From 4e3f69fda7c990845986d72fd6a403a93fab939d Mon Sep 17 00:00:00 2001
From: j2blake <jeb228@cornell.edu>
Date: Wed, 23 Oct 2013 16:03:45 -0400
Subject: [PATCH] Polish the obsolete_URI script.

---
 .../ISF-transition/obsoleteUris/event.rb      |  6 +++
 .../obsoleteUris/known_exceptions.rb          |  6 +++
 .../ISF-transition/obsoleteUris/report.rb     | 24 ++++++++++-
 .../obsoleteUris/vivo_known_exceptions.txt    | 41 +++++++++++++++++++
 4 files changed, 75 insertions(+), 2 deletions(-)
 create mode 100644 utilities/ISF-transition/obsoleteUris/vivo_known_exceptions.txt

diff --git a/utilities/ISF-transition/obsoleteUris/event.rb b/utilities/ISF-transition/obsoleteUris/event.rb
index 5222a123..00f1a598 100644
--- a/utilities/ISF-transition/obsoleteUris/event.rb
+++ b/utilities/ISF-transition/obsoleteUris/event.rb
@@ -1,4 +1,10 @@
 class Event
+  attr_reader :path
+  attr_reader :line_number
+  attr_reader :line
+  attr_reader :string
+  attr_reader :is_localname
+  
   # ------------------------------------------------------------------------------------
   private
   # ------------------------------------------------------------------------------------
diff --git a/utilities/ISF-transition/obsoleteUris/known_exceptions.rb b/utilities/ISF-transition/obsoleteUris/known_exceptions.rb
index b948752c..871782ee 100644
--- a/utilities/ISF-transition/obsoleteUris/known_exceptions.rb
+++ b/utilities/ISF-transition/obsoleteUris/known_exceptions.rb
@@ -68,6 +68,12 @@ class KnownExceptions
   
   def skip?(file, line_number = -1, string = "@!#IMPOSSIBLE#!@")
     @skippers.each() do |skipper|
+      if line_number == -1
+        next if skipper.is_a?(LineSkipper)
+      else
+        next if skipper.is_a?(ExtensionSkipper) || skipper.is_a?(PathSkipper)
+      end
+
       if skipper.skip?(file, line_number, string)
         return true
       end
diff --git a/utilities/ISF-transition/obsoleteUris/report.rb b/utilities/ISF-transition/obsoleteUris/report.rb
index 3794dafe..aa9dce78 100644
--- a/utilities/ISF-transition/obsoleteUris/report.rb
+++ b/utilities/ISF-transition/obsoleteUris/report.rb
@@ -19,8 +19,27 @@ class Report
     @extensions_count.sort.each do |pair|
       puts "    #{pair[0]}  #{pair[1]}"
     end
+    puts "-----------------------------------------------------------------"
+    puts
   end
-  
+ 
+  def collate_and_list_events()
+    hash = Hash.new{|h, k| []}
+    @events.each do |event|
+      hash[event.path] = hash[event.path] << event
+    end
+    
+#    puts "FLAT: #{hash.to_a.flatten}"
+    hash.sort.each do |path, events|
+      puts "#{path}"
+      events.sort{|a, b| a.line_number <=> b.line_number }.each do |e|
+        puts "   #{e.line_number} #{e.line}"
+        puts "       #{e.is_localname ? "Localname" : "URI"} #{e.string}"
+      end
+      puts "--------------------"
+    end
+  end 
+    
   def list_events()
     @events.each do |event|
       puts "Event: #{event}"
@@ -50,6 +69,7 @@ class Report
   def report()
   	state_arguments()
   	file_summary()
-  	list_events()
+#  	list_events()
+  	collate_and_list_events()
   end
 end
diff --git a/utilities/ISF-transition/obsoleteUris/vivo_known_exceptions.txt b/utilities/ISF-transition/obsoleteUris/vivo_known_exceptions.txt
new file mode 100644
index 00000000..3409da4c
--- /dev/null
+++ b/utilities/ISF-transition/obsoleteUris/vivo_known_exceptions.txt
@@ -0,0 +1,41 @@
+.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