Filters/FilterPdf.cs

   1 //
   2 // FilterPdf.cs: Very simplistic PDF filter
   3 //
   4 // Author:
   5 //   Christopher Orr <dashboard@protactin.co.uk>
   6 //
   7 // Copyright 2004 by Christopher Orr
   8 //
   9
  10 using System;
  11 using System.IO;
  12 using System.Diagnostics;
  13
  14 using Beagle.Util;
  15 using Beagle.Daemon;
  16
  17 namespace Beagle.Filters {
  18
  19         public class FilterPdf : Beagle.Daemon.Filter {
  20
  21                 public FilterPdf ()
  22                 {
  23                         AddSupportedFlavor (FilterFlavor.NewFromMimeType ("application/pdf"));
  24                         SnippetMode = true;
  25                 }
  26
  27                 // FIXME: we should have a reasonable failure mode if pdftotext is
  28                 // not installed.
  29
  30                 protected override void DoPullProperties ()
  31                 {
  32                         // create new external process
  33                         SafeProcess pc = new SafeProcess ();
  34                         pc.Arguments = new string [] { "pdfinfo", FileInfo.FullName };
  35                         pc.RedirectStandardOutput = true;
  36
  37                         try {
  38                                 pc.Start ();
  39                         } catch (SafeProcessException e) {
  40                                 Log.Warn (e.Message);
  41                                 Error ();
  42                                 return;
  43                         }
  44
  45                         // add pdfinfo's output to pool
  46                         StreamReader pout = new StreamReader (pc.StandardOutput);
  47                         string str = null;
  48                         string[] tokens = null;
  49                         string strMetaTag = null;
  50                         bool bKeyword = false;
  51
  52                         while ((str = pout.ReadLine ()) != null) {
  53                                 bKeyword = false;
  54                                 strMetaTag = null;
  55                                 tokens = str.Split (':');
  56                                 if (tokens.Length > 1) {
  57                                         switch (tokens[0]) {
  58                                         case "Title":
  59                                                 strMetaTag = "dc:title";
  60                                                 break;
  61                                         case "Author":
  62                                                 strMetaTag = "dc:author";
  63                                                 break;
  64                                         case "Pages":
  65                                                 strMetaTag = "fixme:page-count";
  66                                                 bKeyword = true;
  67                                                 break;
  68                                         case "Creator":
  69                                                 strMetaTag = "dc:creator";
  70                                                 break;
  71                                         case "Producer":
  72                                                 strMetaTag = "dc:appname";
  73                                                 break;
  74                                         }
  75                                         if (strMetaTag != null) {
  76                                                 if (bKeyword)
  77                                                         AddProperty (Beagle.Property.NewUnsearched (strMetaTag,
  78                                                                                                  tokens[1].Trim()));
  79                                                 else
  80                                                         AddProperty (Beagle.Property.New (strMetaTag,
  81                                                                                           tokens[1].Trim()));
  82                                         }
  83
  84                                 }
  85                         }
  86                         pout.Close ();
  87                         pc.Close ();
  88                 }
  89
  90                 protected override void DoPull ()
  91                 {
  92                         // create new external process
  93                         SafeProcess pc = new SafeProcess ();
  94                         pc.Arguments = new string [] { "pdftotext", "-q", "-nopgbrk", "-enc", "UTF-8", FileInfo.FullName, "-" };
  95                         pc.RedirectStandardOutput = true;
  96
  97                         try {
  98                                 pc.Start ();
  99                         } catch (SafeProcessException e) {
 100                                 Log.Warn (e.Message);
 101                                 Error ();
 102                                 return;
 103                         }
 104
 105                         // add pdftotext's output to pool
 106                         StreamReader pout = new StreamReader (pc.StandardOutput);
 107
 108                         // FIXME:  I don't think this is really required
 109                         // Line by line parsing, however, we have to make
 110                         // sure, that "pdftotext" doesn't output any "New-lines".
 111                         string str;
 112                         while ((str = pout.ReadLine()) != null) {
 113                                 AppendText (str);
 114                                 AppendStructuralBreak ();
 115                                 if (! AllowMoreWords ())
 116                                         break;
 117                         }
 118                         pout.Close ();
 119                         pc.Close ();
 120                         Finished ();
 121                 }
 122         }
 123 }
 124