dblp 0.2.0 → 0.3.2

data/History.txt

@@ -1,9 +1,12 @@
-== 0.0.1 2008-05-29
+== 0.3.2 2008-06-29
+    * refactored interface for grabbing data from different sources
+	* added citeseer for finding the right bibtex entry
 
-* added documentation and more command line options
+== 0.2.0 2008-05-29
 
+* added documentation and more command line options
 
-== 0.0.1 2008-05-28
+== 0.1.0 2008-05-28
 
 * 1 major enhancement:
   * Initial release

data/lib/dblp/grabber.rb

@@ -2,28 +2,61 @@ require 'open-uri'
 
 module Dblp
   
-  # Const url to fetch from
-  DBLP_URL = "http://dblp.uni-trier.de/rec/bibtex/"
-  
   class Grabber
     
-    def grab(key)
+    # Const url to fetch from
+    DBLP_URL = "http://dblp.uni-trier.de/rec/bibtex/"
+    
+    def read_html(url)
       content = ""
-      open(DBLP_URL + key) do |f|
+      open(url) do |f|
         content = f.read
       end
+      content
+    end
 
+    def extract_pre(content)
       # extract the bibtex code, that is in pre tags
       result = content.scan(/<pre>(.*?)<.pre>/mix)
       if result
-        result.inject({}) do |m, k|
-          m[k[0].match(/@.*\{(.*?),/)[1].gsub(/(<.*?>)/, "")] = k[0].gsub(/(<.*?>)/, "")
+        result.inject([]) do |m, k|
+          #m[k[0].match(/@.*\{(.*?),/)[1].gsub(/(<.*?>)/, "")] = k[0].gsub(/(<.*?>)/, "")
+          m << k[0].gsub(/(<.*?>)/, "")
           m
         end
       else
-        {}
+        []
+      end
+    end
+    
+    
+    def grab(key)
+      begin 
+        # Check the key 
+        if key =~ /DBLP:/
+          content = read_html(DBLP_URL + key.gsub("DBLP:", ""))
+          extract_pre(content)
+        else
+          CiteseerGrabber.new.grab(key)
+        end
+      rescue
+        []
+      end
+    end
+    
+  end
+  
+  class CiteseerGrabber < Grabber
+    
+    CITESEE_URL = "http://citeseer.ist.psu.edu/"
+    
+    def grab(key)
+      begin
+        content = read_html(CITESEE_URL + key + ".html")
+        extract_pre(content)
+      rescue
+        []
       end
-      
     end
     
   end

data/lib/dblp/parser.rb

@@ -9,9 +9,12 @@ module Dblp
     
     def parse
       File.readlines(@filename).inject([]) {|m, l|
-          result = l.match(/\\citation\{(.*:)?(.*?)\}/)
-          if result
-            m << result[2]
+          cnt = l.match(/\\citation\{(.*?)\}/)
+          
+          if cnt
+            cnt[1].split(",").each do |t|
+              m << t
+            end
           end
           m
         }.uniq

data/lib/dblp/version.rb

@@ -1,8 +1,8 @@
 module Dblp #:nodoc:
   module VERSION #:nodoc:
     MAJOR = 0
-    MINOR = 2
-    TINY  = 0
+    MINOR = 3
+    TINY  = 2
 
     STRING = [MAJOR, MINOR, TINY].join('.')
   end

data/lib/dblp.rb

@@ -17,11 +17,11 @@ module Dblp
       parser = Dblp::Parser.new(file)
       grabber = Dblp::Grabber.new
       File.open(options.output, "w+") do |f|
-        f.puts parser.parse.inject({}) {|m, l| 
-            m.merge!(grabber.grab(l))
+        f.puts parser.parse.inject([]) {|m, l| 
+            m << grabber.grab(l)
             overall_size = m.size
             m
-          }.values.join("\n")
+          }.uniq.join("\n")
       end
       
       if options.bibtex

data/test/test.aux

@@ -13,6 +13,9 @@
 
 \citation{conf/btw/JacobsA07}
 \citation{DBLP:conf/cidr/StonebrakerBCCGHHLRZ07}
+
+\citation{Martin,Grund}
+
 \citation{Stonebraker:2007fk}
 \@writefile{toc}{\contentsline {section}{\numberline {I}Introduction}{1}{section.1}}
 \newlabel{sec:introduction}{{I}{1}{Introduction\relax }{section.1}{}}

data/test/test_dblp.rb

@@ -9,16 +9,35 @@ class TestDblp < Test::Unit::TestCase
     
     p = Dblp::Parser.new(File.dirname(__FILE__) + "/test.aux")
     result = p.parse
-    assert result.size == 3
+    
+    assert_equal 5, result.size
     assert_equal "conf/btw/JacobsA07", result.first
-    assert_equal "conf/cidr/StonebrakerBCCGHHLRZ07", result[1]
+    assert_equal "DBLP:conf/cidr/StonebrakerBCCGHHLRZ07", result[1]
     
   end
   
   def test_grabber
     
     g = Dblp::Grabber.new
-    assert g.grab("conf/btw/JacobsA07").values.size == 2
+    
+    res = g.grab("DBLP:conf/btw/JacobsA07")
+    assert res.size == 2
+    
+    res = g.grab("DBLP:conf/icde/ZukowskiHNB06")
+    
+    res = g.grab("DBLP:conf/btw/JacobsAss07")
+    assert res.size == 0
+  end
+  
+  
+  def test_citeseer
+    
+    g = Dblp::CiteseerGrabber.new
+    res = g.grab("graefe91data")
+    assert res.size == 1
+    
+    res = g.grab("nixnurnix")
+    assert res.size == 0
     
   end
   

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: dblp
 version: !ruby/object:Gem::Version 
-  version: 0.2.0
+  version: 0.3.2
 platform: ruby
 authors: 
 - Martin Grund
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-05-29 00:00:00 +02:00
+date: 2008-06-12 00:00:00 +02:00
 default_executable: 
 dependencies: []