sportdb-parser 0.3.0 → 0.3.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9ebb468318f2b87c33ca66afb6c46611ce5f420258e0c41b40a2cbfabcff7a49
-  data.tar.gz: 0cf1d511f3e936d73531442d1ca6bef94d90a50ae65346b5b57347d4d294dc77
+  metadata.gz: c2deeaae15626c5ed7118035667bf689af0426f3580b9ac50abe48ad310aaf1c
+  data.tar.gz: a3cdff8f2ed1dfb7bf3fa116a0d30928de27f25ecbaf63e75cbcdc90fc60fe0d
 SHA512:
-  metadata.gz: e992ab97d7ae18c514de14078d30eb36adc40f5044242ce9ace089fb88b104c61b29ff86a2aa8101bb7257c3ff2ce32c6150439ff855e195bee1b26032bb0d9d
-  data.tar.gz: 25e66e45e7daf2783bc6507a3cb2c660d9153eab9530210ef51ef6e0d5d3fc531e5891897be3b0492b0ad7ea5fe3d406a0a3dd0559549b85518360d442ed4d8b
+  metadata.gz: 865a3b6844014660fddce03d3236a5a1e3dec003282681c8eb4b6c7ad83e2e91e936d5dd4030188de7578d11258707c72543a1e7b198440d5e9486743e648d9b
+  data.tar.gz: f1449389d729388b1fa29086ad7a89d74690351cd5ead9d561a8f7f5ebbc38dec09487612000055343728094bda859200ecaf71dfef7fd42fb2ecec16a078e98

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.3.0
+### 0.3.2
 
 ### 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] = debug
-  end
-
-  parser.on( "--metal",
-                 "turn off typed parse tree; show to the metal tokens"+
-                   " (default: #{opts[:metal]})" ) do |metal|
-    opts[:metal] = metal
-  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

@@ -64,6 +64,9 @@ ROUND_RE = %r{^(
              [ ] Round
        )
        |
+  ## Playoff Round 1
+     (?:  Play-?off [ ] Round [ ] [1-9][0-9]* )
+       |
   ## starting with preliminary
   #   e.g.  Preliminary round
      (?:  Preliminary  [ ]

data/lib/sportdb/parser/linter.rb

@@ -59,6 +59,7 @@ def read( path, parse: false )
 
   ##  process nodes
   h1 = nil
+  h2 = nil
   orphans = 0    ## track paragraphs's with no heading
 
   attrib_found = false
@@ -69,8 +70,14 @@ def read( path, parse: false )
 
     if type == :h1
         h1 = node[1]  ## get heading text
-        puts
         puts "  = Heading 1 >#{node[1]}<"
+    elsif type == :h2
+        if h1.nil?
+          puts "!! WARN - no heading for subheading; skipping parse"
+          next
+        end
+        h2 = node[1]  ## get heading text
+        puts "  == Heading 2 >#{node[1]}<"
     elsif type == :p
 
        if h1.nil?

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 = 0
+  PATCH = 2
   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: 0.3.0
+  version: 0.3.2
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-09-04 00:00:00.000000000 Z
+date: 2024-09-18 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