#!/usr/bin/python
###########################################################################
#
#   WoS2Pajek - Transforming WoS bibliographies into Pajek
#
#   WoS2Pajek.run(project,WoSbib)
#
#   import sys; sys.path.append(r'D:\Vlado\work\Python\WoS')
#
#   import WoS2Pajek;  WoS2Pajek.run('batagelj','batagelj.WoS',500,False,10)
#   reload(WoS2Pajek); WoS2Pajek.run('batagelj','batagelj.WoS',500,False,1)
#   reload(WoS2Pajek); WoS2Pajek.run('SN','SN.WoS',200000,False,500)
#   reload(WoS2Pajek); WoS2Pajek.run('SN','SN.WoS',200000,False,0)
#   reload(WoS2Pajek); WoS2Pajek.run('VBAF','VBAFall.WoS',50000,False,50)
#   reload(WoS2Pajek); WoS2Pajek.run('VBAF','VBAF.WoS',2000,False,1)
#
#   Vladimir Batagelj,  23-30. march 2007
#
###########################################################################

import string, os, shutil

def ISIname(AU,PY,J9,VL,BP):
  name = AU.upper().replace(' ','').replace(',',' ')
  if PY != '': name = name+', '+PY
  if J9 != '': name = name+', '+J9[:20]
  if VL != '': name = name+', V'+VL
  if BP != '': name = name+', P'+BP
  return name

def nameG(name):
  s = name.split(", ")
  y = 0; v = ''; p = ''; n = ''
  if s[0][0] == '*' : n = s[0]
  else:
    q = s[0].split(" ")
    n = q[0][:8]
    if len(q) > 1 : n = n+'_'+q[1][0]
  if len(s) >= 2 :
    try: y = eval(s[1])
    except: pass
  if len(s) >= 4 :
    if s[3][0] == 'V' :
      v = s[3][1:]
      if len(s) >= 5 :
        if s[4][0] == 'P' : p = s[4][1:]
    elif s[3][0] == 'P' : p = s[3][1:]
  if (y == 0) and (len(s) > 1):
    try:
      y = eval(s[0])
      n = '$'+s[1].replace(' ','_')
    except: pass
  if (v == '') and (p == '') and (y == 0) and (len(s) > 1) :
    p = s[1].replace(' ','_')
  if n != '' : return n+'('+str(y)+')'+v+':'+p
# non-standard
  print "--> ", name
  return name

def Gname(AU,PY,VL,BP):
  name = AU.upper()
  s = name.split(", ")
  name = s[0].replace(' ','')[:8]+'_'
  if len(s) > 1:
    name = name + s[1][0]
  return name+'('+str(PY)+')'+VL+':'+BP

def infVertex(name):
   global nver, vert, maxver, nodes, years, ISI, step
   if vert.has_key(name):
#     print 'Old vertex(', vert[name], '): ' + name
     return vert[name]
   else:
     if ISI :
       s = name.split(", ")
       try:  year = eval(s[1])
       except: year = 0
     else:
       z = name.find('(')+1; k = name.find(')')
       year = name[z:k]
     nver = nver + 1;
     if nver >= maxver :
       print "\n*** max vertex number exceeded\n*** program exited"
       exit(0)
     vert[name] = nver
#     if (nver % 1000) == 0:
#       print 'New vertex(', nver, '): ' + name
     nodes.write(str(nver)+' "'+name+'"\n')
     years.write(str(year)+'\n')
     if name.find('"') >= 0:
       print '***** Bad label ',nver,' : ',name
     return nver

def infAuthor(name):
   global naut, aut, authors
   if aut.has_key(name):
     return aut[name]
   else:
     naut = naut + 1;
     aut[name] = naut
     authors.write(str(naut)+' "'+name+'"\n')
     return naut

def run(project,bibfile,maxvert,ISIn,listep):
   global nver, vert, maxver, nodes, ISI, aut, naut, years, authors, step
   maxver = maxvert; ISI = ISIn
   step = 1000000
   if listep > 0: step = listep
   print "\n*** WoS2Pajek - 0.1"
   print "by V. Batagelj, April 3, 2007 / March 23, 2007\n"
   workdir = 'd:\\vlado\\work\\python\\WoS\\'+project+'\\'
   vertype = [ 0 for i in range(maxver) ]
   try:
     bib = open(workdir+bibfile, 'r')
   except:
     print 'wrong WoS file'
     exit()
   nodes  = open(workdir+'nodes.tmp', 'w')
   arcs   = open(workdir+'arcs.tmp', 'w')
   arcs.write('*arcs \n')
   temp  = open(workdir+'works.tmp', 'w')
   authors = open(workdir+'authors.tmp', 'w')
   years  = open(workdir+'years.tmp', 'w')
   vert = {}; nver = 0
   aut  = {}; naut = 0
   numbib = 0; endbib = False;
   while not endbib:
     line = bib.readline()
     if not line: break
     control = line[:2]
     if control != '  ': state = control
     content = line[3:-1]
     if control =='PT':
       numbib = numbib + 1
       bibAU = ''; bibCR = ''; bibPT = content;
       bibVL = ''; bibJ9 = ''; bibBP = ''; bibPY = '0';
       listCR = []; listAU = []
     elif control =='AU':
       if bibAU == '': bibAU = content
     elif control == 'J9': bibJ9 = content
     elif control == 'PY': bibPY = content
     elif control == 'VL': bibVL = content
     elif control == 'BP': bibBP = content
     elif control == 'ER':
       if listCR != []:
         if ISI : name = ISIname(bibAU,bibPY,bibJ9,bibVL,bibBP)
         else: name = Gname(bibAU,bibPY,bibVL,bibBP)
         if (numbib % step) == 0: print numbib,':',name
         sind = infVertex(name)
         if vertype[sind] :
           print "*** duplicate: ", ISIname(bibAU,bibPY,bibJ9,bibVL,bibBP)
         else:
           vertype[sind] = 1
           for tind in listCR:
             arcs.write(str(sind)+' '+str(tind)+'\n')
           for aind in listAU:
             temp.write(str(sind)+' '+str(aind)+'\n')
     elif control == 'EF':
       endbib = True
     else:
       pass
     if state == 'CR':
       if ISI : name = content
       else: name = nameG(content)
       newWork = not vert.has_key(name)
       work = infVertex(name)
       listCR.append(work)
       if newWork:
         if ISI : author = name[:name.find(',')]
         elif (name[0]=='*') or (name[0]=='$') : author = 'UNKNOWN'
         else: author = name[:name.find('(')]
         iaut = infAuthor(author)
         temp.write(str(work)+' '+str(iaut)+'\n')
     if state == 'AU':
       if ISI : author = content.upper().replace(' ','').replace(',',' ')
       else:
         name = content.upper()
         s = name.split(", ")
         author = s[0].replace(' ','')[:8]+'_'
         if len(s) > 1: author = author + s[1][0]
         listAU.append(infAuthor(author))

   bib.close(); nodes.close(); arcs.close()
   authors.close(); years.close(); temp.close()
   print "number of works   = ",nver
   print "number of authors = ",naut

# year of publication partition
   print "\n*** FILES:\nyear of publication partition: "+project+'Year.clu'
   yinp  = open(workdir+'years.tmp', 'r')
   years  = open(workdir+project+'Year.clu', 'w')
   years.write('*vertices '+str(nver)+'\n')
   shutil.copyfileobj(yinp,years)
   yinp.close(); years.close()
# described / cited only - partition
   print "described / cited only - partition: "+project+'DC.clu'
   part  = open(workdir+project+'DC.clu', 'w')
   part.write('*vertices '+str(nver)+'\n')
   for i in range(nver):
     part.write(str(vertype[i+1])+'\n')
   part.close()
# citation network
   print "citation network: "+project+'Cite.net'
   nodes  = open(workdir+'nodes.tmp', 'r')
   net  = open(workdir+project+'Cite.net', 'w')
   net.write('*vertices '+str(nver)+'\n')
   shutil.copyfileobj(nodes,net)
   arcs   = open(workdir+'arcs.tmp', 'r')
   shutil.copyfileobj(arcs,net)
   arcs.close(); net.close(); nodes.close()
# works X authors network
   print "works X authors network: "+project+'WA.net'
   nodes  = open(workdir+'nodes.tmp', 'r')
   taut  = open(workdir+'authors.tmp', 'r')
   wa  = open(workdir+project+'WA.net', 'w')
   wa.write('*vertices '+str(nver+naut)+' '+str(nver)+'\n')
   shutil.copyfileobj(nodes,wa)
   nodes.close()
   endfile = False;
   while not endfile:
     line = taut.readline()
     if not line: break
     s = line.split(" ")
     wa.write(str(eval(s[0])+nver)+' '+s[1])
   temp  = open(workdir+'works.tmp', 'r')
   wa.write('*arcs\n')
   while not endfile:
     line = temp.readline()
     if not line: break
     s = line.split(" ")
     wa.write(s[0]+' '+str(eval(s[1])+nver)+'\n')
   temp.close(); wa.close(); taut.close()
   try:
     os.remove(workdir+'works.tmp')
     os.remove(workdir+'authors.tmp')
     os.remove(workdir+'years.tmp')
     os.remove(workdir+'arcs.tmp')
     os.remove(workdir+'nodes.tmp')
   except:
     print "unable to delete some temp files"
   print "***\n"