math/mpreduce.h: Missing include files.

[u/mdw/catacomb] / symm / multigen

#! @PYTHON@
###
### Generate files by filling in simple templates
###
### (c) 2013 Straylight/Edgeware
###

###----- Licensing notice ---------------------------------------------------
###
### This file is part of Catacomb.
###
### Catacomb is free software; you can redistribute it and/or modify
### it under the terms of the GNU Library General Public License as
### published by the Free Software Foundation; either version 2 of the
### License, or (at your option) any later version.
###
### Catacomb is distributed in the hope that it will be useful,
### but WITHOUT ANY WARRANTY; without even the implied warranty of
### MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
### GNU Library General Public License for more details.
###
### You should have received a copy of the GNU Library General Public
### License along with Catacomb; if not, write to the Free
### Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
### MA 02111-1307, USA.

from __future__ import with_statement

import itertools as IT
import optparse as OP
import os as OS
import re as RX
from cStringIO import StringIO
from sys import argv, exit, stderr

###--------------------------------------------------------------------------
### Utilities.

QUIS = OS.path.basename(argv[0])

def die(msg):
  stderr.write('%s: %s\n' % (QUIS, msg))
  exit(1)

def indexed(seq):
  return IT.izip(IT.count(), seq)

###--------------------------------------------------------------------------
### Reading the input values.

COLMAP = {}

class Cursor (object):
  def __init__(me, rel):
    me._rel = rel
    me._i = 0
    me._row = rel[0]
  def step(me):
    me._i += 1
    if me._i >= len(me._rel):
      me._i = me._row = None
      return False
    me._row = me._rel[me._i]
    return True
  def reset(me):
    me._i = 0
    me._row = me._rel[0]
  def __getitem__(me, i):
    return me._row[i]
  def __repr__(me):
    return '#<Cursor %r[%d] = %r>' % (me._rel, me._i, me._row)

class CursorSet (object):
  def __init__(me):
    me._map = {}
    me._stack = []
    me._act = None
  def push(me, rels):
    cc = []
    rr = []
    for r in rels:
      if r in me._map: continue
      c = me._map[r] = Cursor(r)
      rr.append(r)
      cc.append(c)
    me._stack.append((me._act, rr))
    me._act = cc
  def step(me):
    i = 0
    while i < len(me._act):
      if me._act[i].step(): return True
      if i >= len(me._act): return False
      me._act[i].reset()
      i += 1
    return False
  def pop(me):
    me._act, rels = me._stack.pop()
    for r in rels: del me._map[r]
  def get(me, rel, i):
    return me._map[rel][i]

class Relation (object):
  def __init__(me, head):
    me._head = head
    me._rows = []
    for i, c in indexed(head): COLMAP[c] = me, i
  def addrow(me, row):
    if len(row) != len(me._head):
      die("mismatch: row `%s' doesn't match heading `%s'" %
          (', '.join(row), ', '.join(head)))
    me._rows.append(row)
  def __len__(me):
    return len(me._rows)
  def __getitem__(me, i):
    return me._rows[i]
  def __repr__(me):
    return '#<Relation %r>' % me._head

def read_immediate(word):
  head, rels = word.split('=', 1)
  rel = Relation([c.strip() for c in head.split(',')])
  for row in rels.split(): rel.addrow([c.strip() for c in row.split(',')])

def read_file(spec):
  file, head = spec.split(':', 1)
  rel = Relation([c.strip() for c in head.split(',')])
  cols = [c.strip() for c in head.split(',')]
  with open(file) as f:
    for line in f:
      line = line.strip()
      if line.startswith('#') or line == '': continue
      rel.addrow(line.split())

def read_thing(spec):
  if spec.startswith('@'): read_file(spec[1:])
  else: read_immediate(spec)

###--------------------------------------------------------------------------
### Template structure.

class BasicTemplate (object):
  pass

class LiteralTemplate (BasicTemplate):
  def __init__(me, text, **kw):
    super(LiteralTemplate, me).__init__(**kw)
    me._text = text
  def relations(me):
    return set()
  def subst(me, out, cs):
    out.write(me._text)
  def __repr__(me):
    return '#<LiteralTemplate %r>' % me._text

class TagTemplate (BasicTemplate):
  def __init__(me, rel, i, op, **kw):
    super(TagTemplate, me).__init__(**kw)
    me._rel = rel
    me._i = i
    me._op = op
  def relations(me):
    return set([me._rel])
  def subst(me, out, cs):
    val = cs.get(me._rel, me._i)
    if me._op is not None: val = me._op(val)
    out.write(val)
  def __repr__(me):
    return '#<TagTemplate %s>' % me._rel._head[me._i]

class SequenceTemplate (BasicTemplate):
  def __new__(cls, seq, **kw):
    if len(seq) == 1:
      return seq[0]
    else:
      me = super(SequenceTemplate, cls).__new__(cls, seq = seq, **kw)
      tt = []
      cls = type(me)
      for t in seq:
        if isinstance(t, cls): tt += t._seq
        else: tt.append(t)
      me._seq = tt
      return me
  def __init__(me, seq, **kw):
    super(SequenceTemplate, me).__init__(**kw)
  def relations(me):
    rr = set()
    for t in me._seq: rr.update(t.relations())
    return rr
  def subst(me, out, cs):
    for t in me._seq: t.subst(out, cs)
  def __repr__(me):
    return '#<SequenceTemplate %r>' % me._seq

class RepeatTemplate (BasicTemplate):
  def __init__(me, sub):
    me._sub = sub
  def relations(me):
    return set()
  def subst(me, out, cs):
    rr = me._sub.relations()
    for r in rr:
      if len(r) == 0: return
    cs.push(rr)
    while True:
      me._sub.subst(out, cs)
      if not cs.step(): break
    cs.pop()
  def __repr__(me):
    return '#<RepeatTemplate %r>' % me._sub

###--------------------------------------------------------------------------
### Some slightly cheesy parsing machinery.

class ParseState (object):
  def __init__(me, file, text):
    me._file = file
    me._i = 0
    me._it = iter(text.splitlines(True))
    me.step()
  def step(me):
    me.curr = next(me._it, None)
    if me.curr is not None: me._i += 1
  def error(me, msg):
    die('%s:%d: %s' % (me._file, me._i, msg))

class token (object):
  def __init__(me, name):
    me._name = name
  def __repr__(me):
    return '#<%s>' % me._name

EOF = token('eof')
END = token('end')

R_SIMPLETAG = RX.compile(r'@ (\w+)', RX.VERBOSE)
R_COMPLEXTAG = RX.compile(r'@ { (\w+) ((?: : \w+)*) }', RX.VERBOSE)

OPMAP = {}

def defop(func):
  name = func.func_name
  if name.startswith('op_'): name = name[3:]
  OPMAP[name] = func
  return func

@defop
def op_u(val): return val.upper()

@defop
def op_l(val): return val.lower()

R_NOTIDENT = RX.compile(r'[^a-zA-Z0-9_]+')
@defop
def op_c(val): return R_NOTIDENT.sub('_', val)

def _pairify(val):
  c = val.find('=')
  if c >= 0: return val[:c], val[c + 1:]
  else: return val, val

@defop
def op_left(val): return _pairify(val)[0]
@defop
def op_right(val): return _pairify(val)[1]

def parse_text(ps):
  tt = []
  lit = StringIO()
  def spill():
    l = lit.getvalue()
    if l: tt.append(LiteralTemplate(l))
    lit.reset()
    lit.truncate()
  while True:
    line = ps.curr
    if line is None: break
    elif line.startswith('%'):
      if line.startswith('%#'): ps.step(); continue
      elif line.startswith('%%'): line = line[1:]
      else: break
    i = 0
    while True:
      j = line.find('@', i)
      if j < 0: break
      lit.write(line[i:j])
      m = R_SIMPLETAG.match(line, j)
      if not m: m = R_COMPLEXTAG.match(line, j)
      if not m: ps.error('invalid tag')
      col = m.group(1)
      try: rel, i = COLMAP[col]
      except KeyError: ps.error("unknown column `%s'" % col)
      wholeop = None
      ops = m.lastindex >= 2 and m.group(2)
      if ops:
        for opname in ops[1:].split(':'):
          try: op = OPMAP[opname]
          except KeyError: ps.error("unknown operation `%s'" % opname)
          if wholeop is None: wholeop = op
          else: wholeop = (lambda f, g: lambda x: f(g(x)))(op, wholeop)
      spill()
      tt.append(TagTemplate(rel, i, wholeop))
      i = m.end()
    lit.write(line[i:])
    ps.step()
  spill()
  return SequenceTemplate(tt)

DIRECT = []

def direct(rx):
  def _(func):
    DIRECT.append((RX.compile(rx, RX.VERBOSE), func))
    return func
  return _

def parse_template(ps):
  while ps.curr is not None and ps.curr.startswith('%#'): ps.step()
  if ps.curr is None: return EOF
  elif ps.curr.startswith('%'):
    if ps.curr.startswith('%%'): return parse_text(ps)
    for rx, func in DIRECT:
      line = ps.curr[1:].strip()
      m = rx.match(line)
      if m:
        ps.step()
        return func(ps, m)
    ps.error("unrecognized directive")
  else:
    return parse_text(ps)

def parse_templseq(ps, nestp):
  tt = []
  while True:
    t = parse_template(ps)
    if t is END:
      if nestp: break
      else: ps.error("unexpected `end' directive")
    elif t is EOF:
      if nestp: ps.error("unexpected end of file")
      else: break
    tt.append(t)
  return SequenceTemplate(tt)

@direct(r'repeat')
def dir_repeat(ps, m):
  return RepeatTemplate(parse_templseq(ps, True))

@direct(r'end')
def dir_end(ps, m):
  return END

def compile_template(file, text):
  ps = ParseState(file, text)
  t = parse_templseq(ps, False)
  return t

###--------------------------------------------------------------------------
### Main code.

op = OP.OptionParser(
  description = 'Generates files by filling in simple templates',
  usage = 'usage: %prog [-gl] FILE [COL,...=VAL,... ... | @FILE:COL,...] ...',
  version = 'Catacomb version @VERSION@')
for short, long, kw in [
  ('-l', '--list', dict(
      action = 'store_const', const = 'list', dest = 'mode',
      help = 'list filenames generated')),
  ('-g', '--generate', dict(
      action = 'store', metavar = 'PATH', dest = 'input',
      help = 'generate output (default)'))]:
  op.add_option(short, long, **kw)
op.set_defaults(mode = 'gen')
opts, args = op.parse_args()

if len(args) < 1: op.error('missing FILE')
filepat = args[0]
for rel in args[1:]: read_thing(rel)
filetempl = compile_template('<output>', filepat)

def filenames(filetempl):
  cs = CursorSet()
  rr = filetempl.relations()
  for r in rr:
    if not len(r): return
  cs.push(rr)
  while True:
    out = StringIO()
    filetempl.subst(out, cs)
    yield out.getvalue(), cs
    if not cs.step(): break
  cs.pop()

if opts.mode == 'list':
  for file, cs in filenames(filetempl): print file
elif opts.mode == 'gen':
  with open(opts.input) as f:
    templ = RepeatTemplate(compile_template(opts.input, f.read()))
  for file, cs in filenames(filetempl):
    new = file + '.new'
    with open(new, 'w') as out:
      templ.subst(out, cs)
    OS.rename(new, file)
else:
  raise Exception, 'What am I doing here?'

###----- That's all, folks --------------------------------------------------