json_raw = res.group(2)
break
return json.loads(json_raw)
@staticmethod
def write_data(fpath, data, delimiter=";"):
with open(fpath, "a", encoding="utf8") as f:
csv_writer = csv.writer(f, delimiter=delimiter)
for row in data:
csv_writer.writerow(row.values())
@staticmethod
def parse(instance, fpath):
data = ParseJavaScriptJsonToCSV.get_data(instance)
ParseJavaScriptJsonToCSV.write_data(fpath, data)
@staticmethod
def path(srconf, jobconf, dpath):
name = Box(json.loads(srconf)).name
data_format = Box(json.loads(jobconf)).data_format
return os.path.join(dpath, "{}.{}".format(name, data_format))
HandlersFactory.register("xlsparse_to_csv", ParseFromExcelToCSV)
HandlersFactory.register("xlsparse_oked_to_csv", ParseOkedToCsv)
HandlersFactory.register("web_api_raw_json_parse_to_csv", ParseFromAPIToCSV)
HandlersFactory.register("web_html_javascript_json_parse_to_csv",
ParseJavaScriptJsonToCSV)
HandlersFactory.register("web_html_table_text_parse_to_csv",
ParseGosRegisterToCSV)
i = -1
for arch in archives:
arch_obj = Utils.get_archive_object(arch)
data_format = Box(json.loads(instance.srconf)).storage.data_format
file_path = path.abspath(path.dirname(arch))
for file in arch_obj.namelist():
if Utils.ext(file) == data_format:
i += 1
arch_obj.extract(file, file_path)
old_path = path.join(file_path, file).replace('/', os.sep)
shutil.move(old_path, targets[i])
return targets
@staticmethod
def path(srconf, dpath):
files_num = Box(json.loads(srconf)).storage.data_files_num
archives_num = Box(json.loads(srconf)).storage.data_archives_num
name = Box(json.loads(srconf)).name
data_format = Box(json.loads(srconf)).storage.data_format
files = list()
for i in range(archives_num):
for j in range(files_num):
files.append(
path.join(dpath,
"{}_{}_{}.{}".format(name, i, j, data_format)))
return files
HandlersFactory.register("extract_file", ExtractorFile)
HandlersFactory.register("extract_files", ExtractorFiles)
return fpath
except Exception as e:
raise e
@staticmethod
def urls(conf):
url = Box(json.loads(conf)).url
base_url = Box(json.loads(conf)).base_url
js = requests.get(url, verify=False).text
regex = Box(json.loads(conf)).storage.html.url_regexp
res = re.findall(regex, js)
urls = ["{}{}".format(base_url, res) for res in res]
return urls
@staticmethod
def path(conf):
name = Box(json.loads(conf)).name
ext = Box(json.loads(conf)).storage.type
urls = DownloaderByUrlStatGovCompanies().urls(conf)
directory = TEMP_PATH
os.makedirs(directory, exist_ok=True)
return [
os.path.join(directory, "{}_{}.{}".format(name, i, ext))
for i, url in enumerate(urls)
]
HandlersFactory.register("download_url", DownloaderByUrlToFile)
HandlersFactory.register("download_urllist_ajax",
DownloaderByUrlStatGovCompanies)