# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1251319889 -7200
# Node ID 17ffd3434b27adec5c26106ce6ea186a1e9f08ab
# Parent  5ef1ed1945938a5504bb45e6b3d98cd833714abe
Switched to a "VInt" (variable-length integer) representation for numbers and
string lengths.

diff -r 5ef1ed194593 -r 17ffd3434b27 iixr.py
--- a/iixr.py	Tue Aug 25 23:53:20 2009 +0200
+++ b/iixr.py	Wed Aug 26 22:51:29 2009 +0200
@@ -53,38 +53,23 @@
         # Special case: one byte containing zero.
 
         elif number == 0:
-            self.f.write(chr(1) + chr(0))
+            self.f.write(chr(0))
             return
 
         # Write the number from least to most significant digits.
 
-        nbytes = 0
         bytes = []
 
         while number != 0:
-            lsd = number & 255
+            lsd = number & 127
+            number = number >> 7
+            if number != 0:
+                lsd |= 128
             bytes.append(chr(lsd))
-            number = number >> 8
-            nbytes += 1
 
-        # Too large numbers are not supported.
-
-        if nbytes > 255:
-            raise ValueError, "Number %r is too large." % number
-
-        bytes.insert(0, chr(nbytes))
         record = "".join(bytes)
         self.f.write(record)
 
-    def write_unsigned_byte(self, number):
-
-        "Write 'number' to the file using a single byte."
-
-        if not (0 <= number <= 255):
-            raise ValueError, "Number %r is out of range." % number
-
-        self.f.write(chr(number))
-
     def write_string(self, s):
 
         "Write 's' to the file, recording its length."
@@ -94,7 +79,7 @@
         if not (0 <= length <= 255):
             raise ValueError, "String %r is too long." % s
 
-        self.write_unsigned_byte(length)
+        self.write_number(length)
         self.f.write(s)
 
 class FileReader(File):
@@ -105,39 +90,31 @@
 
         "Read a number from the file."
 
-        nbytes = self.read_unsigned_byte()
-
         # Read each byte, adding it to the number.
 
-        bytes = self.f.read(nbytes)
-
-        i = 0
         shift = 0
         number = 0
+        more = 1
 
-        while i < nbytes:
-            csd = ord(bytes[i])
+        while more:
+            byte = self.f.read(1)
+            if not byte:
+                raise EOFError
+
+            csd = ord(byte)
+            more = csd & 128 != 0
+            if more:
+                csd &= 127
             number += (csd << shift)
-            shift += 8
-            i += 1
+            shift += 7
 
         return number
 
-    def read_unsigned_byte(self):
-
-        "Read a number from the file, consuming a single byte."
-
-        s = self.f.read(1)
-        if not s:
-            raise EOFError
-
-        return ord(s)
-
     def read_string(self):
 
         "Read a string from the file."
 
-        length = self.read_unsigned_byte()
+        length = self.read_number()
         return self.f.read(length)
 
 # Specific classes.
@@ -282,7 +259,7 @@
         common = len(commonprefix([self.last_term, term]))
         suffix = term[common:]
 
-        self.write_unsigned_byte(common)
+        self.write_number(common)
         self.write_string(suffix)
 
         # Write the offset delta.
@@ -310,7 +287,7 @@
 
         # Read the prefix length and term suffix.
 
-        common = self.read_unsigned_byte()
+        common = self.read_number()
         suffix = self.read_string()
 
         self.last_term = self.last_term[:common] + suffix