xapian-dump: Add a little mor indentation

[notmuch] / xapian-dump.cc

diff --git a/xapian-dump.cc b/xapian-dump.cc
index 0ab5e32c7cb6f4f1810a07ec026e903f0bc38a26..0364481f1d99adae86272b64fa3cbe6a444d53e8 100644 (file)
--- a/xapian-dump.cc
+++ b/xapian-dump.cc
@@ -1,4 +1,4 @@
-/* xapian-dump: Dump all document IDs from a Xapian database
+/* xapian-dump: Create a textual dump of a Xapian database.
  *
  * Copyright © 2009 Carl Worth
  *
@@ -18,26 +18,116 @@
  * Author: Carl Worth <cworth@cworth.org>
  */
 
+/* Currently the dumped data includes:
+ *
+ *     All document IDs
+ *
+ * And for each document ID:
+ *
+ *     Document data
+ *     All document terms
+ *     All document values
+ */
+
 #include <cstdlib>
 #include <iostream>
+#include <algorithm>
 
 #include <xapian.h>
 
 using namespace std;
 
+vector<int> UNSERIALIZE;
+
+static void
+print_document_terms (Xapian::Document doc)
+{
+    Xapian::TermIterator i;
+
+    printf ("    Terms:\n");
+
+    for (i = doc.termlist_begin (); i != doc.termlist_end (); i++)
+       cout << "\t" << *i << endl;
+}
+
+static int
+vector_int_contains (vector<int> v, int i)
+{
+    vector<int>::iterator result;
+
+    result = find (v.begin(), v.end(), i);
+
+    return result != v.end();
+}
+
+static void
+print_document_values (Xapian::Document doc)
+{
+    Xapian::ValueIterator i;
+    int value_no, value_int;
+    double value_float;
+
+    printf ("    Values:\n");
+
+    for (i = doc.values_begin (); i != doc.values_end (); i++) {
+       value_no = i.get_valueno();
+
+       cout << "\t" << i.get_valueno() << ": ";
+
+       if (vector_int_contains (UNSERIALIZE, value_no)) {
+           value_float = Xapian::sortable_unserialise (*i);
+           value_int = value_float;
+           if (value_int == value_float)
+               cout << value_int;
+           else
+               cout << value_float;
+       } else {
+           cout << *i;
+       }
+
+       cout << endl;
+    }
+}
+
+static void
+print_document (Xapian::Database db, Xapian::docid id)
+{
+    Xapian::Document doc;
+
+    printf ("Document %u:\n", id);
+
+    doc = db.get_document (id);
+
+    printf ("    Data:\n");
+    cout << "\t" << doc.get_data () << endl;
+
+    print_document_terms (doc);
+
+    print_document_values (doc);
+}
+
 int
 main (int argc, char *argv[])
 {
     const char *database_path;
+    int i;
 
     if (argc < 2) {
-       fprintf (stderr, "Usage: %s <path-to-xapian-database>\n",
+       fprintf (stderr, "Usage: %s <path-to-xapian-database> [value_nos...]\n",
                 argv[0]);
+       fprintf (stderr, "Dumps data from the given database.\n");
+       fprintf (stderr, "The values corresponding to any value numbers given on the command line\n");
+       fprintf (stderr, "will be unserialized to an before being printed.\n");
        exit (1);
     }
 
     database_path = argv[1];
 
+    UNSERIALIZE = vector<int> ();
+
+    for (i = 2; i < argc; i++)
+       UNSERIALIZE.push_back (atoi (argv[i]));
+
     try {
 
        Xapian::Database db;
@@ -47,7 +137,8 @@ main (int argc, char *argv[])
        db = Xapian::Database (database_path);
        for (i = db.postlist_begin (""); i != db.postlist_end (""); i++) {
            doc_id = *i;
-           printf ("Found document %u\n", doc_id);
+
+           print_document (db, doc_id);
        }
 
     } catch (const Xapian::Error &error) {