def run(total_hours=24, hourly_limit=300, shuffle=False, termlist=None):
if termlist is None:
termlist = space.load_termlist()
total_requests = min(int(total_hours * hourly_limit), len(termlist))
total_time = 60 * 60 * min(total_hours, len(termlist) / hourly_limit)
wait_time = total_time / total_requests
daily_max_requests = hourly_limit * 24
try:
import os
os.mkdir('search_results')
except Exception as e:
# print("could not make directory", e)
pass
# not sure if shuffle is needed, if so try shuffling index
term_indices = list(termlist.index)
if shuffle:
# raise NotImplementedError()
print("shuffling termlist")
random.shuffle(term_indices)
if len(termlist) > daily_max_requests:
print("Warning: termlist length is", len(termlist),
"while max daily requests will be", daily_max_requests)
if len(termlist) > total_requests:
print(
f"Warning: only querying {total_requests} of {len(termlist)} total terms (not enough time specified)"
)
space.write_logs(
f"querying {total_requests} terms for a minimum of {printable_time(seconds=total_time)}",
verbose=True)
google_img_count = 0
baidu_img_count = 0
google_fails = []
baidu_fails = []
results = ResultSetList()
start_ts = time.time()
for i in range(0, total_requests):
start_iter_ts = time.time()
try:
term_idx = term_indices.pop()
english_term = termlist.loc[term_idx].english
chinese_term = termlist.loc[term_idx].chinese
except Exception as e:
print("out of terms", term_idx, str(e))
break
try:
label = termlist.loc[term_idx].label
except Exception as e:
label = "automated_scraper"
result = ResultSet(english_term, chinese_term, label)
print(
f'request {i}, term idx {term_idx}: "{result.combined_term()}", (label: {label})'
)
if not english_term:
print("\tskipping Google for term (English term not present)")
else:
try:
urls = query_google(english_term)
# print(f"\tGoogle got {len(urls)} images")
result.add(urls[:MAX_PICTURES_PER], GOOGLE)
except Exception as e:
google_fails.append(e)
print("\tGoogle fail")
if not chinese_term:
print("\tskipping Baidu for term (Chinese term not present)")
else:
try:
urls = query_baidu(chinese_term)
# print(f"\tbaidu got {len(urls)} images")
result.add(urls[:MAX_PICTURES_PER], BAIDU)
except Exception as e:
baidu_fails.append(e)
print("\tBaidu fail")
# print("done querying search engines for term", english_term)
results.add(result)
# account for the time the calls took
took = time.time() - start_iter_ts
# add in random jitter
time_noise = random.random() * 2 - 1
# print("adding noise to wait time", printable_time(seconds=time_noise))
# cache results. this is a backup and not meant to be a reliable data store
if i % 25 == 24:
try:
update_results(results)
results.clear()
google_img_count += results.wrote[GOOGLE]
baidu_img_count += results.wrote[BAIDU]
except Exception as e:
import traceback
print(
"failed to write search results; waiting until next attempt:",
e)
exc = traceback.format_exc()
print(str(exc))
time.sleep(max(0, wait_time - took + time_noise))
if results.length > 0:
try:
update_results(results)
results.clear()
google_img_count += results.wrote[GOOGLE]
baidu_img_count += results.wrote[BAIDU]
except Exception as e:
import traceback
exc = traceback.format_exc()
print(exc)
print("Failed to update search results, waiting 1 minute")
time.sleep(60)
update_results(results)
results.clear()
google_img_count += results.wrote[GOOGLE]
baidu_img_count += results.wrote[BAIDU]
space.write_logs(
f'wrote {results.wrote["google"]} google images and {results.wrote[BAIDU]} baidu images',
verbose=True)
if len(baidu_fails) > 0 or len(google_fails) > 0:
space.write_error(
f"Baidu failures: {len(baidu_fails)}, Google failures: {len(google_fails)}"
)
print("took", printable_time(seconds=time.time() - start_ts))
return (google_img_count, baidu_img_count, total_requests)
'''
Given a set of searches that the scraper has created, post each individually to
the /createSearch endpoint. spaces_interface.write_search_results should have
created a new list in each result object of the Digital Ocean URLs
'''
search_term_to_id = {}
print(f"saving {results.length} search terms")
for term,result in results.iterterm():
post_result = post_search(result, '192.168.0.1')
if not post_result:
raise Exception("failed to post result for term " + term)
if len(result.urls[GOOGLE]) != len(result.get_datalake_urls(GOOGLE)):
post_images(post_result["search_id"], GOOGLE, result.get_datalake_urls(GOOGLE))
else:
post_images(post_result["search_id"], GOOGLE, result.get_datalake_urls(GOOGLE), result.urls[GOOGLE])
if len(result.urls[BAIDU]) != len(result.get_datalake_urls(BAIDU)):
post_images(post_result["search_id"], BAIDU, result.get_datalake_urls(BAIDU))
else:
post_images(post_result["search_id"], BAIDU, result.get_datalake_urls(BAIDU), result.urls[BAIDU])
search_term_to_id[result.combined_term()] = post_result["search_id"]
return search_term_to_id
if __name__ == "__main__":
from results import ResultSet, ResultSetList
import time
result = ResultSet('bunny', '')
result.add(['google.com', 'bunnies.io'], GOOGLE)
result.set_datalake_urls(['datalake.com/google.com', 'datalake.com/bunnies.io'], GOOGLE)
results = ResultSetList()
results.add(result)
save_search_results(results)
def __init__(self):
self.results = ResultSet() #list of dictionaries containing results
class Connector(object):
def __init__(self):
self.results = ResultSet() #list of dictionaries containing results
def set_auth_type(self, auth_type):
"""Set the authorization type for given web service
Allowed types:
anon: web service only provides anonymous access
login: user must be logged in to use the web service
mixed: web service allows both anonymous and user-based access
"""
pass
def set_throttle(self, limit=None, units=None):
"""Set the request rate for the given service
units: {'requests per second', 'requests per day', etc.}
"""
self.delay = 0
self.max_requests = 1e16
self.made_requests = 0
def throttle(f):
"""wrapper function for throttling web service requests"""
def wrapper(self, *args, **kwargs):
if self.made_requests < self.max_requests:
time.sleep(self.delay)
f(self, *args, **kwargs)
self.made_requests += 1
else:
raise Exception, 'maximum request limit reached'
return wrapper
def set_url(self, url_str):
self.url = Template(url_str)
def set_parser(self, output_format):
"""sets the object variable to the correct output stream parser"""
self.output_parser = output_parsers.get(output_format, lambda x: x)
def get_url_params(self):
p = {}
for attr_name, attr_value in self.__class__.__dict__.items():
if isinstance(attr_value, Input):
p[attr_name] = getattr(attr_value, 'value')
return p
@throttle
def fetch(self, config):
for key, val in config.iteritems():
setattr(getattr(self, key), 'value', val)
url = self.url.substitute(self.get_url_params())
r = requests.get(url)
results = self.output_parser(r.text)
self.parse_results(results)
def fetchmany(self, config_list):
for config in config_list:
self.fetch(config)
def parse_results(self):
raise NotImplementedError
def reset(self):
self.results.clear_results()
def __init__(self):
self.results = ResultSet() # list of dictionaries containing results
class Connector(object):
def __init__(self):
self.results = ResultSet() # list of dictionaries containing results
def set_auth_type(self, auth_type):
"""Set the authorization type for given web service
Allowed types:
anon: web service only provides anonymous access
login: user must be logged in to use the web service
mixed: web service allows both anonymous and user-based access
"""
pass
def set_throttle(self, limit=None, units=None):
"""Set the request rate for the given service
units: {'requests per second', 'requests per day', etc.}
"""
self.delay = 0
self.max_requests = 1e16
self.made_requests = 0
def throttle(f):
"""wrapper function for throttling web service requests"""
def wrapper(self, *args, **kwargs):
if self.made_requests < self.max_requests:
time.sleep(self.delay)
f(self, *args, **kwargs)
self.made_requests += 1
else:
raise Exception, "maximum request limit reached"
return wrapper
def set_url(self, url_str):
self.url = Template(url_str)
def set_parser(self, output_format):
"""sets the object variable to the correct output stream parser"""
self.output_parser = output_parsers.get(output_format, lambda x: x)
def get_url_params(self):
p = {}
for attr_name, attr_value in self.__class__.__dict__.items():
if isinstance(attr_value, Input):
p[attr_name] = getattr(attr_value, "value")
return p
@throttle
def fetch(self, config):
for key, val in config.iteritems():
setattr(getattr(self, key), "value", val)
url = self.url.substitute(self.get_url_params())
r = requests.get(url)
results = self.output_parser(r.text)
self.parse_results(results)
def fetchmany(self, config_list):
for config in config_list:
self.fetch(config)
def parse_results(self):
raise NotImplementedError
def reset(self):
self.results.clear_results()
