sportdb-formats 2.0.2 → 2.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 67f0ea27fe2c7a39512eab9728167881fb1b99b919492b3d513cb1a1ff46c787
-  data.tar.gz: a93978a81ec8cb8c9b639ff4842e5eb46d7ebc018031e4b37e8361b7ed80e8af
+  metadata.gz: c227a8566030829c53c10e3a8ab626b3222653ebcf795143e17c4627609a3e9f
+  data.tar.gz: 25cf88f75828207d783009f7f67f86de3cffab7265cc875fc893133678134bd5
 SHA512:
-  metadata.gz: 94d114bc6772b18e972449cd7ed0f0a890d737203e8e020655e446e4496e8b8e99985ac33c6da2579e27a52a863602ea561e2275f2e1852762355228b3d9383d
-  data.tar.gz: 4e645afc07ca5c73cda0308fe82211293160e2ebd3857770dc959ea41c684cbc2794e1c514f1fcf9e103f22294cc5d7879446632a9be06d2a04399b11daef5ed
+  metadata.gz: 5eb402eb00c0aff8e18a8079807cf39d1ed7d3e4d57786051d3737c39680958917a412a304fd40132b7aa1e7a7cc7d0d0a8e97f15307176e2fb54083e2f16caf
+  data.tar.gz: ba5db77e4b57f72ed3c305cd71bbc5ad7e150971ff9055544a4d3028ff3b24c579ec9c9915e651ed2eae1f2f4d19a7772f8aafc79af7f4d60dcc6e22e33bc930

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 2.0.2
+### 2.1.0
 
 ### 0.0.1 / 2019-10-28
 

data/Manifest.txt

@@ -2,28 +2,10 @@ CHANGELOG.md
 Manifest.txt
 README.md
 Rakefile
-bin/fbx
+bin/fbchk
 lib/sportdb/formats.rb
-lib/sportdb/formats/country/country_reader.rb
-lib/sportdb/formats/csv/goal.rb
-lib/sportdb/formats/csv/goal_parser_csv.rb
-lib/sportdb/formats/csv/match_parser_csv.rb
-lib/sportdb/formats/csv/match_status_parser.rb
-lib/sportdb/formats/datafile.rb
 lib/sportdb/formats/datafile_package.rb
-lib/sportdb/formats/event/event_reader.rb
-lib/sportdb/formats/ground/ground_reader.rb
-lib/sportdb/formats/league/league_outline_reader.rb
-lib/sportdb/formats/league/league_reader.rb
-lib/sportdb/formats/match/conf_parser.rb
-lib/sportdb/formats/match/match_parser.rb
 lib/sportdb/formats/package.rb
-lib/sportdb/formats/search/sport.rb
-lib/sportdb/formats/search/structs.rb
-lib/sportdb/formats/search/world.rb
+lib/sportdb/formats/quick_match_linter.rb
 lib/sportdb/formats/team/club_index_history.rb
-lib/sportdb/formats/team/club_reader.rb
-lib/sportdb/formats/team/club_reader_history.rb
-lib/sportdb/formats/team/club_reader_props.rb
-lib/sportdb/formats/team/wiki_reader.rb
 lib/sportdb/formats/version.rb

data/Rakefile

@@ -20,13 +20,8 @@ Hoe.spec 'sportdb-formats' do
   self.licenses = ['Public Domain']
 
   self.extra_deps = [
-    ['sportdb-structs',  '>= 0.3.1'],
-    ['sportdb-parser',   '>= 0.2.1'],
-    ['sportdb-catalogs', '>= 1.2.2'],
-    ['date-formats',     '>= 1.0.2'],
-    ['cocos',            '>= 0.4.0'],
-    ['logutils',         '>= 0.6.1'],
-
+    ['sportdb-search',   '>= 0.0.1'],
+    ['sportdb-quick',    '>= 0.2.0'],
     ['rubyzip',          '>= 2.3.2' ],
   ]
 

data/bin/fbchk

@@ -0,0 +1,166 @@
+#!/usr/bin/env ruby
+
+## tip: to test run:
+##   ruby -I ./lib bin/fbchk
+
+## our own code
+require 'sportdb/formats'
+
+
+
+require 'optparse'
+
+
+
+## local hack
+##   if exists up-to-date catalog db (use local version NOT built-in)
+catalog_path = '../catalog/catalog.db'
+if File.exist?( catalog_path )
+  SportDb::Import.config.catalog_path = catalog_path
+end
+
+
+ args = ARGV
+ opts = { debug: false,
+          file:  nil,
+        }
+
+ parser = OptionParser.new do |parser|
+  parser.banner = "Usage: #{$PROGRAM_NAME} [options]"
+
+##
+## check if git has a offline option?? (use same)
+##             check for other tools - why? why not?
+#    parser.on( "-q", "--quiet",
+#                 "less debug output/messages - default is (#{!opts[:debug]})" ) do |debug|
+#      opts[:debug] = false
+#    end
+   parser.on( "--verbose", "--debug",
+               "turn on verbose / debug output (default: #{opts[:debug]})" ) do |debug|
+     opts[:debug] = true
+   end
+
+    parser.on( "-f FILE", "--file FILE",
+                    "read datafiles (pathspecs) via .csv file") do |file|
+      opts[:file] = file
+    end
+
+
+end
+parser.parse!( args )
+
+puts "OPTS:"
+p opts
+puts "ARGV:"
+p args
+
+
+    ## todo/check - use packs or projects or such
+    ##                instead of specs - why? why not?
+    specs = []
+    if opts[:file]
+          recs = read_csv( opts[:file] )
+          pp recs
+          ##  note - make pathspecs relative to passed in file arg!!!
+          basedir = File.dirname( opts[:file] )
+          recs.each do |rec|
+             paths = SportDb::Parser::Opts.find( rec['path'], dir: basedir )
+             specs << [paths, rec]
+          end
+    else
+       paths = if args.empty?
+                 []
+              else
+                ## check for directories
+                ##   and auto-expand
+                SportDb::Parser::Opts.expand_args( args )
+              end
+        specs << [paths, {}]
+    end
+
+
+if opts[:debug]
+   SportDb::QuickMatchLinter.debug = true
+   SportDb::QuickMatchReader.debug = true
+   SportDb::MatchParser.debug      = true
+else
+   SportDb::QuickMatchLinter.debug = false
+   SportDb::QuickMatchReader.debug = false
+   SportDb::MatchParser.debug      = false
+   LogUtils::Logger.root.level = :info
+end
+
+
+  specs.each_with_index do |(paths, rec),i|
+    errors = []
+    paths.each_with_index do |path,j|
+      puts "==> [#{j+1}/#{paths.size}] reading >#{path}<..."
+      quick = SportDb::QuickMatchLinter.new( read_text( path ) )
+      matches = quick.parse
+
+      if quick.errors?
+        puts "!! #{quick.errors.size} error(s):"
+        pp quick.errors
+
+        quick.errors.each do |err|
+          errors << [ path, *err ]   # note: use splat (*) to add extra values (starting with msg)
+        end
+      end
+      puts "  #{matches.size} match(es)"
+    end
+
+    if errors.size > 0
+      puts
+      puts "!! #{errors.size} PARSE ERRORS in #{paths.size} datafile(s)"
+      pp errors
+    else
+      puts
+      puts "  OK - no parse errors in #{paths.size} datafile(s)"
+    end
+
+     ## add errors to rec via rec['errors'] to allow
+     ##   for further processing/reporting
+     rec['errors'] = errors
+  end
+
+
+
+###
+## generate a report if --file option used
+if opts[:file]
+
+  buf = String.new
+
+  buf << "# fbchk summary report - #{specs.size} dataset(s)\n\n"
+
+  specs.each_with_index do |(paths, rec),i|
+     errors = rec['errors']
+
+     if errors.size > 0
+       buf << "!! #{errors.size} ERROR(S)  "
+     else
+       buf << "   OK          "
+     end
+     buf << "%-20s" % rec['path']
+     buf << " - #{paths.size} datafile(s)"
+     buf << "\n"
+
+     if errors.size > 0
+         buf << errors.pretty_inspect
+         buf << "\n"
+     end
+ end
+
+  puts
+  puts "SUMMARY:"
+  puts buf
+
+  basedir  = File.dirname( opts[:file] )
+  basename = File.basename( opts[:file], File.extname( opts[:file] ))
+  outpath  = "#{basedir}/fbcheck.#{basename}.txt"
+  write_text( outpath, buf )
+end
+
+
+puts "bye"
+

data/lib/sportdb/formats/quick_match_linter.rb

@@ -0,0 +1,195 @@
+#####
+## quick match linter for datafiles with league outlines
+
+
+module SportDb
+class QuickMatchLinter
+
+  def self.debug=(value) @@debug = value; end
+  def self.debug?() @@debug ||= false; end  ## note: default is FALSE
+  def debug?()  self.class.debug?; 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
+
+
+  include Logging
+
+  def initialize( txt )
+    @errors = []
+    @txt = txt
+  end
+
+  attr_reader :errors
+  def errors?() @errors.size > 0; end
+
+
+
+  CLUB_NAME_RE =  %r{^
+          (?<name>[^()]+?)     ## non-greedy
+          (?:
+             \s+
+             \(
+               (?<code>[A-Z][A-Za-z]{2,3})    ## optional (country) code; support single code e.g. (A) - why? why not?
+             \)
+          )?
+        $}x   ## note - allow (URU) and (Uru) - why? why not
+
+
+  def parse
+    ## note: every (new) read call - resets errors list to empty
+    @errors = []
+
+    data = {}   # return data hash with leagues
+                #    and seasons
+                #   for now merge stage into matches
+
+    secs = QuickLeagueOutlineReader.parse( @txt )
+    pp secs    if debug?
+
+    secs.each do |sec|   ## sec(tion)s
+      season = Season.parse( sec[:season] )   ## convert (str) to season obj!!!
+      league = sec[:league]
+      stage  = sec[:stage]
+      lines  = sec[:lines]
+
+      start =  if season.year?
+                  Date.new( season.start_year, 1, 1 )
+                else
+                  Date.new( season.start_year, 7, 1 )
+                end
+
+###
+##      db check - check league
+##       fix/fix - add season to match_by   !!!!!!!
+      recs =  Import::League.match_by( name: league )
+      league_rec = nil
+      if recs.size == 1
+        league_rec = recs[0]
+        puts "     OK #{league} => #{league_rec.name}"
+      elsif recs.size == 0
+        msg = "NAME ERROR - no league match found for >#{league}<"
+        @errors << [msg]
+        puts "!! #{msg}"
+    else
+        msg = "NAME ERROR - ambigous; too many league matches (#{recs.size}) found for >#{league}<"
+        @errors << [msg]
+        puts "!! #{msg}"
+    end
+
+
+      parser = MatchParser.new( lines,
+                                start )   ## note: keep season start_at date for now (no need for more specific stage date need for now)
+
+      auto_conf_teams,  matches, rounds, groups = parser.parse
+
+      ## auto-add "upstream" errors from parser
+      @errors += parser.errors  if parser.errors?
+
+
+      if debug?
+        puts ">>> #{auto_conf_teams.size} teams:"
+        pp auto_conf_teams
+        puts ">>> #{matches.size} matches:"
+        ## pp matches
+        puts ">>> #{rounds.size} rounds:"
+        pp rounds
+        puts ">>> #{groups.size} groups:"
+        pp groups
+      end
+
+###
+##      db check - check teams
+##     only clubs for now
+##       fix add better support for champs etc
+##               and national teams!!!
+    auto_conf_teams.each do |team|
+      recs =  if league_rec && !league_rec.intl?
+                 Import::Club.match_by( name: team, league: league_rec )
+              else
+                 ###
+                 ##  get country code from name
+                 ##    e.g. Liverpool FC (ENG) or
+                 ##         Liverpool FC (URU) etc.
+
+                 ## check for country code
+                 if m=CLUB_NAME_RE.match( team )
+                   if m[:code]
+                     Import::Club.match_by( name:    m[:name],
+                                            country: m[:code] )
+                   else
+                      msg = "PARSE ERROR - country code missing for club name in int'l tournament; may not be unique >#{team}<"
+                      @errors << [msg]
+                      Import::Club.match_by( name: team )
+                    end
+                 else
+                   msg = "PARSE ERROR - invalid club name; cannot match with CLUB_NAME_RE >#{team}<"
+                   @errors << [msg]
+                   []
+                 end
+              end
+        team_rec = nil
+        if recs.size == 1
+          team_rec = recs[0]
+          puts "     OK #{team} => #{team_rec.name}, #{team_rec.country.name}"
+        elsif recs.size == 0
+          msg = "NAME ERROR - no team match found for >#{team}<"
+          @errors << [msg]
+          puts "!! #{msg}"
+        else
+          msg = "NAME ERROR - ambigous; too many team matches (#{recs.size}) found for >#{team}<"
+          @errors << [msg]
+          puts "!! #{msg}"
+        end
+    end
+
+
+
+
+
+      ## note: pass along stage (if present): stage  - optional from heading!!!!
+      if stage
+        matches.each do |match|
+          match = match.update( stage: stage )
+        end
+      end
+
+      data[ league ] ||= {}
+      data[ league ][ season.key ] ||= []
+
+      data[ league ][ season.key ] += matches
+      ## note - skip teams, rounds, and groups for now
+    end
+
+## check - only one league and one season
+##             allowed in quick style
+
+
+    leagues = data.keys
+    if leagues.size != 1
+        puts "!! (QUICK) PARSE ERROR - expected one league only; got #{leagues.size}:"
+        pp leagues
+        exit 1
+    end
+
+    seasons = data[ leagues[0] ].keys
+    if seasons.size != 1
+        puts "!! (QUICK) PARSE ERROR - expected one #{leagues[0]} season only; got #{seasons.size}:"
+        pp seasons
+        exit 1
+    end
+
+    data[ leagues[0] ][ seasons[0] ]
+  end # method parse
+
+end # class QuickMatchLinter
+end # module SportDb
+

data/lib/sportdb/formats/version.rb

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