[GNUnet-SVN] r2149 - in Extractor: . src/plugins/pdf

[grothoff]

Author: grothoff
Date: 2005-09-26 20:33:37 -0700 (Mon, 26 Sep 2005)
New Revision: 2149

Modified:
   Extractor/ChangeLog
   Extractor/src/plugins/pdf/pdfextractor.cc
Log:
minor improvements to PDF

Modified: Extractor/ChangeLog
===================================================================
--- Extractor/ChangeLog 2005-09-26 14:21:56 UTC (rev 2148)
+++ Extractor/ChangeLog 2005-09-27 03:33:37 UTC (rev 2149)
@@ -1,3 +1,12 @@
+Mon Sep 26 20:44:10 PDT 2005
+       Minor improvements to the PDF extractor:
+       - first change is to avoid outputting keywords with empty values
+         (for now the new check is only effective when the value is seen
+         as a string of 8-bit characters; I'm not sure how to rewrite it
+         for the Unicode case in the if branch just above.)
+        - second change is to remap PDF Creator as 'software' keyword
+          instead of 'creator'.
+
 Sun Sep 25 11:31:51 PDT 2005
        Made sure extract returns error code (1) if some files could not
        be accessed.

Modified: Extractor/src/plugins/pdf/pdfextractor.cc
===================================================================
--- Extractor/src/plugins/pdf/pdfextractor.cc   2005-09-26 14:21:56 UTC (rev 
2148)
+++ Extractor/src/plugins/pdf/pdfextractor.cc   2005-09-27 03:33:37 UTC (rev 
2149)
@@ -87,17 +87,45 @@
          con = (char*) convertToUtf8((const char*) u, 2, "UNICODE");
          strcat(result, con);
          free(con);
-       }               
+       }
        next = addKeyword(type,
                          strdup(result),
                          next);
        free(result);
       } else {
-       next = addKeyword(type,
-                         convertToUtf8(s,
-                                       strlen(s),
+        unsigned int len = (NULL == s) ? 0 : strlen(s);
+
+        while(0 < len) {
+        /*
+         * Avoid outputting trailing spaces.
+         *
+         * The following expression might be rewritten as
+         * (! isspace(s[len - 1]) && 0xA0 != s[len - 1]).
+         * There seem to exist isspace() implementations 
+         * which do return non-zero from NBSP (maybe locale-dependent).
+         * Remove ISO-8859 non-breaking space (NBSP, hex value 0xA0) from
+         * the expression if it looks suspicious (locale issues for instance).
+         *
+         * Squeezing out all non-printable characters might also be useful.
+         */
+          if ( (' '  != s[len - 1]) && ((char)0xA0 != s[len - 1]) &&
+               ('\r' != s[len - 1]) && ('\n' != s[len - 1]) &&
+               ('\t' != s[len - 1]) && ('\v' != s[len - 1]) &&
+               ('\f' != s[len - 1]) )
+             break;
+
+          else
+            len --;
+        }
+
+        /* there should be a check to truncate preposterously long values. */
+
+        if (0 < len) {
+         next = addKeyword(type,
+                         convertToUtf8(s, len,
                                        "ISO-8859-1"),
                          next);
+        }
       }
     }
     obj.free();
@@ -251,9 +279,14 @@
                               "Author",
                               EXTRACTOR_AUTHOR,
                               result);
+      /*
+       * we now believe that Adobe's Creator
+       * is not a person nor an organisation,
+       * but just a piece of software.
+       */
       result = printInfoString(info.getDict(),
                               "Creator",
-                              EXTRACTOR_CREATOR,
+                              EXTRACTOR_SOFTWARE,
                               result);
       result = printInfoString(info.getDict(),
                               "Producer",