Fix handling of boolean flags in elliptical path commands (#183)

* properly parse paths without space after boolean flags (fixes #161)
* omit space after boolean flag to shave off a few bytes when not using renderer workarounds

scour/scour.py

@@ -2580,7 +2580,7 @@ def cleanPolyline(elem, options):
 
 def controlPoints(cmd, data):
     """
-       Checks if there are control points in the path
+       Checks if there are control points in the path data
 
        Returns False if there aren't any
        Returns a list of bools set to True for coordinates in the path data which are control points
@@ -2596,13 +2596,28 @@ def controlPoints(cmd, data):
     return False
 
 
+def flags(cmd, data):
+    """
+       Checks if there are flags in the path data
+
+       Returns the indices of all values in the path data which are flags
+    """
+    if cmd.lower() == 'a':  # a: (rx ry x-axis-rotation large-arc-flag sweep-flag x y)+
+        indices = range(len(data))
+        return [index for index in indices if (index % 7) in [3, 4]]
+
+    return []
+
+
 def serializePath(pathObj, options):
     """
        Reserializes the path data with some cleanups.
     """
     # elliptical arc commands must have comma/wsp separating the coordinates
     # this fixes an issue outlined in Fix https://bugs.launchpad.net/scour/+bug/412754
-    return ''.join([cmd + scourCoordinates(data, options, reduce_precision=controlPoints(cmd, data))
+    return ''.join([cmd + scourCoordinates(data, options,
+                                           reduce_precision=controlPoints(cmd, data),
+                                           flags=flags(cmd, data))
                     for cmd, data in pathObj])
 
 
@@ -2614,7 +2629,7 @@ def serializeTransform(transformObj):
                      for command, numbers in transformObj])
 
 
-def scourCoordinates(data, options, force_whitespace=False, reduce_precision=False):
+def scourCoordinates(data, options, force_whitespace=False, reduce_precision=False, flags=[]):
     """
        Serializes coordinate data with some cleanups:
           - removes all trailing zeros after the decimal
@@ -2636,10 +2651,12 @@ def scourCoordinates(data, options, force_whitespace=False, reduce_precision=Fal
             #   - this number starts with a dot but the previous number had *no* dot or exponent
             #     i.e. '1.3 0.5' -> '1.3.5' or '1e3 0.5' -> '1e3.5' is fine but '123 0.5' -> '123.5' is obviously not
             #   - 'force_whitespace' is explicitly set to 'True'
-            if c > 0 and (force_whitespace
+            # we never need a space after flags (occuring in elliptical arcs), but librsvg struggles without it
+            if (c > 0
+                    and (force_whitespace
                          or scouredCoord[0].isdigit()
-                          or (scouredCoord[0] == '.' and not ('.' in previousCoord or 'e' in previousCoord))
-                          ):
+                         or (scouredCoord[0] == '.' and not ('.' in previousCoord or 'e' in previousCoord)))
+                    and ((c-1 not in flags) or options.renderer_workaround)):
                 newData.append(' ')
 
             # add the scoured coordinate to the path string

scour/svg_regex.py

@@ -247,15 +247,23 @@ class SVGPathParser(object):
             axis_rotation = Decimal(token[1]) * 1
 
             token = next_val_fn()
-            if token[1] not in ('0', '1'):
+            if token[1][0] not in ('0', '1'):
                 raise SyntaxError("expecting a boolean flag; got %r" % (token,))
-            large_arc_flag = Decimal(token[1]) * 1
+            large_arc_flag = Decimal(token[1][0]) * 1
 
+            if len(token[1]) > 1:
+                token = list(token)
+                token[1] = token[1][1:]
+            else:
                 token = next_val_fn()
-            if token[1] not in ('0', '1'):
+            if token[1][0] not in ('0', '1'):
                 raise SyntaxError("expecting a boolean flag; got %r" % (token,))
-            sweep_flag = Decimal(token[1]) * 1
+            sweep_flag = Decimal(token[1][0]) * 1
 
+            if len(token[1]) > 1:
+                token = list(token)
+                token[1] = token[1][1:]
+            else:
                 token = next_val_fn()
             if token[0] not in self.number_tokens:
                 raise SyntaxError("expecting a number; got %r" % (token,))

testscour.py

@@ -1101,6 +1101,24 @@ class ChangeQuadToShorthandInPath(unittest.TestCase):
                          'Did not change quadratic curves into shorthand curve segments in path')
 
 
+class BooleanFlagsInEllipticalPath(unittest.TestCase):
+
+    def test_omit_spaces(self):
+        doc = scourXmlFile('unittests/path-elliptical-flags.svg', parse_args(['--no-renderer-workaround']))
+        paths = doc.getElementsByTagNameNS(SVGNS, 'path')
+        for path in paths:
+            self.assertEqual(path.getAttribute('d'), 'm0 0a100 50 0 00100 50',
+                             'Did not ommit spaces after boolean flags in elliptical arg path command')
+
+    def test_output_spaces_with_renderer_workaround(self):
+        doc = scourXmlFile('unittests/path-elliptical-flags.svg', parse_args(['--renderer-workaround']))
+        paths = doc.getElementsByTagNameNS(SVGNS, 'path')
+        for path in paths:
+            self.assertEqual(path.getAttribute('d'), 'm0 0a100 50 0 0 0 100 50',
+                             'Did not output spaces after boolean flags in elliptical arg path command '
+                             'with renderer workaround')
+
+
 class DoNotOptimzePathIfLarger(unittest.TestCase):
 
     def runTest(self):

unittests/path-elliptical-flags.svg

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<svg viewBox="-100 -50 300 150" xmlns="http://www.w3.org/2000/svg">
+ <path d="m0 0a100 50 0 0 0 100 50" fill="none" stroke="#000"/>
+ <path d="m0 0a100 50 0 0  0100 50" fill="none" stroke="green"/>
+ <path d="m0 0a100 50 0 00  100 50" fill="none" stroke="blue"/>
+ <path d="m0 0a100 50 0   00100 50" fill="none" stroke="red"/>
+</svg>