Data/bpmf-trad-mand.py

   1 #!/usr/bin/python
   2
   3 import sys
   4 import re
   5 import string
   6
   7 def regor(xs):
   8   return "["+("".join([("\\." if x is "." else x) for x in xs]))+"]"
   9
  10 #tree structure: top level chooses between initials (perhaps blank), next level groups, and so on
  11 class Group:
  12   def __init__(self):
  13     self.syms = [];
  14     self.children = {};
  15
  16   def getChild(self, ch):
  17     if (ch not in self.children):
  18       self.children[ch] = Group();
  19     return self.children[ch];
  20
  21   #lists should contain (a list of ordered initials), then (a list of groups), then finals, then tones,
  22   # but only starting with whichever will be required from this level of the tree down.
  23   def write(self, indent, name, lists):
  24     if len(lists)==0:
  25       for sym in self.syms:
  26         s='&#x'+hex(ord(sym))[2:]+';'
  27         print indent + '<s d="'+s+'" t="'+s+'"/>';
  28     else:
  29       for ch in lists[0]:
  30         if ch not in self.children:
  31           continue;
  32         if ch=="":
  33           n=""
  34           desc=indent + '<group label="" visible="no"'
  35         else:
  36           n='&#x'+string.upper(hex(ord(ch))[2:])+';' #hex() returns "0x..."
  37           desc=indent + '<group label="'+n+'"'
  38         if self.children[ch].syms:
  39           desc+=' name="'+name+n+'"'
  40         print desc+'>'
  41         self.children[ch].write(indent+" ",name+n,lists[1:]);
  42         del self.children[ch]
  43         print indent + "</group>";
  44     for ch in self.children:
  45       print "WARNING Did not output",(["tone","final","group","initial"][len(lists)-1]),ch.encode("utf-8")
  46
  47   def addSym(self,sym):
  48     self.syms+=sym
  49
  50 keypat = re.compile("([^ ]+) +([^ \r\n]+)",re.UNICODE);
  51 keys = {}
  52
  53 with open(sys.argv[1]) as file:
  54   for line in file:
  55     if (re.match("%keyname",line) is not None):
  56       break
  57   #line should be the keyname begin.
  58   for line in file:
  59     if (re.match("%keyname",line) is not None): break
  60     m = keypat.match(unicode(line,"utf-8"));
  61     if (m is None):
  62       die("Error, could not parse keyname"+line)
  63     else:
  64       keys[m.group(1)] = m.group(2);
  65   #line should be the keyname end
  66
  67   groups = []
  68   initials = []
  69   finals = []
  70   tones = []
  71
  72   for key,value in keys.items():
  73     v=ord(value)
  74     if v==0x3127 or v==0x3128 or v==0x3129:
  75       groups+=key
  76     elif v>=0x3105 and v<0x311A:
  77       initials+=key
  78     elif v>=0x311A and v<=0x3127:
  79       finals+=key
  80     elif v==0x02CA or v==0x02C7 or v==0x02CB or v==0x02D9: #0x30FB is a big version of 02d9
  81       tones+=key
  82     else:
  83       die("**ERROR** unknown bopomofo "+key+" = "+value)
  84   pat = "("+regor(initials)+"?)("+regor(groups)+"?)("+regor(finals)+"?)("+regor(tones)+"?)[\s\t]*([^\s\t]+)\n"
  85   #print pat
  86   pattern=re.compile(pat,re.UNICODE);
  87
  88   for line in file:
  89     if (re.match("%chardef +begin",line) is not None):
  90       break
  91     die("Expected chardef begin, got "+line)
  92
  93   root = Group()
  94   keys[""]=""
  95   for line in file:
  96     if (re.match("%chardef +end",line) is not None): break
  97     m=pattern.match(unicode(line,"utf-8"))
  98     if m is not None:
  99       root.getChild(keys[m.group(1)]).getChild(keys[m.group(2)]).getChild(keys[m.group(3)]).getChild(keys[m.group(4)]).addSym(m.group(5))
 100     else:
 101       die("**ERROR** matching "+line)
 102
 103 def ch(lst):
 104   lst=[keys[x] for x in lst];
 105   lst.sort()
 106   lst.insert(0,"")
 107   return lst
 108
 109 root.write("","",[ch(x) for x in [initials,groups,finals,tones]])