sportdb-parser 0.3.9 → 0.5.0

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

@@ -1,141 +0,0 @@
-
-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] PATH"
-
-
-      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
-        ## note: for batch (massive) processing auto-set debug (verbose output) to false (as default)
-        opts[:debug] = false
-      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/linter.rb

@@ -1,156 +0,0 @@
-
-module SportDb
-class Parser
-
-###
-## note - Linter for now nested inside Parser - keep? why? why not?
-class Linter
-
-def self.debug=(value) @@debug = value; end
-def self.debug?() @@debug ||= false; end  ## note: default is FALSE
-def debug?()  self.class.debug?; end
-
-
-
-attr_reader :errors
-
-def initialize
-  @errors = []
-  @parser = Parser.new   ## use own parser instance (not shared) - why? why not?
-end
-
-
-def errors?() @errors.size > 0; end
-
-
-
-  ## note:  colon (:) MUST be followed by one (or more) spaces
-  ##      make sure mon feb 12 18:10 will not match
-  ##        allow 1. FC Köln etc.
-  ##               Mainz 05:
-  ##           limit to 30 chars max
-  ##          only allow  chars incl. intl buut (NOT ()[]/;)
-  ##
-  ##   Group A:
-  ##   Group B:   - remove colon
-  ##    or lookup first
-
-  ATTRIB_RE = %r{^
-                   [ ]*?     # slurp leading spaces
-                (?<key>[^:|\]\[()\/; -]
-                       [^:|\]\[()\/;]{0,30}
-                 )
-                   [ ]*?     # slurp trailing spaces
-                   :[ ]+
-                (?<value>.+)
-                    [ ]*?   # slurp trailing spaces
-                   $
-                }ix
-
-
-#########
-## parse - false (default) - tokenize (only)
-##       - true            - tokenize & parse
-def read( path, parse: false )
-  ## note: every (new) read call - resets errors list to empty
-  @errors = []
-
-  nodes = OutlineReader.read( path )
-
-  ##  process nodes
-  h1 = nil
-  h2 = nil
-  orphans = 0    ## track paragraphs's with no heading
-
-  attrib_found = false
-
-
-  nodes.each do |node|
-    type = node[0]
-
-    if type == :h1
-        h1 = node[1]  ## get heading text
-        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?
-         orphans += 1    ## only warn once
-         puts "!! WARN - no heading for #{orphans} text paragraph(s); skipping parse"
-         next
-       end
-
-       lines = node[1]
-
-       tree = []
-       lines.each_with_index do |line,i|
-
-        if debug?
-         puts
-         puts "line >#{line}<"
-        end
-
-
-        ## skip new (experimental attrib syntax)
-        if attrib_found == false &&
-            ATTRIB_RE.match?( line )
-          ## note: check attrib regex AFTER group def e.g.:
-          ##         Group A:
-          ##         Group B:  etc.
-          ##     todo/fix - change Group A: to Group A etc.
-          ##                       Group B: to Group B
-           attrib_found = true
-           ## logger.debug "skipping key/value line - >#{line}<"
-           next
-        end
-
-        if attrib_found
-          ## check if line ends with dot
-          ##  if not slurp up lines to the next do!!!
-          ## logger.debug "skipping key/value line - >#{line}<"
-          attrib_found = false   if line.end_with?( '.' )
-              # logger.debug "skipping key/value line (cont.) - >#{line}<"
-              next
-        end
-
-        t, error_messages  =  if parse
-                                  @parser.parse_with_errors( line )
-                              else
-                                  @parser.tokenize_with_errors( line )
-                              end
-
-
-         if error_messages.size > 0
-            ## add to "global" error list
-            ##   make a triplet tuple (file / msg / line text)
-            error_messages.each do |msg|
-                @errors << [ path,
-                             msg,
-                             line
-                           ]
-            end
-         end
-
-         pp t   if debug?
-
-         tree << t
-       end
-
-       ## pp tree
-    else
-        pp node
-        raise ArgumentError, "unsupported (node) type >#{type}<"
-    end
-  end  # each node
-end  # read
-end  # class Linter
-
-
-end   # class Parser
-end   # module SportDb

data/lib/sportdb/parser/opts.rb

@@ -1,81 +0,0 @@
-
-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?
-    ##   update - note include/allow dot (.) too
-    ##               e.g. 2024-25/at.1.txt
-    ##                        change to at_1 or uefa_cl or such - 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