# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1317477682 -7200
# Node ID 48a194ecc68c42b32cb80f7b43a598c67582cfc0
# Parent  c120a1af7f6c785940cf75d0dfbc9cf879bdd11b
Added short-circuiting of failed searches.
Added term/key conversion support so that numerically sorted keys can be used.

diff -r c120a1af7f6c -r 48a194ecc68c simplex/__init__.py
--- a/simplex/__init__.py	Sat Oct 01 15:07:17 2011 +0200
+++ b/simplex/__init__.py	Sat Oct 01 16:01:22 2011 +0200
@@ -101,9 +101,15 @@
     """
 
     for record in reader.get_records():
-        if term == accessor.get_key(record):
+        key = accessor.get_key(record)
+        if term == key:
             return record
 
+        # Short-circuit failed searches.
+
+        elif term < key:
+            return None
+
     return None
 
 def groups(l, length):
diff -r c120a1af7f6c -r 48a194ecc68c simplex/readers.py
--- a/simplex/readers.py	Sat Oct 01 15:07:17 2011 +0200
+++ b/simplex/readers.py	Sat Oct 01 16:01:22 2011 +0200
@@ -35,20 +35,25 @@
 
     "An accessor using a delimiter to split a record."
 
-    def __init__(self, keys=None, delimiter=None):
+    def __init__(self, keys=None, delimiter=None, numeric=0):
 
         """
         Initialise the accessor using a sequence of 'keys' indicating the
         columns in each record that provide the values in the eventual compound
         key provided by each record, along with a 'delimiter' indicating how
-        such columns are identified.
+        such columns are identified. If 'numeric' is set to a true value, keys
+        will be interpreted as numbers.
         """
 
         self.keys = keys or [0]
         self.delimiter = delimiter
+        self.convert = numeric and self.convert_numeric or (lambda x: x)
+
+    def convert_numeric(self, term):
+        return map(int, term)
 
     def get_key(self, record):
         values = record.split(self.delimiter)
-        return [values[key] for key in self.keys]
+        return self.convert([values[key] for key in self.keys])
 
 # vim: tabstop=4 expandtab shiftwidth=4
diff -r c120a1af7f6c -r 48a194ecc68c test_indexed.py
--- a/test_indexed.py	Sat Oct 01 15:07:17 2011 +0200
+++ b/test_indexed.py	Sat Oct 01 16:01:22 2011 +0200
@@ -5,8 +5,8 @@
 
 try:
     separator = sys.argv.index("--")
-    filename, interval = sys.argv[1:3]
-    keys = map(int, sys.argv[3:separator])
+    filename, numeric, interval = sys.argv[1:4]
+    keys = map(int, sys.argv[4:separator])
     terms = groups(sys.argv[separator+1:], len(keys))
 except (IndexError, ValueError):
     print >>sys.stderr, "Usage: %s <filename> <interval> <key>... -- <term value>..." % sys.argv[0]
@@ -14,7 +14,7 @@
 
 f = open(filename)
 reader = TextFile(f)
-accessor = DelimitedRecord(keys)
+accessor = DelimitedRecord(keys, numeric=(numeric == "true"))
 try:
     t = time.time()
     l = make_index(reader, accessor, int(interval))
@@ -24,7 +24,7 @@
 
     for term in terms:
         t = time.time()
-        line = find_with_index(reader, accessor, l, term)
+        line = find_with_index(reader, accessor, l, accessor.convert(term))
         if line:
             print "Found (at %s seconds)...\n%s" % (time.time() - t, line)
 
diff -r c120a1af7f6c -r 48a194ecc68c test_scan.py
--- a/test_scan.py	Sat Oct 01 15:07:17 2011 +0200
+++ b/test_scan.py	Sat Oct 01 16:01:22 2011 +0200
@@ -5,8 +5,8 @@
 
 try:
     separator = sys.argv.index("--")
-    filename = sys.argv[1]
-    keys = map(int, sys.argv[2:separator])
+    filename, numeric = sys.argv[1:3]
+    keys = map(int, sys.argv[3:separator])
     terms = groups(sys.argv[separator+1:], len(keys))
 except (IndexError, ValueError):
     print >>sys.stderr, "Usage: %s <filename> <key>... -- <term value>..." % sys.argv[0]
@@ -14,13 +14,13 @@
 
 f = open(filename)
 reader = TextFile(f)
-accessor = DelimitedRecord(keys)
+accessor = DelimitedRecord(keys, numeric=(numeric == "true"))
 try:
     for term in terms:
         reader.seek(0)
 
         t = time.time()
-        line = find_in_file(reader, accessor, term)
+        line = find_in_file(reader, accessor, accessor.convert(term))
         if line:
             print "Found (at %s seconds)...\n%s" % (time.time() - t, line)
 finally: