diff --git a/FourmiCrawler/pipelines.py b/FourmiCrawler/pipelines.py index 5f2b68f..e1dadbf 100644 --- a/FourmiCrawler/pipelines.py +++ b/FourmiCrawler/pipelines.py @@ -2,10 +2,11 @@ # # Don't forget to add your pipeline to the ITEM_PIPELINES setting # See: http://doc.scrapy.org/en/latest/topics/item-pipeline.html +import re from scrapy.exceptions import DropItem -class FourmiPipeline(object): +class DuplicatePipeline(object): def __init__(self): self.known_values = set() @@ -17,9 +18,27 @@ class FourmiPipeline(object): :param spider: The spider which scraped the spider :return: :raise DropItem: Returns the item if unique or drops them if it's already known """ - value = item['attribute'], item['value'] + value = (item['attribute'], item['value'], item['conditions']) if value in self.known_values: raise DropItem("Duplicate item found: %s" % item) # #[todo] append sources of first item. else: self.known_values.add(value) return item + +class AttributeSelectionPipeline(object): + + def __init__(self): + pass; + + def process_item(self, item, spider): + """ + The items are processed using the selected attribute list available in the spider, + items that don't match the selected items are dropped. + :param item: The incoming item + :param spider: The spider which scraped the item. Should have an attribute "selected_attributes". + :return: :raise DropItem: Returns item if it matches an selected attribute, else it is dropped. + """ + if [x for x in spider.selected_attributes if re.match(x, item["attribute"])]: + return item + else: + raise DropItem("Attribute not selected by used: %s" % item) \ No newline at end of file diff --git a/FourmiCrawler/settings.py b/FourmiCrawler/settings.py index be91fef..d7ac212 100644 --- a/FourmiCrawler/settings.py +++ b/FourmiCrawler/settings.py @@ -11,7 +11,8 @@ BOT_NAME = 'FourmiCrawler' SPIDER_MODULES = ['FourmiCrawler'] NEWSPIDER_MODULE = 'FourmiCrawler' ITEM_PIPELINES = { - 'FourmiCrawler.pipelines.FourmiPipeline': 100 + 'FourmiCrawler.pipelines.AttributeSelectionPipeline': 100, + 'FourmiCrawler.pipelines.DuplicatePipeline': 200, } FEED_URI = 'results.json' FEED_FORMAT = 'jsonlines' diff --git a/FourmiCrawler/sources/ChemSpider.py b/FourmiCrawler/sources/ChemSpider.py index a62f6dd..2fcd07c 100644 --- a/FourmiCrawler/sources/ChemSpider.py +++ b/FourmiCrawler/sources/ChemSpider.py @@ -47,7 +47,6 @@ class ChemSpider(Source): properties = [] # Predicted - ACD/Labs tab - # [TODO] - test if tab contains data, some chemicals do not have data here td_list = sel.xpath('.//table[@id="acdlabs-table"]//td').xpath( 'normalize-space(string())') prop_names = td_list[::2] @@ -58,6 +57,12 @@ class ChemSpider(Source): prop_value = prop_value.extract().encode('utf-8') prop_conditions = '' + # Test for properties without values, with one hardcoded exception + if (not re.match(r'^\d', prop_value) or + (prop_name == 'Polarizability' and + prop_value == '10-24cm3')): + continue + # Match for condition in parentheses m = re.match(r'(.*) \((.*)\)', prop_name) if m: @@ -192,7 +197,8 @@ class ChemSpider(Source): 'reliability': 'Unknown', 'conditions': '' }) - properties.append(result) + if result['value']: + properties.append(result) return properties def parse_searchrequest(self, response): @@ -200,8 +206,14 @@ class ChemSpider(Source): sel = Selector(response) log.msg('chemspider parse_searchrequest', level=log.DEBUG) sel.register_namespace('cs', 'http://www.chemspider.com/') - csid = sel.xpath('.//cs:int/text()').extract()[0] - # [TODO] - handle multiple csids in case of vague search term + csids = sel.xpath('.//cs:int/text()').extract() + if len(csids) == 0: + log.msg('ChemSpider found nothing', level=log.ERROR) + return + elif len(csids) > 1: + log.msg('ChemSpider found multiple substances, taking first ' + 'element', level=log.DEBUG) + csid = csids[0] structure_url = self.website[:-1] + self.structure % csid extendedinfo_url = self.website[:-1] + self.extendedinfo % csid log.msg('chemspider URL: %s' % structure_url, level=log.DEBUG) @@ -215,4 +227,4 @@ class ChemSpider(Source): return None searchurl = self.website[:-1] + self.search % compound log.msg('chemspider compound', level=log.DEBUG) - return Request(url=searchurl, callback=self.parse_searchrequest) \ No newline at end of file + return Request(url=searchurl, callback=self.parse_searchrequest) diff --git a/FourmiCrawler/sources/WikipediaParser.py b/FourmiCrawler/sources/WikipediaParser.py index c251fca..cb7d0b9 100644 --- a/FourmiCrawler/sources/WikipediaParser.py +++ b/FourmiCrawler/sources/WikipediaParser.py @@ -36,8 +36,8 @@ class WikipediaParser(Source): """ scrape data from infobox on wikipedia. """ items = [] - #be sure to get both chembox (wikipedia template) and drugbox (wikipedia template) to scrape - tr_list = sel.xpath('.//table[@class="infobox bordered" or @class="infobox"]//td[not(@colspan)]').\ + #be sure to get chembox (wikipedia template) + tr_list = sel.xpath('.//table[@class="infobox bordered"]//td[not(@colspan)]'). \ xpath('normalize-space(string())') prop_names = tr_list[::2] prop_values = tr_list[1::2] @@ -46,11 +46,31 @@ class WikipediaParser(Source): 'attribute': prop_name.extract().encode('utf-8'), 'value': prop_values[i].extract().encode('utf-8'), 'source': "Wikipedia", - 'reliability': "", + 'reliability': "Unknown", 'conditions': "" }) items.append(item) log.msg('Wiki prop: |%s| |%s| |%s|' % (item['attribute'], item['value'], item['source']), level=log.DEBUG) + + #scrape the drugbox (wikipedia template) + tr_list2 = sel.xpath('.//table[@class="infobox"]//tr') + log.msg('dit: %s' % tr_list2, level=log.DEBUG) + for tablerow in tr_list2: + log.msg('item: %s' % tablerow.xpath('./th').xpath('normalize-space(string())'), level=log.DEBUG) + if tablerow.xpath('./th').xpath('normalize-space(string())') and tablerow.xpath('./td').xpath( + 'normalize-space(string())'): + item = Result({ + 'attribute': tablerow.xpath('./th').xpath('normalize-space(string())').extract()[0].encode('utf-8'), + 'value': tablerow.xpath('./td').xpath('normalize-space(string())').extract()[0].encode('utf-8'), + 'source': "Wikipedia", + 'reliability': "Unknown", + 'conditions': "" + }) + items.append(item) + log.msg( + 'Wiki prop: |attribute: %s| |value: %s| |%s|' % (item['attribute'], item['value'], item['source']), + level=log.DEBUG) + items = filter(lambda a: a['value'] != '', items) # remove items with an empty value item_list = self.clean_items(items) diff --git a/FourmiCrawler/spider.py b/FourmiCrawler/spider.py index 9f92a84..87f22c6 100644 --- a/FourmiCrawler/spider.py +++ b/FourmiCrawler/spider.py @@ -8,9 +8,10 @@ class FourmiSpider(Spider): __parsers = [] synonyms = [] - def __init__(self, compound=None, *args, **kwargs): + def __init__(self, compound=None, selected_attributes=[".*"], *args, **kwargs): super(FourmiSpider, self).__init__(*args, **kwargs) self.synonyms.append(compound) + self.selected_attributes = selected_attributes; def parse(self, reponse): for parser in self.__parsers: diff --git a/README.md b/README.md new file mode 100644 index 0000000..af012fd --- /dev/null +++ b/README.md @@ -0,0 +1,81 @@ +# Fourmi + +Fourmi is an web scraper for chemical substances. The program is designed to be +used as a search engine to search multiple chemical databases for a specific +substance. The program will produce all available attributes of the substance +and conditions associated with the attributes. Fourmi also attempts to estimate +the reliability of each data point to assist the user in deciding which data +should be used. + +The Fourmi project is open source project licensed under the MIT license. Feel +free to contribute! + +Fourmi is based on the [Scrapy framework](http://scrapy.org/), an open source +web scraping framework for python. Most of the functionality of this project can +be traced to this framework. Should the documentation for this application fall +short, we suggest you take a close look at the [Scrapy architecture] +(http://doc.scrapy.org/en/latest/topics/architecture.html) and the [Scrapy +documentation](http://doc.scrapy.org/en/latest/index.html). + +### Installing + +If you're installing Fourmi, please take a look at our [installation guide](...) +on our wiki. When you've installed the application, make sure to check our +[usage guide](...). + +### Using the Source + +To use the Fourmi source code multiple dependencies are required. Take a look at +the [wiki page](...) on using the application source code for a step by step +installation guide. + +When developing for the Fourmi project keep in mind that code readability is a +must. To maintain the readability, code should be conform with the +[PEP-8](http://legacy.python.org/dev/peps/pep-0008/) style guide for Python +code. More information about the different structures and principles of the +Fourmi application can be found on our [wiki](...). + +### To Do + +The Fourmi project has the following goals for the nearby future: + +__Main goals:__ + +- Improve our documentation and guides. (Assignee: Dekker) +- Build an graphical user interface(GUI) as alternative for the command line +interface(CLI). (Assignee: Harmen) +- Compiling the source into an windows executable. (Assignee: Bas) +- Create an configuration file to hold logins and API keys. +- Determine reliability of our data point. +- Create an module to gather data from NIST. (Assignee: Rob) +- Create an module to gather data from PubChem. (Assignee: Rob) + +__Side goals:__ + +- Clean and unify data. +- Extensive reliability analysis using statistical tests. +- Test data with Descartes 1. + +### Project Origin + +The Fourmi project was started in February of 2014 as part of a software +engineering course at the Radboud University for students studying Computer +Science, Information Science or Artificial Intelligence. Students participate in +a real software development project as part of the +[Giphouse](http://www.giphouse.nl/). + +This particular project was started on behalf of Ivo B. Rietveld. As a chemist +he was in need of an application to automatically search information on chemical +substances and create an phase diagram. The so called "Descrates" project was +split into two teams each creating a different application that has part of the +functionality. We are the team Descartes 2 and as we were responsible for +creating a web crawler, we've named our application Fourmi (Englis: Ants). + +The following people were part of the original team: + +- [Jip J. Dekker](http://jip.dekker.li) +- Rob ten Berge +- Harmen Prins +- Bas van Berkel +- Nout van Deijck +- Michail Kuznetcov \ No newline at end of file diff --git a/README.rst b/README.rst deleted file mode 100644 index c251791..0000000 --- a/README.rst +++ /dev/null @@ -1,16 +0,0 @@ -We are the team Descartes 2. ----------------------------- - -Our team members are: - -+ Rob ten Berge - -+ Bas van Berkel - -+ Nout van Deijck - -+ Jip J. Dekker - -+ Michail Kuznetcov - -+ Harmen Prins \ No newline at end of file diff --git a/fourmi.py b/fourmi.py index e33a833..efa4e54 100755 --- a/fourmi.py +++ b/fourmi.py @@ -12,14 +12,15 @@ Usage: fourmi --version Options: + --attributes= Include only that match these regular expressions split by a comma. [default: .*] -h --help Show this screen. --version Show version. --verbose Verbose logging output. --log= Save log to an file. -o --output= Output file [default: result.*format*] -f --format= Output formats (supported: csv, json, jsonlines, xml) [default: jsonlines] - --include= Include only sources that match the regular these expressions split by a comma. - --exclude= Exclude the sources that match the regular these expressions split by a comma. + --include= Include only sources that match these regular expressions split by a comma. + --exclude= Exclude the sources that match these regular expressions split by a comma. """ from twisted.internet import reactor @@ -32,8 +33,8 @@ from FourmiCrawler.spider import FourmiSpider from sourceloader import SourceLoader -def setup_crawler(searchable, settings, source_loader): - spider = FourmiSpider(compound=searchable) +def setup_crawler(searchable, settings, source_loader, attributes): + spider = FourmiSpider(compound=searchable, selected_attributes=attributes) spider.add_parsers(source_loader.sources) crawler = Crawler(settings) crawler.signals.connect(reactor.stop, signal=signals.spider_closed) @@ -74,12 +75,12 @@ def start_log(docopt_arguments): def search(docopt_arguments, source_loader): start_log(docopt_arguments) settings = scrapy_settings_manipulation(docopt_arguments) - setup_crawler(docopt_arguments[""], settings, source_loader) + setup_crawler(docopt_arguments[""], settings, source_loader, docopt_arguments["--attributes"].split(',')) reactor.run() if __name__ == '__main__': - arguments = docopt.docopt(__doc__, version='Fourmi - V0.2.6') + arguments = docopt.docopt(__doc__, version='Fourmi - V0.3.0') loader = SourceLoader() if arguments["--include"]: