fbtok 0.2.3 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b0cf5b4c14ed4e4466a769b2389e9aaf2add0fe631034010778cfc9a7671d088
-  data.tar.gz: 311e35505a4ec9b729784e4adf415888e64a855fda4a2bed3ab971a51c8ebb40
+  metadata.gz: db7e5c105200820cbce7e9701ed2b8ce0506e366f4215a013288b66b625528ff
+  data.tar.gz: 3f02e52b461d14d014a3a4c3283dd452f0ca6316048969b5cf0af1cd4bd457b8
 SHA512:
-  metadata.gz: d9d1a0b0b736368613f2cae07c69ecc9fec2ca41e3b40a77471e3790ce585a4a4fb5a4559c8005deb59dee2e56fa3f489c2025033b8281910ffec8e77ad94356
-  data.tar.gz: 692715563a64d828eea4bbef1528721773dea1b7c7f4a2103079a804991fcc683ae6afff9aa76dfbf116b5050a77771c85c2ae11f5c950b402391f2eec077d5b
+  metadata.gz: e8c77935b70909ac04697276af902f0597f6eed411b8bddee77248a0145c1fb704eb58a59cd49744ac7e93da97fa5410ca5a35825ea3ec4ca0ea4f834d30543f
+  data.tar.gz: 35ef329be9dddaaf94658252e9e2684b4f3f63a89da8484db3507ab36f8afefc319d2ab4f3154389539a7ad6545779dd39e710f0eb7e82990988690cea107458

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.2.3
+### 0.3.0
 ### 0.0.1 / 2025-01-02
 
 * Everything is new. First release.

data/Rakefile

@@ -2,7 +2,7 @@ require 'hoe'
 
 
 Hoe.spec 'fbtok' do
-  self.version = '0.2.3'
+  self.version = '0.3.0'
 
   self.summary = "fbtok - football.txt lint tools incl. tokenizer, parser & more"
   self.description = summary

data/bin/fbchk

@@ -59,52 +59,51 @@ 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
+  puts "OPTS:"
+  p opts
+  puts "ARGV:"
+  p args
+
+  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
+  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|
+
+
+## todo/check - use packs or projects or such
+##                instead of specs - why? why not?
+specs =  if opts[:file]
+            SportDb::Parser::Opts.read_pathspecs( opts[:file] )
+         else
+           paths =  if args.empty?
+                      [] 
+                    else
+                       ## check for directories
+                       ##   and auto-expand
+                       SportDb::Parser::Opts.expand_args( args )
+                    end
+             ## always return array of specs
+            [SportDb::Parser::Opts.build_pathspec( paths: paths)]  
+         end
+
+
+
+
+  specs.each_with_index do |spec,i|
+    paths = spec.paths
+    rec   = spec.rec
+
     errors = []
     paths.each_with_index do |path,j|
-      puts "==> [#{j+1}/#{paths.size}] reading >#{path}<..."
+      puts "==> [#{i+1}/#{specs.size}, #{j+1}/#{paths.size}] reading >#{path}<..."
       quick = SportDb::QuickMatchLinter.new( read_text( path ),
                                               check_teams: opts[:teams] )
       matches = quick.parse
@@ -140,28 +139,9 @@ 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
+  buf = SportDb::Parser::BatchReport.new( 
+            specs, 
+            title: 'fbchk summary report' ).build
 
   puts
   puts "SUMMARY:"
@@ -169,7 +149,7 @@ if opts[:file]
 
   basedir  = File.dirname( opts[:file] )
   basename = File.basename( opts[:file], File.extname( opts[:file] ))
-  outpath  = "#{basedir}/fbcheck.#{basename}.txt"
+  outpath  = "#{basedir}/fbchk.#{basename}.txt"
   write_text( outpath, buf )
 end
 

data/bin/fbt

@@ -50,51 +50,47 @@ require 'fbtok'
 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
 
 
 if opts[:debug]
-   SportDb::QuickMatchReader.debug = true
-   SportDb::MatchParser.debug      = true
+  puts "OPTS:"
+  p opts
+  puts "ARGV:"
+  p args
+  
+  SportDb::QuickMatchReader.debug = true
+  SportDb::MatchParser.debug      = true
 else
-   SportDb::QuickMatchReader.debug = false
-   SportDb::MatchParser.debug      = false
-   LogUtils::Logger.root.level = :info
+  SportDb::QuickMatchReader.debug = false
+  SportDb::MatchParser.debug      = false
+  LogUtils::Logger.root.level = :info
 end
 
 
-  specs.each_with_index do |(paths, rec),i|
+## todo/check - use packs or projects or such
+##                instead of specs - why? why not?
+specs =  if opts[:file]
+            SportDb::Parser::Opts.read_pathspecs( opts[:file] )
+         else
+           paths =  if args.empty?
+                      ['/sports/openfootball/euro/2021--europe/euro.txt',
+                       '/sports/openfootball/euro/2024--germany/euro.txt'] 
+                    else
+                       ## check for directories
+                       ##   and auto-expand
+                       SportDb::Parser::Opts.expand_args( args )
+                    end
+             ## always return array of specs
+            [SportDb::Parser::Opts.build_pathspec( paths: paths)]  
+         end
+
+
+
+  specs.each_with_index do |spec,i|
+    paths = spec.paths
+    rec   = spec.rec
     errors = []
+
     paths.each_with_index do |path,j|
       puts "==> [#{j+1}/#{paths.size}] reading >#{path}<..."
       quick = SportDb::QuickMatchReader.new( read_text( path ) )
@@ -130,28 +126,9 @@ end
 ###
 ## generate a report if --file option used
 if opts[:file]
-
-  buf = String.new
-
-  buf << "# fbt 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
+  buf = SportDb::Parser::BatchReport.new( 
+            specs, 
+            title: 'fbt summary report' ).build
 
   puts
   puts "SUMMARY:"
@@ -163,5 +140,6 @@ if opts[:file]
 end
 
 
+
 puts "bye"
 

data/bin/fbtok

@@ -57,43 +57,36 @@ if opts[:debug]
 end
 
 
+
 ## 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     if opts[:debug]
-
-      ##  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?
-             [
-               '/sports/openfootball/euro/2021--europe/euro.txt',
-               '/sports/openfootball/euro/2024--germany/euro.txt',
-             ]
-          else
-            ## check for directories
-            ##   and auto-expand
-            SportDb::Parser::Opts.expand_args( args )
-          end
-    specs << [paths, {}]
-end
-
+specs =  if opts[:file]
+            SportDb::Parser::Opts.read_pathspecs( opts[:file] )
+         else
+           paths =  if args.empty?
+                      ['/sports/openfootball/euro/2021--europe/euro.txt',
+                       '/sports/openfootball/euro/2024--germany/euro.txt'] 
+                    else
+                       ## check for directories
+                       ##   and auto-expand
+                       SportDb::Parser::Opts.expand_args( args )
+                    end
+             ## always return array of specs
+            [SportDb::Parser::Opts.build_pathspec( paths: paths)]  
+         end
 
 
 linter = SportDb::Parser::Linter.new
 
 
-specs.each_with_index do |(paths, rec),i|
-   errors = []
+specs.each_with_index do |spec,i|
+    paths = spec.paths
+    rec   = spec.rec
+   
+    errors = []
 
    paths.each_with_index do |path,j|
-      puts "==> [#{j+1}/#{paths.size}] reading >#{path}<..."
+      puts "==> [#{i+1}/#{specs.size}, #{j+1}/#{paths.size}] reading >#{path}<..."
       linter.read( path, parse: false )  ## only tokenize (do NOT parse)
 
       errors += linter.errors    if linter.errors?
@@ -119,35 +112,17 @@ 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
+  buf = SportDb::Parser::BatchReport.new( 
+            specs, 
+            title: 'fbtok summary report' ).build
 
-puts
-puts "SUMMARY:"
-puts buf
+  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] ))
+  #  maybe write out in the future?
+  # basedir  = File.dirname( opts[:file] )
+  # basename = File.basename( opts[:file], File.extname( opts[:file] ))
 end
 
 puts "bye"

data/bin/fbtree

@@ -10,6 +10,7 @@ args = ARGV
 
 opts = {
     debug: true,
+    file:  nil,
 }
 
 parser = OptionParser.new do |parser|
@@ -24,6 +25,14 @@ parser = OptionParser.new do |parser|
                "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
+    ## note: for batch (massive) processing auto-set debug (verbose output) to false (as default)
+    opts[:debug] = false
+  end
+
 end
 parser.parse!( args )
 
@@ -44,16 +53,20 @@ end
 
 ## todo/check - use packs or projects or such
 ##                instead of specs - why? why not?
-   paths = if args.empty?
-             [
-               '/sports/openfootball/euro/2021--europe/euro.txt',
-               '/sports/openfootball/euro/2024--germany/euro.txt',
-             ]
-          else
-            ## check for directories
-            ##   and auto-expand
-            SportDb::Parser::Opts.expand_args( args )
-          end
+specs =  if opts[:file]
+            SportDb::Parser::Opts.read_pathspecs( opts[:file] )
+         else
+           paths =  if args.empty?
+                      ['/sports/openfootball/euro/2021--europe/euro.txt',
+                       '/sports/openfootball/euro/2024--germany/euro.txt'] 
+                    else
+                       ## check for directories
+                       ##   and auto-expand
+                       SportDb::Parser::Opts.expand_args( args )
+                    end
+             ## always return array of specs
+            [SportDb::Parser::Opts.build_pathspec( paths: paths)]  
+         end
 
 
 
@@ -78,16 +91,51 @@ end
 linter = SportDb::Parser::Linter.new
 
 
-paths.each_with_index do |path,i|
-
-  puts "==> [#{i+1}/#{paths.size}] reading >#{path}<..."
+specs.each_with_index do |spec,i|
+    paths = spec.paths
+    rec   = spec.rec
+   
+    errors = []
+
+   paths.each_with_index do |path,j|
+      puts "==> [#{i+1}/#{specs.size}, #{j+1}/#{paths.size}] reading >#{path}<..."
+      tree = linter.read( path, parse: true )
+      dump_tree_stats( tree )
+    
+      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
 
-  tree = linter.read( path, parse: true )
 
-  dump_tree_stats( tree )
-end
+###
+## generate a report if --file option used
+if opts[:file]
 
+  buf = SportDb::Parser::BatchReport.new( 
+            specs, 
+            title: 'fbtree summary report' ).build
 
-puts "bye"
+  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
 
+puts "bye"

data/bin/fbx

@@ -47,13 +47,18 @@ require 'fbtok'
 end
 parser.parse!( args )
 
-puts "OPTS:"
-p opts
-puts "ARGV:"
-p args
-
-
 
+if opts[:debug]
+  puts "OPTS:"
+  p opts
+  puts "ARGV:"
+  p args
+  
+  SportDb::MatchParser.debug = true    
+else
+  SportDb::MatchParser.debug = false    
+  LogUtils::Logger.root.level = :info
+end
 
 
 paths =  if args.empty?
@@ -70,12 +75,6 @@ paths =  if args.empty?
 
 
 
-if opts[:debug]
-  SportDb::MatchParser.debug = true    
-else
-  SportDb::MatchParser.debug = false    
-  LogUtils::Logger.root.level = :info
-end
 
 ## errors = []
 

data/lib/fbtok/opts.rb

@@ -31,7 +31,9 @@ class Opts
                 }x
 
 
-def self.find( path, dir: nil )
+
+
+def self._find( path )
     ## check - rename dir
     ##          use root_dir or work_dir or cd or such - why? why not?
 
@@ -40,11 +42,8 @@ def self.find( path, dir: nil )
     ## 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
+    path =  File.expand_path( path )
+          
 
     ## check all txt files
     ## note: incl. files starting with dot (.)) as candidates
@@ -60,29 +59,109 @@ def self.find( path, dir: nil )
 end
 
 
-def self.expand_args( args )
-    paths = []
-
-    args.each do |arg|
-        ## check if directory
-        if Dir.exist?( arg )
-            datafiles = find( arg )
-            if debug?
-              puts
-              puts "  found #{datafiles.size} match txt datafiles in #{arg}"
-              pp datafiles
-            end
-            paths += datafiles
-        else
-              ## assume it's a file
-            paths << arg
+def self._expand( arg )
+    ## check if directory
+    if Dir.exist?( arg )
+          datafiles = _find( arg )
+          if debug?
+            puts
+            puts "  found #{datafiles.size} match txt datafiles in #{arg}"
+            pp datafiles
+          end
+          datafiles
+    else ## assume it's a file
+        ## make sure path exists; raise error if not 
+        if File.exist?( arg )
+          [arg]   ## note - always return an array - why? why not?
+        else    
+          raise Errno::ENOENT, "No such file or directory - #{arg}" 
         end
     end
+end
+
+
+def self.expand_args( args )
+  paths = []
+  args.each do |arg|
+    paths += _expand( arg )
+  end
+  paths 
+end
+
 
-    paths
+
+## todo/check - find a better name 
+##   e.g.  BatchItem or PackageDef or ???
+##
+##   find a different name for rec for named value props?
+##      why? why not?
+PathspecNode = Struct.new( :paths, :rec )
+
+def self.build_pathspec( paths: )
+    PathspecNode.new( paths: paths, rec: {} )
+end
+
+def self.read_pathspecs( src )
+    specs = []
+
+    recs = read_csv( src )
+    pp recs     if debug?
+  
+    ##  note - make pathspecs relative to passed in file arg!!!
+    basedir = File.dirname( src )
+    recs.each do |rec|
+        path = rec['path']
+        fullpath = File.expand_path( path, basedir ) 
+        ## make sure path exists; raise error if not
+        paths =  if Dir.exist?( fullpath )
+                    _find( fullpath )
+                 else
+                    raise Errno::ENOENT, "No such directory - #{fullpath})" 
+                 end
+        
+        specs << PathspecNode.new( paths: paths, rec: rec )
+    end
+    specs
 end
 end  # class Opts
 
 
+##
+#  BatchReport (a.k.a. PathspecsReport)
+
+class BatchReport
+
+  def initialize( specs, title: )
+     @specs = specs
+     @title = title
+  end
+
+  def build
+     buf = String.new
+     buf << "# #{@title} - #{@specs.size} dataset(s)\n\n"
+  
+     @specs.each_with_index do |spec,i|
+       paths  = spec.paths
+       rec    = spec.rec
+       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
+  
+     buf
+  end   # method build
+end  # class BatchReport
 end   # class Parser
 end   # module SportDb

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fbtok
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.3.0
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-01-22 00:00:00.000000000 Z
+date: 2025-01-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: sportdb-parser