cpython/Lib/lib2to3/patcomp.py
Benjamin Peterson 6118040b7a Merged revisions 72523,72950-72951,72994,73003,73033,73036-73040,73091-73093,73096,73179-73181,73192,73231,73244,73255-73256,73365 via svnmerge from
svn+ssh://pythondev@svn.python.org/sandbox/trunk/2to3/lib2to3

........
  r72523 | benjamin.peterson | 2009-05-09 14:42:26 -0500 (Sat, 09 May 2009) | 1 line

  remove parenthesis
........
  r72950 | benjamin.peterson | 2009-05-26 18:19:45 -0500 (Tue, 26 May 2009) | 1 line

  remove unused imports
........
  r72951 | benjamin.peterson | 2009-05-26 18:27:00 -0500 (Tue, 26 May 2009) | 1 line

  this is no longer executable
........
  r72994 | benjamin.peterson | 2009-05-28 15:32:54 -0500 (Thu, 28 May 2009) | 1 line

  fix test_all_fixers on Windows #6134
........
  r73003 | benjamin.peterson | 2009-05-28 21:57:28 -0500 (Thu, 28 May 2009) | 4 lines

  make 2to3 test utilities easier to use with other applications (3to2)

  Patch by Joe Amenta
........
  r73033 | benjamin.peterson | 2009-05-29 16:58:32 -0500 (Fri, 29 May 2009) | 1 line

  update grammar for multi with statement
........
  r73036 | benjamin.peterson | 2009-05-29 17:33:20 -0500 (Fri, 29 May 2009) | 1 line

  simplify fix_unicode
........
  r73037 | benjamin.peterson | 2009-05-29 17:53:03 -0500 (Fri, 29 May 2009) | 1 line

  add custom error for pattern syntax errors
........
  r73038 | benjamin.peterson | 2009-05-29 17:55:00 -0500 (Fri, 29 May 2009) | 1 line

  complain if details are attached to a token
........
  r73039 | benjamin.peterson | 2009-05-29 18:00:28 -0500 (Fri, 29 May 2009) | 1 line

  add a test for whitespace
........
  r73040 | benjamin.peterson | 2009-05-29 18:01:17 -0500 (Fri, 29 May 2009) | 1 line

  a fix for emacs highlighting
........
  r73091 | benjamin.peterson | 2009-05-31 20:55:25 -0500 (Sun, 31 May 2009) | 1 line

  deprecate set_prefix() and get_prefix() in favor of a prefix property
........
  r73092 | benjamin.peterson | 2009-05-31 21:00:51 -0500 (Sun, 31 May 2009) | 1 line

  change hideous java naming scheme
........
  r73093 | benjamin.peterson | 2009-05-31 21:01:39 -0500 (Sun, 31 May 2009) | 1 line

  remove dated comment
........
  r73096 | benjamin.peterson | 2009-05-31 21:40:53 -0500 (Sun, 31 May 2009) | 1 line

  group tests
........
  r73179 | benjamin.peterson | 2009-06-03 13:09:53 -0500 (Wed, 03 Jun 2009) | 1 line

  handle the case where there's multiple trailers #6185
........
  r73180 | benjamin.peterson | 2009-06-03 13:18:05 -0500 (Wed, 03 Jun 2009) | 1 line

  scrap __main__ section
........
  r73181 | benjamin.peterson | 2009-06-03 13:24:48 -0500 (Wed, 03 Jun 2009) | 1 line

  remove shebang lines and __main__ sections
........
  r73192 | benjamin.peterson | 2009-06-03 19:16:30 -0500 (Wed, 03 Jun 2009) | 4 lines

  actually test something here

  Thanks to Joe Amenta for noticing.y
........
  r73231 | benjamin.peterson | 2009-06-04 13:38:50 -0500 (Thu, 04 Jun 2009) | 1 line

  remove unused variable
........
  r73244 | benjamin.peterson | 2009-06-05 08:39:25 -0500 (Fri, 05 Jun 2009) | 1 line

  allow fixers to give different options in setUp
........
  r73255 | benjamin.peterson | 2009-06-06 11:23:46 -0500 (Sat, 06 Jun 2009) | 1 line

  fix the except fixer on one line suites #6222
........
  r73256 | benjamin.peterson | 2009-06-06 11:27:40 -0500 (Sat, 06 Jun 2009) | 1 line

  test one-line else and finally clauses
........
  r73365 | benjamin.peterson | 2009-06-11 17:01:32 -0500 (Thu, 11 Jun 2009) | 1 line

  normalize whitespace
........
2009-06-11 22:06:46 +00:00

192 lines
6.6 KiB
Python

# Copyright 2006 Google, Inc. All Rights Reserved.
# Licensed to PSF under a Contributor Agreement.
"""Pattern compiler.
The grammer is taken from PatternGrammar.txt.
The compiler compiles a pattern to a pytree.*Pattern instance.
"""
__author__ = "Guido van Rossum <guido@python.org>"
# Python imports
import os
# Fairly local imports
from .pgen2 import driver, literals, token, tokenize, parse
# Really local imports
from . import pytree
from . import pygram
# The pattern grammar file
_PATTERN_GRAMMAR_FILE = os.path.join(os.path.dirname(__file__),
"PatternGrammar.txt")
class PatternSyntaxError(Exception):
pass
def tokenize_wrapper(input):
"""Tokenizes a string suppressing significant whitespace."""
skip = set((token.NEWLINE, token.INDENT, token.DEDENT))
tokens = tokenize.generate_tokens(driver.generate_lines(input).next)
for quintuple in tokens:
type, value, start, end, line_text = quintuple
if type not in skip:
yield quintuple
class PatternCompiler(object):
def __init__(self, grammar_file=_PATTERN_GRAMMAR_FILE):
"""Initializer.
Takes an optional alternative filename for the pattern grammar.
"""
self.grammar = driver.load_grammar(grammar_file)
self.syms = pygram.Symbols(self.grammar)
self.pygrammar = pygram.python_grammar
self.pysyms = pygram.python_symbols
self.driver = driver.Driver(self.grammar, convert=pattern_convert)
def compile_pattern(self, input, debug=False):
"""Compiles a pattern string to a nested pytree.*Pattern object."""
tokens = tokenize_wrapper(input)
try:
root = self.driver.parse_tokens(tokens, debug=debug)
except parse.ParseError as e:
raise PatternSyntaxError(str(e))
return self.compile_node(root)
def compile_node(self, node):
"""Compiles a node, recursively.
This is one big switch on the node type.
"""
# XXX Optimize certain Wildcard-containing-Wildcard patterns
# that can be merged
if node.type == self.syms.Matcher:
node = node.children[0] # Avoid unneeded recursion
if node.type == self.syms.Alternatives:
# Skip the odd children since they are just '|' tokens
alts = [self.compile_node(ch) for ch in node.children[::2]]
if len(alts) == 1:
return alts[0]
p = pytree.WildcardPattern([[a] for a in alts], min=1, max=1)
return p.optimize()
if node.type == self.syms.Alternative:
units = [self.compile_node(ch) for ch in node.children]
if len(units) == 1:
return units[0]
p = pytree.WildcardPattern([units], min=1, max=1)
return p.optimize()
if node.type == self.syms.NegatedUnit:
pattern = self.compile_basic(node.children[1:])
p = pytree.NegatedPattern(pattern)
return p.optimize()
assert node.type == self.syms.Unit
name = None
nodes = node.children
if len(nodes) >= 3 and nodes[1].type == token.EQUAL:
name = nodes[0].value
nodes = nodes[2:]
repeat = None
if len(nodes) >= 2 and nodes[-1].type == self.syms.Repeater:
repeat = nodes[-1]
nodes = nodes[:-1]
# Now we've reduced it to: STRING | NAME [Details] | (...) | [...]
pattern = self.compile_basic(nodes, repeat)
if repeat is not None:
assert repeat.type == self.syms.Repeater
children = repeat.children
child = children[0]
if child.type == token.STAR:
min = 0
max = pytree.HUGE
elif child.type == token.PLUS:
min = 1
max = pytree.HUGE
elif child.type == token.LBRACE:
assert children[-1].type == token.RBRACE
assert len(children) in (3, 5)
min = max = self.get_int(children[1])
if len(children) == 5:
max = self.get_int(children[3])
else:
assert False
if min != 1 or max != 1:
pattern = pattern.optimize()
pattern = pytree.WildcardPattern([[pattern]], min=min, max=max)
if name is not None:
pattern.name = name
return pattern.optimize()
def compile_basic(self, nodes, repeat=None):
# Compile STRING | NAME [Details] | (...) | [...]
assert len(nodes) >= 1
node = nodes[0]
if node.type == token.STRING:
value = unicode(literals.evalString(node.value))
return pytree.LeafPattern(content=value)
elif node.type == token.NAME:
value = node.value
if value.isupper():
if value not in TOKEN_MAP:
raise PatternSyntaxError("Invalid token: %r" % value)
if nodes[1:]:
raise PatternSyntaxError("Can't have details for token")
return pytree.LeafPattern(TOKEN_MAP[value])
else:
if value == "any":
type = None
elif not value.startswith("_"):
type = getattr(self.pysyms, value, None)
if type is None:
raise PatternSyntaxError("Invalid symbol: %r" % value)
if nodes[1:]: # Details present
content = [self.compile_node(nodes[1].children[1])]
else:
content = None
return pytree.NodePattern(type, content)
elif node.value == "(":
return self.compile_node(nodes[1])
elif node.value == "[":
assert repeat is None
subpattern = self.compile_node(nodes[1])
return pytree.WildcardPattern([[subpattern]], min=0, max=1)
assert False, node
def get_int(self, node):
assert node.type == token.NUMBER
return int(node.value)
# Map named tokens to the type value for a LeafPattern
TOKEN_MAP = {"NAME": token.NAME,
"STRING": token.STRING,
"NUMBER": token.NUMBER,
"TOKEN": None}
def pattern_convert(grammar, raw_node_info):
"""Converts raw node information to a Node or Leaf instance."""
type, value, context, children = raw_node_info
if children or type in grammar.number2symbol:
return pytree.Node(type, children, context=context)
else:
return pytree.Leaf(type, value, context=context)
def compile_pattern(pattern):
return PatternCompiler().compile_pattern(pattern)