90 lines
3.8 KiB
Python
Executable File
90 lines
3.8 KiB
Python
Executable File
#!/usr/bin/env python
|
|
"""
|
|
Fourmi, a web scraper build to search specific information for a given compound (and its pseudonyms).
|
|
|
|
Usage:
|
|
fourmi
|
|
fourmi search <compound>
|
|
fourmi [options] search <compound>
|
|
fourmi [options] [-v | -vv | -vvv] [--include=<sourcename> | --exclude=<sourcename>] search <compound>
|
|
fourmi list
|
|
fourmi [--include=<sourcename> | --exclude=<sourcename>] list
|
|
fourmi -h | --help
|
|
fourmi --version
|
|
|
|
Options:
|
|
--attributes=<regex> Include only that match these regular expressions split by a comma. [default: .*]
|
|
-h --help Show this screen.
|
|
--version Show version.
|
|
-v Verbose logging output. (Multiple occurrences increase logging level)
|
|
--log=<file> Save log to an file.
|
|
-o <file> --output=<file> Output file [default: <compound>.*format*]
|
|
-f <format> --format=<format> Output formats (supported: csv, json, jsonlines, xml) [default: csv]
|
|
--include=<regex> Include only sources that match these regular expressions split by a comma.
|
|
--exclude=<regex> Exclude the sources that match these regular expressions split by a comma.
|
|
"""
|
|
|
|
from twisted.internet import reactor
|
|
from scrapy.crawler import Crawler
|
|
from scrapy import signals, log
|
|
import docopt
|
|
|
|
from FourmiCrawler.spider import FourmiSpider
|
|
from utils.configurator import Configurator
|
|
from utils.sourceloader import SourceLoader
|
|
from GUI import gui
|
|
|
|
|
|
def setup_crawler(compound, settings, source_loader, attributes):
|
|
"""
|
|
This function prepares and start the crawler which starts the actual search on the internet
|
|
:param compound: The compound which should be searched
|
|
:param settings: A scrapy settings object
|
|
:param source_loader: A fully functional SourceLoader object which contains only the sources that should be used.
|
|
:param attributes: A list of regular expressions which the attribute names should match.
|
|
"""
|
|
spider = FourmiSpider(compound=compound, selected_attributes=attributes)
|
|
spider.add_sources(source_loader.sources)
|
|
crawler = Crawler(settings)
|
|
crawler.signals.connect(reactor.stop, signal=signals.spider_closed)
|
|
crawler.configure()
|
|
crawler.crawl(spider)
|
|
crawler.start()
|
|
|
|
|
|
def search(docopt_arguments, source_loader):
|
|
"""
|
|
The function that facilitates the search for a specific compound.
|
|
:param docopt_arguments: A dictionary generated by docopt containing all CLI arguments.
|
|
:param source_loader: An initiated SourceLoader object pointed at the directory with the sources.
|
|
"""
|
|
conf = Configurator()
|
|
conf.set_logging(docopt_arguments["--log"], docopt_arguments["-v"])
|
|
conf.set_output(docopt_arguments["--output"], docopt_arguments["--format"], docopt_arguments["<compound>"])
|
|
setup_crawler(docopt_arguments["<compound>"], conf.scrapy_settings,
|
|
source_loader, docopt_arguments["--attributes"].split(','))
|
|
if conf.scrapy_settings.getbool("LOG_ENABLED"):
|
|
log.start(conf.scrapy_settings.get("LOG_FILE"),
|
|
conf.scrapy_settings.get("LOG_LEVEL"), conf.scrapy_settings.get("LOG_STDOUT"))
|
|
reactor.run()
|
|
|
|
|
|
# The start for the Fourmi Command Line interface.
|
|
if __name__ == '__main__':
|
|
arguments = docopt.docopt(__doc__, version='Fourmi - V0.6.0')
|
|
loader = SourceLoader()
|
|
|
|
if arguments["--include"]:
|
|
loader.include(arguments["--include"].split(','))
|
|
elif arguments["--exclude"]:
|
|
loader.exclude(arguments["--exclude"].split(','))
|
|
|
|
if arguments["search"]:
|
|
search(arguments, loader)
|
|
elif arguments["list"]:
|
|
print "-== Available Sources ==-"
|
|
print str(loader)
|
|
else:
|
|
gui_window = gui.GUI(search, sourceloader=SourceLoader())
|
|
gui_window.run()
|