Break another dependency on the 're' module.

Use re2c to translate PLAIN_WORD_RE to a native function, in the exact
same fashion as VAR_NAME_RE -> IsValidVarName().

asdl/pretty.py

@@ -9,10 +9,9 @@
  fastlex = None

# Word characters, - and _, as well as path name characters . and /.
PLAIN_WORD_RE = r'^[a-zA-Z0-9\-_./]+'
PLAIN_WORD_RE = r'[a-zA-Z0-9\-_./]+'

if 0:
#if fastlex:
if fastlex:
  IsPlainWord = fastlex.IsPlainWord
else:
  import re

core/cmd_exec.py

@@ -16,8 +16,6 @@
"""

import os
# Used to test whether we should repr() escape
import re
import resource
import sys
import time

core/lexer_gen.py

@@ -5,10 +5,12 @@
"""

import cStringIO
import pprint
import sys
import sre_parse
import sre_constants

from asdl import pretty  # For PLAIN_WORD_RE
from osh import lex
from osh import meta

@@ -173,6 +175,8 @@ def TranslateTree(re_tree, f, in_char_class=False):

def TranslateRegex(pat):
  re_tree = sre_parse.parse(pat)
  # For debugging
  #print(pprint.pformat(re_tree), file=sys.stderr)
  f = cStringIO.StringIO()
  TranslateTree(re_tree, f)
  return f.getvalue()
@@ -285,10 +289,10 @@ def TranslateLexer(lexer_def):
}
""")

def TranslateOther(var_name_re):
  re2_pat = TranslateRegex(var_name_re)
def TranslateRegexToPredicate(py_regex, func_name):
  re2c_pat = TranslateRegex(py_regex)
  print(r"""
static inline int IsValidVarName(const char* s, int len) {
static inline int %s(const char* s, int len) {
  unsigned char* p = s;  /* modified by re2c */
  unsigned char* end = s + len;

@@ -298,7 +302,7 @@ def TranslateOther(var_name_re):
  *     { return 0; }
  */
}
""" % re2_pat)
""" % (func_name, re2c_pat))


  # note: use YYCURSOR and YYLIMIT
@@ -312,7 +316,8 @@ def main(argv):
  if action == 'c':
    # Print code to stdout.
    TranslateLexer(lex.LEXER_DEF)
    TranslateOther(lex.VAR_NAME_RE)
    TranslateRegexToPredicate(lex.VAR_NAME_RE, 'IsValidVarName')
    TranslateRegexToPredicate(pretty.PLAIN_WORD_RE, 'IsPlainWord')

  elif action == 'print-all':
    # Top level is a switch statement.

native/fastlex.c

@@ -76,11 +76,25 @@ fastlex_IsValidVarName(PyObject *self, PyObject *args) {
  return PyBool_FromLong(IsValidVarName(name, len));
}

static PyObject *
fastlex_IsPlainWord(PyObject *self, PyObject *args) {
  const char *name;
  int len;

  if (!PyArg_ParseTuple(args, "s#", &name, &len)) {
    return NULL;
  }
  return PyBool_FromLong(IsPlainWord(name, len));
}


static PyMethodDef methods[] = {
  {"MatchToken", fastlex_MatchToken, METH_VARARGS,
   "(lexer mode, line, start_pos) -> (id, end_pos)."},
  {"IsValidVarName", fastlex_IsValidVarName, METH_VARARGS,
   "Is it a valid var name?"},
  {"IsPlainWord", fastlex_IsPlainWord, METH_VARARGS,
   "Can the string be pretty-printed without quotes?"},
  {NULL, NULL},
};