def _run_extraction(self, name, templates, page, descriptor, expected_output):
self.trace = None
template_pages = [HtmlPage(None, {}, t) for t in templates]
# extracts with trace enabled in order to generate traceback
extractor = InstanceBasedLearningExtractor([(t, descriptor) for t in template_pages], True)
actual_output, _ = extractor.extract(HtmlPage(None, {}, page))
if actual_output is not None:
actual_output = actual_output[0]
self.trace = ["Extractor:\n%s" % extractor] + actual_output.pop('trace')
# extracts again with trace disabled in order to get the pure output
extractor = InstanceBasedLearningExtractor([(t, descriptor) for t in template_pages])
actual_output, _ = extractor.extract(HtmlPage(None, {}, page))
if actual_output is None:
if expected_output is None:
return
assert False, "failed to extract data for test '%s'" % name
else:
actual_output = actual_output[0]
expected_names = set(expected_output.keys())
actual_names = set(actual_output.keys())
missing_in_output = filter(None, expected_names - actual_names)
error = "attributes '%s' were expected but were not present in test '%s'" % \
("', '".join(missing_in_output), name)
assert len(missing_in_output) == 0, error
unexpected = actual_names - expected_names
error = "unexpected attributes %s in test '%s'" % \
(', '.join(unexpected), name)
assert len(unexpected) == 0, error
for k, v in expected_output.items():
extracted = actual_output[k]
assert v == extracted, "in test '%s' for attribute '%s', " \
"expected value '%s' but got '%s'" % (name, k, v, extracted)
def __init__(self, name, spec, item_schemas, all_extractors, **kw):
super(IblSpider, self).__init__(name, **kw)
spec = deepcopy(spec)
for key, val in kw.items():
if isinstance(val, basestring) and key in ['start_urls', 'exclude_patterns', 'follow_patterns', 'allowed_domains']:
val = val.splitlines()
spec[key] = val
self._item_template_pages = sorted((
[t['scrapes'], dict_to_page(t, 'annotated_body'),
t.get('extractors', [])] \
for t in spec['templates'] if t.get('page_type', 'item') == 'item'
), key=lambda pair: pair[0])
# generate ibl extractor for links pages
_links_pages = [dict_to_page(t, 'annotated_body')
for t in spec['templates'] if t.get('page_type') == 'links']
_links_item_descriptor = create_slybot_item_descriptor({'fields': {}})
self._links_ibl_extractor = InstanceBasedLearningExtractor([(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self._ipages = [page for _, page, _ in self._item_template_pages]
self.html_link_extractor = HtmlLinkExtractor()
self.rss_link_extractor = RssLinkExtractor()
self.build_url_filter(spec)
self.itemcls_info = {}
for itemclass_name, triplets in itertools.groupby(self._item_template_pages, operator.itemgetter(0)):
page_extractors_pairs = map(operator.itemgetter(1, 2), triplets)
schema = item_schemas[itemclass_name]
item_cls = SlybotItem.create_iblitem_class(schema)
page_descriptor_pairs = []
for page, template_extractors in page_extractors_pairs:
item_descriptor = create_slybot_item_descriptor(schema)
apply_extractors(item_descriptor, template_extractors, all_extractors)
page_descriptor_pairs.append((page, item_descriptor))
extractor = InstanceBasedLearningExtractor(page_descriptor_pairs)
self.itemcls_info[itemclass_name] = {
'class': item_cls,
'descriptor': item_descriptor,
'extractor': extractor,
}
self.login_requests = []
self.form_requests = []
self._start_requests = []
self.generic_form = GenericForm(**kw)
self._create_init_requests(spec.get("init_requests", []))
self._process_start_urls(spec)
self.allowed_domains = spec.get('allowed_domains',
self._get_allowed_domains(self._ipages))
if not self.allowed_domains:
self.allowed_domains = None
def setup_bot(self, settings, spec, items, extractors):
"""
Perform any initialization needed for crawling using this plugin
"""
_item_template_pages = sorted(([
t['scrapes'],
dict_to_page(t, 'annotated_body'),
t.get('extractors', [])
] for t in spec['templates'] if t.get('page_type', 'item') == 'item'),
key=lambda pair: pair[0])
self.itemcls_info = {}
if settings.get('AUTO_PAGINATION'):
self.html_link_extractor = PaginationExtractor()
else:
self.html_link_extractor = HtmlLinkExtractor()
for itemclass_name, triplets in groupby(_item_template_pages,
itemgetter(0)):
page_extractors_pairs = map(itemgetter(1, 2), triplets)
schema = items[itemclass_name]
item_cls = SlybotItem.create_iblitem_class(schema)
page_descriptor_pairs = []
for page, template_extractors in page_extractors_pairs:
item_descriptor = create_slybot_item_descriptor(schema)
apply_extractors(item_descriptor, template_extractors,
extractors)
page_descriptor_pairs.append((page, item_descriptor))
extractor = InstanceBasedLearningExtractor(page_descriptor_pairs)
self.itemcls_info[itemclass_name] = {
'class': item_cls,
'descriptor': item_descriptor,
'extractor': extractor,
}
# generate ibl extractor for links pages
_links_pages = [
dict_to_page(t, 'annotated_body') for t in spec['templates']
if t.get('page_type') == 'links'
]
_links_item_descriptor = create_slybot_item_descriptor({'fields': {}})
self._links_ibl_extractor = InstanceBasedLearningExtractor(
[(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self.build_url_filter(spec)
def scrape_page2(self, page, fields_spec):
if self._ex is None:
self._ex = InstanceBasedLearningExtractor(
((t, get_visual_tool_item_descriptor(fields_spec))
for t in self._templates), False, True)
res = self._ex.extract(page)[0]
return res
def scrape(self, url=None, html=None, encoding='utf-8'):
## not version from https://github.com/scrapy/scrapely/blob/master/scrapely/extraction/pageparsing.py
## may need to replace with version from inspect.getsourcelines(Scraper.scrape), as this version is
page = self._get_page(url, encoding, html)
ex = InstanceBasedLearningExtractor(self.templates)
return ex.extract(page)[0]
def test_type_extractor(self):
schema = {
"fields": {
'gender': {
'required': False,
'type': 'number',
'vary': False,
}
}
}
descriptor = create_slybot_item_descriptor(schema)
extractors = {
1: {
"type_extractor": "text"
},
2: {
"regular_expression": "Gender\\s+(Male|Female)"
}
}
apply_extractors(descriptor, {"gender": [1, 2]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template,
descriptor)])
self.assertEqual(
ibl_extractor.extract(self.target)[0][0], {u'gender': [u'Male']})
def get_extractor(site_id):
sds = ScraperDescriptor.objects.filter(site__id=site_id)
if not sds.exists():
return
tmpls = []
for s in sds:
items = s.items.filter(descriptor__target__symbol='ProductInfo')
idesc = ItemDescriptor('', '', [
FieldDescriptor(i.descriptor.symbol,
i.descriptor.desc,
extractor=types[i.descriptor.typ.symbol](i.value))
for i in items
])
ts = load_templates(s.id)
tmpls += [(t, idesc) for _, t in ts]
if tmpls:
ex = InstanceBasedLearningExtractor(tmpls)
def extractor(response):
page = HtmlPage(response.url,
headers=response.headers,
body=response.body.decode(response.encoding),
encoding=response.encoding)
extract = ex.extract(page)
if extract[0] is not None:
for e in extract[0]:
yield e
return extractor
def test_annotate_ignore_unpaired(self):
tm = TemplateMaker(self.PAGE)
tm.annotate('field1', best_match("and that's"), best_match=False)
tpl = tm.get_template()
ex = InstanceBasedLearningExtractor([(tpl, None)])
self.assertEqual(ex.extract(self.PAGE)[0],
[{u'field1': [u"More text with unpaired tag <img />and that's it"]}])
def test_annotate_multiple(self):
tm = TemplateMaker(self.PAGE)
tm.annotate('field1', best_match('text to annotate'), best_match=False)
tpl = tm.get_template()
ex = InstanceBasedLearningExtractor([(tpl, None)])
self.assertEqual(ex.extract(self.PAGE)[0],
[{u'field1': [u'Some text to annotate here', u'Another text to annotate there']}])
def test_extraction(self, name, templates, page, descriptor,
expected_output):
template_pages = [HtmlPage(None, {}, t) for t in templates]
extractor = InstanceBasedLearningExtractor([(t, descriptor)
for t in template_pages])
actual_output, _ = extractor.extract(HtmlPage(None, {}, page))
self.assertEqual(expected_output, actual_output and actual_output[0])
def do_scrape(self, url):
"""scrape <url> - scrape url (alias: s)"""
templates = self._load_templates()
if assert_or_print(templates, "no templates available"):
return
# fall back to the template encoding if none is specified
page = url_to_page(url, default_encoding=templates[0].encoding)
ex = InstanceBasedLearningExtractor((t, None) for t in templates)
pprint.pprint(ex.extract(page)[0])
def test_default_type_extractor(self):
schema = {'fields': {}}
descriptor = create_slybot_item_descriptor(schema)
extractors = {1: {"regular_expression": "Gender\\s+(Male|Female)"}}
apply_extractors(descriptor, {"gender": [1]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template,
descriptor)])
self.assertEqual(
ibl_extractor.extract(self.target)[0][0], {u'gender': [u'Male']})
def test_negative_hit_w_regex(self):
schema = {
'fields': {
'gender': {
'required': False,
'type': 'number',
'vary': False,
}
}
}
descriptor = create_slybot_item_descriptor(schema)
extractors = {1: {"regular_expression": "Gender\\s+(Male|Female)"}}
apply_extractors(descriptor, {"gender": [1]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template,
descriptor)])
self.assertEqual(ibl_extractor.extract(self.target)[0], None)
def test_text_type_w_regex_and_no_groups(self):
schema = {
'fields': {
'gender': {
'required': False,
'type': 'text',
'vary': False,
}
}
}
descriptor = create_slybot_item_descriptor(schema)
extractors = {1: {"regular_expression": "Gender"}}
apply_extractors(descriptor, {"gender": [1]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template,
descriptor)])
self.assertEqual(
ibl_extractor.extract(self.target)[0][0], {u'gender': [u'Gender']})
def test_raw_type_w_regex(self):
schema = {
'fields': {
'gender': {
'required': False,
'type': 'raw',
'vary': False,
}
}
}
descriptor = create_slybot_item_descriptor(schema)
extractors = {
1: {
"regular_expression": "Gender.*(<td\s*>(?:Male|Female)</td>)"
}
}
apply_extractors(descriptor, {"gender": [1]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template,
descriptor)])
self.assertEqual(
ibl_extractor.extract(self.target)[0][0],
{u'gender': [u'<td >Male</td>']})
def test_extractor_w_empty_string_extraction(self):
schema = {
'fields': {
'gender': {
'required': False,
'type': 'text',
'vary': False,
},
'name': {
'required': True,
'type': 'text',
'vary': False,
}
}
}
descriptor = create_slybot_item_descriptor(schema)
extractors = {1: {"regular_expression": "([0-9]+)"}}
apply_extractors(descriptor, {"gender": [1]}, extractors)
ibl_extractor = InstanceBasedLearningExtractor([(self.template2,
descriptor)])
self.assertEqual(
ibl_extractor.extract(self.target2)[0][0],
{u'name': [u'Name Olivia']})
map(
lambda (n, s, desc): (n, s, u'%s' %
(desc.name, ), types[desc.typ.symbol]),
((n, s, KeyValueDescriptor.objects.get(symbol=n))
for n, s, _, _ in items
if KeyValueDescriptor.objects.filter(symbol=n).exists())))
idesc = ItemDescriptor('', '', [
FieldDescriptor(n, desc, extractor=fnc(s)) for n, s, desc, fnc in items
])
ts = load_templates('scraper.json', 'site-%d' % site.id)
if not ts:
ts = annotate(url, 'site-%d' % site.id,
[(n, s) for n, s, _, _ in items])
tmpls += [(tm, idesc) for tm in ts]
ex = InstanceBasedLearningExtractor(tmpls)
urls = (
'http://www.rc-chem.eu/doprava', # should fail
'http://www.rc-chem.eu/produkty/2-fma',
'http://www.rc-chem.eu/produkty/3-fmc',
'http://www.rc-chem.eu/produkty/3-mmc-crystal',
'http://www.rc-chem.eu/produkty/4-fa-crystal',
'http://www.rc-chem.eu/produkty/dimethylone',
'http://www.rc-chem.eu/produkty/ethylphenidate',
'http://www.rc-chem.eu/produkty/mpa',
'http://www.rc-chem.eu/produkty/neb',
'http://www.rc-chem.eu/produkty/pentedrone-velky-crystal',
'http://www.rc-chem.eu/produkty/thio-crystal',
'http://www.rc-chem.eu/produkty/thio-velky-crystal',
'http://mefedronprodej.webnode.cz/produkty-1/',
def setup_bot(self, settings, spec, items, extractors, logger):
"""
Perform any initialization needed for crawling using this plugin
"""
self.logger = logger
templates = map(self._get_annotated_template, spec['templates'])
_item_template_pages = sorted(([
t.get('scrapes'),
dict_to_page(t, 'annotated_body'),
t.get('extractors', []),
t.get('version', '0.12.0')
] for t in templates if t.get('page_type', 'item') == 'item'),
key=lambda x: x[0])
self.item_classes = {}
self.template_scrapes = {
template.get('page_id'): template['scrapes']
for template in templates
}
if (settings.get('AUTO_PAGINATION')
or spec.get('links_to_follow') == 'auto'):
self.html_link_extractor = PaginationExtractor()
else:
self.html_link_extractor = HtmlLinkExtractor()
for schema_name, schema in items.items():
if schema_name not in self.item_classes:
if not schema.get('name'):
schema['name'] = schema_name
item_cls = SlybotItem.create_iblitem_class(schema)
self.item_classes[schema_name] = item_cls
# Create descriptors and apply additional extractors to fields
page_descriptor_pairs = []
self.schema_descriptors = {}
for default, template, template_extractors, v in _item_template_pages:
descriptors = OrderedDict()
for schema_name, schema in items.items():
item_descriptor = create_slybot_item_descriptor(
schema, schema_name)
apply_extractors(item_descriptor, template_extractors,
extractors)
descriptors[schema_name] = item_descriptor
descriptor = descriptors.values() or [{}]
descriptors['#default'] = descriptors.get(default, descriptor[0])
self.schema_descriptors[template.page_id] = descriptors['#default']
page_descriptor_pairs.append((template, descriptors, v))
add_extractors_to_descriptors(descriptors, extractors)
grouped = itertools.groupby(
sorted(page_descriptor_pairs, key=operator.itemgetter(2)),
lambda x: x[2] < '0.13.0')
self.extractors = []
for version, group in grouped:
if version:
self.extractors.append(
InstanceBasedLearningExtractor([
(page, scrapes['#default'])
for page, scrapes, version in group
]))
else:
self.extractors.append(SlybotIBLExtractor(list(group)))
# generate ibl extractor for links pages
_links_pages = [
dict_to_page(t, 'annotated_body') for t in templates
if t.get('page_type') == 'links'
]
_links_item_descriptor = create_slybot_item_descriptor({'fields': {}})
self._links_ibl_extractor = InstanceBasedLearningExtractor(
[(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self.build_url_filter(spec)
# Clustering
self.template_names = [t.get('page_id') for t in spec['templates']]
if settings.get('PAGE_CLUSTERING'):
try:
import page_clustering
self.clustering = page_clustering.kmeans_from_samples(
spec['templates'])
self.logger.info("Clustering activated")
except ImportError:
self.clustering = None
self.logger.warning(
"Clustering could not be used because it is not installed")
else:
self.clustering = None
def scrape_page(self, page):
if self._ex is None:
self._ex = InstanceBasedLearningExtractor((t, None) for t in
self._templates)
return self._ex.extract(page)[0]
def __init__(self, name, spec, item_schemas, all_extractors, **kw):
super(IblSpider, self).__init__(name, **kw)
default_item = spec['scrapes']
self._default_schema = item_schemas[default_item]
if not self._default_schema:
self.log("Scraping unknown default item schema: %s" % default_item, \
log.WARNING)
self._item_template_pages = sorted((
[t.get('scrapes', default_item), dict_to_page(t, 'annotated_body'),
t.get('extractors', [])] \
for t in spec['templates'] if t.get('page_type', 'item') == 'item'
), key=lambda pair: pair[0])
# generate ibl extractor for links pages
_links_pages = [
dict_to_page(t, 'annotated_body') for t in spec['templates']
if t.get('page_type') == 'links'
]
_links_item_descriptor = create_slybot_item_descriptor({
'id': "_links",
'properties': ()
})
self._links_ibl_extractor = InstanceBasedLearningExtractor([(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self._ipages = [page for _, page, _ in self._item_template_pages]
self.start_urls = self.start_urls or spec.get('start_urls')
if isinstance(self.start_urls, basestring):
self.start_urls = self.start_urls.splitlines()
self.link_extractor = LinkExtractor()
self.allowed_domains = self._get_allowed_domains(self._ipages)
self.build_url_filter(spec)
default_item_cls = get_iblitem_class(self._default_schema)
default_item_descriptor = create_slybot_item_descriptor(
self._default_schema)
self.itemcls_info = {}
for itemclass_name, triplets in itertools.groupby(
self._item_template_pages, operator.itemgetter(0)):
page_extractors_pairs = map(operator.itemgetter(1, 2), triplets)
schema = item_schemas[itemclass_name]
item_cls = get_iblitem_class(
schema) if schema else default_item_cls
page_descriptor_pairs = []
for page, extractors in page_extractors_pairs:
item_descriptor = create_slybot_item_descriptor(
schema) if schema else default_item_descriptor
apply_extractors(item_descriptor, extractors, all_extractors)
page_descriptor_pairs.append((page, item_descriptor))
extractor = InstanceBasedLearningExtractor(page_descriptor_pairs)
self.itemcls_info[itemclass_name] = {
'class': item_cls,
'descriptor': item_descriptor,
'extractor': extractor,
}
self.login_requests = []
self.form_requests = []
for rdata in spec.get("init_requests", []):
if rdata["type"] == "login":
request = Request(url=rdata.pop("loginurl"),
meta=rdata,
callback=self.parse_login_page)
self.login_requests.append(request)
elif rdata["type"] == "form":
request = Request(url=rdata.pop("form_url"),
meta=rdata,
callback=self.parse_form_page)
self.form_requests.append(request)
def __init__(self, name, spec, item_schemas, all_extractors, **kw):
super(IblSpider, self).__init__(name, **kw)
self._item_template_pages = sorted((
[t['scrapes'], dict_to_page(t, 'annotated_body'),
t.get('extractors', [])] \
for t in spec['templates'] if t.get('page_type', 'item') == 'item'
), key=lambda pair: pair[0])
# generate ibl extractor for links pages
_links_pages = [dict_to_page(t, 'annotated_body')
for t in spec['templates'] if t.get('page_type') == 'links']
_links_item_descriptor = create_slybot_item_descriptor({'fields': {}})
self._links_ibl_extractor = InstanceBasedLearningExtractor([(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self._ipages = [page for _, page, _ in self._item_template_pages]
self.start_urls = self.start_urls or spec.get('start_urls')
if isinstance(self.start_urls, basestring):
self.start_urls = self.start_urls.splitlines()
self.html_link_extractor = HtmlLinkExtractor()
self.rss_link_extractor = RssLinkExtractor()
self.allowed_domains = spec.get('allowed_domains',
self._get_allowed_domains(self._ipages))
if not self.allowed_domains:
self.allowed_domains = None
self.build_url_filter(spec)
self.itemcls_info = {}
for itemclass_name, triplets in itertools.groupby(self._item_template_pages, operator.itemgetter(0)):
page_extractors_pairs = map(operator.itemgetter(1, 2), triplets)
schema = item_schemas[itemclass_name]
item_cls = get_iblitem_class(schema)
page_descriptor_pairs = []
for page, template_extractors in page_extractors_pairs:
item_descriptor = create_slybot_item_descriptor(schema)
apply_extractors(item_descriptor, template_extractors, all_extractors)
page_descriptor_pairs.append((page, item_descriptor))
extractor = InstanceBasedLearningExtractor(page_descriptor_pairs)
self.itemcls_info[itemclass_name] = {
'class': item_cls,
'descriptor': item_descriptor,
'extractor': extractor,
}
self.login_requests = []
self.form_requests = []
for rdata in spec.get("init_requests", []):
if rdata["type"] == "login":
request = Request(url=rdata.pop("loginurl"), meta=rdata,
callback=self.parse_login_page, dont_filter=True)
self.login_requests.append(request)
elif rdata["type"] == "form":
self.generic_form = GenericForm(**kw)
self.form_requests.append(self.get_generic_form_start_request(rdata))
def setup_bot(self, settings, spec, items, extractors):
"""
Perform any initialization needed for crawling using this plugin
"""
_item_template_pages = sorted(([
t.get('scrapes'),
dict_to_page(t, 'annotated_body'),
t.get('extractors', []),
t.get('version', '0.12.0')
] for t in spec['templates'] if t.get('page_type', 'item') == 'item'),
key=lambda x: x[0])
self.item_classes = {}
self.template_scrapes = {
template.get('page_id'): template['scrapes']
for template in spec.get('templates')
}
self.html_link_extractor = HtmlLinkExtractor()
for schema_name, schema in items.items():
if schema_name not in self.item_classes:
if not schema.get('name'):
schema['name'] = schema_name
item_cls = SlybotItem.create_iblitem_class(schema)
self.item_classes[schema_name] = item_cls
# Create descriptors and apply additional extractors to fields
page_descriptor_pairs = []
self.schema_descriptors = {}
for default, template, template_extractors, v in _item_template_pages:
descriptors = OrderedDict()
for schema_name, schema in items.items():
item_descriptor = create_slybot_item_descriptor(
schema, schema_name)
apply_extractors(item_descriptor, template_extractors,
extractors)
descriptors[schema_name] = item_descriptor
descriptor = descriptors.values() or [{}]
descriptors['#default'] = descriptors.get(default, descriptor[0])
self.schema_descriptors[template.page_id] = descriptors['#default']
page_descriptor_pairs.append((template, descriptors, v))
add_extractors_to_descriptors(descriptors, extractors)
grouped = itertools.groupby(
sorted(page_descriptor_pairs, key=operator.itemgetter(2)),
lambda x: x[2] < '0.13.0')
self.extractors = []
for version, group in grouped:
if version:
self.extractors.append(
InstanceBasedLearningExtractor([
(page, scrapes['#default'])
for page, scrapes, version in group
]))
else:
self.extractors.append(SlybotIBLExtractor(list(group)))
# generate ibl extractor for links pages
_links_pages = [
dict_to_page(t, 'annotated_body') for t in spec['templates']
if t.get('page_type') == 'links'
]
_links_item_descriptor = create_slybot_item_descriptor({'fields': {}})
self._links_ibl_extractor = InstanceBasedLearningExtractor(
[(t, _links_item_descriptor) for t in _links_pages]) \
if _links_pages else None
self.build_url_filter(spec)