Merge pull request #2930 from xonsh/andash

More and/or subprocess token fixes.

news/andash.rst

@@ -0,0 +1,28 @@
+**Added:**
+
+* <news item>
+
+**Changed:**
+
+* <news item>
+
+**Deprecated:**
+
+* <news item>
+
+**Removed:**
+
+* <news item>
+
+**Fixed:**
+
+* Fixed issue with ``and`` & ``or`` being incoreectly tokenized in implicit
+  subprocesses. Auto-wrapping of certain subprocesses will now correctly work.
+  For example::
+
+      $ echo x-and-y
+      x-and-y
+
+**Security:**
+
+* <news item>

tests/sample.xsh

@@ -2,4 +2,4 @@
 aliases['echo'] = lambda args, stdin=None: print(' '.join(args))
 
 $WAKKA = "jawaka"
-x = $(echo "hello mom" $WAKKA)
+x = $(echo "hello mom" $WAKKA)

tests/test_lexer.py

@@ -165,7 +165,10 @@ def test_atdollar_expression():
 
 
 def test_and():
-    assert check_token("and", ["AND", "and", 0])
+    # no preceding whitespace or other tokens, so this
+    # resolves to NAME, since it doesn't make sense for
+    # Python code to start with "and"
+    assert check_token("and", ["NAME", "and", 0])
 
 
 def test_ampersand():

tests/test_tools.py

@@ -374,6 +374,29 @@ def test_subproc_toks_pyeval_nested():
     assert exp == obs
 
 
+@pytest.mark.parametrize('phrase', [
+    'xandy',
+    'xory',
+    'xand',
+    'andy',
+    'xor',
+    'ory',
+    'x-and',
+    'x-or',
+    'and-y',
+    'or-y',
+    'x-and-y',
+    'x-or-y',
+    'in/and/path',
+    'in/or/path',
+])
+def test_subproc_toks_and_or(phrase):
+    s = "echo " + phrase
+    exp = "![{0}]".format(s)
+    obs = subproc_toks(s, lexer=LEXER, returnline=True)
+    assert exp == obs
+
+
 def test_subproc_toks_pyeval_nested_parens():
     s = "echo @(min(1, 42))"
     inp = "({0})".format(s)

xonsh/lexer.py

@@ -4,6 +4,7 @@
 Written using a hybrid of ``tokenize`` and PLY.
 """
 import io
+import re
 
 # 'keyword' interferes with ast.keyword
 import keyword as kwmod
@@ -116,21 +117,33 @@ def token_map():
     return tm
 
 
+NEED_WHITESPACE = frozenset(["and", "or"])
+
+
+@lazyobject
+def RE_NEED_WHITESPACE():
+    pattern = r"\s?(" + "|".join(NEED_WHITESPACE) + r")(\s|[\\]$)"
+    return re.compile(pattern)
+
+
 def handle_name(state, token):
     """Function for handling name tokens"""
     typ = "NAME"
+    state["last"] = token
+    needs_whitespace = token.string in NEED_WHITESPACE
+    has_whitespace = needs_whitespace and RE_NEED_WHITESPACE.match(
+        token.line[max(0, token.start[1] - 1) :]
+    )
     if state["pymode"][-1][0]:
-        if token.string in kwmod.kwlist:
+        if needs_whitespace and not has_whitespace:
+            pass
+        elif token.string in kwmod.kwlist:
             typ = token.string.upper()
-        state["last"] = token
         yield _new_token(typ, token.string, token.start)
     else:
-        prev = state["last"]
-        state["last"] = token
-        has_whitespace = prev.end != token.start
-        if token.string == "and" and has_whitespace:
+        if has_whitespace and token.string == "and":
             yield _new_token("AND", token.string, token.start)
-        elif token.string == "or" and has_whitespace:
+        elif has_whitespace and token.string == "or":
             yield _new_token("OR", token.string, token.start)
         else:
             yield _new_token("NAME", token.string, token.start)