# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1251322562 -7200
# Node ID 247345e7c82fbd8513674fe492fbc4645fe9a604
# Parent  17ffd3434b27adec5c26106ce6ea186a1e9f08ab
Added position readers and writers to term dictionary instances, permitting the
reading and writing of term and position data.

diff -r 17ffd3434b27 -r 247345e7c82f iixr.py
--- a/iixr.py	Wed Aug 26 22:51:29 2009 +0200
+++ b/iixr.py	Wed Aug 26 23:36:02 2009 +0200
@@ -356,9 +356,10 @@
 
     "Writing term dictionaries."
 
-    def __init__(self, info_writer, index_writer, interval):
+    def __init__(self, info_writer, index_writer, position_writer, interval):
         self.info_writer = info_writer
         self.index_writer = index_writer
+        self.position_writer = position_writer
         self.interval = interval
         self.entry = 0
 
@@ -376,17 +377,29 @@
 
         self.entry += 1
 
+    def write_term_positions(self, term, doc_positions):
+
+        """
+        Write the given 'term' and the 'doc_positions' recording the documents
+        and positions at which the term is found.
+        """
+
+        offset = self.position_writer.write_all_positions(doc_positions)
+        self.write_term(term, offset)
+
     def close(self):
         self.info_writer.close()
         self.index_writer.close()
+        self.position_writer.close()
 
 class TermDictionaryReader:
 
     "Reading term dictionaries."
 
-    def __init__(self, info_reader, index_reader):
+    def __init__(self, info_reader, index_reader, position_reader):
         self.info_reader = info_reader
         self.index_reader = index_reader
+        self.position_reader = position_reader
 
         self.terms = []
         try:
@@ -436,8 +449,19 @@
             else:
                 return None
 
+    def find_positions(self, term):
+
+        "Return the documents and positions at which the given 'term' is found."
+
+        offset = self.find(term)
+        if offset is None:
+            return None
+        else:
+            return self.position_reader.read_all_positions(offset)
+
     def close(self):
         self.info_reader.close()
         self.index_reader.close()
+        self.position_reader.close()
 
 # vim: tabstop=4 expandtab shiftwidth=4
diff -r 17ffd3434b27 -r 247345e7c82f test.py
--- a/test.py	Wed Aug 26 22:51:29 2009 +0200
+++ b/test.py	Wed Aug 26 23:36:02 2009 +0200
@@ -2,7 +2,7 @@
 
 import iixr
 
-numbers = [12345678, 0, 1]
+numbers = [12345678, 0, 1, 127, 128, 255, 256]
 
 f = open("test", "wb")
 w = iixr.FileWriter(f)
@@ -115,24 +115,65 @@
 w = iixr.TermWriter(f)
 f2 = open("testI", "wb")
 w2 = iixr.TermIndexWriter(f2)
-w3 = iixr.TermDictionaryWriter(w, w2, 3)
+f3 = open("testP", "wb")
+w3 = iixr.PositionWriter(f3)
+wd = iixr.TermDictionaryWriter(w, w2, w3, 3)
 for term, offset in terms:
-    w3.write_term(term, offset)
-w3.close()
+    wd.write_term(term, offset)
+wd.close()
 
 f = open("test", "rb")
 r = iixr.TermReader(f)
 f2 = open("testI", "rb")
 r2 = iixr.TermIndexReader(f2)
-r3 = iixr.TermDictionaryReader(r, r2)
+f3 = open("testP", "rb")
+r3 = iixr.PositionReader(f3)
+rd = iixr.TermDictionaryReader(r, r2, r3)
 terms_reversed = terms[:]
 terms_reversed.reverse()
 for term, offset in terms_reversed:
-    o = r3.find(term)
+    o = rd.find(term)
     print offset == o, offset, o
 for term in ("dog", "dingo"):
-    o = r3.find(term)
+    o = rd.find(term)
     print o is None, o
-r3.close()
+rd.close()
+
+terms_with_positions = [
+    ("aardvark",  [(1, [2, 45, 96]), (20, [13])]),
+    ("anteater",  [(1, [43, 44])]),
+    ("badger",    [(7, [2, 22, 196]), (19, [55, 1333]), (21, [0])]),
+    ("bull",      [(6, [128]), (16, [12])]),
+    ("bulldog",   [(43, [17, 19, 256, 512])]),
+    ("cat",       [(123, [12, 145, 196]), (1200, [113])])
+    ]
+
+f = open("test", "wb")
+w = iixr.TermWriter(f)
+f2 = open("testI", "wb")
+w2 = iixr.TermIndexWriter(f2)
+f3 = open("testP", "wb")
+w3 = iixr.PositionWriter(f3)
+wd = iixr.TermDictionaryWriter(w, w2, w3, 3)
+for term, doc_positions in terms_with_positions:
+    wd.write_term_positions(term, doc_positions)
+wd.close()
+
+f = open("test", "rb")
+r = iixr.TermReader(f)
+f2 = open("testI", "rb")
+r2 = iixr.TermIndexReader(f2)
+f3 = open("testP", "rb")
+r3 = iixr.PositionReader(f3)
+rd = iixr.TermDictionaryReader(r, r2, r3)
+terms_reversed = terms_with_positions[:]
+terms_reversed.reverse()
+for term, doc_positions in terms_reversed:
+    dp = rd.find_positions(term)
+    print doc_positions == dp, doc_positions, dp
+for term in ("dog", "dingo"):
+    dp = rd.find_positions(term)
+    print dp is None, dp
+rd.close()
 
 # vim: tabstop=4 expandtab shiftwidth=4