def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('PLUGINS', load_plugins(settings))
self.settings = settings
def test_trained(self):
base = 'http://www.daft.ie/ireland/houses-for-sale/?offset={}'.format
daft_url = base(10)
spec = {
'start_urls': [daft_url],
'links_to_follow': 'auto',
'respect_nofollow': False,
'follow_patterns': [],
'exclude_patterns': [],
'init_requests': [],
'templates': [daft_sample]
}
settings = Settings()
settings.set('LOADED_PLUGINS', load_plugins(settings))
spider = IblSpider('hn', spec, {}, {}, settings=settings)
request = Request(daft_url)
response = HtmlResponse(url=daft_url,
body=daft_body,
request=request,
encoding="utf-8")
data = {
r.url
for r in spider.handle_html(response) if isinstance(r, Request)
}
self.assertEqual({base(i) for i in (90, 80, 70)}, data)
def __init__(self, name, spec, item_schemas, all_extractors, settings=None,
**kw):
super(IblSpider, self).__init__(name, **kw)
spec = deepcopy(spec)
for key, val in kw.items():
if isinstance(val, basestring) and key in STRING_KEYS:
val = val.splitlines()
spec[key] = val
self._item_template_pages = sorted(
((t['scrapes'], t) for t in spec['templates']
if t.get('page_type', 'item') == 'item'), key=itemgetter(0))
self._templates = [templ for _, templ in self._item_template_pages]
self.plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, item_schemas, all_extractors)
self.plugins[plugin_name] = instance
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._templates)
)
if not self.allowed_domains:
self.allowed_domains = None
def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('PLUGINS', load_plugins(settings))
self.settings = settings
def _configure_plugins(self, settings, spec, schemas, extractors):
plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, schemas, extractors)
plugins[plugin_name] = instance
return plugins
def _configure_plugins(self, settings, spec, schemas, extractors):
plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, schemas, extractors, self.logger)
plugins[plugin_name] = instance
return plugins
def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
logging.info('Slybot %s Spider', slybot.__version__)
if settings is None:
settings = get_project_settings()
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('LOADED_PLUGINS', load_plugins(settings))
self.settings = settings
def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
logging.info('Slybot %s Spider', slybot.__version__)
if settings is None:
settings = get_project_settings()
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('LOADED_PLUGINS', load_plugins(settings))
self.settings = settings
def __init__(self,
name,
spec,
item_schemas,
all_extractors,
settings=None,
**kw):
super(IblSpider, self).__init__(name, **kw)
self._job_id = settings.get('JOB', '')
spec = deepcopy(spec)
for key, val in kw.items():
if isinstance(val, six.string_types) and key in STRING_KEYS:
val = val.splitlines()
spec[key] = val
self._item_template_pages = sorted(
((t['scrapes'], t) for t in spec['templates']
if t.get('page_type', 'item') == 'item'),
key=itemgetter(0))
self._templates = [templ for _, templ in self._item_template_pages]
self.plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, item_schemas, all_extractors)
self.plugins[plugin_name] = instance
self.js_enabled = False
self.SPLASH_HOST = None
if settings.get('SPLASH_URL'):
self.SPLASH_HOST = urlparse(settings.get('SPLASH_URL')).hostname
self.js_enabled = spec.get('js_enabled', False)
if self.js_enabled and (settings.get('SPLASH_PASS') is not None
or settings.get('SPLASH_USER') is not None):
self.splash_auth = basic_auth_header(
settings.get('SPLASH_USER', ''),
settings.get('SPLASH_PASS', ''))
self._filter_js_urls = self._build_js_url_filter(spec)
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._templates))
self.page_actions = spec.get('page_actions', [])
if not self.allowed_domains:
self.allowed_domains = None
def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
logging.info('Slybot %s Spider', slybot.__version__)
if is_zipfile(datadir):
tempdir = tempfile.mkdtemp(prefix='slybot-')
ZipFile(datadir).extractall(tempdir)
atexit.register(shutil.rmtree, tempdir)
datadir = tempdir
if settings is None:
settings = get_project_settings()
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('LOADED_PLUGINS', load_plugins(settings))
self.settings = settings
def __init__(self, datadir, spider_cls=None, settings=None, **kwargs):
logging.info('Slybot %s Spider', slybot.__version__)
if is_zipfile(datadir):
tempdir = tempfile.mkdtemp(prefix='slybot-')
ZipFile(datadir).extractall(tempdir)
atexit.register(shutil.rmtree, tempdir)
datadir = tempdir
if settings is None:
settings = get_project_settings()
self.spider_cls = load_object(spider_cls) if spider_cls else IblSpider
self._specs = open_project_from_dir(datadir)
settings = settings.copy()
settings.frozen = False
settings.set('LOADED_PLUGINS', load_plugins(settings))
self.settings = settings
def __init__(self, name, spec, item_schemas, all_extractors, settings=None,
**kw):
super(IblSpider, self).__init__(name, **kw)
self._job_id = settings.get('JOB', '')
spec = deepcopy(spec)
for key, val in kw.items():
if isinstance(val, six.string_types) and key in STRING_KEYS:
val = val.splitlines()
spec[key] = val
self._item_template_pages = sorted(
((t['scrapes'], t) for t in spec['templates']
if t.get('page_type', 'item') == 'item'), key=itemgetter(0))
self._templates = [templ for _, templ in self._item_template_pages]
self.plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, item_schemas, all_extractors)
self.plugins[plugin_name] = instance
self.js_enabled = False
self.SPLASH_HOST = None
if settings.get('SPLASH_URL'):
self.SPLASH_HOST = urlparse(settings.get('SPLASH_URL')).hostname
self.js_enabled = spec.get('js_enabled', False)
if self.js_enabled and (settings.get('SPLASH_PASS') is not None or
settings.get('SPLASH_USER') is not None):
self.splash_auth = basic_auth_header(
settings.get('SPLASH_USER', ''),
settings.get('SPLASH_PASS', ''))
self._filter_js_urls = self._build_js_url_filter(spec)
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._templates)
)
self.page_actions = spec.get('page_actions', [])
if not self.allowed_domains:
self.allowed_domains = None
def test_trained(self):
base = 'http://www.daft.ie/ireland/houses-for-sale/?offset={}'.format
daft_url = base(10)
spec = {
'start_urls': [daft_url],
'links_to_follow': 'auto',
'respect_nofollow': False,
'follow_patterns': [],
'exclude_patterns': [],
'init_requests': [],
'templates': [daft_sample]
}
settings = Settings()
settings.set('LOADED_PLUGINS', load_plugins(settings))
spider = IblSpider('hn', spec, {}, {}, settings=settings)
request = Request(daft_url)
response = UTF8HtmlResponse(url=daft_url, body=daft_body,
request=request)
data = {r.url for r in spider.handle_html(response)
if isinstance(r, Request)}
self.assertEqual({base(i) for i in (90, 80, 70)}, data)
def __init__(self,
name,
spec,
item_schemas,
all_extractors,
settings=None,
**kw):
super(IblSpider, self).__init__(name, **kw)
spec = deepcopy(spec)
for key, val in kw.items():
if isinstance(val, basestring) and key in STRING_KEYS:
val = val.splitlines()
spec[key] = val
self._item_template_pages = sorted(
((t['scrapes'], t) for t in spec['templates']
if t.get('page_type', 'item') == 'item'),
key=itemgetter(0))
self._templates = [templ for _, templ in self._item_template_pages]
self.plugins = IndexedDict()
for plugin_class, plugin_name in zip(load_plugins(settings),
load_plugin_names(settings)):
instance = plugin_class()
instance.setup_bot(settings, spec, item_schemas, all_extractors)
self.plugins[plugin_name] = instance
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._templates))
if not self.allowed_domains:
self.allowed_domains = None
