# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1290207674 -3600
# Node ID 00995a70f535af56298371b86b68bb59efb4642c
# Parent  4614ef99dbe102f234f6c50ebf2cd05f1e7aed42
An experiment adding preceding text to position records.

diff -r 4614ef99dbe1 -r 00995a70f535 iixr/phrases.py
--- a/iixr/phrases.py	Sat Nov 20 00:00:01 2010 +0100
+++ b/iixr/phrases.py	Sat Nov 20 00:01:14 2010 +0100
@@ -4,7 +4,7 @@
 Phrase iterators providing navigation over common positions for a number of
 different terms.
 
-Copyright (C) 2009 Paul Boddie <paul@boddie.org.uk>
+Copyright (C) 2009, 2010 Paul Boddie <paul@boddie.org.uk>
 
 This program is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free Software
@@ -168,6 +168,8 @@
             raise StopIteration
 
     def is_phrase_position(self, last, last_token, current, current_token):
-        return current - last <= 1 and current_token > last_token
+        last_position, last_preceding = last
+        position, preceding = current
+        return position - last_position <= 1 and current_token > last_token
 
 # vim: tabstop=4 expandtab shiftwidth=4
diff -r 4614ef99dbe1 -r 00995a70f535 iixr/positions.py
--- a/iixr/positions.py	Sat Nov 20 00:00:01 2010 +0100
+++ b/iixr/positions.py	Sat Nov 20 00:01:14 2010 +0100
@@ -19,7 +19,7 @@
 """
 
 from iixr.files import *
-from iixr.data import vint, vint_to_array
+from iixr.data import vint, vint_to_array, string_to_array
 from array import array
 
 class PositionWriter(FileWriter):
@@ -53,8 +53,11 @@
 
         last = 0
 
-        for position in positions:
+        # Handle tuples incorporating preceding text.
+
+        for position, preceding in positions:
             vint_to_array(position - last, output)
+            string_to_array(preceding, output)
             last = position
 
         output.tofile(self.f)
@@ -89,7 +92,8 @@
 
         while i < npositions:
             last += self.read_number()
-            positions.append(last)
+            preceding = self.read_string()
+            positions.append((last, preceding))
             i += 1
 
         return self.last_docnum, positions