Fix interactive assembly on Python 3

Closes #81
Closes #78
Closes #65
Closes #64
Closes #63

.github/workflows/main.yml

@@ -21,11 +21,14 @@ jobs:
       run: python -V
 
     - name: Install dependencies
-      run: $PIP install setuptools
+      run: $PIP install setuptools pexpect
 
     - name: Run the tests
       run: python setup.py test -q
 
+    - name: Run the end-to-end tests
+      run: END_TO_END=1 python setup.py test -q
+
   build_py34:
     runs-on: ubuntu-20.04
     container: python:3.4
@@ -44,11 +47,14 @@ jobs:
       run: python -V
 
     - name: Install dependencies
-      run: $PIP install setuptools
+      run: $PIP install setuptools pexpect
 
     - name: Run the tests
       run: python setup.py test -q
 
+    - name: Run the end-to-end tests
+      run: END_TO_END=1 python setup.py test -q
+
   build_py35:
     runs-on: ubuntu-20.04
     container: python:3.5
@@ -62,11 +68,14 @@ jobs:
       run: python -V
 
     - name: Install dependencies
-      run: $PIP install setuptools
+      run: $PIP install setuptools pexpect
 
     - name: Run the tests
       run: python setup.py test -q
 
+    - name: Run the end-to-end tests
+      run: END_TO_END=1 python setup.py test -q
+
   build_py3x:
     strategy:
       fail-fast: false
@@ -88,7 +97,10 @@ jobs:
       run: python -V
 
     - name: Install dependencies
-      run: $PIP install setuptools
+      run: $PIP install setuptools pexpect
 
     - name: Run the tests
       run: python setup.py test -q
+
+    - name: Run the end-to-end tests
+      run: END_TO_END=1 python setup.py test -q

CHANGES.rst

@@ -5,6 +5,8 @@
   dropped when pasting in larger amounts of text.  This makes it possible
   to paste programs into EhBASIC and Taliforth.  Patch by SamCoVT.
 
+- Fixed interactive assembly on Python 3.
+
 - Fixed regular expression warnings on Python 3.12.
 
 - The ``fill`` command in the monitor now shows an error message if an

py65/assembler.py

@@ -92,7 +92,7 @@ class Assembler:
             and parsing the address part using AddressParser.  The result of
             the normalization is a tuple of two strings (opcode, operand).
         """
-        statement = ' '.join(str.split(statement))
+        statement = ' '.join(statement.split())
 
         # normalize target in operand
         match = self.Statement.match(statement)

py65/compat.py

@@ -0,0 +1,21 @@
+import sys
+
+PY2 = sys.version_info[0] == 2
+
+if PY2:
+    unicode = unicode
+
+    def as_string(s, encoding='utf-8'):
+        if isinstance(s, unicode):
+            return s
+        else:
+            return s.decode(encoding)
+
+else:
+    unicode = str
+
+    def as_string(s, encoding='utf-8'):
+        if isinstance(s, str):
+            return s
+        else:
+            return s.decode(encoding)

py65/tests/end_to_end.py

@@ -0,0 +1,88 @@
+import os
+import signal
+import sys
+import tempfile
+import unittest
+
+from py65.compat import unicode
+
+# end-to-test tests are slow so only run them when asked
+if 'END_TO_END' in os.environ:
+    if sys.platform == "win32":
+        raise NotImplementedError()
+    else:
+        import pexpect
+        BaseTestCase = unittest.TestCase
+else:
+    BaseTestCase = object
+
+
+class EndToEndTests(BaseTestCase):
+
+    def _spawn(self):
+        mon = pexpect.spawn(
+            sys.executable, 
+            ['-u', '-m', 'py65.monitor'], 
+            encoding='utf-8'
+            )
+        mon.expect_exact(unicode("Py65 Monitor"))
+        self.addCleanup(mon.kill, signal.SIGINT)
+        return mon
+
+    def test_putc(self):
+        mon = self._spawn()
+
+        mon.sendline(unicode("add_label f001 putc"))
+
+        mon.sendline(unicode("a c000 lda #'H"))
+        mon.sendline(unicode("a c002 sta putc"))
+        mon.sendline(unicode("a c005 lda #'I"))
+        mon.sendline(unicode("a c007 sta putc"))
+        mon.sendline(unicode("a c00a brk"))
+
+        mon.sendline(unicode("g c000"))
+        mon.expect_exact(unicode("HI"))
+        mon.sendline(unicode("q"))
+
+    def test_getc(self):
+        mon = self._spawn()
+
+        mon.sendline(unicode("add_label f004 getc"))
+
+        mon.sendline(unicode("a c000 ldx #0"))
+        mon.sendline(unicode("a c002 lda getc"))
+        mon.sendline(unicode("a c005 beq c002"))
+        mon.sendline(unicode("a c007 cmp #'!"))
+        mon.sendline(unicode("a c009 bne c00c"))
+        mon.sendline(unicode("a c00b brk"))
+        mon.sendline(unicode("a c00c sta 1000,x"))
+        mon.sendline(unicode("a c00f inx"))
+        mon.sendline(unicode("a c010 jmp c002"))
+
+        mon.sendline(unicode("g c000"))
+        mon.send(unicode("HELLO!"))
+        mon.expect_exact(unicode("6502:"))
+        mon.sendline(unicode("m 1000:1004"))
+        mon.expect_exact(unicode("48  45  4c  4c  4f"))
+
+    def test_assemble_interactive(self):
+        mon = self._spawn()
+
+        mon.sendline(unicode("assemble 0"))
+        mon.expect_exact(unicode("$0000"))
+
+        mon.sendline(unicode("lda $1234"))
+        mon.expect_exact(unicode("ad 34 12"))
+
+        mon.expect_exact(unicode("$0003"))
+        mon.sendline(unicode("sta $4567"))
+        mon.expect_exact(unicode("8d 67 45"))
+
+        mon.sendline(unicode("invalid"))
+        mon.expect_exact(unicode("?Syntax"))
+
+        mon.sendline()
+        mon.sendline(unicode("quit"))
+
+if __name__ == '__main__':
+    unittest.main()

py65/utils/console.py

@@ -1,5 +1,7 @@
 import sys
 
+from py65.compat import as_string
+
 if sys.platform[:3] == "win":
     import msvcrt
 
@@ -24,10 +26,7 @@ if sys.platform[:3] == "win":
         is available.  Does not echo the character.  The stdin argument is
         for function signature compatibility and is ignored.
         """
-        c = msvcrt.getch()
+        return as_string(msvcrt.getch())
-        if isinstance(c, bytes): # Python 3
-            c = c.decode('latin-1')
-        return c
 
     def getch_noblock(stdin):
         """ Read one character from the Windows console without blocking.
@@ -36,8 +35,8 @@ if sys.platform[:3] == "win":
         available, an empty string is returned.
         """
         if msvcrt.kbhit():
-            return getch(stdin)
+            return as_string(getch(stdin))
-        return ''
+        return u''
 
 else:
     import termios
@@ -157,7 +156,7 @@ else:
                 # use select to make sure there is at least one char to read.
                 rd,wr,er = select([stdin], [], [], 0.01)
                 if rd != []:
-                    char = stdin.read(1)
+                    char = as_string(stdin.read(1))
             except KeyboardInterrupt:
                 # Pass along a CTRL-C interrupt.
                 raise
@@ -180,7 +179,7 @@ else:
             # use select to make sure there is at least one char to read.
             rd,wr,er = select([stdin], [], [], 0.01)
             if rd != []:
-                char = stdin.read(1)
+                char = as_string(stdin.read(1))
         except KeyboardInterrupt:
             # Pass along a CTRL-C interrupt.
             raise
@@ -200,6 +199,7 @@ def line_input(prompt='', stdin=sys.stdin, stdout=sys.stdout):
     is useful in modes like the interactive assembler.
     """
     stdout.write(prompt)
+    stdout.flush()
     line = ''
     while True:
         char = getch(stdin)
@@ -211,6 +211,7 @@ def line_input(prompt='', stdin=sys.stdin, stdout=sys.stdout):
                 line = line[:-1]
                 stdout.write("\r%s\r%s%s" %
                              (' ' * (len(prompt + line) + 5), prompt, line))
+                stdout.flush()
         elif code == 0x1b:  # escape
             pass
         else: