Add Python3 compatibility for RiveScript

Noah Petherbridge · Noah Petherbridge · commit d6d4567f6ebc · 2013-06-12T12:09:03.000-07:00
Make the RiveScript library simultaneously compatible with Python 2 and
Python 3 by importing the print_function from __future__ and reworking
parts of the code to make it compatible with both versions.

Implement the "null character" based placeholders for substitutions that
replaces the old ROT13-based method (bringing the lib up to speed with
the Perl version).

Only call msg.decode('utf8') for Python 2, as this isn't needed in
Python 3 due to strings being Unicode by default.

Make rivescript.python more robust, giving the compile-time AND run-time
error messages as console output when errors occur calling the Python
object macros.

Add example3.py as a Python 3 version of example.py.

Update brain/python.rs's Base64 example to make it compatible with both
Python 2 and Python 3. Add an "add" object macro to demonstrate the
casting to str of the responses from object macros. Make
rivescript.python cast to str to fix any possible sources of crashes
when an object returns an int or other data type.
diff --git a/brain/python.rs b/brain/python.rs
@@ -3,9 +3,33 @@
 ! version = 2.0
 
 > object base64 python
-	import base64 as b64
-	return b64.b64encode(" ".join(args))
+    import base64 as b64
+    mess = " ".join(args)
+
+    # Make this function work in Python3 as well.
+    import sys
+    if sys.version_info[0] == 3:
+        # Python3's Base64 requires bytes, not a str,
+        # so encode the str into bytes.
+        mess = mess.encode()
+        base = b64.b64encode(mess)
+
+        # Return the base64 result, decoded back into a str.
+        return base.decode()
+    else:
+        # Python2 is simple.
+        return b64.b64encode(mess)
+< object
+
+> object add python
+    # This function returns an int, and shows that the results
+    # from python object macros are always casted to str.
+    a, b = args
+    return int(a) + int(b)
 < object
 
 + encode * in base64
 - OK: <call>base64 <star></call>
+
++ what is # plus #
+- <star1> + <star2> = <call>add <star1> <star2></call>
diff --git a/example.py b/example.py
diff --git a/example3.py b/example3.py
@@ -0,0 +1,31 @@
+#!/usr/bin/python3
+
+# Python 3 example
+
+from rivescript import RiveScript
+
+rs = RiveScript()
+rs.load_directory("./brain")
+rs.sort_replies()
+
+print("""This is a bare minimal example for how to write your own RiveScript bot!
+
+For a more full-fledged example, try running: `python rivescript brain`
+This will run the built-in Interactive Mode of the RiveScript library. It has
+some more advanced features like supporting JSON for communication with the
+bot. See `python rivescript --help` for more info.
+
+example.py is just a simple script that loads the RiveScript documents from
+the 'brain/' folder, and lets you chat with the bot.
+
+Type /quit when you're done to exit this example.
+""")
+
+while True:
+    msg = input("You> ")
+    if msg == '/quit':
+        quit()
+    reply = rs.reply("localuser", msg)
+    print("Bot>", reply)
+
+# vim:expandtab
diff --git a/rivescript/__init__.py b/rivescript/__init__.py
@@ -1,6 +1,9 @@
 #!/usr/bin/env python
 # pyRiveScript - A RiveScript interpreter written in Python.
 
+# Python 3 compat
+from __future__ import print_function
+
 __author__     = 'Noah Petherbridge'
 __copyright__  = 'Copyright 2013, Noah Petherbridge'
 __credits__    = [
@@ -15,6 +18,7 @@
 __all__      = ['rivescript']
 __version__  = '1.05'
 
+import sys
 import os
 import glob
 import re
@@ -24,7 +28,7 @@
 import copy
 import codecs
 
-from python import PyRiveObjects
+from rivescript.python import PyRiveObjects
 
 # Common regular expressions.
 re_equals  = re.compile('\s*=\s*')
@@ -33,7 +37,6 @@
 re_weight  = re.compile('\{weight=(\d+)\}')
 re_inherit = re.compile('\{inherits=(\d+)\}')
 re_wilds   = re.compile('[\s\*\#\_]+')
-re_rot13   = re.compile('<rot13sub>(.+?)<bus31tor>')
 re_nasties = re.compile('[^A-Za-z0-9 ]')
 
 # Version of RiveScript we support.
@@ -93,7 +96,7 @@ def VERSION(self=None):
 
     def _say(self, message):
         if self._debug:
-            print "[RS]", message
+            print("[RS]", message)
         if self._log:
             # Log it to the file.
             fh = open(self._log, 'a')
@@ -102,13 +105,13 @@ def _say(self, message):
 
     def _warn(self, message, fname='', lineno=0):
         if self._debug:
-            print "[RS::Warning]",
+            print("[RS::Warning]"),
         else:
-            print "[RS]",
+            print("[RS]"),
         if len(fname) and lineno > 0:
-            print message, "at", fname, "line", lineno
+            print(message, "at", fname, "line", lineno)
         else:
-            print message
+            print(message)
 
     ############################################################################
     # Loading and Parsing Methods                                              #
@@ -877,7 +880,7 @@ def by_length(word1, word2):
         # Sort them.
         output = []
         for count in sorted(track.keys(), reverse=True):
-            sort = sorted(track[count], cmp=by_length)
+            sort = sorted(track[count], key=len, reverse=True)
             output.extend(sort)
 
         self._sorted["lists"][name] = output
@@ -1129,8 +1132,8 @@ def reply(self, user, msg):
     def _format_message(self, msg):
         """Format a user's message for safe processing."""
 
-        # Make sure the string is Unicode.
-        if isinstance(msg, str):
+        # Make sure the string is Unicode for Python 2.
+        if sys.version_info[0] < 3 and isinstance(msg, str):
             msg = msg.decode('utf8')
 
         # Lowercase it.
@@ -1440,19 +1443,29 @@ def _substitute(self, msg, list):
         else:
             subs = self._person
 
+        # Make placeholders each time we substitute something.
+        ph = []
+        i  = 0
+
         for pattern in self._sorted["lists"][list]:
-            result = "<rot13sub>" + self._rot13(subs[pattern]) + "<bus31tor>"
+            result = subs[pattern]
+
+            # Make a placeholder.
+            ph.append(result)
+            placeholder = "\x00%d\x00" % i
+            i += 1
+
             qm     = re.escape(pattern)
             msg    = re.sub(r'^' + qm + "$", result, msg)
             msg    = re.sub(r'^' + qm + r'(\W+)', result+r'\1', msg)
             msg    = re.sub(r'(\W+)' + qm + r'(\W+)', r'\1'+result+r'\2', msg)
             msg    = re.sub(r'(\W+)' + qm + r'$', r'\1'+result, msg)
 
-        placeholders = re.findall(re_rot13, msg)
+        placeholders = re.findall(r'\x00(\d+)\x00', msg)
         for match in placeholders:
-            rot13   = match
-            decoded = self._rot13(match)
-            msg     = re.sub('<rot13sub>' + re.escape(rot13) + '<bus31tor>', decoded, msg)
+            i = match
+            result = ph[i]
+            msg = re.sub(r'\x00' + i + r'\x00', result, msg)
 
         # Strip & return.
         return msg.strip()
@@ -1917,13 +1930,6 @@ def _word_count(self, trigger, all=False):
 
         return wc
 
-    def _rot13(self, n):
-        """Encode and decode a string into ROT13."""
-        trans = string.maketrans(
-            "ABCDEFGHIJKLMabcdefghijklmNOPQRSTUVWXYZnopqrstuvwxyz",
-            "NOPQRSTUVWXYZnopqrstuvwxyzABCDEFGHIJKLMabcdefghijklm")
-        return string.translate(str(n), trans)
-
     def _strip_nasties(self, s):
         """Formats a string for ASCII regex matching."""
         s = re.sub(re_nasties, '', s)
@@ -1933,30 +1939,30 @@ def _dump(self):
         """For debugging, dump the entire data structure."""
         pp = pprint.PrettyPrinter(indent=4)
 
-        print "=== Variables ==="
-        print "-- Globals --"
+        print("=== Variables ===")
+        print("-- Globals --")
         pp.pprint(self._gvars)
-        print "-- Bot vars --"
+        print("-- Bot vars --")
         pp.pprint(self._bvars)
-        print "-- Substitutions --"
+        print("-- Substitutions --")
         pp.pprint(self._subs)
-        print "-- Person Substitutions --"
+        print("-- Person Substitutions --")
         pp.pprint(self._person)
-        print "-- Arrays --"
+        print("-- Arrays --")
         pp.pprint(self._arrays)
 
-        print "=== Topic Structure ==="
+        print("=== Topic Structure ===")
         pp.pprint(self._topics)
-        print "=== %Previous Structure ==="
+        print("=== %Previous Structure ===")
         pp.pprint(self._thats)
 
-        print "=== Includes ==="
+        print("=== Includes ===")
         pp.pprint(self._includes)
 
-        print "=== Inherits ==="
+        print("=== Inherits ===")
         pp.pprint(self._lineage)
 
-        print "=== Sort Buffer ==="
+        print("=== Sort Buffer ===")
         pp.pprint(self._sorted)
 
 ################################################################################
diff --git a/rivescript/python.py b/rivescript/python.py
@@ -1,5 +1,8 @@
 #!/usr/bin/env python
 
+# Python3 compat
+from __future__ import print_function
+
 """Python object macro support for RiveScript.
 
 This class provides built-in support for your RiveScript documents to include
@@ -34,11 +37,14 @@ def load(self, name, code):
         for line in code:
             source = source + "\t" + line + "\n"
 
+        source += "self._objects[name] = RSOBJ\n"
+
         try:
-            exec source
-            self._objects[name] = RSOBJ
-        except:
-            print "Failed to load code from object " + name
+            exec(source)
+            #self._objects[name] = RSOBJ
+        except Exception as e:
+            print("Failed to load code from object", name)
+            print("The error given was: ", e)
 
     def call(self, rs, name, fields):
         """Invoke a previously loaded object."""
@@ -49,6 +55,7 @@ def call(self, rs, name, fields):
             reply = func(rs, fields)
             if reply == None:
                 reply = ''
-        except:
+        except Exception as e:
+            print("Error executing Python object:", e)
             reply = '[ERR: Error when executing Python object]'
-        return reply
+        return str(reply)
