def __init__(self, _id, content=None, **kwargs):
self._id = str(_id)
self.paper = Paper()
page_data = col_paper.find_one({"url_id": self._id})
if page_data:
# 数据库中已经存在，直接返回
return
if content is None:
self.content = Downloader(host + self._id)()
if self.content:
self.valid = True
else:
logger.error("当前网页为空，无法进行解析\turl_id:" + self._id)
self.valid = False
return
else:
self.valid = True
self.content = content
self.selector = etree.HTML(self.content)
self.paper.url_id = self._id
def get_paper(self, paper_url=None, paper_id=None):
"""
Load paper details for the paper given by detail page URL
or numeric ID
"""
paper_url = '%svo020.asp?VOLFDNR=%s' % (self.config.BASE_URL, paper_id)
logging.info("Getting paper %d from %s", paper_id, paper_url)
# stupid re-try concept because AllRis sometimes misses start < at tags at first request
try_counter = 0
while True:
try:
response = self.get_url(paper_url)
if not response:
return
if "noauth" in response.url:
logging.warn("Paper %s in %s seems to private" % (paper_id, paper_url))
return
text = self.preprocess_text(response.text)
doc = html.fromstring(text)
data = {}
# Beratungsfolge-Table checken
table = self.table_css(doc)[0] # lets hope we always have this table
self.consultation_list_start = False
last_headline = ''
for line in table:
if line.tag == 'tr':
headline = line[0].text
elif line.tag == 'td':
headline = line.text
else:
logging.error("ERROR: Serious error in data table. Unable to parse.")
if headline:
headline = headline.split(":")[0].lower()
if headline[-1]==":":
headline = headline[:-1]
if headline == "betreff":
value = line[1].text_content().strip()
value = value.split("-->")[1] # there is some html comment with a script tag in front of the text which we remove
data[headline] = " ".join(value.split()) # remove all multiple spaces from the string
elif headline in ['verfasser', u'federführend', 'drucksache-art']:
data[headline] = line[1].text.strip()
elif headline in ['status']:
data[headline] = line[1].text.strip()
# related papers
if len(line) > 2:
if len(line[3]):
# gets identifier. is there something else at this position? (will break)
data['paper'] = [{'paper': Paper(numeric_id=line[3][0][0][1][0].get('href').split('=')[1].split('&')[0] , identifier=line[3][0][0][1][0].text)}]
elif headline == "beratungsfolge":
# the actual list will be in the next row inside a table, so we only set a marker
data = self.parse_consultation_list_headline(line, data) # for parser which have the consultation list here
elif self.consultation_list_start:
data = self.parse_consultation_list(line, data) # for parser which have the consultation list in the next tr
self.consultation_list_start = False # set the marker to False again as we have read it
last_headline = headline
# we simply ignore the rest (there might not be much more actually)
# the actual text comes after the table in a div but it's not valid XML or HTML this using regex
data['docs'] = self.body_re.findall(response.text)
first_date = False
for single_consultation in data['consultation']:
if first_date:
if single_consultation['date'] < first_date:
first_date = single_consultation['date']
else:
first_date = single_consultation['date']
paper = Paper(numeric_id = paper_id)
paper.original_url = paper_url
paper.title = data['betreff']
paper.description = data['docs']
paper.type = data['drucksache-art']
if first_date:
paper.date = first_date.strftime("%Y-%m-%d")
if 'identifier' in data:
paper.identifier = data['identifier']
paper.document = []
# get the attachments step 1 (Drucksache)
document_1 = self.attachment_1_css(doc)
if len(document_1):
if document_1[0].value:
href = '%sdo027.asp' % self.config.BASE_URL
identifier = document_1[0].value
title = 'Drucksache'
document = Document(
identifier=identifier,
numeric_id=int(identifier),
title=title)
document = self.get_document_file(document, href, True)
paper.document.append({'document': document, 'relation': 'misc'})
# get the attachments step 2 (additional attachments)
documents = self.attachments_css(doc)
if len(documents) > 0:
if len(documents[0]) > 1:
if documents[0][1][0].text.strip() == "Anlagen:":
for tr in documents[0][2:]:
link = tr[0][0]
href = "%s%s" % (self.config.BASE_URL, link.attrib["href"])
title = link.text
identifier = str(int(link.attrib["href"].split('/')[4]))
document = Document(
identifier=identifier,
numeric_id=int(identifier),
title=title)
document = self.get_document_file(document, href)
paper.document.append({'document': document, 'relation': 'misc'})
oid = self.db.save_paper(paper)
return
except (KeyError, IndexError):
if try_counter < 3:
logging.info("Try again: Getting paper %d from %s", paper_id, paper_url)
try_counter += 1
else:
logging.error("Failed getting paper %d from %s", paper_id, paper_url)
return
def get_paper(self, paper_url=None, paper_id=None):
"""
Load paper details for the paper given by detail page URL
or numeric ID
"""
# Read either paper_id or paper_url from the opposite
if paper_id is not None:
paper_url = self.urls['SUBMISSION_DETAIL_PRINT_PATTERN'] % paper_id
elif paper_url is not None:
parsed = parse.search(self.urls['SUBMISSION_DETAIL_PARSE_PATTERN'], paper_url)
paper_id = parsed['paper_id']
logging.info("Getting paper %d from %s", paper_id, paper_url)
paper = Paper(numeric_id=paper_id)
try_until = 1
try_counter = 0
try_found = False
while (try_counter < try_until):
try_counter += 1
try_found = False
time.sleep(self.config.WAIT_TIME)
try:
response = self.user_agent.open(paper_url)
except urllib2.HTTPError, e:
if e.code == 404:
sys.stderr.write("URL not found (HTTP 404) error caught: %s\n" % paper_url)
sys.stderr.write("Please check BASE_URL in your configuration.\n")
sys.exit(1)
elif e.code == 502:
try_until = 4
try_found = True
if try_until == try_counter:
logging.error("Permanent error in %s after 4 retrys.", paper_url)
return
else:
logging.info("Original RIS Server Bug, restart scraping paper %s", paper_url)
mechanize_forms = mechanize.ParseResponse(response, backwards_compat=False)
response.seek(0)
html = response.read()
html = html.replace(' ', ' ')
parser = etree.HTMLParser()
dom = etree.parse(StringIO(html), parser)
# Hole die Seite noch einmal wenn unbekannter zufällig auftretender Fehler ohne Fehlermeldung ausgegeben wird (gefunden in Duisburg, vermutlich kaputte Server Config)
try:
page_title = dom.xpath('//h1')[0].text
if 'Fehler' in page_title:
try_until = 4
try_found = True
if try_until == try_counter:
logging.error("Permanent error in %s after 3 retrys, proceed.", paper_url)
else:
logging.info("Original RIS Server Bug, restart scraping paper %s", paper_url)
except:
pass
if (try_found == False):
# check for page errors
try:
if 'Fehlermeldung' in page_title:
logging.info("Page %s cannot be accessed due to server error", paper_url)
return
if 'Berechtigungsfehler' in page_title:
logging.info("Page %s cannot be accessed due to permissions", paper_url)
return
except:
pass
paper.original_url = paper_url
paper_related = []
# Paper title
try:
stitle = dom.xpath(self.xpath['SUBMISSION_DETAIL_TITLE'])
paper.title = stitle[0].text
except:
logging.critical('Cannot find paper title element using XPath SUBMISSION_DETAIL_TITLE')
raise TemplateError('Cannot find paper title element using XPath SUBMISSION_DETAIL_TITLE')
# Submission identifier, date, type etc
tds = dom.xpath(self.xpath['SUBMISSION_DETAIL_IDENTIFIER_TD'])
if len(tds) == 0:
logging.critical('Cannot find table fields using XPath SUBMISSION_DETAIL_IDENTIFIER_TD')
logging.critical('HTML Dump:' + html)
raise TemplateError('Cannot find table fields using XPath SUBMISSION_DETAIL_IDENTIFIER_TD')
else:
current_category = None
for n in range(0, len(tds)):
try:
tdcontent = tds[n].text.strip()
except:
continue
if tdcontent == 'Name:':
paper.identifier = tds[n + 1].text.strip()
elif tdcontent == 'Art:':
paper.type = tds[n + 1].text.strip()
elif tdcontent == 'Datum:':
paper.date = tds[n + 1].text.strip()
elif tdcontent == 'Name:':
paper.identifier = tds[n + 1].text.strip()
elif tdcontent == 'Betreff:':
paper.subject = '; '.join(tds[n + 1].xpath('./text()'))
elif tdcontent == 'Aktenzeichen:':
paper.reference_number = tds[n + 1].text.strip()
elif tdcontent == 'Referenzvorlage:':
link = tds[n + 1].xpath('a')[0]
href = link.get('href')
parsed = parse.search(self.urls['SUBMISSION_DETAIL_PARSE_PATTERN'], href)
superordinated_paper = Paper(numeric_id=parsed['paper_id'], identifier=link.text.strip())
paper_related.append({ 'relation': 'superordinated',
'paper': superordinated_paper})
# add superordinate paper to queue
if hasattr(self, 'paper_queue'):
self.paper_queue.add(parsed['paper_id'])
# subordinate papers are added to the queue
elif tdcontent == 'Untergeordnete Vorlage(n):':
current_category = 'subordinates'
for link in tds[n + 1].xpath('a'):
href = link.get('href')
parsed = parse.search(self.urls['SUBMISSION_DETAIL_PARSE_PATTERN'], href)
subordinated_paper = Paper(numeric_id=parsed['paper_id'], identifier=link.text.strip())
paper_related.append({ 'relation': 'subordinated',
'paper': subordinated_paper})
if hasattr(self, 'paper_queue') and parsed is not None:
# add subordinate paper to queue
self.paper_queue.add(parsed['paper_id'])
else:
if current_category == 'subordinates' and len(tds) > n+1:
for link in tds[n + 1].xpath('a'):
href = link.get('href')
parsed = parse.search(self.urls['SUBMISSION_DETAIL_PARSE_PATTERN'], href)
subordinated_paper = Paper(numeric_id=parsed['paper_id'], identifier=link.text.strip())
paper_related.append({ 'relation': 'subordinated',
'paper': subordinated_paper})
if hasattr(self, 'paper_queue') and parsed is not None:
self.paper_queue.add(parsed['paper_id'])
if len(paper_related):
paper.paper = paper_related
if not hasattr(paper, 'identifier'):
logging.critical('Cannot find paper identifier using SESSION_DETAIL_IDENTIFIER_TD_XPATH')
raise TemplateError('Cannot find paper identifier using SESSION_DETAIL_IDENTIFIER_TD_XPATH')
# "Beratungsfolge"(list of sessions for this paper)
# This is currently not parsed for scraping, but only for
# gathering session-document ids for later exclusion
found_documents = []
rows = dom.xpath(self.xpath['SUBMISSION_DETAIL_AGENDA_ROWS'])
for row in rows:
# find forms
formfields = row.xpath('.//input[@type="hidden"][@name="DT"]')
for formfield in formfields:
document_id = formfield.get('value')
if document_id is not None:
found_documents.append(document_id)
# find links
links = row.xpath('.//a[contains(@href,"getfile.")]')
for link in links:
if not link.xpath('.//img'):
file_link = self.config.BASE_URL + link.get('href')
document_id = file_link.split('id=')[1].split('&')[0]
found_documents.append(document_id)
# paper-related documents
documents = []
paper.document = []
containers = dom.xpath(self.xpath['SUBMISSION_DETAIL_ATTACHMENTS'])
for container in containers:
try:
classes = container.get('class').split(' ')
except:
continue
if self.xpath['SUBMISSION_DETAIL_ATTACHMENTS_CONTAINER_CLASSNAME'] not in classes:
continue
rows = container.xpath('.//tr')
for row in rows:
# seems that we have direct links
if not row.xpath('.//form'):
links = row.xpath('.//a')
for link in links:
# ignore additional pdf icon links
if not link.xpath('.//img'):
title = ' '.join(link.xpath('./text()')).strip()
file_link = self.config.BASE_URL + link.get('href')
document_id = file_link.split('id=')[1].split('&')[0]
if document_id in found_documents:
continue
document = Document(
identifier=document_id,
numeric_id=document_id,
title=title,
original_url=file_link)
document = self.get_document_file(document=document, link=file_link)
if 'Einladung' in title:
document_type = 'invitation'
elif 'Niederschrift' in title:
document_type = 'results_protocol'
else:
document_type = 'misc'
paper.document.append({'relation': document_type, 'document': document})
found_documents.append(document_id)
# no direct link, so we have to handle forms
else:
forms = row.xpath('.//form')
for form in forms:
title = " ".join(row.xpath('./td/text()')).strip()
for hidden_field in form.xpath('input[@name="DT"]'):
document_id = hidden_field.get('value')
if document_id in found_documents:
continue
document = Document(
identifier=document_id,
numeric_id=document_id,
title=title)
# Traversing the whole mechanize response to submit this form
for mform in mechanize_forms:
for control in mform.controls:
if control.name == 'DT' and control.value == document_id:
document = self.get_document_file(document=document, form=mform)
if 'Einladung' in title:
document_type = 'invitation'
elif 'Niederschrift' in title:
document_type = 'results_protocol'
else:
document_type = 'misc'
paper.document.append({'relation': document_type, 'document': document})
found_documents.append(document_id)
if len(documents):
paper.document = documents
# forcing overwrite=True here
oid = self.db.save_paper(paper)
def get_paper(self, paper_url=None, paper_id=None):
"""
Load paper details for the paper given by detail page URL
or numeric ID
"""
# Read either paper_id or paper_url from the opposite
if paper_id is not None:
paper_url = self.urls['PAPER_DETAIL_PRINT_PATTERN'] % (self.config["scraper"]["base_url"], paper_id)
elif paper_url is not None:
parsed = parse.search(self.urls['PAPER_DETAIL_PARSE_PATTERN'], paper_url)
paper_id = parsed['paper_id']
logging.info("Getting paper %d from %s", paper_id, paper_url)
paper = Paper(originalId=paper_id)
try_until = 1
try_counter = 0
try_found = False
while (try_counter < try_until):
try_counter += 1
try_found = False
time.sleep(self.config['scraper']['wait_time'])
try:
response = self.user_agent.open(paper_url)
except urllib2.HTTPError, e:
if e.code == 404:
sys.stderr.write("URL not found (HTTP 404) error caught: %s\n" % paper_url)
sys.stderr.write("Please check BASE_URL in your configuration.\n")
sys.exit(1)
elif e.code == 502 or e.code == 500:
try_until = 4
try_found = True
if try_until == try_counter:
logging.error("Permanent error in %s after 4 retrys.", paper_url)
return
else:
logging.info("Original RIS Server Bug, restart fetching paper %s", paper_url)
if not response:
return
mechanize_forms = mechanize.ParseResponse(response, backwards_compat=False)
response.seek(0)
html = response.read()
html = html.replace(' ', ' ')
parser = etree.HTMLParser()
dom = etree.parse(StringIO(html), parser)
# Hole die Seite noch einmal wenn unbekannter zufällig auftretender Fehler ohne Fehlermeldung ausgegeben wird (gefunden in Duisburg, vermutlich kaputte Server Config)
try:
page_title = dom.xpath('//h1')[0].text
if 'Fehler' in page_title:
try_until = 4
try_found = True
if try_until == try_counter:
logging.error("Permanent error in %s after 3 retrys, proceed.", paper_url)
else:
logging.info("Original RIS Server Bug, restart scraping paper %s", paper_url)
except:
pass
if (try_found == False):
# check for page errors
try:
if 'Fehlermeldung' in page_title:
logging.info("Page %s cannot be accessed due to server error", paper_url)
return
if 'Berechtigungsfehler' in page_title:
logging.info("Page %s cannot be accessed due to permissions", paper_url)
return
except:
pass
paper.originalUrl = paper_url
superordinated_papers = []
subordinated_papers = []
# Paper title
try:
stitle = dom.xpath(self.xpath['PAPER_DETAIL_TITLE'])
paper.title = stitle[0].text
except:
logging.critical('Cannot find paper title element using XPath PAPER_DETAIL_TITLE')
raise TemplateError('Cannot find paper title element using XPath PAPER_DETAIL_TITLE')
# Paper identifier, date, type etc
tds = dom.xpath(self.xpath['PAPER_DETAIL_IDENTIFIER_TD'])
if len(tds) == 0:
logging.critical('Cannot find table fields using XPath PAPER_DETAIL_IDENTIFIER_TD')
logging.critical('HTML Dump:' + html)
raise TemplateError('Cannot find table fields using XPath PAPER_DETAIL_IDENTIFIER_TD')
else:
current_category = None
for n in range(0, len(tds)):
try:
tdcontent = tds[n].text.strip()
except:
continue
if tdcontent == 'Name:':
paper.nameShort = tds[n + 1].text.strip()
# TODO: Dereferenzierung von Paper Type Strings
elif tdcontent == 'Art:':
paper.paperType = tds[n + 1].text.strip()
elif tdcontent == 'Datum:':
paper.publishedDate = tds[n + 1].text.strip()
elif tdcontent == 'Betreff:':
paper.name = '; '.join(tds[n + 1].xpath('./text()'))
elif tdcontent == 'Aktenzeichen:':
paper.reference = tds[n + 1].text.strip()
elif tdcontent == 'Referenzvorlage:':
link = tds[n + 1].xpath('a')[0]
href = link.get('href')
parsed = parse.search(self.urls['PAPER_DETAIL_PARSE_PATTERN'], href)
superordinated_paper = Paper(originalId=parsed['paper_id'], nameShort=link.text.strip())
superordinated_papers.append(superordinated_paper)
# add superordinate paper to queue
if hasattr(self, 'paper_queue'):
self.paper_queue.add(parsed['paper_id'])
# subordinate papers are added to the queue
elif tdcontent == 'Untergeordnete Vorlage(n):':
current_category = 'subordinates'
for link in tds[n + 1].xpath('a'):
href = link.get('href')
parsed = parse.search(self.urls['PAPER_DETAIL_PARSE_PATTERN'], href)
subordinated_paper = Paper(originalId=parsed['paper_id'], nameShort=link.text.strip())
subordinated_papers.append(subordinated_paper)
if hasattr(self, 'paper_queue') and parsed is not None:
# add subordinate paper to queue
self.paper_queue.add(parsed['paper_id'])
elif tdcontent == u'Anträge zur Vorlage:':
current_category = 'todo'
pass #TODO: WTF is this?
else:
if current_category == 'subordinates' and len(tds) > n+1:
for link in tds[n + 1].xpath('a'):
href = link.get('href')
parsed = parse.search(self.urls['PAPER_DETAIL_PARSE_PATTERN'], href)
subordinated_paper = Paper(originalId=parsed['paper_id'], nameShort=link.text.strip())
subordinated_papers.append(subordinated_paper)
if hasattr(self, 'paper_queue') and parsed is not None:
self.paper_queue.add(parsed['paper_id'])
if len(subordinated_papers):
paper.subordinatedPaper = subordinated_papers
if len(superordinated_papers):
paper.superordinatedPaper = superordinated_papers
if not hasattr(paper, 'originalId'):
logging.critical('Cannot find paper identifier using MEETING_DETAIL_IDENTIFIER_TD')
raise TemplateError('Cannot find paper identifier using MEETING_DETAIL_IDENTIFIER_TD')
# "Beratungsfolge"(list of sessions for this paper)
# This is currently not parsed for scraping, but only for
# gathering session-document ids for later exclusion
found_files = [] #already changed: found_files, files. todo: document_foo
rows = dom.xpath(self.xpath['PAPER_DETAIL_AGENDA_ROWS'])
for row in rows:
# find forms
formfields = row.xpath('.//input[@type="hidden"][@name="DT"]')
for formfield in formfields:
file_id = formfield.get('value')
if file_id is not None:
found_files.append(file_id)
# find links
links = row.xpath('.//a[contains(@href,"getfile.")]')
for link in links:
if not link.xpath('.//img'):
file_link = self.config['scraper']['base_url'] + link.get('href')
file_id = file_link.split('id=')[1].split('&')[0]
found_files.append(file_id)
# paper-related documents
files = []
containers = dom.xpath(self.xpath['PAPER_DETAIL_FILES'])
for container in containers:
try:
classes = container.get('class').split(' ')
except:
continue
if self.xpath['PAPER_DETAIL_FILES_CONTAINER_CLASSNAME'] not in classes:
continue
rows = container.xpath('.//tr')
for row in rows:
# seems that we have direct links
if not row.xpath('.//form'):
links = row.xpath('.//a')
for link in links:
# ignore additional pdf icon links
if not link.xpath('.//img'):
name = ' '.join(link.xpath('./text()')).strip()
file_link = self.config['scraper']['base_url'] + link.get('href')
file_id = file_link.split('id=')[1].split('&')[0]
if file_id in found_files:
continue
file = File(
originalId=file_id,
name=name,
originalUrl=file_link,
originalDownloadPossible = True)
file = self.get_file(file=file, link=file_link)
files.append(file)
found_files.append(file_id)
# no direct link, so we have to handle forms
else:
forms = row.xpath('.//form')
for form in forms:
name = " ".join(row.xpath('./td/text()')).strip()
for hidden_field in form.xpath('input[@name="DT"]'):
file_id = hidden_field.get('value')
if file_id in found_files:
continue
file = File(
originalId=file_id,
name=name,
originalDownloadPossible = False)
# Traversing the whole mechanize response to submit this form
for mform in mechanize_forms:
for control in mform.controls:
if control.name == 'DT' and control.value == file_id:
file = self.get_file(file=file, form=mform)
files.append(file)
found_files.append(file_id)
if len(files):
paper.mainFile = files[0]
if len(files) > 1:
paper.auxiliaryFile = files[1:]
oid = self.db.save_paper(paper)
def get_paper(self, paper_url=None, paper_id=None):
"""
Load paper details for the paper given by detail page URL
or numeric ID
"""
paper_url = ('%svo020.asp?VOLFDNR=%s' %
(self.config['scraper']['base_url'], paper_id))
logging.info("Getting paper %d from %s", paper_id, paper_url)
# Stupid re-try concept because AllRis sometimes misses
# start < at tags at first request.
try_counter = 0
while True:
try:
response = self.get_url(paper_url)
if not response:
return
if "noauth" in response.url:
logging.warn("Paper %s in %s seems to private", paper_id,
paper_url)
return
text = response.text
doc = html.fromstring(text)
data = {}
# Beratungsfolge-Table checken
# lets hope we always have this table
table = self.table_css(doc)[0]
self.consultation_list_start = False
last_headline = ''
for line in table:
if line.tag == 'tr':
headline = line[0].text
elif line.tag == 'td':
headline = line.text
else:
logging.error("ERROR: Serious error in data table. "
"Unable to parse.")
if headline:
headline = headline.split(":")[0].lower()
if headline[-1] == ":":
headline = headline[:-1]
if headline == "betreff":
value = line[1].text_content().strip()
# There is some html comment with a script
# tag in front of the text which we remove.
value = value.split("-->")[1]
# remove all multiple spaces from the string
data[headline] = " ".join(value.split())
elif headline in [
'verfasser', u'federführend', 'drucksache-art'
]:
data[headline] = line[1].text.strip()
elif headline in ['status']:
data[headline] = line[1].text.strip()
# related papers
if len(line) > 2:
if len(line[3]):
# Gets originalId. is there something
# else at this position? (will break)
paper_id = line[3][0][0][1][0].get(
'href').split('=')[1].split('&')[0]
data['relatedPaper'] = [
Paper(originalId=paper_id)
]
# Lot's of scraping just because of the date (?)
elif headline == "beratungsfolge":
# The actual list will be in the next row
# inside a table, so we only set a marker.
self.consultation_list_start = True
elif self.consultation_list_start:
elem = line[0][0]
# The first line is pixel images, so skip
# it, then we need to jump in steps of two.
amount = (len(elem) - 1) / 2
consultations = []
date_list = []
i = 0
item = None
for elem_line in elem:
if i == 0:
i += 1
continue
"""
Here we need to parse the actual list which can have different forms. A complex example
can be found at http://ratsinfo.aachen.de/bi/vo020.asp?VOLFDNR=10822
The first line is some sort of headline with the committee in question and the type of consultation.
After that 0-n lines of detailed information of meetings with a date, transscript and decision.
The first line has 3 columns (thanks to colspan) and the others have 7.
Here we make every meeting a separate entry, we can group them together later again if we want to.
"""
# now we need to parse the actual list
# those lists
new_consultation = Consultation()
new_consultation.status = \
elem_line[0].attrib['title'].lower()
if len(elem_line) == 3:
# The order is "color/status", name of
# committee / link to TOP, more info we
# define a head dict here which can be
# shared for the other lines once we find
# another head line we will create a new
# one here.
new_consultation.role = \
elem_line[2].text.strip()
# Name of committee, e.g.
# "Finanzausschuss", unfort. without id
#'committee' : elem_line[1].text.strip(),
# For some obscure reasons sometimes action
# is missing.
elif len(elem_line) == 2:
# The order is "color/status", name of
# committee / link to TOP, more info.
status = \
elem_line[0].attrib['title'].lower()
# We define a head dict here which can be
# shared for the other lines once we find
# another head line we will create a new
# one here.
# name of committee, e.g.
# "Finanzausschuss", unfort. without id
#'committee' : elem_line[1].text.strip(),
elif len(elem_line) == 7:
try:
# This is about line 2 with lots of
# more stuff to process.
# Date can be text or a link with that
# text.
# We have a link (and ignore it).
if len(elem_line[1]) == 1:
date_text = elem_line[1][0].text
else:
date_text = elem_line[1].text
date_list.append(
datetime.datetime.strptime(
date_text.strip(), "%d.%m.%Y"))
if len(elem_line[2]):
# Form with silfdnr and toplfdnr
# but only in link (action=
# "to010.asp?topSelected=57023")
form = elem_line[2][0]
meeting_id = form[0].attrib[
'value']
new_consultation.meeting = [
Meeting(originalId=meeting_id)
]
# Full name of meeting, e.g.
# "A/31/WP.16 öffentliche/
# nichtöffentliche Sitzung des
# Finanzausschusses"
#item['meeting'] = \
# elem_line[3][0].text.strip()
else:
# No link to TOP. Should not be
# possible but happens.
# (TODO: Bugreport?)
# Here we have no link but the text
# is in the TD directly - will be
# scaped as meeting.
#item['meeting'] = \
# elem_line[3].text.strip()
logging.warn(
"AgendaItem in consultation "
"list on the web page does not "
"contain a link to the actual "
"meeting at paper %s",
paper_url)
toplfdnr = None
if len(elem_line[6]) > 0:
form = elem_line[6][0]
toplfdnr = form[0].attrib['value']
if toplfdnr:
new_consultation.originalId = \
"%s-%s" % (toplfdnr,
paper_id)
# actually the id of the transcript
new_consultation.agendaItem = \
AgendaItem(
originalId=toplfdnr)
# e.g. "ungeändert beschlossen"
new_consultation.agendaItem.result \
= elem_line[4].text.strip()
consultations.append(
new_consultation)
else:
logging.error(
"missing agendaItem ID in "
"consultation list at %s",
paper_url)
except (IndexError, KeyError):
logging.error(
"ERROR: Serious error in "
"consultation list. Unable to "
"parse.")
logging.error(
"Serious error in consultation "
"list. Unable to parse.")
return []
i += 1
# Theory: we don't need this at all, because it's
# scraped at meeting.
#data['consultations'] = consultations
# set the marker to False again as we have read it
self.consultation_list_start = False
last_headline = headline
# We simply ignore the rest (there might not be much more
# actually).
# The actual text comes after the table in a div but it's not
# valid XML or HTML this using regex.
data['docs'] = self.body_re.findall(response.text)
first_date = False
for single_date in date_list:
if first_date:
if single_date < first_date:
first_date = single_date
else:
first_date = single_date
paper = Paper(originalId=paper_id)
paper.originalUrl = paper_url
paper.name = data['betreff']
paper.description = data['docs']
if 'drucksache-art' in data:
paper.paperType = data['drucksache-art']
if first_date:
paper.publishedDate = first_date.strftime("%d.%m.%Y")
# see theory above
#if 'consultations' in data:
# paper.consultation = data['consultations']
paper.auxiliaryFile = []
# get the attachments step 1 (Drucksache)
file_1 = self.attachment_1_css(doc)
if len(file_1):
if file_1[0].value:
href = ('%sdo027.asp' %
self.config['scraper']['base_url'])
original_id = file_1[0].value
name = 'Drucksache'
main_file = File(originalId=original_id, name=name)
main_file = self.get_file(main_file, href, True)
paper.mainFile = main_file
# get the attachments step 2 (additional attachments)
files = self.attachments_css(doc)
if len(files) > 0:
if len(files[0]) > 1:
if files[0][1][0].text.strip() == "Anlagen:":
for tr in files[0][2:]:
link = tr[0][0]
href = ("%s%s" %
(self.config['scraper']['base_url'],
link.attrib["href"]))
name = link.text
path_tokens = link.attrib["href"].split('/')
original_id = "%d-%d" % (int(
path_tokens[4]), int(path_tokens[6]))
aux_file = File(originalId=original_id,
name=name)
aux_file = self.get_file(aux_file, href)
paper.auxiliaryFile.append(aux_file)
print paper.auxiliaryFile
if not len(paper.auxiliaryFile):
del paper.auxiliaryFile
oid = self.db.save_paper(paper)
return
except (KeyError, IndexError):
if try_counter < 3:
logging.info("Try again: Getting paper %d from %s",
paper_id, paper_url)
try_counter += 1
else:
logging.error("Failed getting paper %d from %s", paper_id,
paper_url)
return
def get_meeting(self, meeting_url=None, meeting_id=None):
""" Load meeting details (e.g. agendaitems) for the given detail page
URL or numeric ID
"""
meeting_url = ("%sto010.asp?selfaction=ws&template=xyz&SILFDNR=%s" %
(self.config['scraper']['base_url'], meeting_id))
logging.info("Getting meeting %d from %s", meeting_id, meeting_url)
r = self.get_url(meeting_url)
if not r:
return
# If r.history has an item we have a problem
if len(r.history):
if r.history[0].status_code == 302:
logging.info("Meeting %d from %s seems to be private",
meeting_id, meeting_id)
else:
logging.error(
"Strange redirect %d from %s with status code %s",
meeting_id, meeting_url, r.history[0].status_code)
return
h = HTMLParser.HTMLParser()
xml = str(r.text.encode('ascii', 'xmlcharrefreplace'))
parser = etree.XMLParser(recover=True)
root = etree.fromstring(xml, parser=parser)
meeting = Meeting(originalId=meeting_id)
# special area
special = {}
for item in root[0].iterchildren():
special[item.tag] = item.text
# Woher kriegen wir das Datum? Nur über die Übersicht?
#if 'sisb' in special:
#if 'sise' in special:
if 'saname' in special:
meeting.type = special['saname']
# head area
head = {}
for item in root[1].iterchildren():
if item.text:
head[item.tag] = h.unescape(item.text)
else:
head[item.text] = ''
if 'sitext' in head:
meeting.name = head['sitext']
if 'raname' in head:
meeting.room = head['raname']
if 'raort' in head:
meeting.address = head['raort']
agendaitems = []
for item in root[2].iterchildren():
elem = {}
for e in item.iterchildren():
elem[e.tag] = e.text
section = [elem['tofnum'], elem['tofunum'], elem['tofuunum']]
section = [x for x in section if x != "0"]
elem['section'] = ".".join(section)
agendaitem = AgendaItem()
agendaitem.originalId = int(elem['tolfdnr'])
agendaitem.public = (elem['toostLang'] == u'öffentlich')
#agendaitem.name = elem['totext1']
# get agenda detail page
# TODO: Own Queue
time.sleep(self.config['scraper']['wait_time'])
agendaitem_url = (
'%sto020.asp?selfaction=ws&template=xyz&TOLFDNR=%s' %
(self.config['scraper']['base_url'], agendaitem.originalId))
logging.info("Getting agendaitem %d from %s",
agendaitem.originalId, agendaitem_url)
agendaitem_r = self.get_url(agendaitem_url)
if not agendaitem_r:
return
if len(agendaitem_r.history):
logging.info("Agenda item %d from %s seems to be private",
meeting_id, meeting_url)
else:
agendaitem_xml = agendaitem_r.text.encode(
'ascii', 'xmlcharrefreplace')
# TODO: mixup of agendaitem_parser / parser below?
agendaitem_parser = etree.XMLParser(recover=True)
agendaitem_root = etree.fromstring(agendaitem_xml,
parser=parser)
add_agenda_item = {}
for add_item in agendaitem_root[0].iterchildren():
if add_item.tag == "rtfWP" and len(add_item) > 0:
try:
agendaitem.resolution_text = h.unescape(
etree.tostring(add_item[0][1][0]))
except:
logging.warn(
"Unable to parse resolution text at "
"%s", agendaitem_url)
else:
if add_item.text:
add_agenda_item[add_item.tag] = h.unescape(
add_item.text)
if 'toptext' in add_agenda_item:
agendaitem.name = add_agenda_item['toptext']
# there are papers with id = 0. we don't need them.
if int(elem['volfdnr']):
consult_id = (unicode(agendaitem.originalId) +
unicode(int(elem['volfdnr'])))
consultation = Consultation(originalId=consult_id)
paper_id = int(elem['volfdnr'])
if 'voname' in add_agenda_item:
consultation.paper = Paper(
originalId=paper_id,
name=add_agenda_item['voname'])
else:
consultation.paper = Paper(originalId=paper_id)
agendaitem.consultation = [consultation]
if 'vobetr' in add_agenda_item:
if add_agenda_item['vobetr'] != agendaitem.name:
logging.warn(
"different values for name: %s and %s",
agendaitem.name, add_agenda_item['vobetr'])
if hasattr(self, 'paper_queue'):
self.paper_queue.add(int(elem['volfdnr']))
if 'totyp' in add_agenda_item:
agendaitem.result = add_agenda_item['totyp']
agendaitems.append(agendaitem)
meeting.agendaItem = agendaitems
oid = self.db.save_meeting(meeting)
logging.info("Meeting %d stored with _id %s", meeting_id, oid)
class PaperPage(object):
valid = False # 网页数据的有效性，无效则不进行解析与储存
def __init__(self, _id, content=None, **kwargs):
self._id = str(_id)
self.paper = Paper()
page_data = col_paper.find_one({"url_id": self._id})
if page_data:
# 数据库中已经存在，直接返回
return
if content is None:
self.content = Downloader(host + self._id)()
if self.content:
self.valid = True
else:
logger.error("当前网页为空，无法进行解析\turl_id:" + self._id)
self.valid = False
return
else:
self.valid = True
self.content = content
self.selector = etree.HTML(self.content)
self.paper.url_id = self._id
def run(self):
if not self.valid:
logger.info("该paper已存在\turl_id:" + self._id)
return
self.main_page()
self.get_in_citation()
if len(self.paper.in_citations) < 1:
logger.info("该paper参考文献小于1，已进行排除\turl_id:" + self._id)
return
self.get_citing_sentence()
self.get_out_citation()
self.paper.save()
logger.info("完成解析\turl:%s\tpaper_id:%s" % (self._id, self.paper._id))
def main_page(self):
try:
self.paper.year = to_int(
deep_select(self.selector, 0, "//table//tr[5]/td/text()"))
# if self.paper.year < 2013:
# # 对数据集进行筛选
# return
# 列出所有作者
authors = deep_select(self.selector,
return_type="list",
xpath="//table//tr[6]/td//a/@href")
if not authors:
# 如果self.paper中无作者，则直接剔除数据
self.valid = False
return
authors_id = [to_num(x) for x in authors]
self.paper.authors_full_name = deep_select(
self.selector,
return_type="list",
xpath="//table//tr[6]/td//a/text()")
self.paper.authors = authors_id
self.paper._id = deep_select(self.selector, 0,
"//table//tr[1]/td/text()")
if not self.paper._id:
self.valid = False
return
self.paper.title = deep_select(self.selector, 0,
"//table//tr[2]/td/text()")
self.paper.venue = deep_select(self.selector, 0,
"//table//tr[3]/td/text()")
self.paper.session = deep_select(self.selector, 0,
"//table//tr[4]/td/text()")
self.paper.abstract = clean(
deep_select(self.selector, 0,
'//div[@id="abstract"]/p/text()'))
except Exception as e:
logger.error("id:%s\t%s" % (self._id, e))
def get_out_citation(self):
if not self.valid:
return
self.selector = etree.HTML(
Downloader('http://aan.how/browse/outgoing_citations/' +
self._id)())
out_citations = deep_select(self.selector,
return_type="list",
xpath='//a/@href')
if out_citations:
self.paper.out_citations = [to_num(x) for x in out_citations]
def get_in_citation(self):
if not self.valid:
return
self.selector = etree.HTML(
Downloader('http://aan.how/browse/incoming_citations/' +
self._id)())
in_citations = deep_select(self.selector,
return_type="list",
xpath='//a/@href')
if in_citations:
self.paper.in_citations = [to_num(x) for x in in_citations]
def get_citing_sentence(self):
if not self.valid:
return
self.selector = etree.HTML(
Downloader('http://aan.how/browse/citing_sentences/' + self._id)())
paper_id = deep_select(self.selector,
return_type="list",
xpath='//a/text()')
sentence = deep_select(self.selector,
return_type="list",
xpath="//tr/td[4]/div/text()")
line = deep_select(self.selector,
return_type="list",
xpath="//tr/td[3]/text()")
if paper_id and sentence:
for x in range(len(paper_id)):
citing_sentences = {
"paper_id": paper_id[x],
"sentence": clean(sentence[x]),
"line": line[x]
}
self.paper.citing_sentences.append(citing_sentences)
def get_paper(self, paper_url=None, paper_id=None):
"""
Load paper details for the paper given by detail page URL
or numeric ID
"""
paper_url = ('%svo020.asp?VOLFDNR=%s'
% (self.config['scraper']['base_url'], paper_id))
logging.info("Getting paper %d from %s", paper_id, paper_url)
# Stupid re-try concept because AllRis sometimes misses
# start < at tags at first request.
try_counter = 0
while True:
try:
response = self.get_url(paper_url)
if not response:
return
if "noauth" in response.url:
logging.warn("Paper %s in %s seems to private",
paper_id, paper_url)
return
text = response.text
doc = html.fromstring(text)
data = {}
# Beratungsfolge-Table checken
# lets hope we always have this table
table = self.table_css(doc)[0]
self.consultation_list_start = False
last_headline = ''
for line in table:
if line.tag == 'tr':
headline = line[0].text
elif line.tag == 'td':
headline = line.text
else:
logging.error("ERROR: Serious error in data table. "
"Unable to parse.")
if headline:
headline = headline.split(":")[0].lower()
if headline[-1] == ":":
headline = headline[:-1]
if headline == "betreff":
value = line[1].text_content().strip()
# There is some html comment with a script
# tag in front of the text which we remove.
value = value.split("-->")[1]
# remove all multiple spaces from the string
data[headline] = " ".join(value.split())
elif headline in ['verfasser', u'federführend',
'drucksache-art']:
data[headline] = line[1].text.strip()
elif headline in ['status']:
data[headline] = line[1].text.strip()
# related papers
if len(line) > 2:
if len(line[3]):
# Gets originalId. is there something
# else at this position? (will break)
paper_id = line[3][0][0][1][0].get(
'href').split('=')[1].split('&')[0]
data['relatedPaper'] = [Paper(
originalId=paper_id)]
# Lot's of scraping just because of the date (?)
elif headline == "beratungsfolge":
# The actual list will be in the next row
# inside a table, so we only set a marker.
self.consultation_list_start = True
elif self.consultation_list_start:
elem = line[0][0]
# The first line is pixel images, so skip
# it, then we need to jump in steps of two.
amount = (len(elem) - 1) / 2
consultations = []
date_list = []
i = 0
item = None
for elem_line in elem:
if i == 0:
i += 1
continue
"""
Here we need to parse the actual list which can have different forms. A complex example
can be found at http://ratsinfo.aachen.de/bi/vo020.asp?VOLFDNR=10822
The first line is some sort of headline with the committee in question and the type of consultation.
After that 0-n lines of detailed information of meetings with a date, transscript and decision.
The first line has 3 columns (thanks to colspan) and the others have 7.
Here we make every meeting a separate entry, we can group them together later again if we want to.
"""
# now we need to parse the actual list
# those lists
new_consultation = Consultation()
new_consultation.status = \
elem_line[0].attrib['title'].lower()
if len(elem_line) == 3:
# The order is "color/status", name of
# committee / link to TOP, more info we
# define a head dict here which can be
# shared for the other lines once we find
# another head line we will create a new
# one here.
new_consultation.role = \
elem_line[2].text.strip()
# Name of committee, e.g.
# "Finanzausschuss", unfort. without id
#'committee' : elem_line[1].text.strip(),
# For some obscure reasons sometimes action
# is missing.
elif len(elem_line) == 2:
# The order is "color/status", name of
# committee / link to TOP, more info.
status = \
elem_line[0].attrib['title'].lower()
# We define a head dict here which can be
# shared for the other lines once we find
# another head line we will create a new
# one here.
# name of committee, e.g.
# "Finanzausschuss", unfort. without id
#'committee' : elem_line[1].text.strip(),
elif len(elem_line) == 7:
try:
# This is about line 2 with lots of
# more stuff to process.
# Date can be text or a link with that
# text.
# We have a link (and ignore it).
if len(elem_line[1]) == 1:
date_text = elem_line[1][0].text
else:
date_text = elem_line[1].text
date_list.append(
datetime.datetime.strptime(
date_text.strip(), "%d.%m.%Y"))
if len(elem_line[2]):
# Form with silfdnr and toplfdnr
# but only in link (action=
# "to010.asp?topSelected=57023")
form = elem_line[2][0]
meeting_id = form[0].attrib['value']
new_consultation.meeting = [Meeting(
originalId=meeting_id)]
# Full name of meeting, e.g.
# "A/31/WP.16 öffentliche/
# nichtöffentliche Sitzung des
# Finanzausschusses"
#item['meeting'] = \
# elem_line[3][0].text.strip()
else:
# No link to TOP. Should not be
# possible but happens.
# (TODO: Bugreport?)
# Here we have no link but the text
# is in the TD directly - will be
# scaped as meeting.
#item['meeting'] = \
# elem_line[3].text.strip()
logging.warn(
"AgendaItem in consultation "
"list on the web page does not "
"contain a link to the actual "
"meeting at paper %s",
paper_url)
toplfdnr = None
if len(elem_line[6]) > 0:
form = elem_line[6][0]
toplfdnr = form[0].attrib['value']
if toplfdnr:
new_consultation.originalId = \
"%s-%s" % (toplfdnr,
paper_id)
# actually the id of the transcript
new_consultation.agendaItem = \
AgendaItem(
originalId=toplfdnr)
# e.g. "ungeändert beschlossen"
new_consultation.agendaItem.result \
= elem_line[4].text.strip()
consultations.append(
new_consultation)
else:
logging.error(
"missing agendaItem ID in "
"consultation list at %s",
paper_url)
except (IndexError, KeyError):
logging.error(
"ERROR: Serious error in "
"consultation list. Unable to "
"parse.")
logging.error(
"Serious error in consultation "
"list. Unable to parse.")
return []
i += 1
# Theory: we don't need this at all, because it's
# scraped at meeting.
#data['consultations'] = consultations
# set the marker to False again as we have read it
self.consultation_list_start = False
last_headline = headline
# We simply ignore the rest (there might not be much more
# actually).
# The actual text comes after the table in a div but it's not
# valid XML or HTML this using regex.
data['docs'] = self.body_re.findall(response.text)
first_date = False
for single_date in date_list:
if first_date:
if single_date < first_date:
first_date = single_date
else:
first_date = single_date
paper = Paper(originalId=paper_id)
paper.originalUrl = paper_url
paper.name = data['betreff']
paper.description = data['docs']
if 'drucksache-art' in data:
paper.paperType = data['drucksache-art']
if first_date:
paper.publishedDate = first_date.strftime("%d.%m.%Y")
# see theory above
#if 'consultations' in data:
# paper.consultation = data['consultations']
paper.auxiliaryFile = []
# get the attachments step 1 (Drucksache)
file_1 = self.attachment_1_css(doc)
if len(file_1):
if file_1[0].value:
href = ('%sdo027.asp'
% self.config['scraper']['base_url'])
original_id = file_1[0].value
name = 'Drucksache'
main_file = File(originalId=original_id, name=name)
main_file = self.get_file(main_file, href, True)
paper.mainFile = main_file
# get the attachments step 2 (additional attachments)
files = self.attachments_css(doc)
if len(files) > 0:
if len(files[0]) > 1:
if files[0][1][0].text.strip() == "Anlagen:":
for tr in files[0][2:]:
link = tr[0][0]
href = ("%s%s"
% (self.config['scraper']['base_url'],
link.attrib["href"]))
name = link.text
path_tokens = link.attrib["href"].split('/')
original_id = "%d-%d" % (int(path_tokens[4]),
int(path_tokens[6]))
aux_file = File(originalId=original_id,
name=name)
aux_file = self.get_file(aux_file, href)
paper.auxiliaryFile.append(aux_file)
print paper.auxiliaryFile
if not len(paper.auxiliaryFile):
del paper.auxiliaryFile
oid = self.db.save_paper(paper)
return
except (KeyError, IndexError):
if try_counter < 3:
logging.info("Try again: Getting paper %d from %s",
paper_id, paper_url)
try_counter += 1
else:
logging.error("Failed getting paper %d from %s",
paper_id, paper_url)
return
