Factor out asdl/pretty.py because it uses the 're' module.

We want convert regexes to re2c and break the dependency.

asdl/format.py

@@ -15,9 +15,8 @@
 - abbreviated, unnamed fields
"""

import re

from asdl import asdl_ as asdl
from asdl import pretty
from core import util

import os
@@ -320,23 +319,6 @@ def MakeTree(obj, abbrev_hook=None, omit_empty=True):
  return out_node


# This is word characters, - and _, as well as path name characters . and /.
_PLAIN_RE = re.compile(r'^[a-zA-Z0-9\-_./]+$')

# NOTE: Turning JSON back on can be a cheap hack for detecting invalid unicode.
# But we want to write our own AST walker for that.

def _PrettyString(s):
  if '\n' in s:
    #return json.dumps(s)  # account for the fact that $ matches the newline
    return repr(s)
  if _PLAIN_RE.match(s):
    return s
  else:
    #return json.dumps(s)
    return repr(s)


INDENT = 2

def _PrintWrappedArray(array, prefix_len, f, indent, max_col):
@@ -483,11 +465,11 @@ def PrintTree(node, f, indent=0, max_col=100):
    return

  if isinstance(node, str):
    f.write(ind + _PrettyString(node))
    f.write(ind + pretty.Str(node))

  elif isinstance(node, _ColoredString):
    f.PushColor(node.str_type)
    f.write(_PrettyString(node.s))
    f.write(pretty.Str(node.s))
    f.PopColor()

  elif isinstance(node, _Obj):
@@ -540,11 +522,11 @@ def _TrySingleLine(node, f, max_chars):
      If False, you can't use the value of f.
  """
  if isinstance(node, str):
    f.write(_PrettyString(node))
    f.write(pretty.Str(node))

  elif isinstance(node, _ColoredString):
    f.PushColor(node.str_type)
    f.write(_PrettyString(node.s))
    f.write(pretty.Str(node.s))
    f.PopColor()

  elif isinstance(node, list):  # Can we fit the WHOLE list on the line?

asdl/pretty.py

@@ -0,0 +1,45 @@
#!/usr/bin/python
"""
pretty.py
"""

try:
  import fastlex
except ImportError:
  fastlex = None

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

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

  _PLAIN_WORD_RE = re.compile(PLAIN_WORD_RE + '$')

  def IsPlainWord(s):
    if '\n' in s:  # account for the fact that $ matches the newline
      return False
    return _PLAIN_WORD_RE.match(s)


# NOTE: bash prints \' for single quote, repr() prints "'".  Gah.  This is also
# used for printf %q and ${var@q} (bash 4.4).

def Str(s):
  """Return a human-friendly representation of an arbitrary shell string.

  Used for ASDL pretty printing as well as the 'xtrace' feature in
  core/cmd_exec.py.
  """
  if IsPlainWord(s):
    return s
  else:
    return repr(s)


# NOTE: Converting strings to JSON and can be a cheap hack for detecting
# invalid unicode.  But we want to write our own AST walker for that.

core/cmd_exec.py

@@ -23,6 +23,7 @@
import time

from asdl import const
from asdl import pretty

from core import alloc
from core import args
@@ -1497,7 +1498,7 @@ def OnSimpleCommand(self, argv):
      return

    first_char, prefix = self._EvalPS4()
    cmd = ' '.join(_PrettyString(a) for a in argv)
    cmd = ' '.join(pretty.Str(a) for a in argv)
    print('%s%s%s' % (first_char, prefix, cmd), file=sys.stderr)

  def OnAssignment(self, lval, val, flags, lookup_mode):
@@ -1520,23 +1521,3 @@ def Event(self):
      - We should desugar to SetVar like mksh
    """
    pass


# Copied from asdl/format.py.  We're not using it directly because that is
# debug output, and this is real input.
# TODO: Is this slow?

# NOTE: bash prints \' for single quote, repr() prints "'".  Gah.  This is also
# used for printf %q and ${var@q} (bash 4.4).

_PLAIN_RE = re.compile(r'^[a-zA-Z0-9\-_./]+$')

def _PrettyString(s):
  if '\n' in s:
    #return json.dumps(s)  # account for the fact that $ matches the newline
    return repr(s)
  if _PLAIN_RE.match(s):
    return s
  else:
    #return json.dumps(s)
    return repr(s)