X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=xapian-dump.cc;h=39c1c3404b02c13e6d01d1c3dfbe944503d63778;hp=0ab5e32c7cb6f4f1810a07ec026e903f0bc38a26;hb=9bc4253fa804b62ff31e8de82a139b2cb12b118f;hpb=26795d64e6150b543d850c1e882a9d1395b58d9e

diff --git a/xapian-dump.cc b/xapian-dump.cc
index 0ab5e32c..39c1c340 100644
--- 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
  *
@@ -18,38 +18,183 @@
  * 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;
+
+unsigned int MAX_TERMS = 0;
+
+static void
+print_escaped_string (const char *s)
+{
+    printf ("\"");
+
+    while (*s) {
+	if (*s == '"')
+	    printf ("\\");
+	printf ("%c", *s);
+	s++;
+    }
+
+    printf ("\"");
+}
+
+static void
+print_document_terms (Xapian::Document doc)
+{
+    Xapian::TermIterator it;
+    unsigned int i;
+
+    printf ("    {\n");
+
+    for (it = doc.termlist_begin (), i = 0;
+	 it != doc.termlist_end ();
+	 it++, i++)
+    {
+	printf ("        ");
+	print_escaped_string ((*it).c_str());
+	printf (",\n");
+    }
+
+    for ( ; i < MAX_TERMS; i++)
+	printf ("        \"\",\n");
+
+    printf ("    },\n");
+}
+
+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;
+
+    for (i = doc.values_begin (); i != doc.values_end (); i++) {
+	value_no = i.get_valueno();
+
+	printf ("    ");
+
+	if (vector_int_contains (UNSERIALIZE, value_no)) {
+	    value_float = Xapian::sortable_unserialise (*i);
+	    value_int = value_float;
+	    if (value_int == value_float)
+		printf ("%d", value_int);
+	    else
+		printf ("\"%f\"", value_float);
+	} else {
+	    print_escaped_string ((*i).c_str ());
+	}
+
+	printf (",\n");
+    }
+
+}
+
+static void
+print_document (Xapian::Database db, Xapian::docid id)
+{
+    Xapian::Document doc;
+
+    printf ("{\n");
+
+    doc = db.get_document (id);
+
+    printf ("    \"%s\",\n", doc.get_data ().c_str());
+
+    print_document_terms (doc);
+
+    print_document_values (doc);
+
+    printf ("},\n");
+}
+
 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];
 
-    try {
+    UNSERIALIZE = vector<int> ();
 
+    for (i = 2; i < argc; i++)
+	UNSERIALIZE.push_back (atoi (argv[i]));
+
+    try {
 	Xapian::Database db;
         Xapian::PostingIterator i;
 	Xapian::docid doc_id;
 
 	db = Xapian::Database (database_path);
+
 	for (i = db.postlist_begin (""); i != db.postlist_end (""); i++) {
+	    Xapian::Document doc;
+
 	    doc_id = *i;
-	    printf ("Found document %u\n", doc_id);
+
+	    doc = db.get_document (doc_id);
+
+	    if (doc.termlist_count () > MAX_TERMS)
+		MAX_TERMS = doc.termlist_count ();
 	}
 
+	printf ("#define MAX_TERMS %d\n\n", MAX_TERMS);
+
+	printf ("typedef struct {\n"
+		"    char data[255];\n"
+		"    char terms[MAX_TERMS][255];\n"
+		"    char message_id[255];\n"
+		"    char thread_id[4096];\n"
+		"    time_t time;\n"
+		"} document_dump_t;\n\n");
+
+	printf ("document_dump_t dump[] = {\n");
+
+	for (i = db.postlist_begin (""); i != db.postlist_end (""); i++) {
+	    doc_id = *i;
+
+	    print_document (db, doc_id);
+	}
+
+	printf ("};\n");
+
     } catch (const Xapian::Error &error) {
 	cerr << "A Xapian exception occurred: " << error.get_msg () << endl;
 	exit (1);