rsssf 0.2.0 → 0.3.0

data/lib/rsssf/fmtfix/fmtfix.rb

@@ -0,0 +1,101 @@
+
+module Rsssf
+class  Fmtfix    ## todo: find a better name e.g. Format or Fixer or ??
+
+
+
+## convenience helper
+def self.fmtfix_pages( pages, outdir:, path:, heading_patches: nil )
+      @@fmtfix ||= new   ## use a "shared" built-in fmtfix
+      @@fmtfix.fmtfix_pages( pages, outdir: outdir,
+                                    path: path,
+                                    heading_patches: heading_patches )
+end
+
+def fmtfix_pages( pages, outdir:,
+                         path:,         ## (lookup search) path (array expected!!!)
+                         heading_patches: nil )
+
+     pages.each_with_index do |config,i|
+
+            puts "==> #{i+1}/#{pages.size} #{config.pretty_inspect}..."
+
+            page = config['page']
+            dirname  = File.dirname( page )
+            basename = File.basename( page, File.extname( page ) )
+            extname  = File.extname( page )
+
+            inname = "#{dirname}/#{basename}.txt"
+            filename = find_file!( inname, path: path )
+
+            txt = read_text( filename )
+            newtxt = fmtfix( txt, heading_patches: heading_patches )
+
+            outfile = File.join(  outdir, "#{basename}.txt" )
+            write_text( outfile, newtxt )
+     end
+end
+
+
+
+## convenience helper
+def self.fmtfix( txt, heading_patches: nil )
+      @@fmtfix ||= new   ## use a "shared" built-in fmtfix
+      @@fmtfix.fmtfix( txt, heading_patches: heading_patches )
+end
+
+
+
+def fmtfix( txt,  heading_patches: nil )
+
+        ### note - step 1
+        ##      autofix-outline
+        ##  and patch headings/outline if empty
+        ##        with at_headings.txt, de_headings.txt etc.
+
+        ## get title
+        meta = Page.parse_meta( txt )
+        title = meta[:title] || 'n/a'
+
+        newtxt = autofix_outline( txt, title: title )
+
+
+        if heading_patches
+            ##
+            ## check if any headings / outline
+             headings = _scan_outline( newtxt )
+             if headings.size == 0
+                newtxt = patch_headings( newtxt, heading_patches )
+             end
+        end
+
+
+        newtxt = autofix( newtxt )
+
+
+
+=begin
+        ##
+        ## add (quick) outline
+        outline = build_outline( newtxt )
+
+        ## add inside  <!-- source: ...  [auto-add here] -->
+        ## e.g.
+        ##   <!--
+        ##      source: https://rsssf.org/tableso/oost98.html
+        ##    -->
+
+        newtxt = newtxt.sub( %r{^[ ]*<!--
+                       [ \n]*
+                         (source: .+?)
+                        [ \n]*
+                      -->
+                   }ix,
+               "<!--\n  \\1\n\n#{outline} -->" )
+=end
+         newtxt
+end
+
+
+end    ## class Fmtfix
+end    ## module Rsssf

data/lib/rsssf/fmtfix/goals.rb

@@ -0,0 +1,173 @@
+module Rsssf
+class  Fmtfix    ## todo: find a better name e.g. Format or Fixer or ??
+
+
+
+###############
+## todo - fix/fix/fix/fix
+
+=begin
+
+
+(Match winners plus best 3 overall qualify for Quarterfinals)
+(Os vencedores dos jogos e os outros 3 melhores passam às Quartas-de-Final)
+
+-- rework goal line!!!!
+   only if following score line  must include  4-4 or such !!!
+=end
+
+
+##
+##  note -  exclude colon (:) too!!
+##    will note match props such as  [red card: ...]
+##                                   [ref: ...]
+##    and others
+#
+##  note - [^] by default always excludes newline (\n)
+
+
+GOALS_ = %q{
+                 [^:\[\]\n]*?
+                    \b
+                    \d{1,3}  '?  ## incl. minute
+                 [^\[\]\n]*?
+           }
+
+
+
+def handle_goals( txt, opts: {} )
+
+
+##
+##  quick fix - change [pen] to (pen) and
+##                     [og] to (og)
+##   e.g. [Parkin 57 [og] - Nogan 47]
+##        [McIndoe 11 [pen] Green 20, Blundell 90 - Robinson 74]
+
+
+   txt = txt.gsub( '[pen]', '(pen)')
+   txt = txt.gsub( '[og]', '(og)')
+
+
+
+=begin
+   ##   [15' Barisic, 80' Gilewicz; 10' (og) Barisic]
+   ##  try (simple) goal line
+   ##   note keep leading spaces / indent
+
+##  note - first line must include a score!!
+###      change to named captures!! - use \k<> !!!
+   txt = txt.gsub( %r{^
+                        (  .+?
+                             \d{1,2}-\d{1,2}
+                           .*?
+                          \n
+                         )
+                     ([ ]*)
+                       \[
+                        ( .*?
+                           \b\d{1,3}'  ## incl. minute
+                          .*?
+                        )
+                      \]
+                     [ ]*
+                    $}ix,
+                    '\1\2(\3)' )
+=end
+
+
+
+
+  ##  try (simple double) goal line
+   ##   note keep leading spaces / indent
+  ## [21' Dospel, 42' and 64' Mayrleb, 51' Datoru, 72' Sobczak; 25' and
+  ## 90' B.Akwuegbu]
+  ##  -or-
+  ###  [Jose Manuel Jurado 12, Diego Forlán 40, 63,
+  ##   "Simao" Pedro Fonseca 90]
+  ##  [Rubén Suárez 10; Abdoulay Konko 12, 63, Alvaro Negredo 27,
+  ##   "Renato" Dirnei Florencio 87]
+
+
+##  ["Edmilson" Gomes de Moraes 40, Marco Perez 68,
+##   Ander Herrera 82; Fernando Fernandez 1, 27,
+##   Juan Miguel Jimenez "Juanmi" 6, 28, Quincy Owusu-abeyie 35]
+##  or
+##  [Jose Manuel Casado 16,Emiliano Armenteros 20,
+##   Jorge Andujar Moreno "Coke" 60; Jose Javier Barkero 14pen,
+##   Jose Antonio Culebras 90+].
+##    note - remove optional
+
+   txt = txt.gsub( %r{^
+                     ([ ]*)
+                       \[
+                        (            #{GOALS_}
+                              \n     #{GOALS_}
+                             (?:
+                                 \n  #{GOALS_}
+                             )?
+                        )
+                      \]
+                      \.?  ## optional trailing dot
+                      [ ]*
+                    $}ix,
+                    '\1(\2)' )
+
+
+## note - match for single line goes last !!
+
+
+###
+###    [Fernando Llorente 47]
+##   [Sebastián Fernández 44; Aritz Aduriz 9, Joaquín Sanchez 71, 75]
+   ##  try (simple) goal line with number only!!!
+   ##   note keep leading spaces / indent
+
+## Fluminense     3-0   0-2  São Caetano
+##    [Magno Alves 70', 88', Roni 75']
+##    [Daniel 15', Magrão 46'p]
+
+## fix/fix/fix - merge with rule above!!!
+##                make minute optional!!!
+
+   txt = txt.gsub( %r{^
+                      (?<match>  .+?
+                                \d{1,2}-\d{1,2}
+                                .*?
+                         )
+                          \n
+                     (?<indent1>  [ ]*)
+                       \[
+                        (?<goals1> #{GOALS_})
+                      \]
+                     [ ]*
+                (?:   ## check for second goal line following
+                      ##   used in br for aggregate matches
+                    \n
+                    (?<indent2> [ ]*)
+                       \[
+                        (?<goals2> #{GOALS_})
+                      \]
+                     [ ]*
+                )?
+                    $}ix ) do |match|
+
+                     if opts[:goals]
+                        puts "  match:"
+                        puts match
+                     end
+
+                      m = Regexp.last_match
+                      buf = String.new
+                      buf += "#{m[:match]}\n"
+                      buf += "#{m[:indent1]}(#{m[:goals1]})"
+                      buf += "\n#{m[:indent2]}(#{m[:goals2]})"   if m[:indent2] && m[:goals2]
+                      buf
+                    end
+
+   txt
+end
+
+
+end    ## class Fmtfix
+end    ## module Rsssf

data/lib/rsssf/fmtfix/headers.rb

@@ -0,0 +1,326 @@
+module Rsssf
+class  Fmtfix    ## todo: find a better name e.g. Format or Fixer or ??
+
+
+
+## let's you check optional ref e.g. ‹§fin›
+OPT_REF = %q{
+            (?: [ ]*
+              ‹ (?<ref> §[^›]+?) ›
+            )?
+         }
+
+
+
+=begin
+## let's you check optional ref e.g. ‹§fin›
+OPT_REF = %q{
+            (?: [ ]*
+              ‹§ (?<ref> [^›]+?) ›
+            )?
+         }
+=end
+
+
+###
+### note - allow optional colon e.g.
+##     Playoff:
+##     Round 21:
+
+HEADER_ROUND_RE = %r{\A
+        [ ]*
+         (?<round> #{ROUND_PAT})
+              :?   ## note - allow optional colon (:)  e.g. Playoff:
+            #{OPT_REF}
+         [ ]*
+\z}ix
+
+
+
+## date header (w/ brackets)
+##   [Aug 7]
+##   [Oct 23]
+##
+###  note - might be date range or date list!!!
+##   [Aug 7-9]
+##   [Aug 7, 8]
+
+
+
+## helper for inline regexes (with union) and escaped
+def self.date_( *re )
+      raise ArgumentError, "more than one date regex expected, got #{re}"  if re.size < 1
+
+      ## (auto-)wrap in non-capature group - why? why not?
+      "(?: #{Regexp.union( *re ).source})"
+end
+
+
+HEADER_DATE_RE = %r{\A
+      [ ]*
+      \[  #{date_(DATE_I_RE, DATE_IB_RE,
+                  DATE_II_RE,
+                  DATE_RANGE_RE,
+                  DATE_LIST_RE, DATE_LEGS_RE,
+                  )}
+      \]
+      [ ]*
+\z}ix
+
+## pp HEADER_DATE_RE
+## pp DATE_I_RE
+## pp DATE_I_RE.source  ## note - will NOT include re flags (e.g. +i/insensitive)
+## exit 1
+
+
+
+## alternate date header (no brackets incl. year)
+##     Aug 7 1999
+##     Sep 4 1999
+##    Oct 23 1999
+##    Nov 20 1999
+##    Apr 1 2000
+
+HEADER_DATE_II_RE = %r{\A
+      [ ]*
+         #{date_(DATE_I_RE, DATE_II_RE)}
+      [ ]*
+\z}ix
+
+
+
+##
+## [Sep 16, Berchtold 26, Glasner 54, Kuljic 60]
+## --- note - exclude  numbers in follow-up text!!!
+##
+##   use a shared pattern for city-like text !!
+##      maybe allow more and make more specific later
+#
+##   exclude comma (,) - why? why not?
+##    split in CITY_ and CITY_PLUS_ or such?
+##    or find a better name ??
+##
+## allow number if:
+##    Happel-Stadion, Wien, att: 9,200
+##    Happel-Stadion, Wien; att: 7000
+##    Innsbruck; att: 6700
+##    Wörthersee-Stadion, Klagenfurt; att: 30,000
+##    Wörthersee Stadion, Klagenfurt; att: 20,500
+##    Hayward, Calif.; att: 5.528   -- note: dot (.) NOT comma (,)
+##
+##
+## Apr 30, 28 Black Arena, Klagenfurt; att: 30,000
+###   Wörthersee Stadion, known as 28 Black Arena for sponsorship reasons
+##
+##   Ernst-Happel-Stadion, Wien; att: 20100; ref: Hofmann
+
+CITY_ = %q{   (?<city>  (?:   [^0-9:;\[\]]+?
+                            | .+?
+                                [ ] att: [ ] [0-9,.]+
+                                (?: [;,] [ ] ref: [ ] .+?  ## w/ optional ref:
+                                )?
+                         )
+               )
+          }
+
+
+##  [Jun 3, Ferrol]
+##  [Apr 2, Wembley]
+##  -or-
+##  [Sat May 17 - at Millennium Stadium, Cardiff]
+##  [Sun May 25 - at Millennium Stadium, Cardiff]
+
+HEADER_DATE_N_CITY_RE = %r{\A
+      [ ]*
+      \[  #{date_(DATE_I_RE,
+                  DATE_II_RE)}
+           (?:       , [ ]*
+               | [ ] - [ ] at [ ]
+            )
+           #{CITY_}
+      \]
+      [ ]*
+\z}ix
+
+
+###
+##  alternate date header with brackets (in oost02.txt)
+##   [31-08]  change to _ 31/08 _
+##   [07-09]
+##   [07-09]
+##   [30-05, Thaur]
+
+HEADER_DATE_ALT_RE = %r{\A
+      [ ]*
+      \[  (?<date>
+             (?<day> \d{1,2}) - (?<month> \d{1,2})
+          )
+          (?:
+              , [ ]*
+              #{CITY_}
+          )?
+      \]
+      [ ]*
+\z}ix
+
+
+
+
+
+
+
+## Round 24 [Mar 21]
+##  note - might be date range or date list
+## Round 24 [Mar 21, 22]
+## Round 24 [Mar 21-23]
+## Round 2 [Aug 4-6]
+## Round 1 [Aug 13-16]
+## Round 2 [Aug 20-23]
+##
+###  note - with optional ref/anchor
+## Preliminary Round [Nov 20]  ‹§inplay›
+
+
+HEADER_ROUND_N_DATE_RE = %r{\A
+        [ ]*
+         (?<round> #{ROUND_PAT})
+         [ ]+
+        \[
+           #{date_(DATE_I_RE, DATE_IB_RE, DATE_II_RE,
+                   DATE_RANGE_RE,
+                   DATE_LIST_RE, DATE_LEGS_RE)}
+        \]
+        #{OPT_REF}
+        [ ]*
+\z}ix
+
+
+## Final [May 1, Klagenfurt]
+HEADER_ROUND_N_DATE_N_CITY_RE = %r{\A
+        [ ]*
+         (?<round> #{ROUND_PAT})
+         [ ]+
+        \[  #{date_(DATE_I_RE, DATE_II_RE)}
+             , [ ]*
+           #{CITY_}
+        \]
+        [ ]*
+\z}ix
+
+
+###
+## Final [in Völs]
+## Final [in Kundl]
+HEADER_ROUND_N_CITY_RE = %r{\A
+        [ ]*
+         (?<round> #{ROUND_PAT})
+         [ ]+
+        \[in [ ]+ #{CITY_}
+        \]
+        [ ]*
+\z}ix
+
+
+##
+## reverse
+##  Final [Graz, May 12]
+## Super Cup Final [Graz, Jul 6]
+## Final [London, Feb 27]
+HEADER_ROUND_N_CITY_N_DATE_RE = %r{\A
+        [ ]*
+         (?<round> #{ROUND_PAT})
+         [ ]+
+        \[ #{CITY_}
+             , [ ]*
+            #{date_(DATE_I_RE, DATE_II_RE)}
+        \]
+        [ ]*
+\z}ix
+
+
+
+
+#####
+## note - line-by-line processing / matching
+def _norm_date( m, format: nil )
+   ## quick fix for undefined group name reference
+   m = m.named_captures.transform_keys(&:to_sym)  if m.is_a?(MatchData)
+
+  if m[:date_list]
+    _fmt_date_list(_build_date_list( m ), format: format )
+  elsif m[:date_legs]
+    _fmt_date_legs(_build_date_legs( m ), format: format )
+  elsif m[:date_range]
+    _fmt_date_range(_build_date_range( m ), format: format )
+  else   ## assume m[:date]
+    _fmt_date(_build_date( m ), format: format )
+  end
+end
+
+
+def handle_header( line )
+      ## note - returns    newline (matched header line reformatted)
+      ##                    or nil (if no match!!)
+      ##
+       line = line.rstrip   ## expect chomp of newline "upstream" - why? why not?
+
+
+      if m = HEADER_ROUND_RE.match(line.rstrip)
+                   "▪ #{m[:round]} ▪\n"
+      elsif m = HEADER_DATE_RE.match(line.rstrip)
+                   ## e.g. [Nov 20]
+                   ## e.g. [April 1]
+                   date = _norm_date( m )
+                   "_ #{date} _\n"
+      elsif m = HEADER_DATE_N_CITY_RE.match(line.rstrip)
+                   ## e.g. [Jun 3, Ferrol]
+                   ## e.g. [Apr 2, Wembley]
+                   ##   [Sat May 17 - at Millennium Stadium, Cardiff]
+                   ##   [Sun May 25 - at Millennium Stadium, Cardiff]
+
+                   date = _norm_date( m )
+
+                   ##  note - check for special case
+                   ##     [Dec 10, replay]
+                   ##  change to  ▪ Replay ▪   _ Dec 10 _
+                   if m[:city] == 'replay'
+                      "▪ Replay ▪  _ #{date} _\n"
+                   else
+                      "_ #{date} _ @ #{m[:city]}\n"
+                   end
+      elsif m = HEADER_DATE_II_RE.match(line.rstrip)
+                    ##  note - no enclosing brackets []!!!
+                    ## e.g. Nov 20 1999  or Nov 20, 1999
+                    ##      Apr 1 2000   or Apr 1, 2000
+                     date = _norm_date( m )
+                   "_ #{date} _\n"
+      elsif m = HEADER_DATE_ALT_RE.match(line.rstrip)
+                    ## e.g. [07-09]
+                    ##      [30-05, Thaur]
+                    ## date = _norm_date( m, format: 'numeric' )
+                    date = _norm_date( m  )
+                    buf = String.new
+                    buf += "_ #{date} _"
+                    buf += " @ #{m[:city]}"    if m[:city]
+                    buf += "\n"
+                    buf
+      elsif m = HEADER_ROUND_N_DATE_RE.match(line.strip)
+                     date = _norm_date( m )
+                   "▪ #{m[:round]} ▪  #{date}\n"
+      elsif m = HEADER_ROUND_N_DATE_N_CITY_RE.match(line.strip)
+                     date = _norm_date( m )
+                   "▪ #{m[:round]} ▪  #{date} @ #{m[:city]}\n"
+      elsif m = HEADER_ROUND_N_CITY_RE.match(line.strip)
+                   "▪ #{m[:round]} ▪  @ #{m[:city]}\n"
+      elsif m = HEADER_ROUND_N_CITY_N_DATE_RE.match(line.strip)
+                     date = _norm_date( m )
+                    ## note - reverse (rotate) date & city
+                   "▪ #{m[:round]} ▪  #{date} @ #{m[:city]}\n"
+       else
+         nil
+       end
+end
+
+
+end    ## class Fmtfix
+end    ## module Rsssf