Source Code for Module Bio.ExPASy.Prodoc

  1  # Copyright 2000 by Jeffrey Chang.  All rights reserved. 
  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  """ 
  7  This module provides code to work with the prosite.doc file from 
  8  Prosite. 
  9  http://www.expasy.ch/prosite/ 
 10   
 11  Tested with: 
 12  Release 15.0, July 1998 
 13  Release 16.0, July 1999 
 14  Release 20.22, 13 November 2007 
 15  Release 20.43, 10 February 2009 
 16   
 17   
 18  Functions: 
 19  read               Read a Prodoc file containing exactly one Prodoc entry. 
 20  parse              Iterates over entries in a Prodoc file. 
 21   
 22  Classes: 
 23  Record             Holds Prodoc data. 
 24  Reference          Holds data from a Prodoc reference. 
 25  """ 
 26   
 27   
 28 -def read(handle): 
 29      record = __read(handle) 
 30      # We should have reached the end of the record by now 
 31      line = handle.readline() 
 32      if line: 
 33          raise ValueError("More than one Prodoc record found") 
 34      return record 
 35   
 36   
 37 -def parse(handle): 
 38      while True: 
 39          record = __read(handle) 
 40          if not record: 
 41              return 
 42          yield record 
 43   
 44   
 45 -class Record(object): 
 46      """Holds information from a Prodoc record. 
 47   
 48      Members: 
 49      accession      Accession number of the record. 
 50      prosite_refs   List of tuples (prosite accession, prosite name). 
 51      text           Free format text. 
 52      references     List of reference objects. 
 53   
 54      """ 
 55 -    def __init__(self): 
 56          self.accession = '' 
 57          self.prosite_refs = [] 
 58          self.text = '' 
 59          self.references = [] 
 60   
 61   
 62 -class Reference(object): 
 63      """Holds information from a Prodoc citation. 
 64   
 65      Members: 
 66      number     Number of the reference. (string) 
 67      authors    Names of the authors. 
 68      citation   Describes the citation. 
 69   
 70      """ 
 71 -    def __init__(self): 
 72          self.number = '' 
 73          self.authors = '' 
 74          self.citation = '' 
 75   
 76  # Below are private functions 
 77   
 78   
 79 -def __read_prosite_reference_line(record, line): 
 80      line = line.rstrip() 
 81      if line[-1] != '}': 
 82          raise ValueError("I don't understand the Prosite reference on line\n%s" % line) 
 83      acc, name = line[1:-1].split('; ') 
 84      record.prosite_refs.append((acc, name)) 
 85   
 86   
 87 -def __read_text_line(record, line): 
 88      record.text += line 
 89      return True 
 90   
 91   
 92 -def __read_reference_start(record, line): 
 93      # Read the references 
 94      reference = Reference() 
 95      reference.number = line[1:3].strip() 
 96      if line[1] == 'E': 
 97          # If it's an electronic reference, then the URL is on the 
 98          # line, instead of the author. 
 99          reference.citation = line[4:].strip() 
100      else: 
101          reference.authors = line[4:].strip() 
102      record.references.append(reference) 
103   
104   
105 -def __read_reference_line(record, line): 
106      if not line.strip(): 
107          return False 
108      reference = record.references[-1] 
109      if line.startswith('     '): 
110          if reference.authors[-1]==',': 
111              reference.authors += line[4:].rstrip() 
112          else: 
113              reference.citation += line[5:] 
114          return True 
115      raise Exception("I don't understand the reference line\n%s" % line) 
116   
117   
118 -def __read_copyright_line(record, line): 
119      # Skip the copyright statement 
120      if line.startswith('+----'): 
121          return False 
122      return True 
123   
124   
125 -def __read(handle): 
126      # Skip blank lines between records 
127      for line in handle: 
128          line = line.rstrip() 
129          if line and not line.startswith("//"): 
130              break 
131      else: 
132          return None 
133      record = Record() 
134      # Read the accession number 
135      if not line.startswith("{PDOC"): 
136          raise ValueError("Line does not start with '{PDOC':\n%s" % line) 
137      if line[-1] != '}': 
138          raise ValueError("I don't understand accession line\n%s" % line) 
139      record.accession = line[1:-1] 
140      # Read the Prosite references 
141      for line in handle: 
142          if line.startswith('{PS'): 
143              __read_prosite_reference_line(record, line) 
144          else: 
145              break 
146      else: 
147          raise ValueError("Unexpected end of stream.") 
148      # Read the actual text 
149      if not line.startswith('{BEGIN'): 
150          raise ValueError("Line does not start with '{BEGIN':\n%s" % line) 
151      read_line = __read_text_line 
152      for line in handle: 
153          if line.startswith('{END}'): 
154              # Clean up the record and return 
155              for reference in record.references: 
156                  reference.citation = reference.citation.rstrip() 
157                  reference.authors = reference.authors.rstrip() 
158              return record 
159          elif line[0] == '[' and line[3] == ']' and line[4] == ' ': 
160              __read_reference_start(record, line) 
161              read_line = __read_reference_line 
162          elif line.startswith('+----'): 
163              read_line = __read_copyright_line 
164          elif read_line: 
165              if not read_line(record, line): 
166                  read_line = None 
167      raise ValueError("Unexpected end of stream.") 
168