Fix parser bug with escaped characters

This is the proper fix for #28
fkie · Apr 16, 2016 · f767a39 · f767a39
1 parent f5a63e0
commit f767a39
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 5 deletions.
diff --git a/src/catkin_lint/cmake.py b/src/catkin_lint/cmake.py
@@ -68,9 +68,9 @@ def _resolve_vars(s, var, env_var):
     ( 'SKIP', r'[ \t]+' ),
     ( 'LPAREN', r'\(' ),
     ( 'RPAREN', r'\)' ),
-    ( 'STRING', r'"[^\\"]*(?:\\.[^\\"]*)*"' ),
+    ( 'STRING', r'"(?:\\.|[^\\"])*"' ),
     ( 'SEMICOLON', r';'),
-    ( 'WORD', r'[^\(\)"# \t\r\n;]+' ),
+    ( 'WORD', r'(?:\\.|[^\\\(\)"# \t\r\n;])+' ),
     ( 'PRAGMA', r'#catkin_lint:.*?$' ),
     ( 'COMMENT', r'#.*?$' ),
 ]
@@ -99,17 +99,30 @@ def _lexer(s):
     if pos != len(s):
         raise SyntaxError("Unexpected character %r on line %d" % (s[pos], line))
 
+_arg_spec = [
+    ( 'SKIP', r';' ),
+    ( 'ARG', r'(?:\\.|[^;])+' ),
+]
+_next_arg = re.compile('|'.join('(?P<%s>%s)' % pair for pair in _arg_spec)).match
 
 def _resolve_args(arg_tokens, var, env_var):
     args = []
     for typ, val in arg_tokens:
         if typ == "STRING":
             val = _resolve_vars(val, var, env_var)
+            # Treat quoted strings as a single word
             args.append(_unescape(val))
         elif typ == "WORD":
             val = _resolve_vars(val, var, env_var)
-            if val:
-                args += re.split(r";|[ \t]+", _unescape(val))
+            # Split unquoted text into list items
+            mo = _next_arg(val)
+            while mo is not None:
+                typ = mo.lastgroup
+                if typ != "SKIP":
+                    arg = mo.group(typ)
+                    args.append(_unescape(arg))
+                pos = mo.end()
+                mo = _next_arg(val, pos)
         elif typ != "SEMICOLON":
             args.append(val)
     return args

diff --git a/test/test_cmake_parser.py b/test/test_cmake_parser.py
@@ -54,6 +54,10 @@ def test_string(self):
             self.parse_all('cmd("string that spans\nmultiple lines")'),
             [ ("cmd", [ 'string that spans\nmultiple lines' ]) ]
         )
+        self.assertEqual(
+            self.parse_all('cmd("\\\\"\\")'),
+            [ ("cmd", [ '\\', '"' ]) ]
+        )
 
     def test_macro(self):
         self.assertEqual(
@@ -224,6 +228,14 @@ def test_arguments(self):
             self.parse_all('cmd("\\"")'),
             [ ("cmd", [ '"' ]) ]
         )
+        self.assertEqual(
+            self.parse_all('cmd(\\")'),
+            [ ("cmd", [ '"' ]) ]
+        )
+        self.assertEqual(
+            self.parse_all('cmd(a\\ b)'),
+            [ ("cmd", [ 'a b' ]) ]
+        )
         self.assertEqual(
             self.parse_all("cmd(ENV{PATH})"),
             [ ("cmd", [ "ENV{PATH}" ]) ]
@@ -249,7 +261,7 @@ def test_substitution(self):
         )
         self.assertEqual(
             self.parse_all("cmd(${args})", var={ "args" : "one two three"}),
-            [ ("cmd", [ "one", "two", "three" ]) ]
+            [ ("cmd", [ "one two three" ]) ]
         )
         self.assertEqual(
             self.parse_all('cmd("${args}")', var={ "args" : "one;two;three"}),