Index: core/scripts/make_token_matcher.py |
diff --git a/core/scripts/make_token_matcher.py b/core/scripts/make_token_matcher.py |
deleted file mode 100755 |
index 10fae628c1cf4dc12686330747e2bf0e8aa9faba..0000000000000000000000000000000000000000 |
--- a/core/scripts/make_token_matcher.py |
+++ /dev/null |
@@ -1,231 +0,0 @@ |
-#!/usr/bin/env python |
-# Copyright (C) 2013 Google Inc. All rights reserved. |
-# |
-# Redistribution and use in source and binary forms, with or without |
-# modification, are permitted provided that the following conditions are |
-# met: |
-# |
-# * Redistributions of source code must retain the above copyright |
-# notice, this list of conditions and the following disclaimer. |
-# * Redistributions in binary form must reproduce the above |
-# copyright notice, this list of conditions and the following disclaimer |
-# in the documentation and/or other materials provided with the |
-# distribution. |
-# * Neither the name of Google Inc. nor the names of its |
-# contributors may be used to endorse or promote products derived from |
-# this software without specific prior written permission. |
-# |
-# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
-# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
-# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
-# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
-# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
-# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
-# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
-# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
-# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
-# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
-# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
- |
-import io |
-import itertools |
-import re |
-import sys |
- |
- |
-class BadInput(Exception): |
- """Unsupported input has been found.""" |
- |
- |
-class SwitchCase(object): |
- """Represents a CASE block.""" |
- def __init__(self, identifier, block): |
- self.identifier = identifier |
- self.block = block |
- |
- |
-class Optimizer(object): |
- """Generates optimized identifier matching code.""" |
- def __init__(self, output_file, array_variable, length_variable): |
- self.output_file = output_file |
- self.array_variable = array_variable |
- self.length_variable = length_variable |
- |
- def inspect(self, cases): |
- lengths = list(set([len(c.identifier) for c in cases])) |
- lengths.sort() |
- |
- def response(length): |
- self.inspect_array([c for c in cases if len(c.identifier) == length], range(length)) |
- self.write_selection(self.length_variable, lengths, str, response) |
- |
- def score(self, alternatives): |
- return -sum([len(list(count)) ** 2 for _, count in itertools.groupby(sorted(alternatives))]) |
- |
- def choose_selection_pos(self, cases, pending): |
- candidates = [pos for pos in pending if all(alternative.isalpha() for alternative in [c.identifier[pos] for c in cases])] |
- if not candidates: |
- raise BadInput('Case-insensitive switching on non-alphabetic characters not yet implemented') |
- return sorted(candidates, key=lambda pos: self.score([c.identifier[pos] for c in cases]))[0] |
- |
- def inspect_array(self, cases, pending): |
- assert len(cases) >= 1 |
- if pending: |
- common = [pos for pos in pending |
- if len(set([c.identifier[pos] for c in cases])) == 1] |
- if common: |
- identifier = cases[0].identifier |
- for index in xrange(len(common)): |
- if index == 0: |
- self.output_file.write(u'if (LIKELY(') |
- else: |
- self.output_file.write(u' && ') |
- pos = common[index] |
- if identifier[pos].isalpha(): |
- self.output_file.write("(%s[%d] | 0x20) == '%s'" % |
- (self.array_variable, pos, identifier[pos])) |
- else: |
- self.output_file.write("%s[%d] == '%s'" % |
- (self.array_variable, pos, identifier[pos])) |
- self.output_file.write(u')) {\n') |
- next_pending = list(set(pending) - set(common)) |
- next_pending.sort() |
- self.inspect_array(cases, next_pending) |
- self.output_file.write(u'}\n') |
- else: |
- pos = self.choose_selection_pos(cases, pending) |
- next_pending = filter(lambda p: p != pos, pending) |
- |
- alternatives = list(set([c.identifier[pos] for c in cases])) |
- alternatives.sort() |
- |
- def literal(alternative): |
- if isinstance(alternative, int): |
- return str(alternative) |
- else: |
- return "'%s'" % alternative |
- |
- def response(alternative): |
- self.inspect_array([c for c in cases if c.identifier[pos] == alternative], |
- next_pending) |
- |
- expression = '(%s[%d] | 0x20)' % (self.array_variable, pos) |
- self.write_selection(expression, alternatives, literal, response) |
- else: |
- assert len(cases) == 1 |
- for block_line in cases[0].block: |
- self.output_file.write(block_line) |
- |
- def write_selection(self, expression, alternatives, literal, response): |
- if len(alternatives) == 1: |
- self.output_file.write(u'if (LIKELY(%s == %s)) {\n' % (expression, literal(alternatives[0]))) |
- response(alternatives[0]) |
- self.output_file.write(u'}\n') |
- elif len(alternatives) == 2: |
- self.output_file.write(u'if (%s == %s) {\n' % (expression, literal(alternatives[0]))) |
- response(alternatives[0]) |
- self.output_file.write(u'} else if (LIKELY(%s == %s)) {\n' % (expression, literal(alternatives[1]))) |
- response(alternatives[1]) |
- self.output_file.write(u'}\n') |
- else: |
- self.output_file.write('switch (%s) {\n' % expression) |
- for alternative in alternatives: |
- self.output_file.write(u'case %s: {\n' % literal(alternative)) |
- response(alternative) |
- self.output_file.write(u'} break;\n') |
- self.output_file.write(u'}\n') |
- |
- |
-class LineProcessor(object): |
- def process_line(self, line): |
- pass |
- |
- |
-class MainLineProcessor(LineProcessor): |
- """Processes the contents of an input file.""" |
- SWITCH_PATTERN = re.compile(r'\s*SWITCH\s*\((\w*),\s*(\w*)\) \{$') |
- |
- def __init__(self, output_file): |
- self.output_file = output_file |
- |
- def process_line(self, line): |
- match_switch = MainLineProcessor.SWITCH_PATTERN.match(line) |
- if match_switch: |
- array_variable = match_switch.group(1) |
- length_variable = match_switch.group(2) |
- return SwitchLineProcessor(self, self.output_file, array_variable, length_variable) |
- else: |
- self.output_file.write(line) |
- return self |
- |
- |
-class SwitchLineProcessor(LineProcessor): |
- """Processes the contents of a SWITCH block.""" |
- CASE_PATTERN = re.compile(r'\s*CASE\s*\(\"([a-z0-9_\-\(]*)\"\) \{$') |
- CLOSE_BRACE_PATTERN = re.compile(r'\s*\}$') |
- EMPTY_PATTERN = re.compile(r'\s*$') |
- |
- def __init__(self, parent, output_file, array_variable, length_variable): |
- self.parent = parent |
- self.output_file = output_file |
- self.array_variable = array_variable |
- self.length_variable = length_variable |
- self.cases = [] |
- |
- def process_line(self, line): |
- match_case = SwitchLineProcessor.CASE_PATTERN.match(line) |
- match_close_brace = SwitchLineProcessor.CLOSE_BRACE_PATTERN.match(line) |
- match_empty = SwitchLineProcessor.EMPTY_PATTERN.match(line) |
- if match_case: |
- identifier = match_case.group(1) |
- return CaseLineProcessor(self, self.output_file, identifier) |
- elif match_close_brace: |
- Optimizer(self.output_file, self.array_variable, self.length_variable).inspect(self.cases) |
- return self.parent |
- elif match_empty: |
- return self |
- else: |
- raise BadInput('Invalid line within SWITCH: %s' % line) |
- |
- def add_case(self, latest_case): |
- if latest_case.identifier in [c.identifier for c in self.cases]: |
- raise BadInput('Repeated case: %s' % latest_case.identifier) |
- self.cases.append(latest_case) |
- |
- |
-class CaseLineProcessor(LineProcessor): |
- """Processes the contents of a CASE block.""" |
- CLOSE_BRACE_PATTERN = re.compile(r'\s*\}$') |
- BREAK_PATTERN = re.compile(r'break;') |
- |
- def __init__(self, parent, output_file, identifier): |
- self.parent = parent |
- self.output_file = output_file |
- self.identifier = identifier |
- self.block = [] |
- |
- def process_line(self, line): |
- match_close_brace = CaseLineProcessor.CLOSE_BRACE_PATTERN.match(line) |
- match_break = CaseLineProcessor.BREAK_PATTERN.search(line) |
- if match_close_brace: |
- self.parent.add_case(SwitchCase(self.identifier, self.block)) |
- return self.parent |
- elif match_break: |
- raise BadInput('break within CASE not supported: %s' % line) |
- else: |
- self.block.append(line) |
- return self |
- |
- |
-def process_file(input_name, output_name): |
- """Transforms input file into legal C++ source code.""" |
- with io.open(input_name, 'r', -1, 'utf-8') as input_file: |
- with io.open(output_name, 'w', -1, 'utf-8') as output_file: |
- processor = MainLineProcessor(output_file) |
- input_lines = input_file.readlines() |
- for line in input_lines: |
- processor = processor.process_line(line) |
- |
- |
-if __name__ == '__main__': |
- process_file(sys.argv[1], sys.argv[2]) |