X-Git-Url: https://code.kerkeslager.com/?p=fur;a=blobdiff_plain;f=tokenization.py;h=819a0de9ea7fd5e07a9c2c0a5c7d93b55d6f0fda;hp=3c4dc6fb561a2fe08fbed90f23d945bc98873adf;hb=c7f381fbcb57ba1b7e33558a28fdb34f31234c07;hpb=d6af7d074bf65e782e42055623a197863b5f8000

diff --git a/tokenization.py b/tokenization.py
index 3c4dc6f..819a0de 100644
--- a/tokenization.py
+++ b/tokenization.py
@@ -5,12 +5,19 @@ import util
 
 Token = collections.namedtuple(
     'Token',
-    [
+    (
         'type',
         'match',
+        'metadata',
+    ),
+)
+
+NodeMetadata = collections.namedtuple(
+    'NodeMetadata',
+    (
         'index',
         'line',
-    ],
+    ),
 )
 
 def _make_token_matcher(definition):
@@ -26,22 +33,34 @@ def _make_token_matcher(definition):
         return (
             True,
             index + len(match.group()),
-            Token(type=name, match=match.group(), index=index, line=line),
+            Token(
+                type=name,
+                match=match.group(),
+                metadata=NodeMetadata(
+                    index=index,
+                    line=line,
+                ),
+            ),
         )
 
     return token_matcher
 
 _TOKEN_MATCHERS = [
-    ('keyword',                         r'(def|end)(?![a-z_])'),
+    ('keyword',                         r'(def|do|else|end|if|lambda)(?![a-z_])'),
+    ('open_bracket',                    r'\['),
+    ('close_bracket',                   r'\]'),
     ('open_parenthese',                 r'\('),
     ('close_parenthese',                r'\)'),
     ('comma',                           r','),
+    ('colon',                           r':'),
+    ('period',                          r'\.'),
     ('integer_literal',                 r'\d+'),
     ('symbol',                          r'[a-z_]+'),
     ('single_quoted_string_literal',    r"'.*?'"),
-    ('comparison_level_operator',         r'(<=|>=|==|!=|<|>)'),
+    ('double_quoted_string_literal',    r'".*?"'),
+    ('comparison_level_operator',       r'(<=|>=|==|!=|<|>)'),
     ('assignment_operator',             r'='),
-    ('addition_level_operator',         r'(\+|-)'),
+    ('addition_level_operator',         r'(\+\+|\+|-)'),
     ('multiplication_level_operator',   r'(\*|//|%)'),
     ('newline',                         r'\n'),
 ]