From af84b5d968dcceaf74dc76b9ee720b384d8277e5 Mon Sep 17 00:00:00 2001 From: Dnomd343 Date: Sun, 16 Oct 2022 01:07:41 +0800 Subject: [PATCH] feat: fetch script for `ixsw.la` --- src/ixsw.la/fetch.py | 53 +++++++++++++++++++++++++++++++++++++++++++ src/ixsw.la/logger.py | 47 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 100 insertions(+) create mode 100644 src/ixsw.la/fetch.py create mode 100644 src/ixsw.la/logger.py diff --git a/src/ixsw.la/fetch.py b/src/ixsw.la/fetch.py new file mode 100644 index 0000000..0c235ac --- /dev/null +++ b/src/ixsw.la/fetch.py @@ -0,0 +1,53 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + +""" +Download raw html content as `.html` files. + + USAGE: python3 fetch.py [CATALOG] [OUTPUT_DIR] +""" + +import os +import sys +import json +import time +import requests +from logger import logger + +userAgent = ( # default user agent + 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) ' + 'AppleWebKit/537.36 (KHTML, like Gecko) ' + 'Chrome/106.0.0.0 Safari/537.36 Edg/106.0.1370.47' +) + + +def httpRequest(fileUrl: str, fileName: str) -> bool: # save html content + try: + logger.debug('Http request `%s` -> `%s`' % (fileUrl, fileName)) + request = requests.get(fileUrl, timeout = 30, # timeout -> 30s + headers = { + 'user-agent': userAgent, # with fake user-agent + } + ) + if request.status_code not in range(200, 300): # http status code 2xx + logger.warning('Http request failed -> `%s`' % fileUrl) + return False + logger.debug('Http request success -> `%s`' % fileUrl) + with open(fileName, 'wb') as fileObj: # save html content + fileObj.write(request.content) + logger.debug('File save success -> `%s`' % fileName) + except: + return False + return True + + +catalog = json.loads(open(sys.argv[1]).read()) # load catalog + +for _, chapterId in catalog.items(): # traverse all chapters + pageUrl = 'https://www.ixsw.la/ks82668/%s.html' % chapterId + pageFile = os.path.join(sys.argv[2], '%s.html' % chapterId) + if httpRequest(pageUrl, pageFile): # save html content + logger.info('Page request success -> %s' % pageUrl) + else: + logger.error('Page request failed -> %s' % pageUrl) + time.sleep(1) # avoid being blocked by the server diff --git a/src/ixsw.la/logger.py b/src/ixsw.la/logger.py new file mode 100644 index 0000000..36cfaf8 --- /dev/null +++ b/src/ixsw.la/logger.py @@ -0,0 +1,47 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + +""" +Logger module: + INFO level -> stderr + DEBUG level -> runtime.log +""" + +import sys +import logging +import colorlog + +logColor = { # log color + 'DEBUG': 'white', + 'INFO': 'green', + 'WARNING': 'yellow', + 'ERROR': 'red', + 'CRITICAL': 'bold_red', +} + +dateFormat = '%Y-%m-%d %H:%M:%S' +timeFormat = '%(asctime)s.%(msecs)03d' +logFormat = '[%(levelname)s] %(message)s (%(module)s.%(funcName)s:%(lineno)d)' + +# load fileHandler -> log file +fileHandler = logging.FileHandler('runtime.log', encoding = 'utf-8') +fileHandler.setFormatter(logging.Formatter( + '[' + timeFormat + '] ' + logFormat, + datefmt = dateFormat +)) +fileHandler.setLevel(logging.DEBUG) # debug level for log file + +# load stdHandler -> stderr +stdHandler = colorlog.StreamHandler() +stdHandler.setFormatter(colorlog.ColoredFormatter( + '%(light_black)s' + timeFormat + '%(log_color)s ' + logFormat, + datefmt = dateFormat, + log_colors = logColor, + stream = sys.stderr +)) +stdHandler.setLevel(logging.INFO) # info level for stderr + +logger = logging.getLogger() +logger.addHandler(stdHandler) +logger.addHandler(fileHandler) +logger.setLevel(logging.DEBUG) # set log level in handler