gistfile1.txt

def _decodepdf(filename):

    rsrcmgr = PDFResourceManager()
    retstr = StringIO()
    codec = 'utf-8'
    laparams = LAParams()
    device = TextConverter(rsrcmgr, retstr, codec=codec, laparams=laparams)

    fp = file(filename, 'rb')
    process_pdf(rsrcmgr, device, fp)
    fp.close()
    device.close()

    pdfstr = retstr.getvalue()
    retstr.close()

    labels = ['Book Dt:','Book Typ:','Cus Typ:','Bail:','Bond:','Court:','Judge:',
              'Arr Agy:','Arr Typ:','ROC:','Chg:','Indict:','Adj Dt:','Term:']

    for label in labels:
        pdfstr = pdfstr.replace(label,"\n{0} ".foramt(label)

    pdfstr = re.sub(' +',' ',pdfstr)
    pdfstr = re.sub('\n+','\n',pdfstr)

    return pdfstr,True