Tag interpretation implementation

ptulsconv/transformations.py

@@ -1,4 +1,5 @@
 from . import broadcast_timecode
+from parsimonious import Grammar, NodeVisitor
 import math
 
 class Transformation:
@@ -55,3 +56,71 @@ class TimecodeInterpreter(Transformation):
         return dict(frame_count=frame_count, logical_fps=lfps, drop_frame=drop_frame)
 
 
+class TagInterpreter:
+    tag_grammar = Grammar(
+        r"""
+        document  = modifier? line? word_sep? tag_list?
+        line      = word (word_sep word)*
+        tag_list  = tag*
+        tag       = key_tag / short_tag / full_text_tag / tag_junk
+        key_tag        = "[" key "]" word_sep?
+        short_tag      = "$" key "=" word 
+        full_text_tag  = "{" key "=" value "}" word_sep?
+        key            = ~"[A-Za-z][A-Za-z0-9_]*"
+        value          = ~"[^}]+"
+        tag_junk       = (word word_sep)*
+        word           = ~"[^ \[\{\$][^ ]*"
+        word_sep       = ~" +"
+        modifier       = ("@" / "&") word_sep?
+        """
+    )
+
+    class TagListVisitor(NodeVisitor):
+        def visit_document(self, _, visited_children):
+            modifier_opt, line_opt, _, tag_list_opt = visited_children
+
+            return dict(line= next(iter(line_opt), None),
+                        tags= next(iter(tag_list_opt), None),
+                        mode= next(iter(modifier_opt), 'Normal')
+                        )
+
+        def visit_line(self, node, _):
+            return str.strip(node.text, " ")
+
+        def visit_modifier(self, node, _):
+            if node.text.startswith('@'):
+                return 'Timespan'
+            elif node.text.startswith('&'):
+                return 'Append'
+            else:
+                return 'Normal'
+
+        def visit_tag_list(self, _, visited_children):
+            retdict = dict()
+            for child in visited_children:
+                k, v = child[0]
+                retdict[k] = v
+            return retdict
+
+        def visit_key_tag(self, _, children):
+            return children[1].text, children[1].text
+
+        def visit_short_tag(self, _, children):
+            return children[1].text, children[3].text
+
+        def visit_full_text_tag(self, _, children):
+            return children[1].text, children[3].text
+
+        def generic_visit(self, node, visited_children):
+            return visited_children or node
+
+    def __init__(self):
+        pass
+
+    def parse_tags(self, source):
+        parse_tree = self.tag_grammar.parse(source)
+        v = TagInterpreter.TagListVisitor()
+        return v.visit(parse_tree)
+
+
+

tests/test_tag_interpreter.py

@@ -0,0 +1,41 @@
+import unittest
+
+from ptulsconv.transformations import TagInterpreter
+
+class MyTestCase(unittest.TestCase):
+    def test_line(self):
+        ti = TagInterpreter()
+        s1 = ti.parse_tags("this is a test")
+        self.assertEqual(s1['line'], "this is a test")
+        self.assertEqual(s1['mode'], 'Normal')
+        self.assertEqual(len(s1['tags']), 0)
+
+        s2 = ti.parse_tags("this! IS! Me! ** Typing! 123 <> |||")
+        self.assertEqual(s2['line'], "this! IS! Me! ** Typing! 123 <> |||")
+        self.assertEqual(s2['mode'], 'Normal')
+        self.assertEqual(len(s2['tags']), 0)
+
+    def test_tags(self):
+        ti = TagInterpreter()
+        s1 = ti.parse_tags("{a=100}")
+        self.assertIn('tags', s1)
+        self.assertEqual(s1['tags']['a'], "100")
+
+        s2 = ti.parse_tags("{b=This is a test} [option] $X=9")
+        self.assertEqual(s2['tags']['b'], 'This is a test')
+        self.assertEqual(s2['tags']['option'], 'option')
+        self.assertEqual(s2['tags']['X'], "9")
+
+    def test_modes(self):
+        ti = TagInterpreter()
+        s1 = ti.parse_tags("@ Monday Tuesday {a=1}")
+        self.assertEqual(s1['mode'], 'Timespan')
+
+        s2 = ti.parse_tags("Monday Tuesday {a=1}")
+        self.assertEqual(s2['mode'], 'Normal')
+
+        s3 = ti.parse_tags("&Monday Tuesday {a=1}")
+        self.assertEqual(s3['mode'], 'Append')
+
+if __name__ == '__main__':
+    unittest.main()