def printlastmovements(num, html): print '\n\nNúmero de compte: "%s"\n' % num soup = BeautifulSoup(html, convertEntities=BeautifulSoup.HTML_ENTITIES) table = soup.find('table', {'class': 'sortable'}) rows = [] for tr in table.findAll('tr'): row = [] for x in tr.findAll(['th', 'td']): text = x.text.encode('utf8').strip() row.append(text) rows.append(row) print indent(rows, True)
def parse(self, response, pprint): try: if pprint: reader = csv.reader(response) return indent([x for x in reader], hasHeader=True) else: return csv.DictReader(response) except Exception, e: print e raise Exception(e)
def posicioglobal(html): soup = BeautifulSoup(html, convertEntities=BeautifulSoup.HTML_ENTITIES) accountnums = [] for table in soup.findAll('table', {'class': 'TaulaPosGlb'}): print table.caption.span.text.strip().encode('utf8') print rows = [] for tr in table.findAll('tr'): row = [] for i, x in enumerate(tr.findAll(['th', 'td'])): text = x.text.encode('utf8').strip() if i != 3: row.append(text) else: # Extract account numbers if re.search(r'expandeix.+LLI', text): m = re.search(r'\((.+)\)', text) accountnums.append(m.group(0).strip().strip('()').split(',')[0].strip().strip('\'')) rows.append(row) print indent(rows, True) print return accountnums
def posicioglobal(html): soup = BeautifulSoup(html, convertEntities=BeautifulSoup.HTML_ENTITIES) accountnums = [] for table in soup.findAll('table', {'class': 'TaulaPosGlb'}): print table.caption.span.text.strip().encode('utf8') print rows = [] for tr in table.findAll('tr'): row = [] for i, x in enumerate(tr.findAll(['th', 'td'])): text = x.text.encode('utf8').strip() if i != 3: row.append(text) else: # Extract account numbers if re.search(r'expandeix.+LLI', text): m = re.search(r'\((.+)\)', text) accountnums.append( m.group(0).strip().strip('()').split(',') [0].strip().strip('\'')) rows.append(row) print indent(rows, True) print return accountnums