def extract_action(cli_args):
if cli_args.glob is None and cli_args.input_dir is None:
cli_args.input_dir = DEFAULT_CONTENT_FOLDER
input_data = cli_args.report
if cli_args.glob is not None:
input_data = dummy_csv_file_from_glob(cli_args.glob, cli_args.input_dir)
enricher = casanova.enricher(
input_data,
cli_args.output,
keep=cli_args.select,
add=OUTPUT_ADDITIONAL_HEADERS
)
loading_bar = LoadingBar(
desc='Extracting content',
total=cli_args.total,
unit='doc'
)
def on_irrelevant_row(reason, row, i):
loading_bar.update()
loading_bar.print('Row n°{n} could not be processed: {reason}'.format(n=i + 1, reason=reason))
enricher.writerow(row, format_error(reason))
if (
cli_args.glob is None and
'raw_contents' not in enricher.headers and
not isdir(cli_args.input_dir)
):
loading_bar.die([
'Could not find the "%s" directory!' % cli_args.input_dir,
'Did you forget to specify it with -i/--input-dir?'
])
files = create_report_iterator(
cli_args,
enricher,
on_irrelevant_row=on_irrelevant_row
)
pool = LazyPool(cli_args.processes)
loading_bar.update_stats(p=pool.processes)
with pool:
for error, row, result in pool.imap_unordered(worker, files):
loading_bar.update()
if error is not None:
enricher.writerow(row, format_error(report_error(error)))
continue
if result is None:
enricher.writerow(row, format_error('no-result'))
continue
enricher.writerow(row, result)
def extract_action(namespace):
output_file = open_output_file(namespace.output)
enricher = casanova.enricher(namespace.report,
output_file,
keep=namespace.select,
add=OUTPUT_ADDITIONAL_HEADERS)
loading_bar = tqdm(desc='Extracting content',
total=namespace.total,
dynamic_ncols=True,
unit=' docs')
files = create_report_iterator(namespace,
enricher,
loading_bar=loading_bar)
with Pool(namespace.processes) as pool:
for error, row, result in pool.imap_unordered(worker, files):
loading_bar.update()
if error is not None:
enricher.writerow(row, [report_error(error)] + PADDING)
continue
if result is None:
enricher.writerow(row, ['no-content'] + PADDING)
continue
enricher.writerow(row, result)
output_file.close()
def extract_action(namespace):
output_file = open_output_file(namespace.output)
enricher = casanova.enricher(
namespace.report,
output_file,
keep=namespace.select,
add=OUTPUT_ADDITIONAL_HEADERS
)
loading_bar = LoadingBar(
desc='Extracting content',
total=namespace.total,
unit='doc'
)
def on_irrelevant_row(reason, row):
loading_bar.update()
enricher.writerow(row, format_error(reason))
try:
files = create_report_iterator(
namespace,
enricher,
on_irrelevant_row=on_irrelevant_row
)
except NotADirectoryError:
loading_bar.die([
'Could not find the "%s" directory!' % namespace.input_dir,
'Did you forget to specify it with -i/--input-dir?'
])
pool = LazyPool(namespace.processes)
loading_bar.update_stats(p=pool.processes)
with pool:
for error, row, result in pool.imap_unordered(worker, files):
loading_bar.update()
if error is not None:
enricher.writerow(row, format_error(report_error(error)))
continue
if result is None:
enricher.writerow(row, format_error('no-content'))
continue
enricher.writerow(row, result)
loading_bar.close()
output_file.close()
def format_job_for_csv(result):
if result.error is not None:
return [
result.job.spider, result.job.url, '', '',
report_error(result.error), '', '', '0', result.job.level
]
resolved = result.response.geturl()
return [
result.job.spider, result.job.url,
resolved if resolved != result.job.url else '', result.response.status,
'', '',
result.meta.get('encoding', ''),
len(result.next_jobs) if result.next_jobs is not None else '0',
result.job.level
]
def extract_action(namespace):
input_headers, pos, reader = custom_reader(
namespace.report, ('status', 'filename', 'encoding'))
selected_fields = namespace.select.split(',') if namespace.select else None
selected_pos = [input_headers.index(h)
for h in selected_fields] if selected_fields else None
output_headers = (list(input_headers) if not selected_pos else
[input_headers[i] for i in selected_pos])
output_headers += OUTPUT_ADDITIONAL_HEADERS
output_file = open_output_file(namespace.output)
output_writer = csv.writer(output_file)
output_writer.writerow(output_headers)
loading_bar = tqdm(desc='Extracting content',
total=namespace.total,
dynamic_ncols=True,
unit=' docs')
namespace.report.close()
namespace.report = open(namespace.report.name)
files = create_report_iterator(namespace, loading_bar=loading_bar)
with Pool(namespace.processes) as pool:
for error, line, content in pool.imap_unordered(worker, files):
loading_bar.update()
if error is not None:
message = report_error(error)
line.extend([message, ''])
output_writer.writerow(line)
continue
line.extend(['', content])
output_writer.writerow(line)
output_file.close()
def fetch_action(cli_args, resolve=False, defer=None):
# If we are hitting a single url we enable contents_in_report by default
if not resolve and isinstance(cli_args.file, StringIO) and cli_args.contents_in_report is None:
cli_args.contents_in_report = True
if not resolve and cli_args.contents_in_report and cli_args.compress:
raise InvalidArgumentsError('Cannot both --compress and output --contents-in-report!')
# HTTP method
http_method = cli_args.method
# Cookie grabber
get_cookie = None
if cli_args.grab_cookies:
get_cookie = grab_cookies(cli_args.grab_cookies)
# Global headers
global_headers = None
if cli_args.headers:
global_headers = {}
for header in cli_args.headers:
k, v = parse_http_header(header)
global_headers[k] = v
# Resume listener
skipped_rows = 0
resuming_reader_loading = None
if cli_args.resume and cli_args.output.can_resume():
resuming_reader_loading = LoadingBar(
desc='Resuming',
unit='line'
)
def output_read_listener(event, row):
nonlocal skipped_rows
if event != 'output.row':
return
skipped_rows += 1
resuming_reader_loading.update()
cli_args.output.listener = output_read_listener
if resolve:
additional_headers = RESOLVE_ADDITIONAL_HEADERS
else:
additional_headers = FETCH_ADDITIONAL_HEADERS
if cli_args.contents_in_report:
additional_headers = additional_headers + ['raw_contents']
# Enricher
multiplex = None
if cli_args.separator is not None:
multiplex = (cli_args.column, cli_args.separator)
enricher = casanova.threadsafe_enricher(
cli_args.file,
cli_args.output,
add=additional_headers,
keep=cli_args.select,
total=cli_args.total,
prebuffer_bytes=DEFAULT_PREBUFFER_BYTES,
multiplex=multiplex
)
if resuming_reader_loading is not None:
resuming_reader_loading.close()
if cli_args.column not in enricher.headers:
raise InvalidArgumentsError('Could not find the "%s" column containing the urls in the given CSV file.' % cli_args.column)
url_pos = enricher.headers[cli_args.column]
filename_pos = None
if not resolve and cli_args.filename is not None:
if cli_args.filename not in enricher.headers:
raise InvalidArgumentsError('Could not find the "%s" column containing the filenames in the given CSV file.' % cli_args.filename)
filename_pos = enricher.headers[cli_args.filename]
# Loading bar
loading_bar = LoadingBar(
desc='Fetching pages',
total=enricher.total,
unit='url',
initial=skipped_rows
)
defer(loading_bar.close) # NOTE: it could be dangerous with multithreaded execution, not to close it ourselves
def update_loading_bar(result):
nonlocal errors
if result.error is not None:
errors += 1
else:
if resolve:
status = result.stack[-1].status
else:
status = result.response.status
if status >= 400:
status_codes[status] += 1
stats = {'errors': errors}
for code, count in status_codes.most_common(1):
stats[str(code)] = count
loading_bar.update_stats(**stats)
loading_bar.update()
only_shortened = getattr(cli_args, 'only_shortened', False)
def url_key(item):
url = item[1][url_pos].strip()
if not url:
return
if only_shortened and not is_shortened_url(url):
return
# Url templating
if cli_args.url_template:
return cli_args.url_template.format(value=url)
return url
def request_args(domain, url, item):
cookie = None
# Cookie
if get_cookie:
cookie = get_cookie(url)
# Headers
headers = None
if global_headers:
headers = global_headers
return {
'method': http_method,
'cookie': cookie,
'headers': headers
}
# Worker callback internals
filename_builder = None
files_writer = None
if not resolve:
try:
filename_builder = FilenameBuilder(
folder_strategy=cli_args.folder_strategy,
template=cli_args.filename_template
)
except TypeError:
die([
'Invalid "%s" --folder-strategy!' % cli_args.folder_strategy,
'Check the list at the end of the command help:',
' $ minet fetch -h'
])
files_writer = ThreadSafeFilesWriter(cli_args.output_dir)
def worker_callback(result):
# NOTE: at this point the callback is only fired on success
row = result.item[1]
response = result.response
meta = result.meta
if cli_args.keep_failed_contents and response.status != 200:
return
# First we need to build a filename
filename_cell = row[filename_pos] if filename_pos else None
formatter_kwargs = {}
if cli_args.filename_template and 'line' in cli_args.filename_template:
formatter_kwargs['line'] = enricher.wrap(row)
try:
filename = filename_builder(
result.resolved,
filename=filename_cell,
ext=meta.get('ext'),
formatter_kwargs=formatter_kwargs,
compressed=cli_args.compress
)
except FilenameFormattingError as e:
result.error = e
return
meta['filename'] = filename
# Decoding the response data?
is_text = meta.get('is_text', False)
original_encoding = meta.get('encoding', 'utf-8')
data = response.data
binary = True
if is_text and (cli_args.standardize_encoding or cli_args.contents_in_report):
data = data.decode(original_encoding, errors='replace')
binary = False
if cli_args.contents_in_report:
meta['decoded_contents'] = data
# Writing the file?
# TODO: specify what should happen when contents are empty (e.g. POST queries)
if data and not cli_args.contents_in_report:
files_writer.write(
filename,
data,
binary=binary,
compress=cli_args.compress
)
def write_fetch_output(index, row, resolved=None, status=None, error=None,
filename=None, encoding=None, mimetype=None, data=None):
addendum = [
resolved or '',
status or '',
error or '',
filename or '',
mimetype or '',
encoding or ''
]
if cli_args.contents_in_report:
addendum.append(data or '')
enricher.writerow(index, row, addendum)
def write_resolve_output(index, row, resolved=None, status=None, error=None,
redirects=None, chain=None):
addendum = [
resolved or '',
status or '',
error or '',
redirects or '',
chain or ''
]
enricher.writerow(index, row, addendum)
errors = 0
status_codes = Counter()
common_kwargs = {
'key': url_key,
'insecure': cli_args.insecure,
'threads': cli_args.threads,
'throttle': cli_args.throttle,
'domain_parallelism': cli_args.domain_parallelism,
'max_redirects': cli_args.max_redirects,
'wait': False,
'daemonic': True
}
if cli_args.timeout is not None:
common_kwargs['timeout'] = cli_args.timeout
# Normal fetch
if not resolve:
multithreaded_iterator = multithreaded_fetch(
enricher,
request_args=request_args,
callback=worker_callback,
**common_kwargs
)
for result in multithreaded_iterator:
index, row = result.item
if not result.url:
write_fetch_output(
index,
row
)
loading_bar.update()
continue
# Updating stats
update_loading_bar(result)
# No error
if result.error is None:
meta = result.meta
# Final url target
resolved_url = result.resolved
if resolved_url == result.url:
resolved_url = None
# Reporting in output
write_fetch_output(
index,
row,
resolved=resolved_url,
status=result.response.status,
filename=meta.get('filename'),
encoding=meta.get('encoding'),
mimetype=meta.get('mimetype'),
data=meta.get('decoded_contents')
)
# Handling potential errors
else:
error_code = report_error(result.error)
resolved = None
if isinstance(result.error, InvalidURLError):
resolved = result.error.url
if isinstance(result.error, FilenameFormattingError):
loading_bar.print(report_filename_formatting_error(result.error))
write_fetch_output(
index,
row,
error=error_code,
resolved=resolved
)
# Resolve
else:
multithreaded_iterator = multithreaded_resolve(
enricher,
resolve_args=request_args,
follow_meta_refresh=cli_args.follow_meta_refresh,
follow_js_relocation=cli_args.follow_js_relocation,
infer_redirection=cli_args.infer_redirection,
**common_kwargs
)
for result in multithreaded_iterator:
index, row = result.item
if not result.url:
write_resolve_output(
index,
row
)
loading_bar.update()
continue
# Updating stats
update_loading_bar(result)
# No error
if result.error is None:
# Reporting in output
last = result.stack[-1]
write_resolve_output(
index,
row,
resolved=last.url,
status=last.status,
redirects=len(result.stack) - 1,
chain='|'.join(step.type for step in result.stack)
)
# Handling potential errors
else:
error_code = report_error(result.error)
write_resolve_output(
index,
row,
error=error_code,
redirects=(len(result.stack) - 1) if result.stack else None,
chain='|'.join(step.type for step in result.stack) if result.stack else None
)
def fetch_action(namespace):
# Are we resuming
resuming = namespace.resume
if resuming and not namespace.output:
die(['Cannot --resume without specifying -o/--output.'])
# Do we need to fetch only a single url?
if namespace.file is sys.stdin and is_url(namespace.column):
namespace.file = StringIO('url\n%s' % namespace.column)
namespace.column = 'url'
# If we are hitting a single url we enable contents_in_report
if namespace.contents_in_report is None:
namespace.contents_in_report = True
input_headers, pos, reader = custom_reader(namespace.file,
namespace.column)
filename_pos = input_headers.index(
namespace.filename) if namespace.filename else None
indexed_input_headers = {h: p for p, h in enumerate(input_headers)}
selected_fields = namespace.select.split(',') if namespace.select else None
selected_pos = [input_headers.index(h)
for h in selected_fields] if selected_fields else None
# HTTP method
http_method = namespace.method
# Cookie grabber
get_cookie = None
if namespace.grab_cookies:
get_cookie = grab_cookies(namespace.grab_cookies)
# Global headers
global_headers = None
if namespace.headers:
global_headers = {}
for header in namespace.headers:
k, v = parse_http_header(header)
global_headers = v
# Reading output
output_headers = (list(input_headers) if not selected_pos else
[input_headers[i] for i in selected_pos])
output_headers += OUTPUT_ADDITIONAL_HEADERS
if namespace.contents_in_report:
output_headers.append('raw_content')
flag = 'w'
if namespace.output is not None and resuming and isfile(namespace.output):
flag = 'r+'
output_file = open_output_file(namespace.output, flag=flag)
output_writer = csv.writer(output_file)
if not resuming:
output_writer.writerow(output_headers)
else:
# Reading report to know what need to be done
_, rpos, resuming_reader = custom_reader(output_file, 'line')
resuming_reader_loading = tqdm(resuming_reader,
desc='Resuming',
dynamic_ncols=True,
unit=' lines')
already_done = ContiguousRangeSet()
for line in resuming_reader_loading:
index = line[rpos]
already_done.add(int(index))
# Loading bar
total = namespace.total
if total is not None and resuming:
total -= len(already_done)
loading_bar = tqdm(desc='Fetching pages',
total=total,
dynamic_ncols=True,
unit=' urls')
def url_key(item):
line = item[1]
url = line[pos].strip()
if not url:
return
# Url templating
if namespace.url_template:
return namespace.url_template.format(value=url)
return url
def request_args(url, item):
cookie = None
# Cookie
if get_cookie:
cookie = get_cookie(url)
# Headers
headers = None
if global_headers:
headers = global_headers
return {'method': http_method, 'cookie': cookie, 'headers': headers}
def write_output(index,
line,
resolved=None,
status=None,
error=None,
filename=None,
encoding=None,
data=None):
if selected_pos:
line = [line[p] for p in selected_pos]
line.extend([
index, resolved or '', status or '', error or '', filename or '',
encoding or ''
])
if namespace.contents_in_report:
line.append(data or '')
output_writer.writerow(line)
errors = 0
status_codes = Counter()
target_iterator = enumerate(reader)
if resuming:
target_iterator = (pair for pair in target_iterator
if not already_done.stateful_contains(pair[0]))
multithreaded_iterator = multithreaded_fetch(target_iterator,
key=url_key,
request_args=request_args,
threads=namespace.threads,
throttle=namespace.throttle)
for result in multithreaded_iterator:
line_index, line = result.item
if not result.url:
write_output(line_index, line)
loading_bar.update()
continue
response = result.response
data = response.data if response is not None else None
content_write_flag = 'wb'
# Updating stats
if result.error is not None:
errors += 1
else:
if response.status >= 400:
status_codes[response.status] += 1
postfix = {'errors': errors}
for code, count in status_codes.most_common(1):
postfix[str(code)] = count
loading_bar.set_postfix(**postfix)
loading_bar.update()
# No error
if result.error is None:
filename = None
# Building filename
if data:
if filename_pos is not None or namespace.filename_template:
if namespace.filename_template:
filename = CUSTOM_FORMATTER.format(
namespace.filename_template,
value=line[filename_pos]
if filename_pos is not None else None,
ext=result.meta['ext'],
line=LazyLineDict(indexed_input_headers, line))
else:
filename = line[filename_pos] + result.meta['ext']
else:
# NOTE: it would be nice to have an id that can be sorted by time
filename = str(uuid4()) + result.meta['ext']
# Standardize encoding?
encoding = result.meta['encoding']
if data and namespace.standardize_encoding or namespace.contents_in_report:
if encoding is None or encoding != 'utf-8' or namespace.contents_in_report:
data = data.decode(
encoding if encoding is not None else 'utf-8',
errors='replace')
encoding = 'utf-8'
content_write_flag = 'w'
# Writing file on disk
if data and not namespace.contents_in_report:
if namespace.compress:
filename += '.gz'
resource_path = join(namespace.output_dir, filename)
resource_dir = dirname(resource_path)
os.makedirs(resource_dir, exist_ok=True)
with open(resource_path, content_write_flag) as f:
# TODO: what if standardize_encoding + compress?
f.write(
gzip.compress(data) if namespace.compress else data)
# Reporting in output
resolved_url = response.geturl()
write_output(
line_index,
line,
resolved=resolved_url if resolved_url != result.url else None,
status=response.status,
filename=filename,
encoding=encoding,
data=data)
# Handling potential errors
else:
error_code = report_error(result.error)
write_output(line_index, line, error=error_code)
# Closing files
if namespace.output is not None:
output_file.close()
def fetch_action(namespace):
# Are we resuming
resuming = namespace.resume
if resuming and not namespace.output:
die(['Cannot --resume without specifying -o/--output.'])
# Do we need to fetch only a single url?
single_url = namespace.file is sys.stdin and is_url(namespace.column)
if single_url:
edit_namespace_with_csv_io(namespace, 'url')
# If we are hitting a single url we enable contents_in_report
if namespace.contents_in_report is None:
namespace.contents_in_report = True
# HTTP method
http_method = namespace.method
# Cookie grabber
get_cookie = None
if namespace.grab_cookies:
get_cookie = grab_cookies(namespace.grab_cookies)
# Global headers
global_headers = None
if namespace.headers:
global_headers = {}
for header in namespace.headers:
k, v = parse_http_header(header)
global_headers = v
flag = 'w'
if namespace.output is not None and resuming and isfile(namespace.output):
flag = 'r+'
output_file = open_output_file(namespace.output, flag=flag)
# Resume listener
listener = None
resuming_reader_loading = None
skipped = 0
if resuming:
resuming_reader_loading = tqdm(desc='Resuming',
dynamic_ncols=True,
unit=' lines')
def listener(event, row):
nonlocal skipped
if event == 'resume.output':
resuming_reader_loading.update()
if event == 'resume.input':
skipped += 1
loading_bar.set_postfix(skipped=skipped)
loading_bar.update()
# Enricher
enricher = casanova.threadsafe_enricher(
namespace.file,
output_file,
resumable=resuming,
auto_resume=False,
add=OUTPUT_ADDITIONAL_HEADERS +
(['raw_contents'] if namespace.contents_in_report else []),
keep=namespace.select,
listener=listener)
if namespace.column not in enricher.pos:
die([
'Could not find the "%s" column containing the urls in the given CSV file.'
% namespace.column
])
url_pos = enricher.pos[namespace.column]
filename_pos = None
if namespace.filename is not None:
if namespace.filename not in enricher.pos:
die([
'Could not find the "%s" column containing the filenames in the given CSV file.'
% namespace.filename
])
filename_pos = enricher.pos[namespace.filename]
indexed_input_headers = {h: i for i, h in enumerate(enricher.fieldnames)}
if resuming:
enricher.resume()
resuming_reader_loading.close()
# Loading bar
total = namespace.total
loading_bar = tqdm(desc='Fetching pages',
total=total,
dynamic_ncols=True,
unit=' urls')
def url_key(item):
url = item[1][url_pos].strip()
if not url:
return
# Url templating
if namespace.url_template:
return namespace.url_template.format(value=url)
return url
def request_args(url, item):
cookie = None
# Cookie
if get_cookie:
cookie = get_cookie(url)
# Headers
headers = None
if global_headers:
headers = global_headers
return {'method': http_method, 'cookie': cookie, 'headers': headers}
def write_output(index,
row,
resolved=None,
status=None,
error=None,
filename=None,
encoding=None,
data=None):
addendum = [
resolved or '', status or '', error or '', filename or '', encoding
or ''
]
if namespace.contents_in_report:
addendum.append(data or '')
enricher.writerow(index, row, addendum)
errors = 0
status_codes = Counter()
fetch_kwargs = {
'threads': namespace.threads,
'throttle': namespace.throttle,
'domain_parallelism': namespace.domain_parallelism
}
if namespace.timeout is not None:
fetch_kwargs['timeout'] = namespace.timeout
multithreaded_iterator = multithreaded_fetch(enricher,
key=url_key,
request_args=request_args,
**fetch_kwargs)
for result in multithreaded_iterator:
index, row = result.item
if not result.url:
write_output(index, row)
loading_bar.update()
continue
response = result.response
data = response.data if response is not None else None
content_write_flag = 'wb'
# Updating stats
if result.error is not None:
errors += 1
else:
if response.status >= 400:
status_codes[response.status] += 1
postfix = {'errors': errors}
for code, count in status_codes.most_common(1):
postfix[str(code)] = count
loading_bar.set_postfix(**postfix)
loading_bar.update()
# No error
if result.error is None:
filename = None
# Building filename
if data:
if filename_pos is not None or namespace.filename_template:
if namespace.filename_template:
filename = CUSTOM_FORMATTER.format(
namespace.filename_template,
value=row[filename_pos]
if filename_pos is not None else None,
ext=result.meta['ext'],
line=LazyLineDict(indexed_input_headers, row))
else:
filename = row[filename_pos] + result.meta['ext']
else:
# NOTE: it would be nice to have an id that can be sorted by time
filename = str(uuid4()) + result.meta['ext']
# Standardize encoding?
encoding = result.meta['encoding']
if data and namespace.standardize_encoding or namespace.contents_in_report:
if encoding is None or encoding != 'utf-8' or namespace.contents_in_report:
data = data.decode(
encoding if encoding is not None else 'utf-8',
errors='replace')
encoding = 'utf-8'
content_write_flag = 'w'
# Writing file on disk
if data and not namespace.contents_in_report:
if namespace.compress:
filename += '.gz'
resource_path = join(namespace.output_dir, filename)
resource_dir = dirname(resource_path)
os.makedirs(resource_dir, exist_ok=True)
with open(resource_path, content_write_flag) as f:
# TODO: what if standardize_encoding + compress?
f.write(
gzip.compress(data) if namespace.compress else data)
# Reporting in output
resolved_url = response.geturl()
write_output(
index,
row,
resolved=resolved_url if resolved_url != result.url else None,
status=response.status,
filename=filename,
encoding=encoding,
data=data)
# Handling potential errors
else:
error_code = report_error(result.error)
write_output(index, row, error=error_code)
# Closing files
output_file.close()
