Source Code for Module Chem.FastSDMolSupplier

 1  # $Id: FastSDMolSupplier.py 391 2007-10-20 06:26:50Z glandrum $ 
 2  # 
 3  #  Copyright (C) 2007 Greg Landrum 
 4  #   @@ All Rights Reserved @@ 
 5  # 
 6  import sys 
 7  import Chem 
 8   
 9 -class FastSDMolSupplier(object): 
10    """ A wrapper around an SDMolSupplier that precomputes and stores  
11        molecular indices (via text processing) to allow quick length  
12        calculations and random access. 
13   
14        NOTE that this class needs to have the entire SD data in memory, 
15        so it's probably not particularly useful with large files. 
16    """ 
17    suppl=None 
18    data=None 
19    sanitize=True 
20 -  def __init__(self,fileN=None,data=None,sanitize=True,removeHs=True): 
21      if fileN: 
22        data = open(fileN,'r').read() 
23      self.sanitize=sanitize 
24      self.removeHs=removeHs 
25      if data: 
26        data = data.replace('\r\n','\n') 
27        self.init(data) 
28         
29 -  def init(self,data,recogTxt='$$$$\n'): 
30      if not data: 
31        raise ValueError,'no data' 
32      # FIX: it'd be nice to not be caching data locally like this, but it's the easiest 
33      # way to handle pickle support. 
34      self.data=data 
35      self.suppl = Chem.SDMolSupplier() 
36      self.suppl.SetData(data,sanitize=self.sanitize,removeHs=self.removeHs) 
37   
38      self._pos = [0] 
39      p = 0 
40      while 1: 
41        try: 
42          p = data.index(recogTxt,p+1) 
43          p+=len(recogTxt) 
44        except: 
45          break 
46        else: 
47          self._pos.append(p) 
48      self._pos.pop(-1) 
49      self.suppl._SetStreamIndices(self._pos) 
50      self._idx=0 
51       
52 -  def GetItemText(self,idx): 
53      startOfItem = self._pos[idx] 
54      if idx+1<len(self._pos): 
55        endOfItem = self._pos[idx+1] 
56      else: 
57        endOfItem = -1 
58      return self.data[startOfItem:endOfItem] 
59     
60 -  def reset(self): 
61      self.suppl.reset() 
62      self._idx=0 
63   
64    # ---------------------------------------------------------------- 
65    # support random access and an iterator interface: 
66 -  def __iter__(self): 
67      self.suppl.reset() 
68      return self 
69 -  def next(self): 
70      self._idx+=1 
71      return self.suppl.next() 
72     
73 -  def __len__(self): 
74      return len(self.suppl) 
75 -  def __getitem__(self,idx): 
76      return self.suppl[idx] 
77