Polish the obsolete_URI script, so it can be used to scan an active workspace, not just a released directory.

utilities/ISF-transition/obsoleteUris/doit2

@@ -0,0 +1,7 @@
+ruby obsoleteUriChecker.rb /Users/jeb228/git/VIVO  \
+   ../../../productMods/WEB-INF/ontologies/update/diff.tab.txt \
+   vivo_known_exceptions_2.txt complete > scan_VIVO_maint_branch
+ruby obsoleteUriChecker.rb /Users/jeb228/git/Vitro  \
+   ../../../productMods/WEB-INF/ontologies/update/diff.tab.txt \
+   vivo_known_exceptions_2.txt complete > scan_Vitro_maint_branch
+

utilities/ISF-transition/obsoleteUris/event.rb

@@ -5,10 +5,16 @@ class Event
   attr_reader :string
   attr_reader :is_localname
   
+  require 'pathname'
+
   # ------------------------------------------------------------------------------------
   private
   # ------------------------------------------------------------------------------------
   
+  def relativize(root, path)
+    Pathname.new(path).relative_path_from(Pathname.new(root)).to_s
+  end
+
   # ------------------------------------------------------------------------------------
   public
   # ------------------------------------------------------------------------------------
@@ -21,7 +27,11 @@ class Event
     @is_localname = string[0] == ?:
   end
   
-  def to_s()
+  def to_s(directory_root = nil)
-    "#{@path} \n   #{@line_number} #{@line} \n   #{@string} #{@is_localname ? "Localname" : "URI"}"
+    if directory_root
+      "#{relativize(directory_root, @path)} \n   #{@line_number} #{@line} \n   #{@string} #{@is_localname ? "Localname" : "URI"}"
+    else
+      "#{@path} \n   #{@line_number} #{@line} \n   #{@string} #{@is_localname ? "Localname" : "URI"}"
+    end
   end
 end

utilities/ISF-transition/obsoleteUris/known_exceptions.rb

@@ -1,5 +1,22 @@
+require 'pathname'
+
 class KnownExceptionsError < StandardError; end
 
+class GlobSkipper
+  def initialize(root_path, glob)
+    @root_path = root_path
+    @glob = glob
+  end
+  
+  def relativize(path)
+    Pathname.new(path).relative_path_from(Pathname.new(@root_path)).to_s
+  end
+
+  def skip?(path, line, uri)
+    return File.fnmatch(@glob, relativize(path))
+  end
+end
+
 class ExtensionSkipper
   def initialize(extension)
     @extension = extension
@@ -44,10 +61,11 @@ class KnownExceptions
       line.strip!
       next if line.length == 0 || line[0..0] == '#' || line[0] == ?!
 
-      if line =~ /^\.[^\/]*$/
+#      if line =~ /^\.[^\/]*$/
-        skippers << ExtensionSkipper.new(line)
+#        skippers << ExtensionSkipper.new(line)
-      elsif line =~ /^(\S+)\s*$/
+      if line =~ /^(\S+)\s*$/
-        skippers << PathSkipper.new(@root_path, $1)
+#        skippers << PathSkipper.new(@root_path, $1)
+        skippers << GlobSkipper.new(@root_path, $1)
       elsif line =~ /^(\S+)\s*(\d+)\s*$/
         skippers << LineSkipper.new(@root_path, $1, $2.to_i)
       else
@@ -71,7 +89,7 @@ class KnownExceptions
       if line_number == -1
         next if skipper.is_a?(LineSkipper)
       else
-        next if skipper.is_a?(ExtensionSkipper) || skipper.is_a?(PathSkipper)
+        next if skipper.is_a?(ExtensionSkipper) || skipper.is_a?(PathSkipper) || skipper.is_a?(GlobSkipper)
       end
 
       if skipper.skip?(file, line_number, string)

utilities/ISF-transition/obsoleteUris/obsoleteUriChecker.rb

@@ -80,8 +80,8 @@ class ObsoleteUriChecker
   # ------------------------------------------------------------------------------------
 
   def initialize(args)
-    @report = Report.new(args) 
     @directory_root, @obsolete_uris, @known_exceptions, @complete = parse_arguments(args)
+    @report = Report.new(args, @directory_root) 
   rescue UsageError => e
     puts "\n----------------\nUsage error\n----------------\n\n#{e}\n\n----------------\n\n"
     exit

utilities/ISF-transition/obsoleteUris/report.rb

@@ -3,6 +3,10 @@ class Report
   private
   # ------------------------------------------------------------------------------------
 
+  def relativize(path)
+    Pathname.new(path).relative_path_from(Pathname.new(@directory_root)).to_s
+  end
+
   def state_arguments()
   	puts
   	puts "-----------------------------------------------------------------"
@@ -29,9 +33,8 @@ class Report
       hash[event.path] = hash[event.path] << event
     end
     
-#    puts "FLAT: #{hash.to_a.flatten}"
     hash.sort.each do |path, events|
-      puts "#{path}"
+      puts "#{relativize(path)}"
       events.sort{|a, b| a.line_number <=> b.line_number }.each do |e|
         trimmed = 
           if e.line.size <= 100
@@ -46,18 +49,13 @@ class Report
     end
   end 
     
-  def list_events()
-    @events.each do |event|
-      puts "Event: #{event}"
-    end
-  end
-  
   # ------------------------------------------------------------------------------------
   public
   # ------------------------------------------------------------------------------------
 
-  def initialize(args)
+  def initialize(args, directory_root)
   	@args = args;
+  	@directory_root = directory_root
     @file_count = 0
     @extensions_count = Hash.new(0)
     @events = []
@@ -75,7 +73,6 @@ class Report
   def report()
   	state_arguments()
   	file_summary()
-#  	list_events()
   	collate_and_list_events()
   end
 end

utilities/ISF-transition/obsoleteUris/vivo_known_exceptions_2.txt

@@ -0,0 +1,46 @@
+.*
+bin
+
+*.GIF
+*.as
+*.class
+*.fla
+*.gif
+*.gz
+*.ico
+*.jar
+*.jpg
+*.psd
+*.png
+*.swf
+*.war
+*.zip
+
+**/.*
+
+#
+# first_pass: no excluded files. everything was duplicated in the .bin directory, and
+#          probably in the .build directory also. Ran against the repository, so VIVO
+#          only.
+#
+#    7498   89730 2277668 first_pass.output
+#
+
+test/edu/cornell/mannlib/vitro/webapp/search/solr/NIHVIVO3853_DataSet1.rdf
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-event-1.5.owl
+productMods/WEB-INF/ontologies/update/oldAnnotations/vivo-core-1.5-annotations.rdf
+productMods/WEB-INF/ontologies/update/diff.tab.txt
+
+#
+# second_pass: excluded these nasty files. Ran against a clean distro, with VIVO and
+#          Vitro, but no .bin or .build
+#
+#    1798    5159  170092 second_pass.output
+#
+
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-foaf-1.5.owl
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-core-1.5.owl
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-bibo-1.5.owl
+productMods/WEB-INF/ontologies/update/oldVersion/scires-1.5.owl 
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-dcterms-1.5.owl
+productMods/WEB-INF/ontologies/update/oldVersion/vivo-dcelements-1.5.owl