &blk ); end
-  def each_match( &blk )  each( pattern: MATCH_RE, &blk ); end
-
-
-
-  def read_clubs_props
-    each_read( pattern: CLUBS_PROPS_RE ) do |name, txt|
-      ## todo/fix:  add/use SportDb.parse_club_props helper !!!!!!
-      SportDb::Import::ClubPropsReader.parse( txt )
-    end
-  end
-
-  def read_conf( *names,
-                 season: nil, sync: true )
-    if names.empty?   ## no (entry) names passed in; read in all
-      each_read( pattern: CONF_RE ) do |name, txt|
-        SportDb.parse_conf( txt, season: season, sync: sync )
-      end
-    else
-      names.each do |name|
-        txt = read_entry( name )
-        SportDb.parse_conf( txt, season: season, sync: sync )
-      end
-    end
-  end
-
-  def read_match( *names,
-                  season: nil, sync: true )
-    if names.empty?   ## no (entry) names passed in; read in all
-      each_read( pattern: MATCH_RE ) do |name, txt|
-        SportDb.parse_match( txt, season: season, sync: sync )
-      end
-    else
-      names.each do |name|
-        txt = read_entry( name )
-        SportDb.parse_match( txt, season: season, sync: sync )
-      end
-    end
-  end
-
-
-  def read( *names,
-            season: nil, sync: true )
-    if names.empty?   ##  read all datafiles
-      read_clubs_props()   if sync
-      read_conf( season: season, sync: sync )
-      read_match( season: season, sync: sync )
-    else
-      names.each do |name|
-        txt = read_entry( name )
-        ## fix/todo: add read_clubs_props too!!!
-        if Datafile.match_conf( name )      ## check if datafile matches conf(iguration) naming (e.g. .conf.txt)
-          SportDb.parse_conf( txt, season: season, sync: sync )
-        else                                ## assume "regular" match datafile
-          SportDb.parse_match( txt, season: season, sync: sync )
-        end
-      end
-    end
+class DirPackage    ## todo/check: find a better name e.g. UnzippedPackage, FilesystemPackage, etc. - why? why not?
+class Entry
+  def initialize( pack, path )
+    @pack = pack  ## parent package
+    @path = path
+    ## todo/fix!!!!: calculate @name (cut-off pack.path!!!)
+    @name =  path
   end
-end   # class PackageBase
-
-
+  def name()  @name; end
+  def read()  File.open( @path, 'r:utf-8' ).read; end
+end  # class DirPackage::Entry
 
 
-class DirPackage < PackageBase    ## todo/check: find a better name e.g. UnzippedPackage, FilesystemPackage, etc. - why? why not?
+  attr_reader :name, :path
 
   def initialize( path )
     @path = path   ## rename to root_path or base_path or somehting - why? why not?
-  end
 
+    basename = File.basename( path )   ## note: ALWAYS keeps "extension"-like name if present (e.g. ./austria.zip => austria.zip)
+    @name = basename
+  end
 
-  def each_file( pattern: )    ## todo/check: rename to glob or something - why? why not?
+  def each( pattern:,  extension: 'txt' )    ## todo/check: rename to glob or something - why? why not?
+    ##   use just .* for extension or remove and check if File.file? and skip File.directory? - why? why not?
     ## note: incl. files starting with dot (.)) as candidates (normally excluded with just *)
-    Dir.glob( "#{@path}/**/{*,.*}.txt" ).each do |path|
+    Dir.glob( "#{@path}/**/{*,.*}.#{extension}" ).each do |path|
       ## todo/fix: (auto) skip and check for directories
       if pattern.match( path )
-        yield( path )
+        yield( Entry.new( self, path ))
       else
         ## puts "  skipping >#{path}<"
       end
     end
   end
 
+  def find( name )
+    Entry.new( self, "#{@path}/#{name}" )
+  end
+end  # class DirPackage
 
-  Entry = Struct.new( :name )
 
-  def each( pattern: )   ## todo/check: rename to each_entry - why? why not?
-    each_file( pattern: pattern ) do |path|
-      ## fix: split path like a "virtual" zip like entry
-      yield( Entry.new( path ) )
-    end
+## helper wrapper for datafiles in zips
+class ZipPackage
+class Entry
+  def initialize( pack, entry )
+    @pack  = pack
+    @entry = entry
   end
 
-  def read_entry( name )
-    txt = File.open( "#{@path}/#{name}", 'r:utf-8').read
+  def name()  @entry.name; end
+  def read
+    txt = @entry.get_input_stream.read
+    ##  puts "** encoding: #{txt.encoding}"  #=> encoding: ASCII-8BIT
+    txt = txt.force_encoding( Encoding::UTF_8 )
     txt
   end
+end # class ZipPackage::Entry
 
-  def each_read( pattern: )
-    each_file( pattern: pattern ) do |path|
-      txt = File.open( path, 'r:utf-8').read
-      yield( path, txt )  ## only pass along txt - why? why not? or pass along entry and not just entry.name?
-    end
-  end
-end  # class DirPackage
-
-
+  attr_reader :name, :path
 
-## helper wrapper for datafiles in zips
-class ZipPackage < PackageBase
   def initialize( path )
     @path = path
-  end
 
+    extname  = File.extname( path )    ## todo/check: double check if extension is .zip - why? why not?
+    basename = File.basename( path, extname )
+    @name = basename
+  end
 
   def each( pattern: )
     Zip::File.open( @path ) do |zipfile|
@@ -150,21 +97,35 @@ class ZipPackage < PackageBase
           next ## skip
         elsif entry.file?
           if pattern.match( entry.name )
-            yield( entry )
+            yield( Entry.new( self, entry ) )   # wrap entry in uniform access interface / api
           else
             ## puts "  skipping >#{entry.name}<"
           end
         else
-          puts "** !! ERROR !! #{entry.name} is unknown zip file type in >#{@path}<, sorry"
+          puts "** !!! ERROR !!! #{entry.name} is unknown zip file type in >#{@path}<, sorry"
           exit 1
         end
       end
     end
   end
 
+  def find( name )
+     entries = match_entry( name )
+     if entries.empty?
+       puts "** !!! ERROR !!! zip entry >#{name}< not found in >#{@path}<; sorry"
+       exit 1
+     elsif entries.size > 1
+       puts "** !!! ERROR !!! ambigious zip entry >#{name}<; found #{entries.size} entries in >#{@path}<:"
+       pp entries
+       exit 1
+     else
+       Entry.new( self, entries[0] )    # wrap entry in uniform access interface / api
+     end
+  end
 
+private
   def match_entry( name )
-    ## todo/fix:  use Zip::File.glob or find_entry or ?  why? why not?
+    ## todo/fix:  use Zip::File.glob or find_entry or something better/faster?  why? why not?
 
     pattern = %r{ #{Regexp.escape( name )}    ## match string if ends with name
                    $
@@ -180,39 +141,12 @@ class ZipPackage < PackageBase
             entries << entry
           end
         else
-          puts "** !! ERROR !! #{entry.name} is unknown zip file type in >#{@path}<, sorry"
+          puts "** !!! ERROR !!! #{entry.name} is unknown zip file type in >#{@path}<, sorry"
           exit 1
         end
       end
     end
     entries
   end
-
-  def read_entry( name )
-     entries = match_entry( name )
-     if entries.empty?
-       puts "** !!! ERROR !!! zip entry >#{name}< not found in >#{@path}<; sorry"
-       exit 1
-     elsif entries.size > 1
-       puts "** !!! ERROR !!! ambigious zip entry >#{name}<; found #{entries.size} entries in >#{@path}<:"
-       pp entries
-       exit 1
-     else
-       entry = entries[0]
-       txt = entry.get_input_stream.read
-       ##  puts "** encoding: #{txt.encoding}"  #=> encoding: ASCII-8BIT
-       txt = txt.force_encoding( Encoding::UTF_8 )
-     end
-  end
-
-
-  def each_read( pattern: )
-    each( pattern: pattern ) do |entry|
-      txt = entry.get_input_stream.read
-      ##  puts "** encoding: #{txt.encoding}"  #=> encoding: ASCII-8BIT
-      txt = txt.force_encoding( Encoding::UTF_8 )
-      yield( "#{@path}!/#{entry.name}", txt )    ## only pass along txt - why? why not? or pass along entry and not just entry.name?
-    end
-  end
 end  # class ZipPackage
 end  # module Datafile

data/lib/sportdb/readers/package.rb

@@ -0,0 +1,90 @@
+
+module SportDb
+  class Package
+
+    CONF_RE        = Datafile::CONF_RE
+    CLUB_PROPS_RE  = Datafile::CLUB_PROPS_RE
+
+    ## note: if pattern includes directory add here (otherwise move to more "generic" datafile) - why? why not?
+    MATCH_RE = %r{ /\d{4}-\d{2}        ## season folder e.g. /2019-20
+                   /[a-z0-9_-]+\.txt$  ## txt e.g /1-premierleague.txt
+                }x
+
+    def initialize( path )
+      if !File.exist?( path )  ## file or directory
+        puts "** !!! ERROR !!! file NOT found >#{path}<; cannot open package"
+        exit 1
+      end
+
+      if File.directory?( path )
+        @pack = Datafile::DirPackage.new( path )     ## delegate to "generic" package
+      elsif File.file?( path ) && File.extname( path ) == '.zip'  # note: includes dot (.) eg .zip
+        @pack = Datafile::ZipPackage.new( path )
+      else
+        puts "** !!! ERROR !!! cannot open package - directory or file with .zip extension required"
+        exit 1
+      end
+    end
+
+    def each_conf( &blk )       @pack.each( pattern: CONF_RE, &blk ); end
+    def each_match( &blk )      @pack.each( pattern: MATCH_RE, &blk ); end
+    def each_club_props( &blk ) @pack.each( pattern: CLUB_PROPS_RE, &blk ); end
+
+
+    def read_club_props( sync: true )
+      each_club_props do |entry|
+        SportDb.parse_club_props( entry.read, sync: sync )
+      end
+    end
+
+    def read_conf( *names,
+                   season: nil, sync: true )
+      if names.empty?   ## no (entry) names passed in; read in all
+        each_conf do |entry|
+          SportDb.parse_conf( entry.read, season: season, sync: sync )
+        end
+      else
+        names.each do |name|
+          entry = @pack.find( name )
+          SportDb.parse_conf( entry.read, season: season, sync: sync )
+        end
+      end
+    end
+
+    def read_match( *names,
+                    season: nil, sync: true )
+      if names.empty?   ## no (entry) names passed in; read in all
+        each_match do |entry|
+          SportDb.parse_match( entry.read, season: season, sync: sync )
+        end
+      else
+        names.each do |name|
+          entry = @pack.find( name )
+          SportDb.parse_match( entry.read, season: season, sync: sync )
+        end
+      end
+    end
+
+
+    def read( *names,
+              season: nil, sync: true )
+      if names.empty?   ##  read all datafiles
+        read_club_props( sync: sync )
+        read_conf( season: season, sync: sync )
+        read_match( season: season, sync: sync )
+      else
+        names.each do |name|
+          entry = @pack.find( name )
+          ## fix/todo: add read_clubs_props too!!!
+          if Datafile.match_conf( name )      ## check if datafile matches conf(iguration) naming (e.g. .conf.txt)
+            SportDb.parse_conf( entry.read, season: season, sync: sync )
+          elsif Datafile.match_club_props( name )
+            SportDb.parse_club_props( entry.read, sync: sync )
+          else   ## assume "regular" match datafile
+            SportDb.parse_match( entry.read, season: season, sync: sync )
+          end
+        end
+      end
+    end
+  end   # class Package
+end   # module SportDb

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 = 2
+  PATCH = 3
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/lib/sportdb/readers.rb

@@ -1,6 +1,6 @@
 # encoding: utf-8
 
-require 'zip'     ## todo/check: if zip is alreay included in a required module?
+require 'zip'     ## todo/check: if zip is alreay included in a required module - move into sportdb-formats and add datafile!!!
 
 
 
@@ -20,6 +20,8 @@ require 'sportdb/readers/match_reader'
 require 'sportdb/readers/match_linter'
 require 'sportdb/readers/club_props_reader'
 require 'sportdb/readers/datafile'
+require 'sportdb/readers/package'
+
 
 
 
@@ -32,39 +34,46 @@ module SportDb
     sync ? ConfReaderV2.read( path, season: season )
          : ConfLinter.read( path, season: season )
   end
-
-  def self.read_match( path, season: nil, sync: true )  ### todo/check: add alias read_matches - why? why not?
-    sync ? MatchReaderV2.read( path, season: season )
-         : MatchLinter.read( path, season: season )
-  end
-
-
   def self.parse_conf( txt, season: nil, sync: true )
     sync ? ConfReaderV2.parse( txt, season: season )
          : ConfLinter.parse( txt, season: season )
   end
 
+  def self.read_match( path, season: nil, sync: true )  ### todo/check: add alias read_matches - why? why not?
+    sync ? MatchReaderV2.read( path, season: season )
+         : MatchLinter.read( path, season: season )
+  end
   def self.parse_match( txt, season: nil, sync: true )  ### todo/check: add alias read_matches - why? why not?
     sync ? MatchReaderV2.parse( txt, season: season )
          : MatchLinter.parse( txt, season: season )
   end
 
+  def self.read_club_props( path, sync: true )
+    ## note: for now run only if sync (e.g. run with db updates)
+    SportDb::Import::ClubPropsReader.read( path )   if sync
+  end
+  def self.parse_club_props( txt, sync: true )
+    ## note: for now run only if sync (e.g. run with db updates)
+    SportDb::Import::ClubPropsReader.parse( txt )   if sync
+  end
+
 
   def self.read( path, season: nil, sync: true )
     pack = if File.directory?( path )          ## if directory assume "unzipped" package
-              Datafile::DirPackage.new( path )
-           elsif Datafile.match_zip( path )    ## check if file is a .zip (archive) file
-              Datafile::ZipPackage.new( path )
+              Package.new( path )
+           elsif File.file?( path ) && Datafile.match_zip( path )  ## check if file is a .zip (archive) file
+              Package.new( path )
            else                                ## no package; assume single (standalone) datafile
              nil
            end
 
     if pack
-       pack.read_conf( season: season, sync: sync )
-       pack.read_match( season: season, sync: sync )
+       pack.read( season: season, sync: sync )
     else
       if Datafile.match_conf( path )      ## check if datafile matches conf(iguration) naming (e.g. .conf.txt)
         read_conf( path, season: season, sync: sync )
+      elsif Datafile.match_club_props( path )
+        read_club_props( path, sync: sync )
       else                                ## assume "regular" match datafile
         read_match( path, season: season, sync: sync )
       end
@@ -72,6 +81,7 @@ module SportDb
   end  # method read
 
 
+
   ## (more) convenience helpers for lint(ing)
   def self.lint( path, season: nil )        read( path, season: season, sync: false ); end
   def self.lint_conf( path, season: nil )   read_conf( path, season: season, sync: false ); end

data/test/test_package.rb

@@ -11,15 +11,17 @@ require 'helper'
 class TestPackage < MiniTest::Test
 
   def test_read
-    eng = Datafile::DirPackage.new( '../../../openfootball/england' )
-    ## eng = Datafile::ZipPackage.new( 'tmp/england-master.zip' )
-    assert eng.read_entry( '2015-16/.conf.txt' ).start_with?( '= English Premier League 2015/16' )
-    assert eng.read_entry( '2017-18/.conf.txt' ).start_with?( '= English Premier League 2017/18' )
-    assert eng.read_entry( '2015-16/1-premierleague-i.txt' ).start_with?( '= English Premier League 2015/16' )
-
-    at  = Datafile::DirPackage.new( '../../../openfootball/austria' )
-    ## at  = Datafile::ZipPackage.new( 'tmp/austria-master.zip' )
-    assert at.read_entry( '2018-19/.conf.txt' ).start_with?( '= Österr. Bundesliga 2018/19' )
+    [Datafile::DirPackage.new( '../../../openfootball/england' ),
+     Datafile::ZipPackage.new( 'tmp/england-master.zip' )].each do |eng|
+       assert eng.find( '2015-16/.conf.txt' ).read.start_with?( '= English Premier League 2015/16' )
+       assert eng.find( '2017-18/.conf.txt' ).read.start_with?( '= English Premier League 2017/18' )
+       assert eng.find( '2015-16/1-premierleague-i.txt' ).read.start_with?( '= English Premier League 2015/16' )
+    end
+
+    [Datafile::DirPackage.new( '../../../openfootball/austria' ),
+     Datafile::ZipPackage.new( 'tmp/austria-master.zip' )].each do |at|
+       assert at.find( '2018-19/.conf.txt' ).read.start_with?( '= Österr. Bundesliga 2018/19' )
+    end
   end  # method test_read
 
 end  # class TestPackage

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-readers
 version: !ruby/object:Gem::Version
-  version: 0.3.2
+  version: 0.3.3
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-11-17 00:00:00.000000000 Z
+date: 2019-11-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: sportdb-config
@@ -103,6 +103,7 @@ files:
 - lib/sportdb/readers/match_linter.rb
 - lib/sportdb/readers/match_parser.rb
 - lib/sportdb/readers/match_reader.rb
+- lib/sportdb/readers/package.rb
 - lib/sportdb/readers/version.rb
 - test/helper.rb
 - test/test_match_parser.rb