sportdb-parser 0.3.1 → 0.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3721d30a9ec1145f59d6bc84bb9a6cf81330fdafa00314decc2c165f2c6b92c1
-  data.tar.gz: a50b917e0bc5db3ac21cb4ead5507233cbd91f4b6f3b52668ef74ba8c6db6140
+  metadata.gz: 8c4d7c1c55b5122f11fc1fdd863e2b8a5cc929b9ae9c57bcc0322400516ab4b8
+  data.tar.gz: f5fcae93a1010e1d74fd4f453cb6a201f8f6e11de1ad0d0c9a572c9bb00bd540
 SHA512:
-  metadata.gz: 3e286842bcb5c163d841d414a15d4cbd91359324b88106603e22ebbec19b4cb8ffff3119d92a5c9e302e6aeabbc8a7ba74975efebee11ba275ce1d49f0286c1f
-  data.tar.gz: f237aedda025eb35bb08b31345941a3f1f073cbf8c3495c1afcf6e45de73f067e47a8a4c55981d5881079a5ca45a04af427908f379555208c98851c5d4751aa1
+  metadata.gz: 93cb2dc95acf74b1a26088afe3162f0f87a25e949d3f81cd6408c35ec9c42d7309a7c5f59a4c29870146b72e77ed6d5bc3e14c754abf2925b1564b55cafe0ac3
+  data.tar.gz: 6de22cc1e40f8e423786cef7386ae890533a627b6877f6f17c73d47e6956b2de181c6f3d357a731be4476781c363abaab0de9b21c83a4f88f8b058f8a3a7d005

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.3.1
+### 0.3.3
 
 ### 0.0.1 / 2024-07-12
 

data/Manifest.txt

@@ -4,8 +4,10 @@ README.md
 Rakefile
 bin/fbtok
 lib/sportdb/parser.rb
+lib/sportdb/parser/fbtok/main.rb
 lib/sportdb/parser/lang.rb
 lib/sportdb/parser/linter.rb
+lib/sportdb/parser/opts.rb
 lib/sportdb/parser/outline_reader.rb
 lib/sportdb/parser/parser.rb
 lib/sportdb/parser/token-date.rb

data/bin/fbtok

@@ -6,61 +6,7 @@
 require 'sportdb/parser'
 
 
-require 'optparse'   ## check - already auto-required in cocos? keep? why? why not?
-
-
-args=ARGV
-
-
-opts = {
-    debug: true,
-    metal: false,
-}
-
-parser = OptionParser.new do |parser|
-  parser.banner = "Usage: #{$PROGRAM_NAME} [options]"
-
-  parser.on( "--verbose", "--debug",
-               "turn on verbose / debug output (default: #{opts[:debug]})" ) do |debug|
-    opts[:debug] = true
-  end
-
-  parser.on( "--metal",
-                 "turn off typed parse tree; show to the metal tokens"+
-                   " (default: #{opts[:metal]})" ) do |metal|
-    opts[:metal] = true
-  end
-end
-parser.parse!( args )
-
-puts "OPTS:"
-p opts
-puts "ARGV:"
-p args
-
-
-SportDb::Parser::Linter.debug = true    if opts[:debug]
-
-linter = SportDb::Parser::Linter.new
-errors = []
-
-paths = args
-paths.each_with_index do |path,i|
-    puts "==> [#{i+1}/#{paths.size}] reading >#{path}<..."
-    linter.read( path, parse: !opts[:metal] )
-
-    errors += linter.errors    if linter.errors?
-end
-
-if errors.size > 0
-    puts
-    pp errors
-    puts
-    puts "!!   #{errors.size} parse error(s) in #{paths.size} datafiles(s)"
-else
-    puts
-    puts "OK   no parse errors found in #{paths.size} datafile(s)"
-end
+Fbtok.main( ARGV )
 
 
 puts "bye"

data/lib/sportdb/parser/fbtok/main.rb

@@ -0,0 +1,139 @@
+
+module Fbtok
+def self.main( args=ARGV )
+
+    opts = {
+        debug: true,
+        metal: false,
+        file:  nil,
+    }
+
+    parser = OptionParser.new do |parser|
+      parser.banner = "Usage: #{$PROGRAM_NAME} [options]"
+
+
+      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( "--metal",
+                     "turn off typed parse tree; show to the metal tokens"+
+                       " (default: #{opts[:metal]})" ) do |metal|
+        opts[:metal] = 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?
+                 [
+                   '../../../openfootball/euro/2021--europe/euro.txt',
+                   '../../../openfootball/euro/2024--germany/euro.txt',
+                 ]
+              else
+                ## check for directories
+                ##   and auto-expand
+                SportDb::Parser::Opts.expand_args( args )
+              end
+        specs << [paths, {}]
+    end
+
+
+    SportDb::Parser::Linter.debug = true    if opts[:debug]
+
+    linter = SportDb::Parser::Linter.new
+
+
+    specs.each_with_index do |(paths, rec),i|
+       errors = []
+
+       paths.each_with_index do |path,j|
+          puts "==> [#{j+1}/#{paths.size}] reading >#{path}<..."
+          linter.read( path, parse: !opts[:metal] )
+
+          errors += linter.errors    if linter.errors?
+       end
+
+       if errors.size > 0
+          puts
+          pp errors
+          puts
+          puts "!!   #{errors.size} parse error(s) in #{paths.size} datafiles(s)"
+       else
+          puts
+          puts "OK   no parse errors found 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 << "# fbtok 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
+
+  #  maybe write out in the future?
+  # basedir  = File.dirname( opts[:file] )
+  # basename = File.basename( opts[:file], File.extname( opts[:file] ))
+end
+
+
+
+end  # method self.main
+end  #  module Fbtok

data/lib/sportdb/parser/lang.rb

@@ -26,11 +26,13 @@ end
 
 
 ROUND_RE = %r{^(
-
    ## add special case for group play-off rounds!
    ##  group 2 play-off   (e.g. worldcup 1954, 1958)
-     (?:   Group [ ] [a-z0-9]+ [ ]
-           Play-?offs?
+   ##
+   ### note - allow Group ("stand-alone") as "generic" round for now
+   ##      BUT do NOT allow Group 1, Group 2, Group A, Group B, etc.
+     (?: Group [ ] [A-Z0-9]+ [ ] Play-?offs?  |
+         Group
      )
         |
    # round  - note - requiers number e.g. round 1,2, etc.
@@ -44,29 +46,30 @@ ROUND_RE = %r{^(
       )
        |
    ##  starting with qual(ification)
-   ## Qual. Round 1 / Qual. Round 2 / Qual. Round 3
-     (?:  Qual \. [ ]
-          Round
-           [ ] [1-9][0-9]*
-      )
+   ##   Qual. Round 1 / Qual. Round 2 / Qual. Round 3
+   ##  or
+   ##  Playoff Round 1
+   ##  Play-in Round 1
+     (?:  (?: Qual \. |
+              Play-?off |
+              Play-?in
+          )
+           [ ] Round [ ] [1-9][0-9]* )
        |
    ## 1. Round / 2. Round / 3. Round / etc.
-   ##  Play-off Round
    ##  First Round
+   ##  Play-off Round
    ##  Final Round   (e.g. Worldcup 1950)
       (?:
            (?: [1-9][0-9]* \.  |
-                Play-?off   |
                 1st | First   |
                 2nd | Second  |
+                Play-?off   |
                 Final
            )
              [ ] Round
        )
        |
-  ## Playoff Round 1
-     (?:  Play-?off [ ] Round [ ] [1-9][0-9]* )
-       |
   ## starting with preliminary
   #   e.g.  Preliminary round
      (?:  Preliminary  [ ]
@@ -135,6 +138,11 @@ ROUND_RE = %r{^(
          )
         [ ] Replays?
       )
+     |
+  ## more
+     (?:
+          Reclassification
+     )
 )$}ix
 
 

data/lib/sportdb/parser/opts.rb

@@ -0,0 +1,78 @@
+
+module SportDb
+class Parser
+
+
+###
+## note - Opts Helpers for now nested inside Parser - keep here? why? why not?
+class Opts
+
+    SEASON_RE = %r{ (?:
+                       \d{4}-\d{2}
+                     | \d{4}(--[a-z0-9_-]+)?
+                    )
+                  }x
+    SEASON = SEASON_RE.source    ## "inline" helper for embedding in other regexes - keep? why? why not?
+
+
+    ## note: if pattern includes directory add here
+    ##     (otherwise move to more "generic" datafile) - why? why not?
+    MATCH_RE = %r{ (?: ^|/ )      # beginning (^) or beginning of path (/)
+                       #{SEASON}
+                     /[a-z0-9_-]+\.txt$  ## txt e.g /1-premierleague.txt
+                }x
+
+
+def self.find( path, dir: nil )
+    ## check - rename dir
+    ##          use root_dir or work_dir or cd or such - why? why not?
+
+    datafiles = []
+
+    ## note: normalize path - use File.expand_path ??
+    ##    change all backslash to slash for now
+    ## path = path.gsub( "\\", '/' )
+    path =  if dir
+              File.expand_path( path, File.expand_path( dir ))
+            else
+              File.expand_path( path )
+            end
+
+    ## check all txt files
+    ## note: incl. files starting with dot (.)) as candidates
+    ##     (normally excluded with just *)
+    candidates = Dir.glob( "#{path}/**/{*,.*}.txt" )
+    ## pp candidates
+    candidates.each do |candidate|
+      datafiles << candidate    if MATCH_RE.match( candidate )
+    end
+
+    ## pp datafiles
+    datafiles
+end
+
+
+def self.expand_args( args )
+    paths = []
+
+    args.each do |arg|
+        ## check if directory
+        if Dir.exist?( arg )
+            datafiles = find( arg )
+            puts
+            puts "  found #{datafiles.size} match txt datafiles in #{arg}"
+            pp datafiles
+            paths += datafiles
+        else
+              ## assume it's a file
+            paths << arg
+        end
+    end
+
+    paths
+end
+end  # class Opts
+
+
+end   # class Parser
+end   # module SportDb

data/lib/sportdb/parser/version.rb

@@ -4,7 +4,7 @@ module SportDb
     module Parser
   MAJOR = 0    ## todo: namespace inside version or something - why? why not??
   MINOR = 3
-  PATCH = 1
+  PATCH = 3
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/lib/sportdb/parser.rb

@@ -3,6 +3,10 @@ require 'cocos'
 require 'season/formats'  # e.g. Season() support machinery
 
 
+## more stdlibs
+require 'optparse'   ## check - already auto-required in cocos? keep? why? why not?
+
+
 
 ####
 # try a (simple) tokenizer/parser with regex
@@ -28,7 +32,12 @@ require_relative 'parser/parser'
 ##  todo/check - move outline reader upstream to cocos - why? why not?
 ##       use  read_outline(), parse_outline()  - why? why not?
 require_relative 'parser/outline_reader'
+
+
+require_relative 'parser/opts'
 require_relative 'parser/linter'
+require_relative 'parser/fbtok/main'
+
 
 ###
 #  make parser api (easily) available - why? why not?

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-parser
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.3.3
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-09-18 00:00:00.000000000 Z
+date: 2024-09-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: cocos
@@ -88,8 +88,10 @@ files:
 - Rakefile
 - bin/fbtok
 - lib/sportdb/parser.rb
+- lib/sportdb/parser/fbtok/main.rb
 - lib/sportdb/parser/lang.rb
 - lib/sportdb/parser/linter.rb
+- lib/sportdb/parser/opts.rb
 - lib/sportdb/parser/outline_reader.rb
 - lib/sportdb/parser/parser.rb
 - lib/sportdb/parser/token-date.rb