Symbol match on alphanumerics for parameter references instead of trying to

specify negative set.

Author: Peter Amstutz

cwltool/expression.py

@@ -86,13 +86,13 @@ class DR(object):
 
     raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
 
-seg_symbol = r"""[^[\].(){} ]+"""
+seg_symbol = r"""\w+"""
 seg_single = r"""\['([^']|\\')+'\]"""
 seg_double = r"""\["([^"]|\\")+"\]"""
 seg_index  = r"""\[[0-9]+\]"""
 segments = r"(\.%s|%s|%s|%s)" % (seg_symbol, seg_single, seg_double, seg_index)
-segment_re = re.compile(segments)
-param_re = re.compile(r"\$\((%s)%s*\)" % (seg_symbol, segments))
+segment_re = re.compile(segments, flags=re.UNICODE)
+param_re = re.compile(r"\$\((%s)%s*\)" % (seg_symbol, segments), flags=re.UNICODE)
 
 def next_seg(remain, obj):
     if remain:
@@ -115,11 +115,13 @@ def param_interpolate(ex, obj, strip=True):
         if strip and len(ex.strip()) == len(m.group(0)):
             return leaf
         else:
-            leaf = json.dumps(leaf)
+            leaf = json.dumps(leaf, sort_keys=True)
             if leaf[0] == '"':
                 leaf = leaf[1:-1]
             return ex[0:m.start(0)] + leaf + param_interpolate(ex[m.end(0):], obj, False)
     else:
+        if "$(" in ex:
+            _logger.warn("Possible bug: found '$(' in '%s' but did not match valid parameter reference.")
         return ex
 
 

cwltool/sandboxjs.py

@@ -103,6 +103,7 @@ def scanner(scan):
     else:
         return None
 
+
 def interpolate(scan, jslib):
     scan = scan.strip()
     parts = []
@@ -114,7 +115,10 @@ def interpolate(scan, jslib):
             e = execjs(scan[w[0]+1:w[1]], jslib)
             if w[0] == 0 and w[1] == len(scan):
                 return e
-            parts.append(json.dumps(e))
+            leaf = json.dumps(e, sort_keys=True)
+            if leaf[0] == '"':
+                leaf = leaf[1:-1]
+            parts.append(leaf)
         elif scan[w[0]] == '\\':
             e = scan[w[1]-1]
             parts.append(e)

tests/test_examples.py

@@ -13,6 +13,7 @@ def test_params(self):
         self.assertTrue(expr.param_re.match("$(foo['bar']['baz'])"))
         self.assertTrue(expr.param_re.match("$(foo['b\\'ar']['baz'])"))
         self.assertTrue(expr.param_re.match("$(foo['b ar']['baz'])"))
+        self.assertTrue(expr.param_re.match("$(foo_bar)"))
 
         self.assertFalse(expr.param_re.match("$(foo.[\"bar\"])"))
         self.assertFalse(expr.param_re.match("$(.foo[\"bar\"])"))
@@ -26,6 +27,8 @@ def test_params(self):
         self.assertFalse(expr.param_re.match("$foo.bar)"))
         self.assertFalse(expr.param_re.match("$foo.b ar)"))
         self.assertFalse(expr.param_re.match("$foo.b\'ar)"))
+        self.assertFalse(expr.param_re.match("$(foo+bar"))
+        self.assertFalse(expr.param_re.match("$(foo bar"))
 
         inputs = {
             "foo": {