Source Code for Module Bio.PDB.MMCIF2Dict

  1  # Copyright (C) 2002, Thomas Hamelryck (thamelry@binf.ku.dk) 
  2  # This code is part of the Biopython distribution and governed by its 
  3  # license.  Please see the LICENSE file that should have been included 
  4  # as part of this package. 
  5   
  6  """Turn an mmCIF file into a dictionary.""" 
  7   
  8  import os.path 
  9  import warnings 
 10  import Bio.PDB.mmCIF.MMCIFlex as MMCIFlex 
 11   
 12   
 13 -class MMCIF2Dict(dict): 
 14      # The token identifiers 
 15      NAME=1 
 16      LOOP=2 
 17      DATA=3 
 18      SEMICOLONS=4 
 19      DOUBLEQUOTED=5 
 20      QUOTED=6 
 21      SIMPLE=7 
 22   
 23 -    def __init__(self, filename): 
 24          if not os.path.isfile(filename): 
 25              raise IOError("File not found.") 
 26          MMCIFlex.open_file(filename) 
 27          dict.__init__(self, **self._make_mmcif_dict()) 
 28          MMCIFlex.close_file() 
 29   
 30 -    def _make_mmcif_dict(self): 
 31          """ 
 32          Loop through PLY token (type, value) pairs, return a dict. 
 33   
 34          """ 
 35          # this dict will contain the name/data pairs 
 36          mmcif_dict = {} 
 37          # entry for garbage 
 38          mmcif_dict[None] = [] 
 39          # local copies 
 40          NAME=self.NAME 
 41          LOOP=self.LOOP 
 42          DATA=self.DATA 
 43          SEMICOLONS=self.SEMICOLONS 
 44          DOUBLEQUOTED=self.DOUBLEQUOTED 
 45          QUOTED=self.QUOTED 
 46          SIMPLE=self.SIMPLE 
 47          get_token=MMCIFlex.get_token 
 48          # are we looping? 
 49          loop_flag=0 
 50          # list of names in loop 
 51          temp_list=[] 
 52          # last encountered name 
 53          current_name=None 
 54          # get first token/value pair 
 55          token, value=get_token() 
 56          # print token, value 
 57          # loop until EOF (token==0) 
 58          while token: 
 59              if token==NAME: 
 60                  if loop_flag: 
 61                      # Make lists for all the names in the loop 
 62                      while token==NAME: 
 63                          # create  a list for each name encountered in loop 
 64                          new_list=mmcif_dict[value]=[] 
 65                          temp_list.append(new_list) 
 66                          token, value=get_token() 
 67                          # print token, value 
 68                      loop_flag=0 
 69                      # nr of data items parsed 
 70                      data_counter=0 
 71                      # corresponding data name 
 72                      pos=0 
 73                      nr_fields=len(temp_list) 
 74                      # Now fill all lists with the data 
 75                      while token>3: 
 76                          pos=data_counter%nr_fields 
 77                          data_counter=data_counter+1 
 78                          temp_list[pos].append(value) 
 79                          token, value=get_token() 
 80                          # print token, value 
 81                      if pos!=nr_fields-1: 
 82                          warnings.warn("ERROR: broken name-data pair " 
 83                                        "(data missing)!", RuntimeWarning) 
 84                      # The last token was not used, so 
 85                      # don't set token to None! (this means the 
 86                      # last parsed token goes through the loop again) 
 87                  else: 
 88                      # simple name-data pair (no loop) 
 89                      # so next token should be the data 
 90                      next_token, data=get_token() 
 91                      # print token, value 
 92                      mmcif_dict[value]=data 
 93                      if next_token<4: 
 94                          warnings.warn("ERROR: broken name-data pair " 
 95                                        "(name-non data pair)!", RuntimeWarning) 
 96                          # print token, value 
 97                      else: 
 98                          # get next token 
 99                          token=None 
100              elif token==LOOP: 
101                  loop_flag=1 
102                  temp_list=[] 
103                  # get next token 
104                  token=None 
105              elif token==DATA: 
106                  mmcif_dict[value[0:5]]=value[5:] 
107                  token=None 
108              else: 
109                  # we found some complete garbage 
110                  warnings.warn("ERROR: broken name-data pair " 
111                                "(missing name)!\n%s %s" % (token, value), 
112                                RuntimeWarning) 
113                  mmcif_dict[None].append(value) 
114                  # get next token 
115                  token=None 
116              if token is None: 
117                  token, value=get_token() 
118                  # print token, value 
119          return mmcif_dict 
120   
121   
122  if __name__=="__main__": 
123   
124      import sys 
125   
126      if len(sys.argv)!=2: 
127          print "Usage: python MMCIF2Dict filename." 
128   
129      filename=sys.argv[1] 
130   
131      mmcif_dict = MMCIF2Dict(filename) 
132   
133      entry = "" 
134      print "Now type a key ('q' to end, 'k' for a list of all keys):" 
135      while(entry != "q"): 
136          entry = raw_input("MMCIF dictionary key ==> ") 
137          if entry == "q": 
138              sys.exit() 
139          if entry == "k": 
140              for key in mmcif_dict: 
141                  print key 
142              continue 
143          try: 
144              value=mmcif_dict[entry] 
145              if isinstance(value, list): 
146                  for item in value: 
147                      print item 
148              else: 
149                  print value 
150          except KeyError: 
151              print "No such key found." 
152