sportdb-writers 0.1.1 → 0.2.0

data/lib/sportdb/fbgen/main.rb

@@ -0,0 +1,211 @@
+
+module Fbgen
+def self.main( args=ARGV )
+
+opts = {
+  source_path: [],
+  push:     false,
+  dry:      false,  ## dry run (no write)
+  debug:    true,
+  file:     nil,
+}
+
+parser = OptionParser.new do |parser|
+  parser.banner = "Usage: #{$PROGRAM_NAME} [options] [args]"
+
+    parser.on( "-p", "--push",
+               "fast forward sync and commit & push changes to git repo - default is (#{opts[:push]})" ) do |push|
+      opts[:push] = push
+    end
+    parser.on( "--dry",
+                "dry run; do NOT write - default is (#{opts[:dry]})" ) do |dry|
+      opts[:dry] = dry
+    end
+    parser.on( "-q", "--quiet",
+               "less debug output/messages - default is (#{!opts[:debug]})" ) do |debug|
+      opts[:debug] = !debug
+    end
+
+    parser.on( "-f FILE", "--file FILE",
+                "read leagues via .csv file") do |file|
+      opts[:file] = file
+    end
+end
+parser.parse!( args )
+
+
+
+if opts[:source_path].empty? &&
+  File.exist?( '/sports/cache.api.fbdat')  &&
+  File.exist?( '/sports/cache.wfb' )
+ opts[:source_path] << '/sports/cache.api.fbdat'
+ opts[:source_path] << '/sports/cache.wfb'
+end
+
+
+puts "OPTS:"
+p opts
+puts "ARGV:"
+p args
+
+datasets =   if opts[:file]
+                  read_datasets( opts[:file] )
+             else
+                  parse_datasets( args )
+             end
+
+puts "datasets:"
+pp datasets
+
+
+source_path = opts[:source_path]
+source_path = ['.']    if source_path.empty?  ## use ./ as default
+
+root_dir =  if opts[:push]
+               GitHubSync.root   # e.g. "/sports"
+            else
+               './o'
+            end
+
+puts "  (output) root_dir: >#{root_dir}<"
+
+sync  = if opts[:push]
+           repos = GitHubSync.find_repos( datasets )
+           puts "  #{repos.size} repo(s):"
+           pp repos
+           GitHubSync.new( repos )
+        else
+           nil
+        end
+puts "  sync:"
+pp sync
+
+sync.git_fast_forward_if_clean      if sync
+
+
+
+datasets.each do |league_key, seasons|
+    seasons = [ Season('2024/25') ]   if seasons.empty?
+
+    puts "==> gen #{league_key} - #{seasons.size} seasons(s)..."
+
+    league_info = Writer::LEAGUES[ league_key ]
+    pp league_info
+
+    seasons.each do |season|
+      ### get matches
+
+      filename = "#{season.to_path}/#{league_key}.csv"
+      path = find_file( filename, path: source_path )
+
+      if path.nil?
+        puts "!! no source found for #{filename}; sorry"
+        exit 1
+      end
+
+      puts "  ---> reading matches in #{path} ..."
+      matches = SportDb::CsvMatchParser.read( path )
+      puts "     #{matches.size} matches"
+
+      ## build
+      txt = SportDb::TxtMatchWriter.build( matches )
+      puts txt   if opts[:debug]
+
+      league_name  = league_info[ :name ]      # e.g. Brasileiro Série A
+      basename     = league_info[ :basename]   #.e.g  1-seriea
+
+      league_name =  league_name.call( season )   if league_name.is_a?( Proc )  ## is proc/func - name depends on season
+      basename    =  basename.call( season )      if basename.is_a?( Proc )  ## is proc/func - name depends on season
+
+      buf = String.new
+      buf << "= #{league_name} #{season}\n\n"
+      buf << txt
+
+      repo  = GitHubSync::REPOS[ league_key ]
+      repo_path = "#{repo['owner']}/#{repo['name']}"
+      repo_path << "/#{repo['path']}"    if repo['path']  ## note: do NOT forget to add optional extra path!!!
+
+      outpath = "#{root_dir}/#{repo_path}/#{season.to_path}/#{basename}.txt"
+      if opts[:dry]
+        puts "   (dry) writing to >#{outpath}<..."
+      else
+        write_text( outpath, buf )
+      end
+    end
+end
+
+sync.git_push_if_changes   if sync
+
+end  # method self.main
+
+
+
+def self.parse_datasets( args )
+### split args in datasets with leagues and seasons
+datasets = []
+args.each do |arg|
+   if arg =~ %r{^[0-9/-]+$}   ##  season
+       if datasets.empty?
+         puts "!! ERROR - league required before season arg; sorry"
+         exit 1
+       end
+
+       season = Season.parse( arg )  ## check season
+       datasets[-1][1] << season
+   else ## assume league key
+       key = arg.downcase
+       league_info = Writer::LEAGUES[ key ]
+
+       if league_info.nil?
+         puts "!! ERROR - no league found for >#{key}<; sorry"
+         exit 1
+       end
+
+       datasets << [key, []]
+   end
+end
+datasets
+end
+
+
+def self.read_datasets( path )
+### split args in datasets with leagues and seasons
+datasets = []
+recs = read_csv( path )
+recs.each do |rec|
+    league_code  = rec['league']
+    key = league_code.downcase
+    league_info = Writer::LEAGUES[ key ]
+
+    if league_info.nil?
+      puts "!! ERROR - no league found for >#{key}<; sorry"
+      exit 1
+    end
+
+    datasets << [key, []]
+
+    seasons_str = rec['seasons']
+    seasons = seasons_str.split( /[ ]+/ )
+
+    seasons.each do |season_str|
+        season = Season.parse( season_str )  ## check season
+        datasets[-1][1] << season
+    end
+end
+datasets
+end
+
+
+def self.find_file( filename, path: )
+    path.each do |src_dir|
+       path = "#{src_dir}/#{filename}"
+       return path   if File.exist?( path )
+    end
+
+    ##  fix - raise file not found error!!!
+    nil  ## not found - raise filenot found error - why? why not?
+end
+
+
+
+end  # module Fbgen

data/lib/sportdb/fbtxt/main.rb

@@ -0,0 +1,37 @@
+
+module Fbtxt
+def self.main( args=ARGV )
+
+
+opts = {
+}
+
+parser = OptionParser.new do |parser|
+  parser.banner = "Usage: #{$PROGRAM_NAME} [options]"
+end
+parser.parse!( args )
+
+puts "OPTS:"
+p opts
+puts "ARGV:"
+p args
+
+
+matches = []
+
+## step 1 - get all matches via csv
+args.each do |arg|
+   path = arg
+   puts "==> reading matches in #{path} ..."
+   more_matches = SportDb::CsvMatchParser.read( path )
+   matches += more_matches
+end
+
+puts "#{matches.size} matches"
+puts
+
+txt = SportDb::TxtMatchWriter.build( matches )
+puts txt
+puts
+end
+end # module Fbtxt

data/lib/sportdb/writers/league_config.rb

@@ -0,0 +1,83 @@
+
+module SportDb
+class LeagueConfig
+
+def self.read( path )
+   recs = read_csv( path )
+   new( recs )
+end
+
+def initialize( recs=nil )
+   @table = {}
+   add( recs )  if recs
+end
+
+
+class LeagueItem
+    def initialize
+       @recs = []
+    end
+    def add( rec ) @recs << rec; end
+    alias_method :<<, :add
+
+    def find_by_season( season )
+        @recs.each do |rec|
+            start_season = rec['start_season']
+            end_season   = rec['end_season']
+            return rec  if (start_season.nil? || start_season <= season) &&
+                           (end_season.nil? || end_season >= season)
+        end
+        nil
+    end
+
+
+    def name_by_season( season )
+        rec = find_by_season( season )
+        rec ? rec['name'] : nil
+    end
+
+    def basename_by_season( season )
+        rec = find_by_season( season )
+        rec ? rec['basename'] : nil
+    end
+
+
+    def [](key)
+        ## short cut - if only one or zero rec
+        ##   return directly
+        if @recs.empty?
+            nil
+        elsif @recs.size == 1 &&
+              @recs[0]['start_season'].nil? &&
+              @recs[0]['end_season'].nil?
+            @recs[0][key.to_s]
+        else   ### return proc that requires season arg
+            case key.to_sym
+            when :name      then  method(:name_by_season).to_proc
+            when :basename  then  method(:basename_by_season).to_proc
+            else
+                nil  ## return nil - why? why not?
+                ## raise ArgumentError, "invalid key #{key}; use :name or :basename"
+            end
+        end
+    end
+end
+
+def add( recs )
+   recs.each do |rec|
+      @table[ rec['key'] ] ||= LeagueItem.new
+
+      ## note: auto-change seasons to season object or nil
+      @table[ rec['key'] ] << {  'name'         => rec['name'],
+                                 'basename'     => rec['basename'],
+                                 'start_season' => rec['start_season'].empty? ? nil : Season.parse( rec['start_season'] ),
+                                 'end_season'   => rec['end_season'].empty?   ? nil : Season.parse( rec['end_season'] ),
+                              }
+   end
+end
+
+
+def [](key) @table[ key.to_s.downcase ]; end
+
+end # class LeagueConfig
+end # module SportDb

data/lib/sportdb/writers/txt_writer.rb

@@ -19,12 +19,116 @@ def self.build( matches, rounds: true )
   ## note: make sure rounds is a bool, that is, true or false  (do NOT pass in strings etc.)
   raise ArgumentError, "rounds flag - bool expected; got: #{rounds.inspect}"    unless rounds.is_a?( TrueClass ) || rounds.is_a?( FalseClass )
 
-  ## note: for now always english  
+
+  ### check for stages & stats
+  stats = {  'stage' => Hash.new(0),
+             'date' =>  { 'start_date' => nil,
+                          'end_date'   => nil, },
+             'teams' => Hash.new(0),
+              }
+
+## add matches played stats too??
+
+  matches.each do |match|
+     stage = match.stage
+     stage = 'Regular Season'   if stage.nil? || stage.empty?
+     stats['stage'][ stage ] += 1
+
+     if match.date
+
+      ## todo/fix - norm date (parse as Date)
+      ##   check format etc.
+      date = if match.date.is_a?( String )
+                Date.strptime( match.date, '%Y-%m-%d' )
+              else  ## assume it's already a date (object)
+                match.date
+              end
+       stats['date']['start_date'] ||= date
+       stats['date']['end_date']   ||= date
+
+       stats['date']['start_date'] = date  if date < stats['date']['start_date']
+       stats['date']['end_date']   = date  if date > stats['date']['end_date']
+      end
+
+     [match.team1, match.team2].each do |team|
+        stats['teams'][ team ] += 1    if team && !['N.N.'].include?( team )
+     end
+  end
+
+  use_stages =  if stats['stage'].size >= 2  ||
+                   (stats['stage'].size == 1  &&
+                    stats['stage'].keys[0] != 'Regular Season')
+                  true
+                else
+                  false
+                end
+
+
+   ### add comment header
+    buf = String.new
+    # e.g. 13 April – 25 September 2024
+    #  or  16 August 2024 – 25 May 2025
+    buf << "# Date       "
+    start_date = stats['date']['start_date']
+    end_date   = stats['date']['end_date']
+    if start_date.year != end_date.year
+      buf << "#{start_date.strftime('%a %b/%-d %Y')} - #{end_date.strftime('%a %b/%-d %Y')}"
+    else
+      buf << "#{start_date.strftime('%a %b/%-d')} - #{end_date.strftime('%a %b/%-d %Y')}"
+    end
+    buf << " (#{end_date.jd-start_date.jd}d)"   ## add days
+    buf << "\n"
+
+    buf << "# Teams      #{stats['teams'].size}\n"
+    buf << "# Matches    #{matches.size}\n"
+
+    if use_stages
+      buf << "# Stages     "
+      stages = stats['stage'].map { |name,count| "#{name} (#{count})" }.join( '  ' )
+      buf << stages
+      buf << "\n"
+    end
+    buf << "\n\n"
+
+
+  if use_stages
+    ## split matches by stage
+    matches_by_stage = {}
+    matches.each do |match|
+      stage = match.stage || ''
+      matches_by_stage[stage] ||= []
+      matches_by_stage[stage] << match
+    end
+
+    ## todo/fix
+    ## note - empty stage must go first!!!!
+    matches_by_stage.each_with_index do |(name, matches),i|
+      buf << "\n"  if i != 0   # add extra new line (if not first stage)
+      if name.empty?
+        buf << "# Regular Season\n"   ## empty stage
+      else
+        buf << "== #{name}\n"
+      end
+      buf +=  _build_batch( matches, rounds: rounds )
+      buf << "\n"    if i+1 != matches_by_stage.size
+    end
+    buf
+  else
+    buf += _build_batch( matches, rounds: rounds )
+    buf
+  end
+end
+
+
+def self._build_batch( matches, rounds: true )
+  ## note: make sure rounds is a bool, that is, true or false  (do NOT pass in strings etc.)
+  raise ArgumentError, "rounds flag - bool expected; got: #{rounds.inspect}"    unless rounds.is_a?( TrueClass ) || rounds.is_a?( FalseClass )
+
+  ## note: for now always english
   round              = 'Matchday'
   format_date        = ->(date) { date.strftime( '%a %b/%-d' )  }
   format_score       = ->(match) { match.score.to_s( lang: 'en' ) }
   round_translations = ROUND_TRANSLATIONS
-  
 
   buf = String.new
 
@@ -33,12 +137,12 @@ def self.build( matches, rounds: true )
   last_time  = nil
 
 
-  matches.each do |match|
+  matches.each_with_index do |match,i|
 
      ## note: make rounds optional (set rounds flag to false to turn off)
      if rounds
        if match.round != last_round
-         buf << "\n\n"
+         buf << (i == 0 ? "\n" : "\n\n")    ## start with single empty line
          if match.round.is_a?( Integer ) ||
             match.round =~ /^[0-9]+$/   ## all numbers/digits
              ## default "class format

data/lib/sportdb/writers/version.rb

@@ -3,8 +3,8 @@ module SportDb
 module Module
   module Writers
     MAJOR = 0    ## todo: namespace inside version or something - why? why not??
-    MINOR = 1
-    PATCH = 1
+    MINOR = 2
+    PATCH = 0
     VERSION = [MAJOR,MINOR,PATCH].join('.')
 
     def self.version

data/lib/sportdb/writers/write.rb

@@ -53,7 +53,7 @@ def self.write( league:, season:,
                 source:,
                 extra: nil,
                 split: false,
-                normalize: false,   
+                normalize: false,
                 rounds: true )
   season = Season( season )  ## normalize season
 
@@ -71,7 +71,7 @@ def self.write( league:, season:,
        exit 1
   end
   source_info = { path: source }   ## wrap in "plain" source dir in source info
-  
+
   source_path = source_info[:path]
 
   ## format lets you specify directory layout
@@ -106,14 +106,14 @@ def self.write( league:, season:,
      if normalize.is_a?(Proc)
         matches = normalize.call( matches, league: league,
                                            season: season )
-     else 
+     else
        puts "!! ERROR - normalize; expected proc got #{normalize.inspect}"
        exit 1
-     end 
+     end
   end
- 
 
-  
+
+
   league_name  = league_info[ :name ]      # e.g. Brasileiro Série A
   basename     = league_info[ :basename]   #.e.g  1-seriea
 
@@ -122,7 +122,10 @@ def self.write( league:, season:,
 
   ## note - repo_path moved!!!
   ## repo_path    = league_info[ :path ]      # e.g. brazil or world/europe/portugal etc.
-  repo_path  = SportDb::GitHubSync::REPOS[ league ]
+  repo  = SportDb::GitHubSync::REPOS[ league ]
+  repo_path = "#{repo['owner']}/#{repo['name']}"
+  repo_path << "/#{repo['path']}"    if repo['path']  ## note: do NOT forget to add optional extra path!!!
+
 
 
   season_path = String.new     ## note: allow extra path for output!!!! e.g. archive/2000s etc.
@@ -175,7 +178,7 @@ def self.write( league:, season:,
                       )
 
     ## note: might be empty!!! if no matches skip (do NOT write)
-    write_text( "#{config.out_dir}/#{repo_path}/#{season_path}/#{stage_basename}.txt", 
+    write_text( "#{config.out_dir}/#{repo_path}/#{season_path}/#{stage_basename}.txt",
                   buf )   unless buf.empty?
   end
   else  ## no stages - assume "regular" plain vanilla season

data/lib/sportdb/writers.rb

@@ -2,6 +2,9 @@
 require 'sportdb/quick'
 
 
+require 'optparse'    ## command-line processing; check if included updstream?
+
+
 
 module Writer
   class Configuration
@@ -29,30 +32,74 @@ require_relative 'writers/goals'
 require_relative 'writers/write'
 
 
+## setup  leagues (info) table
+require_relative 'writers/league_config'
+
+module Writer
+  LEAGUES = SportDb::LeagueConfig.new
+
+  ['leagues_europe',
+   'leagues_america',
+   'leagues_world'
+  ].each do |name|
+    recs = read_csv( "#{SportDb::Module::Writers.root}/config/#{name}.csv" )
+    LEAGUES.add( recs )
+  end
+end  # module Writer
+
+
+###
+#  fbtxt tool
+require_relative 'fbtxt/main'
+
+
 
 ########################
 #  push & pull github scripts
 require 'gitti'    ## note - requires git machinery
 
-require_relative 'writers/github'   ## github helpers/update machinery
+require_relative 'fbgen/github_config'
+require_relative 'fbgen/github'   ## github helpers/update machinery
 
 
+module Fbgen
+class  GitHubSync
+  REPOS = GitHubConfig.new
+  recs = read_csv( "#{SportDb::Module::Writers.root}/config/openfootball.csv" )
+  REPOS.add( recs )
 
-## setup empty leagues (info) hash
-module Writer
-  LEAGUES = {}
+## todo/check: find a better name for helper?
+## note: datasets of format
+##
+## DATASETS = [
+##   ['it.1',    %w[2020/21 2019/20]],
+##  ['it.2',    %w[2019/20]],
+##  ['es.1',    %w[2019/20]],
+##  ['es.2',    %w[2019/20]],
+## ]
+
+def self.find_repos( datasets )
+  repos = []
+  datasets.each do |dataset|
+    league_key = dataset[0]
+    repo  = REPOS[ league_key ]
+    ## pp repo
+    if repo.nil?
+       puts "!! ERROR - no repo config/path found for league >#{league_key}<; sorry"
+       exit 1
+    end
+
+    repos <<  "#{repo['owner']}/#{repo['name']}"
+  end
+
+  pp repos
+  repos.uniq   ## note: remove duplicates (e.g. europe or world or such)
 end
+end  #  class  GitHubSync
+end  # module Fbgen
 
-require_relative 'leagues/leagues_at'
-require_relative 'leagues/leagues_de'
-require_relative 'leagues/leagues_eng'
-require_relative 'leagues/leagues_es'
-require_relative 'leagues/leagues_europe'
-require_relative 'leagues/leagues_it'
-require_relative 'leagues/leagues_mx'
-require_relative 'leagues/leagues_south_america'
-require_relative 'leagues/leagues_world'
 
+require_relative 'fbgen/main'
 
 
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-writers
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.0
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-09-04 00:00:00.000000000 Z
+date: 2024-09-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: sportdb-quick
@@ -90,18 +90,17 @@ files:
 - Rakefile
 - bin/fbgen
 - bin/fbtxt
-- lib/sportdb/leagues/leagues_at.rb
-- lib/sportdb/leagues/leagues_de.rb
-- lib/sportdb/leagues/leagues_eng.rb
-- lib/sportdb/leagues/leagues_es.rb
-- lib/sportdb/leagues/leagues_europe.rb
-- lib/sportdb/leagues/leagues_it.rb
-- lib/sportdb/leagues/leagues_mx.rb
-- lib/sportdb/leagues/leagues_south_america.rb
-- lib/sportdb/leagues/leagues_world.rb
+- config/leagues_america.csv
+- config/leagues_europe.csv
+- config/leagues_world.csv
+- config/openfootball.csv
+- lib/sportdb/fbgen/github.rb
+- lib/sportdb/fbgen/github_config.rb
+- lib/sportdb/fbgen/main.rb
+- lib/sportdb/fbtxt/main.rb
 - lib/sportdb/writers.rb
-- lib/sportdb/writers/github.rb
 - lib/sportdb/writers/goals.rb
+- lib/sportdb/writers/league_config.rb
 - lib/sportdb/writers/txt_writer.rb
 - lib/sportdb/writers/version.rb
 - lib/sportdb/writers/write.rb