Source Code for Module Dbase.Pubmed.PubmedUtils

  1  # $Id: PubmedUtils.py 2 2006-05-06 22:54:39Z glandrum $ 
  2  # 
  3  #  Copyright (C) 2003-2006  Rational Discovery LLC 
  4  # 
  5  #   @@ All Rights Reserved  @@ 
  6  # 
  7  import re,urllib 
  8  import QueryParams 
  9   
 10   
 11 -def FillParagraph(text,width=70,indentLinesBy=0): 
 12    """ 
 13   
 14    >>> txt = "123 567 90 123\\n456 89\\n" 
 15    >>> print FillParagraph(txt,width=10) 
 16    123 567 90 
 17    123 456 89 
 18     
 19    >>> print FillParagraph(txt,width=10,indentLinesBy=3) 
 20    123 567 90 
 21       123 456 
 22       89 
 23    >>> txt = '123 567 \\n' 
 24    >>> print FillParagraph(txt,width=10) 
 25    123 567 
 26   
 27    """ 
 28    expr = re.compile('[\ \t\n\r]+') 
 29    words = expr.split(text) 
 30    res = [] 
 31    line = '' 
 32    for word in words: 
 33      if word: 
 34        if len(line)+len(word)>=width: 
 35          res.append(line) 
 36          line = ' '*indentLinesBy 
 37          line += word 
 38        else: 
 39          if line: 
 40            line += ' %s'%word 
 41          else: 
 42            line = word 
 43    if line: 
 44      res.append(line)     
 45    return '\n'.join(res) 
 46   
 47 -def RecordsToPubmedText(records): 
 48    """  Exports a sequence of JournalArticleRecords to a Medline text 
 49    format (not XML) that can be imported by EndNote 
 50   
 51    """ 
 52    res = [] 
 53    for i in range(len(records)): 
 54      rec = records[i] 
 55      res.append('UI  - %d'%i) 
 56      res.append('PMID- %s'%rec.PubMedId) 
 57      for auth in rec.authors: 
 58        res.append('AU  - %s %s'%(auth[0],auth[2])) 
 59      res.append('TI  - %s'%FillParagraph(rec.Title,indentLinesBy=6)) 
 60      res.append('TA  - %s'%rec.Source) 
 61      res.append('DP  - %s'%rec.PubYear) 
 62      res.append('PG  - %s'%rec.Pages) 
 63      res.append('VI  - %s'%rec.Volume) 
 64      res.append('IP  - %s'%rec.Issue) 
 65      res.append('AB  - %s'%FillParagraph(rec.Abstract,indentLinesBy=6)) 
 66   
 67      for keyword in rec.keywords: 
 68        res.append('MH  - %s'%(keyword)) 
 69                    
 70    return '\n'.join(res) 
 71       
 72   
 73   
 74 -def GetSearchURL(term,db="PubMed"): 
 75    queryParams = {'CMD':'search', 
 76                   'DB':db, 
 77                   'term':term} 
 78    url = '%s?%s'%(QueryParams.queryBase, 
 79                   urllib.urlencode(queryParams)) 
 80    return url 
 81     
 82   
 83   
 84  #------------------------------------ 
 85  # 
 86  #  doctest boilerplate 
 87  # 
 88 -def _test(): 
 89    import doctest,sys 
 90    return doctest.testmod(sys.modules["__main__"]) 
 91   
 92  if __name__ == '__main__': 
 93    import sys,os.path 
 94    failed,tried = _test() 
 95    sys.exit(failed) 
 96   
 97    import Searches 
 98    recs = Searches.GetRecords(['11960484','10893315'], 
 99                               conn=open('test_data/records.xml','r')) 
100    res = RecordsToPubmedText(recs) 
101    print res 
102