从输入文件中提取分子式、分子式信息

  1 # -*- coding: utf-8 -*-
  2 """
  3 Created on Mon Jul 11 18:12:11 2016
  4 
  5 @author: hugo
  6 
  7 分析输入文件，提取分子式，分子量信息
  8 """
  9 
 10 import os
 11 import linecache
 12 #import numpy
 13 #import sys
 14 
 15 PeriodicTable = {'H' : 1, 'C' : 12, 'O' : 16, 'N' : 14, 'F' : 19}
 16 
 17 
 18 def getFileName(path):
 19 #     获取指定目录下的所有指定后缀的文件名,并将文件名赋给filenamelist列表
 20     
 21      filenamelist = []
 22      f_list = os.listdir(path)
 23      for i in f_list:
 24 #        os.path.splitext(): 分离文件名与扩展名 
 25          if os.path.splitext(i)[1] == '.gjf':
 26 #             print i
 27              filenamelist.append(i)
 28 
 29      return filenamelist
 30 
 31                     
 32 if __name__ == '__main__':
 33     path = '.'
 34     #getFileName(path)
 35 
 36 # 写入文件名，分子式，分子量
 37 f = open('filename-formula-mass.txt', 'a')
 38 f.truncate()                                     # 清空文件
 39 f.write('Filename')
 40 f.write('\t')
 41 f.write('\t')
 42 f.write('Molecular Formula')
 43 f.write('\t')
 44 f.write('Mass of Molecule')
 45 f.write('\n')
 46 
 47 for filename in getFileName(path):
 48 #    print filename
 49     lines = linecache.getlines(filename)[9:]      # 读取filename第9行至末尾
 50 #    print lines    
 51     lines2 = lines[:-2]                           # 去掉最后两个空行            
 52     
 53     
 54     num_of_c = 0                                  # C原子个数
 55     num_of_h = 0                                  # H原子个数
 56     num_of_o = 0                                  # O原子个数
 57     num_of_n = 0                                  # N原子个数
 58     num_of_f = 0                                  # N原子个数
 59     mass_of_atom = 0                              # 原子质量
 60     mass_of_molecule = 0                          # 分子质量    
 61     
 62     for symbol_of_atom in lines2:
 63 #       symbol_of_atom[2]：每一行为一个str,读取str第二个字符，即为原子
 64         mass_of_atom = PeriodicTable[symbol_of_atom[2]]
 65         mass_of_molecule = mass_of_molecule + mass_of_atom
 66         
 67         if symbol_of_atom[2] == 'C':
 68             num_of_c = num_of_c + 1
 69         elif symbol_of_atom[2] == 'H':
 70             num_of_h = num_of_h + 1
 71         elif symbol_of_atom[2] == 'O':
 72             num_of_o = num_of_o + 1
 73         elif symbol_of_atom[2] == 'N' :
 74             num_of_n = num_of_n + 1
 75         else:
 76             num_of_f = num_of_f + 1
 77     
 78     # 分子式
 79     if num_of_f == 0:        
 80         formula = 'C' + str(num_of_c) + 'H' + str(num_of_h) + \
 81                   'O' + str(num_of_o) + 'N' + str(num_of_n)
 82     else:
 83         formula = 'C' + str(num_of_c) + 'H' + str(num_of_h) + \
 84                   'O' + str(num_of_o) + 'N' + str(num_of_n) + \
 85                   'F' + str(num_of_f)
 86               
 87 #    print formula
 88     
 89     # 分子量    
 90 #    print mass_of_molecule
 91     linecache.clearcache()
 92     
 93     # 写入文件名，分子式，分子量
 94 
 95     f.write(filename)
 96     f.write('\t')
 97     f.write('\t')
 98     f.write(formula)
 99     f.write('\t')
100     f.write('\t')
101     f.write(str(mass_of_molecule))
102     f.write('\n')
103 f.close()

 

 

输入文件例子：

%chk=sep1_c.chk
%mem=8GB
%nprocshared=8
#p nosymm Freq RB3LYP 6-31G(d)

N30:F398:A98:B98 bis-(2,2,2-trinitroethyl)-nitramine

0 1
N 0.050166 0.606549 0.111024
C -0.991352 -0.153881 0.759816
C -2.359635 -0.271483 0.040587
N -2.266367 -0.291738 -1.482893
O -1.713944 0.679205 -1.977460
O -2.729313 -1.259234 -2.061533
N -3.053844 -1.605171 0.444658
O -2.312353 -2.554503 0.637319
O -4.267284 -1.582700 0.492355
N -3.361456 0.825265 0.460123
O -3.970323 1.397169 -0.419621
O -3.464790 0.953113 1.667286
N -0.101221 2.000887 0.065310
O 0.778708 2.630096 -0.501944
O -1.102134 2.442779 0.611345
C 1.066123 0.033761 -0.743528
C 2.418849 -0.242480 -0.056897
N 3.024094 0.998601 0.606765
O 2.415534 1.388378 1.588008
O 4.019495 1.471328 0.093119
N 2.340188 -1.339446 1.017288
O 1.494052 -2.203660 0.811024
O 3.132751 -1.282849 1.932600
N 3.427657 -0.754862 -1.111580
O 3.311667 -0.266010 -2.222400
O 4.236583 -1.582004 -0.736132
H -0.617446 -1.170900 0.881856
H -1.206334 0.262848 1.746634
H 0.714386 -0.922014 -1.136239
H 1.261770 0.705098 -1.578729

转载于:https://www.cnblogs.com/hugo07/p/5666394.html