From 3da458ab108ae31c846fb2aeb3efe67e0d9ae925 Mon Sep 17 00:00:00 2001
From: Sascha Silbe <sascha-pgp@silbe.org>
Date: Mon, 22 Aug 2011 18:26:13 +0000
Subject: Don't index terms longer than the hardcoded Xapian limit

Xapian has a hardcoded limit of 245 characters for terms.
---
diff --git a/gdatastore/index.py b/gdatastore/index.py
index 616eb81..5730f54 100644
--- a/gdatastore/index.py
+++ b/gdatastore/index.py
@@ -108,10 +108,12 @@ class TermGenerator(xapian.TermGenerator):
         elif not isinstance(value, str):
             value = str(value)
 
-        # We need to add the full value (i.e. not split into words), too, so
-        # we can enumerate unique values. It also simplifies setting up
-        # dictionary-based queries.
-        self._document.add_term(_PREFIX_FULL_VALUE + prefix + value)
+        # Hardcoded Xapian term length limit
+        if len(prefix + value) < 240:
+            # We need to add the full value (i.e. not split into words), too,
+            # so we can enumerate unique values. It also simplifies setting up
+            # dictionary-based queries.
+            self._document.add_term(_PREFIX_FULL_VALUE + prefix + value)
 
         self.index_text(value, 1, prefix)
         self.increase_termpos()
--
cgit v0.9.1