def test_all_bankruptcy_dktrpts_for_parties_basics(self): count =0 no_parties_dockets = [] one_parties_dockets = [] for filename in os.listdir(BANK_TEST_DOCKET_PATH): court, casenum = filename.split(".")[:2] soup = _open_soup("/".join([BANK_TEST_DOCKET_PATH, filename])) parties = PP._get_parties_info_from_dkrpt(soup,court) if len(parties)==0: no_parties_dockets.append(filename) if len(parties) == 1: one_parties_dockets.append(filename) print "" print "Dockets with no parties:" for filename in no_parties_dockets: print filename print "Dockets with only one party (possible error): " for filename in one_parties_dockets: print filename
def test_get_parties_info_from_dkrpt_multiple_tables(self): the_soup = _open_soup(TEST_DOCKET_PATH + "mad.137971.html") parties = PP._get_parties_info_from_dkrpt(the_soup, "mad") self.assertEquals(len(parties), 4) self.assertEquals(parties[0]["name"], "Aaron Swartz") self.assertEquals(parties[0]["extra_info"], "TERMINATED: 01/14/2013") self.assertEquals(len(parties[0]["attorneys"]), 6) self.assertEquals(parties[1]["name"], "Massachusetts Institute of Technology") self.assertEquals(len(parties[1]["attorneys"]), 2) self.assertEquals(parties[1]["type"], "Interested Party") self.assertEquals(parties[2]["name"], "JSTOR") self.assertEquals(len(parties[2]["attorneys"]), 1) self.assertEquals(parties[2]["type"], "Interested Party") self.assertEquals(parties[3]["name"], "USA") self.assertEquals(len(parties[3]["attorneys"]), 3) self.assertEquals(parties[3]["type"], "Plaintiff") self.assertEquals(parties[3]["attorneys"][0]["attorney_name"], "Jack W. Pirozzolo") self.assertEquals(parties[3]["attorneys"][1]["attorney_name"], "Scott Garland") self.assertEquals(parties[3]["attorneys"][2]["attorney_name"], "Stephen P. Heymann")
def test_get_parties_info_from_dkrpt(self): testdockets = {} for docket in TEST_DOCKET_LIST: testdockets[docket] = _open_soup(TEST_DOCKET_PATH + docket + "docket.html") parties = PP._get_parties_info_from_dkrpt(testdockets["txed"], "txed") self.assertEquals(len(parties), 14) self.assertEquals(parties[0]["name"], "AOL LLC") self.assertEquals(parties[0]["extra_info"], "TERMINATED: 03/26/2008") self.assertEquals(parties[2]["type"], "Mediator") self.assertEquals(parties[2]["name"], "James W. Knowles") parties = PP._get_parties_info_from_dkrpt(testdockets["deb"], "deb") self.assertEquals(len(parties), 9) self.assertEquals(parties[0]["name"], "American Business Financial Services, Inc., a Delaware Corporation") self.assertEquals(parties[0]["type"], "Debtor") self.assertEquals(len(parties[0]["attorneys"]), 9) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Bonnie Glantz Fatell") self.assertEquals(parties[0]["attorneys"][0]["attorney_role"], "TERMINATED: 04/11/2006") self.assertEquals(len(parties[1]["attorneys"]), 11) self.assertEquals(len(parties[1]["attorneys"]), 11) parties = PP._get_parties_info_from_dkrpt(testdockets["almb"], "almb") self.assertEquals(len(parties), 2) self.assertEquals(parties[0]["name"], "Ruthie Harris") # Should be no attorneys object self.assertEquals(parties[1].get("attorneys"), None) parties = PP._get_parties_info_from_dkrpt(testdockets["almd"], "almd") self.assertEquals(len(parties), 2) self.assertEquals(parties[0]["name"], "Joyce Efurd") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(len(parties[0]["attorneys"]), 3) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Allen Durham Arnold") self.assertEquals(parties[0]["attorneys"][0]["attorney_role"], "LEAD ATTORNEY\nATTORNEY TO BE NOTICED") self.assertEquals(len(parties[1]["attorneys"]), 3) # Should be no extra_info self.assertEquals(parties[0].get("extra_info"), None) parties = PP._get_parties_info_from_dkrpt(testdockets["cit"], "cit") self.assertEquals(len(parties), 4) self.assertEquals(parties[0]["name"], "New Hampshire Ball Bearing, Inc.") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Frank H. Morgan") self.assertEquals(parties[0]["attorneys"][0]["attorney_role"], "LEAD ATTORNEY\nATTORNEY TO BE NOTICED") # self.assertEquals(parties[2]["name"], "United States Customs and Border Protection") self.assertEquals(len(parties[2]["attorneys"]), 1) # This document has no parties, but it shouldn't break anything when doing that parties = PP._get_parties_info_from_dkrpt(testdockets["cit2"], "cit") self.assertEquals(len(parties), 0) """ this docket doesn't work - errors in creating beautiful soup parties = PP._get_parties_info_from_dkrpt(testdockets["akd"], "akd") print_parties(parties) self.assertEquals(len(parties), 4) self.assertEquals(parties[0]["name"], "West American Insurance Company") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Brewster H. Jamieson") self.assertEquals(parties[0]["attorneys"][0]["attorney_role"], "LEAD ATTORNEY\nATTORNEY TO BE NOTICED") self.assertEquals(len(parties[1]["attorneys"]), 0) """ parties = PP._get_parties_info_from_dkrpt(testdockets["cand"], "cand") self.assertEquals(len(parties), 3) self.assertEquals(parties[0]["name"], "John Michael Balbo") self.assertEquals(parties[0]["type"], "Petitioner") self.assertTrue("PRO SE" in parties[0]["attorneys"][0]["contact"]) self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[1]["extra_info"], "Secretary CDCR") parties = PP._get_parties_info_from_dkrpt(testdockets["cand2"], "cand") self.assertEquals(len(parties), 4) self.assertEquals(parties[0]["name"], "James Brady") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(parties[1]["name"], "Sarah Cavanagh") self.assertEquals(parties[1]["type"], "Plaintiff" ) self.assertEquals(parties[1]["extra_info"], "individually and on behalf of all others similarly situated" ) self.assertEquals(len(parties[0]["attorneys"]), 3) self.assertEquals(parties[2]["name"], "Deloitte & Touche LLP") self.assertEquals(parties[2]["extra_info"], "a limited liability partnership") self.assertEquals(parties[3]["name"], "Deloitte Tax LLP") self.assertEquals(parties[3]["extra_info"], "TERMINATED: 08/14/2008") # There is extra metadata in this one that doesn't appear in others - Pending courts, highest offense level, disposition, etc, not collecting those currently # parties = PP._get_parties_info_from_dkrpt(testdockets["cand3"], "cand") # self.assertEquals(len(parties), 2) # self.assertEquals(len(parties[0]["attorneys"]), 1) # self.assertEquals(parties[0]["name"], "Gustavo Alfaro-Medina") parties = PP._get_parties_info_from_dkrpt(testdockets["caed"], "caed") self.assertEquals(len(parties), 6) self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["name"], "Corey Mitchell") self.assertEquals(parties[1]["extra_info"], "Correctional Officer") parties = PP._get_parties_info_from_dkrpt(testdockets["ded"], "ded") self.assertEquals(len(parties), 4) self.assertEquals(len(parties[0]["attorneys"]), 2) self.assertEquals(parties[0]["name"], "Cubist Pharmaceuticals Inc.") parties = PP._get_parties_info_from_dkrpt(testdockets["cacd"], "cacd") self.assertEquals(len(parties), 18) self.assertEquals(len(parties[0]["attorneys"]), 2) self.assertEquals(parties[0]["name"], "LA Printex Industries Inc")
def test_bankruptcy_parties_info_from_dkrpt(self): bank_dockets_list = ["njb.658906", "mnb.325447", "mdb.532409", "nvb.242643", "mieb.600286", "mdb.541423"] bank_soups = {} for docket in bank_dockets_list: bank_soups[docket] = _open_soup(BANK_TEST_DOCKET_PATH + docket + ".html") # Normal bankruptcy proceedings parties = PP._get_parties_info_from_dkrpt(bank_soups["mdb.532409"], "mdb") self.assertEquals(len(parties), 3) self.assertEquals(parties[0]["name"], "Rodney K. Cunningham") self.assertEquals(parties[0]["type"], "Debtor") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Sopo Ngwa") self.assertEquals(parties[1]["name"], "Karen S. Cunningham") self.assertEquals(parties[1]["type"], "Debtor") self.assertEquals(len(parties[1]["attorneys"]), 1) self.assertEquals(parties[1]["attorneys"][0]["attorney_name"], "Sopo Ngwa") self.assertEquals(parties[2]["type"], "Trustee") parties = PP._get_parties_info_from_dkrpt(bank_soups["nvb.242643"], "nvb") self.assertEquals(len(parties), 4) self.assertEquals(parties[0]["name"], "PAUL OGALESCO") self.assertTrue("PRO SE" in parties[0]["attorneys"][0]["contact"]) self.assertEquals(parties[0]["type"], "Debtor") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[1]["name"], "RICK A. YARNALL") self.assertEquals(parties[1]["type"], "Trustee") self.assertTrue("TERMINATED" in parties[1]["extra_info"]) # Adversary Proceeding type docket parties = PP._get_parties_info_from_dkrpt(bank_soups["njb.658906"], "njb") self.assertEquals(len(parties), 6) self.assertEquals(parties[0]["name"], "Richard A. Spair") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Eugene D. Roth") self.assertEquals(parties[1]["attorneys"][1]["attorney_role"], "LEAD ATTORNEY") self.assertEquals(parties[3]["name"], "Albert Russo") self.assertEquals(parties[3]["type"], "Trustee") parties = PP._get_parties_info_from_dkrpt(bank_soups["mnb.325447"], "mnb") self.assertEquals(len(parties), 2) self.assertEquals(parties[0]["name"], "RANDALL L SEAVER") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(parties[0]["extra_info"], "101 W. Burnsville Pkwy., Suite 201\nBurnsville, MN 55337") self.assertEquals(len(parties[0]["attorneys"]), 1) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Matthew R. Burton") parties = PP._get_parties_info_from_dkrpt(bank_soups["mieb.600286"], "mieb") miebfilebits = open(BANK_TEST_DOCKET_PATH+ "mieb.600286" + ".html").read() miebdocket = PP.parse_dktrpt(miebfilebits, "mieb", "600286") # mdb Adversary proceedings have slightly different formats, more similar to normal bank, but still different enough to crash parsepacer parties = PP._get_parties_info_from_dkrpt(bank_soups["mdb.541423"], "mdb") self.assertEquals(len(parties), 3) self.assertEquals(parties[0]["name"], "Metamorphix, Inc.") self.assertEquals(parties[0]["type"], "Plaintiff") self.assertEquals(parties[0]["extra_info"], "Metamorphix, Inc.\nAttn: Dr. Edwin Quattlebaum\n8000 Virginia Manor Road\nBeltsville, MD 20705") self.assertEquals(len(parties[0]["attorneys"]), 2) self.assertEquals(parties[0]["attorneys"][0]["attorney_name"], "Peter D. Guattery") self.assertEquals(parties[1]["name"], "Edwin Quattlebaum") self.assertEquals(parties[1]["type"], "Plaintiff") self.assertEquals(parties[2]["name"], "Theresa Brady") self.assertEquals(parties[2]["type"], "Defendant")