def get_money_range(s, logger): res = ["0", "-1"] s = s.strip() if re.match('^.*\+$', s): money_val = themortgagemeter_utils.get_money(s, logger) res[0] = money_val elif re.match('^.*up to.*$', s) or re.match('^.*under.*$', s): money_val = themortgagemeter_utils.get_money(s, logger) res[1] = money_val elif re.match('^.*or more$', s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None, ',') elif re.match('^.* - .*$', s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None, ',') res[1] = s.split()[2][2:].translate(None, ',') elif re.match('^.*[0-9] to .*$', s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None, ',') res[1] = s.split()[2][2:].translate(None, ',') else: #logger.info(t) #logger.info('value not handled: ' + s) themortgagemeter_utils.record_alert( 'ERROR: value not properly parsed by get_money_range: ' + s, logger, themortgagemeter_db.db_connection, themortgagemeter_db.cursor) exit() return res
def get_money_range(s,logger): res = ["0","-1"] s = s.strip() if re.match('^.*\+$',s): money_val = themortgagemeter_utils.get_money(s,logger) res[0] = money_val elif re.match('^.*up to.*$',s) or re.match('^.*under.*$',s): money_val = themortgagemeter_utils.get_money(s,logger) res[1] = money_val elif re.match('^.*or more$',s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None,',') elif re.match('^.* - .*$',s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None,',') res[1] = s.split()[2][2:].translate(None,',') elif re.match('^.*[0-9] to .*$',s): index = s.find("£") s = s[index:] res[0] = s.split()[0][2:].translate(None,',') res[1] = s.split()[2][2:].translate(None,',') else: #logger.info(t) #logger.info('value not handled: ' + s) themortgagemeter_utils.record_alert('ERROR: value not properly parsed by get_money_range: ' + s,logger,themortgagemeter_db.db_connection,themortgagemeter_db.cursor) exit() return res
def get_product_page(static,url): logger = logging.getLogger('retrieve') bsobj = themortgagemeter_utils.get_page(static,'static_html/first_direct/mortgage-rates',url,logger) print bsobj sections = bsobj.find_all(attrs={'class':'section'}) for section in sections: #print section #print "=============================" tbodys = section.find_all("tbody") for tbody in tbodys: print tbody trs = tbody.find_all("tr") for tr in trs: tds = tr.find_all("td") booking_fee_int = 0 count = 0 for td in tds: # assume default of 25 years term = str(25 * 12) td_text = td.text.strip().encode('utf-8') #print count #print td if count == 0: #initial_period initial_period = themortgagemeter_utils.get_months(td_text,logger) #mortgage_type F/D/T/O/V mortgage_type = mc_util.get_mortgage_type(td_text,logger) #eligibility print td_text pass elif count == 1: #ltv_percent ltv_percent = themortgagemeter_utils.get_percentage(td_text,logger) elif count == 2: #rate_percent rate_percent = themortgagemeter_utils.get_percentage(td_text,logger) elif count == 3: #svr_percent svr_percent = themortgagemeter_utils.get_percentage(td_text,logger) elif count == 4: #apr_percent apr_percent = themortgagemeter_utils.get_percentage(td_text,logger) elif count == 5: booking_fee_int = booking_fee_int + int(themortgagemeter_utils.get_money(td_text,logger)) elif count == 6: booking_fee_int = booking_fee_int + int(themortgagemeter_utils.get_money(td_text,logger)) count = count + 1 booking_fee = str(booking_fee_int) mc_util.handle_mortgage_insert(institution_code,mortgage_type,rate_percent,svr_percent,apr_percent,ltv_percent,initial_period,booking_fee,term,url,eligibility,logger)
def get_product_pages(static,base_url,ext_url,logger): url = base_url + ext_url urls_seen = [] bsobj = themortgagemeter_utils.get_page(static,'static_html/halifax/savings-accounts.html',url,logger) # let's see how much info we can extract from the page # Get all the sortable tables and divine as much info as possible from that. sortable_tables = doormatCols = bsobj.find_all(attrs={'class' : 'sortableTable'}) for table in sortable_tables: #print table for tr in table.find_all('tr'): td_idx = 0 savings_data = savings_util.get_savings_data_object() for td in tr.find_all('td'): td_text = td.text.encode('utf-8').strip().lower() if td_idx == 0: # title of account - Junior == child #re.match('/product/A[0-9]+.*',href) ##0 ##<td style="text-align: left;"><a href="/savings/accounts/cash-isas/isa-saver-online/">ISA Saver Online</a></td> ##1 ##<td style="text-align: left;"><strong class="apr">1.35%</strong> tax free/AER variable including 12 month fixed bonus of<strong> </strong>1.10%</td> ##2 ##<td>£1</td> ##3 ##<td>Variable</td> ##4 ##<td>Unlimited</td> ##5 ##<td style="text-align: center;"><img alt="Online" src="/common/images/icons/mousegrey.gif" title="Online"/></td> ##6 ##<td><a href="/savings/accounts/cash-isas/isa-saver-online/"><img alt="Find out more" src="/common/images/Buttons/primary_find_out_more.gif"/></a></td> #print "0: " + td_text if re.match('^.*isa.*$',td_text): savings_data['isa'] = 'Y' if re.match('^.*junior.*$',td_text): savings_data['child'] = 'Y' elif td_idx == 1: #print "1: " + td_text # We don't bother with this at the moment - TODO - sort this out #if re.match('.*bonus.*',td_text): # savings_data['bonus'] = 'Y' pass elif td_idx == 2: #print "2: " + td_text # minimum investment, max is always infinity min_amt = themortgagemeter_utils.get_money(td_text,logger) savings_data['min_amt'] = min_amt elif td_idx == 3: #print "3: " + td_text # Variable/Fixed if re.match('.*variable.*',td_text): savings_data['variability'] = 'V' elif re.match('.*fixed.*',td_text): savings_data['variability'] = 'F' else: themortgagemeter_utils.record_alert('ERROR: unknown variability: ' + td_text,logger,themortgagemeter_db.db_connection,themortgagemeter_db.cursor) exit() elif td_idx == 4: #print "4: " + td_text # Let's assume we'll get this info from the sub-page. # Withdrawals allowed: "None, by closure only", "Unlimited", "None, until child is 18" pass elif td_idx == 5: #print "5: " + td_text for img in td.find_all('img'): title = img.get('title').lower().strip() if title == 'online': savings_data['online'] = 'Y' elif title == 'branch': savings_data['branch'] = 'Y' # I'm going to ignore "phone" elif title == 'phone': pass elif td_idx == 6: #print "6: " + td_text # more details link new_url = base_url + td.find_all('a')[0].get('href') if new_url in urls_seen: continue savings_array = process_more_info_page(savings_data,new_url,logger) print new_url for this_savings_data in savings_array: # insert savings here TODO. print this_savings_data isa = this_savings_data['isa'] regular_saver = this_savings_data['regular_saver'] regular_saver_frequency_period = this_savings_data['regular_saver_frequency_period'] regular_saver_frequency_type = this_savings_data['regular_saver_frequency_type'] regular_saver_min_amt = this_savings_data['regular_saver_min_amt'] regular_saver_max_amt = this_savings_data['regular_saver_max_amt'] bonus = this_savings_data['bonus'] bonus_frequency_period = this_savings_data['bonus_frequency_period'] bonus_frequency_type = this_savings_data['bonus_frequency_type'] online = this_savings_data['online'] branch = this_savings_data['branch'] variability = this_savings_data['variability'] min_amt = this_savings_data['min_amt'] max_amt = this_savings_data['max_amt'] gross_percent = this_savings_data['gross_percent'] aer_percent = this_savings_data['aer_percent'] interest_paid = this_savings_data['interest_paid'] child = this_savings_data['child'] savings_period = this_savings_data['savings_period'] savings_util.handle_savings_insert(institution_code, isa, regular_saver, regular_saver_frequency_period, regular_saver_frequency_type, regular_saver_min_amt, regular_saver_max_amt, bonus, bonus_frequency_period, bonus_frequency_type, online, branch, variability, savings_period, min_amt, max_amt, gross_percent, aer_percent, child, interest_paid, url, logger) urls_seen.insert(0,new_url) else: themortgagemeter_utils.record_alert('ERROR: too many tds in tr: ' + tr,logger,themortgagemeter_db.db_connection,themortgagemeter_db.cursor) exit() td_idx = td_idx + 1
def get_product_pages(static, base_url, ext_url, logger): url = base_url + ext_url urls_seen = [] bsobj = themortgagemeter_utils.get_page( static, 'static_html/halifax/savings-accounts.html', url, logger) # let's see how much info we can extract from the page # Get all the sortable tables and divine as much info as possible from that. sortable_tables = doormatCols = bsobj.find_all( attrs={'class': 'sortableTable'}) for table in sortable_tables: #print table for tr in table.find_all('tr'): td_idx = 0 savings_data = savings_util.get_savings_data_object() for td in tr.find_all('td'): td_text = td.text.encode('utf-8').strip().lower() if td_idx == 0: # title of account - Junior == child #re.match('/product/A[0-9]+.*',href) ##0 ##<td style="text-align: left;"><a href="/savings/accounts/cash-isas/isa-saver-online/">ISA Saver Online</a></td> ##1 ##<td style="text-align: left;"><strong class="apr">1.35%</strong> tax free/AER variable including 12 month fixed bonus of<strong> </strong>1.10%</td> ##2 ##<td>£1</td> ##3 ##<td>Variable</td> ##4 ##<td>Unlimited</td> ##5 ##<td style="text-align: center;"><img alt="Online" src="/common/images/icons/mousegrey.gif" title="Online"/></td> ##6 ##<td><a href="/savings/accounts/cash-isas/isa-saver-online/"><img alt="Find out more" src="/common/images/Buttons/primary_find_out_more.gif"/></a></td> #print "0: " + td_text if re.match('^.*isa.*$', td_text): savings_data['isa'] = 'Y' if re.match('^.*junior.*$', td_text): savings_data['child'] = 'Y' elif td_idx == 1: #print "1: " + td_text # We don't bother with this at the moment - TODO - sort this out #if re.match('.*bonus.*',td_text): # savings_data['bonus'] = 'Y' pass elif td_idx == 2: #print "2: " + td_text # minimum investment, max is always infinity min_amt = themortgagemeter_utils.get_money(td_text, logger) savings_data['min_amt'] = min_amt elif td_idx == 3: #print "3: " + td_text # Variable/Fixed if re.match('.*variable.*', td_text): savings_data['variability'] = 'V' elif re.match('.*fixed.*', td_text): savings_data['variability'] = 'F' else: themortgagemeter_utils.record_alert( 'ERROR: unknown variability: ' + td_text, logger, themortgagemeter_db.db_connection, themortgagemeter_db.cursor) exit() elif td_idx == 4: #print "4: " + td_text # Let's assume we'll get this info from the sub-page. # Withdrawals allowed: "None, by closure only", "Unlimited", "None, until child is 18" pass elif td_idx == 5: #print "5: " + td_text for img in td.find_all('img'): title = img.get('title').lower().strip() if title == 'online': savings_data['online'] = 'Y' elif title == 'branch': savings_data['branch'] = 'Y' # I'm going to ignore "phone" elif title == 'phone': pass elif td_idx == 6: #print "6: " + td_text # more details link new_url = base_url + td.find_all('a')[0].get('href') if new_url in urls_seen: continue savings_array = process_more_info_page( savings_data, new_url, logger) print new_url for this_savings_data in savings_array: # insert savings here TODO. print this_savings_data isa = this_savings_data['isa'] regular_saver = this_savings_data['regular_saver'] regular_saver_frequency_period = this_savings_data[ 'regular_saver_frequency_period'] regular_saver_frequency_type = this_savings_data[ 'regular_saver_frequency_type'] regular_saver_min_amt = this_savings_data[ 'regular_saver_min_amt'] regular_saver_max_amt = this_savings_data[ 'regular_saver_max_amt'] bonus = this_savings_data['bonus'] bonus_frequency_period = this_savings_data[ 'bonus_frequency_period'] bonus_frequency_type = this_savings_data[ 'bonus_frequency_type'] online = this_savings_data['online'] branch = this_savings_data['branch'] variability = this_savings_data['variability'] min_amt = this_savings_data['min_amt'] max_amt = this_savings_data['max_amt'] gross_percent = this_savings_data['gross_percent'] aer_percent = this_savings_data['aer_percent'] interest_paid = this_savings_data['interest_paid'] child = this_savings_data['child'] savings_period = this_savings_data['savings_period'] savings_util.handle_savings_insert( institution_code, isa, regular_saver, regular_saver_frequency_period, regular_saver_frequency_type, regular_saver_min_amt, regular_saver_max_amt, bonus, bonus_frequency_period, bonus_frequency_type, online, branch, variability, savings_period, min_amt, max_amt, gross_percent, aer_percent, child, interest_paid, url, logger) urls_seen.insert(0, new_url) else: themortgagemeter_utils.record_alert( 'ERROR: too many tds in tr: ' + tr, logger, themortgagemeter_db.db_connection, themortgagemeter_db.cursor) exit() td_idx = td_idx + 1