Make cache work with non-default line lenghts (#163)

[etc/vim.git] / blib2to3 / pgen2 / driver.py

diff --git a/blib2to3/pgen2/driver.py b/blib2to3/pgen2/driver.py
index cbc58e759d5374e3980f1402ab708d0888cda59c..2265c26443d2f45280112653f5f25eda59360b07 100644 (file)
--- a/blib2to3/pgen2/driver.py
+++ b/blib2to3/pgen2/driver.py
@@ -43,6 +43,7 @@ class Driver(object):
         p.setup()
         lineno = 1
         column = 0
+        indent_columns = []
         type = value = start = end = line_text = None
         prefix = ""
         for quintuple in tokens:
@@ -69,11 +70,26 @@ class Driver(object):
             if debug:
                 self.logger.debug("%s %r (prefix=%r)",
                                   token.tok_name[type], value, prefix)
+            if type in {token.INDENT, token.DEDENT}:
+                _prefix = prefix
+                prefix = ""
+            if type == token.DEDENT:
+                _indent_col = indent_columns.pop()
+                prefix, _prefix = self._partially_consume_prefix(_prefix, _indent_col)
             if p.addtoken(type, value, (prefix, start)):
                 if debug:
                     self.logger.debug("Stop.")
                 break
             prefix = ""
+            if type == token.INDENT:
+                indent_columns.append(len(value))
+                if _prefix.startswith(value):
+                    # Don't double-indent.  Since we're delaying the prefix that
+                    # would normally belong to INDENT, we need to put the value
+                    # at the end versus at the beginning.
+                    _prefix = _prefix[len(value):] + value
+            if type in {token.INDENT, token.DEDENT}:
+                prefix = _prefix
             lineno, column = end
             if value.endswith("\n"):
                 lineno += 1
@@ -103,6 +119,35 @@ class Driver(object):
         tokens = tokenize.generate_tokens(io.StringIO(text).readline)
         return self.parse_tokens(tokens, debug)
 
+    def _partially_consume_prefix(self, prefix, column):
+        lines = []
+        current_line = ""
+        current_column = 0
+        wait_for_nl = False
+        for char in prefix:
+            current_line += char
+            if wait_for_nl:
+                if char == '\n':
+                    if current_line.strip() and current_column < column:
+                        res = ''.join(lines)
+                        return res, prefix[len(res):]
+
+                    lines.append(current_line)
+                    current_line = ""
+                    current_column = 0
+                    wait_for_nl = False
+            elif char == ' ':
+                current_column += 1
+            elif char == '\t':
+                current_column += 4
+            elif char == '\n':
+                # enexpected empty line
+                current_column = 0
+            else:
+                # indent is finished
+                wait_for_nl = True
+        return ''.join(lines), current_line
+
 
 def _generate_pickle_name(gt):
     head, tail = os.path.splitext(gt)