src/common_dmv.py

   1 # common-dmv.py, stuff common to cnf_ and loc_h_-files
   2
   3 # non-tweakable/constant "lookup" globals
   4
   5 LEFT, RIGHT, ADJ, NON = 0, 1, True, False
   6
   7
   8 GOR = 0  # (RIGHT,0) as an alternate possibility
   9 RGOL = 1 # (LEFT, 1)
  10 SEAL = 2 # 2
  11 GOL = 3  # (LEFT, 0)
  12 LGOR = 4 # (RIGHT,1)
  13 SEALS = [GOR, RGOL, SEAL, GOL, LGOR]
  14
  15 # CNF_DMV uses these:
  16 NGOR = 5  # we don't care about left-first attachment,
  17 NRGOL = 6 # but have to add the non-adjacent info
  18
  19 ROOTNUM = -1
  20 MPPROOT = ('ROOT', ROOTNUM)
  21 ROOT = (SEAL, ROOTNUM) # rather arbitrary, doesn't actually make a difference.
  22 STOP = (GOR, -2)
  23
  24 DEBUG = set(['TODO'])
  25
  26 def dirseal(dir):
  27     if dir == LEFT:
  28         return (GOL, RGOL)
  29     elif dir == RIGHT:
  30         return (GOR, LGOR)
  31     else:
  32         raise ValueError
  33
  34 def xgo_left(gt, lteq): # i < k <= loc_l(h)
  35     return xrange(gt+1, lteq+1)
  36
  37 def xgo_right(gt, lt): # loc_l(h) < k < j (but loc_r(h) <= k)
  38     return xrange(gt+1, lt)
  39
  40 def xlteq(lteq):
  41     return xrange(0,lteq+1)
  42
  43 def xlt(lt):
  44     return xrange(0,lt)
  45
  46 def xgt(gt, sent):
  47     '''Include the position "behind" the last word, ie. len(sent).'''
  48     lt = len(sent) + 1
  49     return xrange(gt+1, lt)
  50
  51 def xeq(eq, sent=[]):
  52     return [eq]
  53
  54 def xgteq(gteq, sent):
  55     '''Include the position "behind" the last word, ie. len(sent).'''
  56     lt = len(sent) + 1
  57     return xrange(gteq, lt)
  58
  59 def xtween(gt, lt):
  60     ''' For use with eg k:i<k<loc_h. Makes sure we get all and only the
  61     integers between gt and lt:
  62
  63     >>> [x for x in xtween(3,7)]
  64     [4, 5, 6]
  65     >>> [x for x in xtween(3.5,7)]
  66     [4, 5, 6]
  67     >>> [x for x in xtween(3.5,6.5)]
  68     [4, 5, 6]    '''
  69     return xrange(gt+1, lt)
  70
  71 def seals(node):
  72     return node[0]
  73
  74 def POS(node):
  75     "Actually just a number representing the POS-tag"
  76     return node[1]
  77
  78 def node_str(node, tag=lambda x:x):
  79     if(node == ROOT):
  80         return 'ROOT '
  81     elif(node == STOP):
  82         return 'STOP '
  83     else:
  84         s_h = seals(node)
  85         h = POS(node)
  86         if(s_h == RGOL):
  87             return "  %s><" % tag(h)
  88         elif(s_h == SEAL):
  89             return " _%s_ " % tag(h)
  90         elif(s_h == GOR):
  91             return "  %s> " % tag(h)
  92         elif(s_h == GOL):
  93             return " <%s  " % tag(h)
  94         elif(s_h == LGOR):
  95             return "><%s  " % tag(h)
  96         elif(s_h == NGOR):
  97             return " >%s> " % tag(h)
  98         elif(s_h == NRGOL):
  99             return " <%s><" % tag(h)
 100         else:
 101             raise ValueError("in node_str, got as node: %s" % node)
 102
 103
 104 def test(wanted, got, wanted_name="it", got_name=""):
 105     if not wanted == got:
 106         raise Warning, "Wanted %s to be %s, but %s was %s" % (wanted_name, wanted, got_name, got)
 107