From 76ffe8bed83a32860e35923fcd172e408f092a84 Mon Sep 17 00:00:00 2001
From: Giovanni Giorgi <jj@gioorgi.com>
Date: Wed, 16 May 2012 14:45:06 +0200
Subject: [PATCH] First implementation of wild character '*'

---
 lib/code_zauker.rb           | 40 ++++++++++++++++++++++++++++++++++--
 lib/code_zauker/constants.rb |  8 +++++++-
 test/fixture/wildtest.txt    |  1 +
 test/test_wild_search.rb     | 40 ++++++++++++++++++++++++++++++++++++
 4 files changed, 86 insertions(+), 3 deletions(-)
 create mode 100644 test/fixture/wildtest.txt
 create mode 100644 test/test_wild_search.rb

diff --git a/lib/code_zauker.rb b/lib/code_zauker.rb
index 10157b1..cf6cf27 100644
--- a/lib/code_zauker.rb
+++ b/lib/code_zauker.rb
@@ -160,7 +160,12 @@ def initialize(redisConnection=nil)
 
     
     def disconnect()    
-      @redis.quit
+      begin
+        @redis.quit
+      rescue Errno::EAGAIN =>e
+        # Nothing to do...
+        puts "Ignored EAGAIN ERROR during disconnect..."
+      end
     end
 
 
@@ -292,7 +297,7 @@ def split_in_trigrams(term, prefix)
       trigramInAnd=Set.new()
       # Search=> Sea AND ear AND arc AND rch
       for j in 0...term.length
-        currentTrigram=term[j,GRAM_SIZE]        
+        currentTrigram=term[j,GRAM_SIZE]
         if currentTrigram.length <GRAM_SIZE
           # We are at the end...
           break
@@ -330,6 +335,37 @@ def isearch(term)
       return map_ids_to_files(fileIds)      
     end
 
+    # = wild cards search
+    # You can search trigram in the form
+    # public*class*Apple
+    # will match java declaration of MyApple  but not
+    # YourAppManager
+    def wsearch(term,case_sensitive=true)
+      # Split stuff
+      puts "Wild Search request:#{term}"
+      m=term.split("*")
+      if m.length>0
+        trigramInAnd=Set.new()
+        puts "*= Found:#{m.length}"
+        m.each do | wt |
+          puts "Splitting  #{wt}"
+          trigSet=split_in_trigrams(wt,"trigram")
+          trigramInAnd=trigramInAnd.merge(trigSet)
+        end
+        puts "Trigrams: #{trigramInAnd.length}"
+        if trigramInAnd.length==0
+          return []
+        end      
+        fileIds=    @redis.sinter(*trigramInAnd)
+        fileNames=map_ids_to_files(fileIds)
+        puts "DEBUG #{fileIds} #{fileNames}"
+        return fileNames     
+      else
+        puts "Warn no Wild!"
+        return search(term)
+      end
+    end
+
 
     # = search
     # Find a list of file candidates to a search string
diff --git a/lib/code_zauker/constants.rb b/lib/code_zauker/constants.rb
index bae3458..68d596e 100644
--- a/lib/code_zauker/constants.rb
+++ b/lib/code_zauker/constants.rb
@@ -13,6 +13,12 @@ module CodeZauker
                               ".zip",".7z","rar",
                               # MS Office zip-like files...
                               ".pptx",".docx",".xlsx",
+                              # MS Visual Studio big bad files"
+                              ".scc",".datasource",".pdb","vspscc",".settings",
+                              #"Telerik.Web.UI.xml",
+                              ".Web.UI.xml",
+                              # Auto-generated stuff...is suggested to be avoided
+                              ".designer.cs",
                               # Avoid slurping text document too...
                               ".doc",
                               ".ppt",".xls",".rtf",".vsd", ".odf",
@@ -27,7 +33,7 @@ module CodeZauker
                               ".tar",
                               ".gz",".Z",
                               ".dropbox",
-                              ".svn-base",".pdb",".cache", 
+                              ".svn-base",".cache", 
                               #IDE STUFF
                               ".wlwLock",
                               # Music exclusion
diff --git a/test/fixture/wildtest.txt b/test/fixture/wildtest.txt
new file mode 100644
index 0000000..e6ea4b7
--- /dev/null
+++ b/test/fixture/wildtest.txt
@@ -0,0 +1 @@
+Wild Wild West Movie
diff --git a/test/test_wild_search.rb b/test/test_wild_search.rb
new file mode 100644
index 0000000..5f83b2f
--- /dev/null
+++ b/test/test_wild_search.rb
@@ -0,0 +1,40 @@
+# -*- encoding: utf-8 -*-
+# To test use 
+# rake TEST=test/test_wild_search.rb
+require 'test/unit'
+require 'code_zauker'
+
+# See ri Test::Unit::Assertions
+# for assertion documentation
+class FileScannerBasicSearch < Test::Unit::TestCase
+  #This test can search very uinque things...
+  def test_foolish_wild1
+    fs=CodeZauker::FileScanner.new()
+    fs.load("./test/fixture/wildtest.txt")
+    files=fs.wsearch("Wild*West")
+    assert(files.include?("./test/fixture/wildtest.txt")== true, 
+           "Expected file not found. Files found:#{files}")
+    assert(files.length==1)
+  end
+
+  def test_foolish_wild2
+    fs=CodeZauker::FileScanner.new()
+    fs.load("./test/fixture/wildtest.txt")
+    files=fs.wsearch("Wild*West*Movie")
+    assert(files.include?("./test/fixture/wildtest.txt")== true, 
+           "Expected file not found. Files found:#{files}")
+    assert(files.length==1)
+  end
+
+  # Also unordered match will work
+  # So the negative match is difficult
+  def test_foolish_wild3
+    fs=CodeZauker::FileScanner.new()
+    fs.load("./test/fixture/wildtest.txt")
+    files=fs.wsearch("West*Wild*NotOnTheSameLineForSure")
+    assert(files.include?("./test/fixture/wildtest.txt")== false, 
+           "Expected not matching wildtest.txt file. Matches:#{files}")    
+  end
+
+
+end