lib/docsplit/transparent_pdfs.rb in docsplit-0.7.4 vs lib/docsplit/transparent_pdfs.rb in docsplit-0.7.5

@@ -6,21 +6,24 @@
 
     # Temporarily convert any non-PDF documents to PDFs before running them
     # through further extraction.
     def ensure_pdfs(docs)
       [docs].flatten.map do |doc|
-        ext = File.extname(doc)
-        if ext.downcase == '.pdf'
+        if is_pdf?(doc)
           doc
         else
           tempdir = File.join(Dir.tmpdir, 'docsplit')
           extract_pdf([doc], {:output => tempdir})
-          File.join(tempdir, File.basename(doc, ext) + '.pdf')
+          File.join(tempdir, File.basename(doc, File.extname(doc)) + '.pdf')
         end
       end
     end
 
+    def is_pdf?(doc)
+      File.extname(doc).downcase == '.pdf' || File.open(doc, 'rb', &:readline) =~ /\A\%PDF-\d+(\.\d+)?/
+    end
+
   end
 
   extend TransparentPDFs
 
-end
\ No newline at end of file
+end