def requires(self):
return [
Executable(name='metha-sync',
message='https://github.com/miku/metha'),
Executable(name='span-import',
message='https://github.com/miku/span'),
]
def requires(self):
"""
If we have more DOI sources, we could add them as requirements here.
"""
return {
'input':
CrossrefDOIList(date=datetime.date.today()),
'hurrly':
Executable(name='hurrly', message='http://github.com/miku/hurrly'),
'pigz':
Executable(name='pigz', message='http://zlib.net/pigz/')
}
def requires(self):
return {
'input': CrossrefIntermediateSchema(begin=self.begin,
date=self.date),
'jq': Executable(name='jq',
message='https://github.com/stedolan/jq')
}
def requires(self):
return {
'span-import':
Executable(name='span-import', message='http://git.io/vI8NV'),
'input':
DOAJHarvest(date=self.date),
}
def requires(self):
return [(Executable(name='bin/grobid2json.py'),
Executable(name='jq')),
ItemGrobidTarballs(crawl=self.crawl, item=self.item)]
def requires(self):
return [(Executable(name='bin/grobid2json.py'),
Executable(name='jq')),
ItemGrobidServiceExtract(crawl=self.crawl, item=self.item)]
def requires(self):
return [(Executable(name='java'),
Executable(name='/usr/bin/time')),
ItemPdfs(crawl=self.crawl, item=self.item)]
def requires(self):
return [Executable(name='bin/warc_extract_filter.py'),
ItemWarcDownload(crawl=self.crawl, item=self.item)]
def requires(self):
return Executable('rsync', message='https://rsync.samba.org/')
def requires(self):
return {
'input': ElsevierJournalsIntermediateSchema(date=self.date),
'jq': Executable(name='jq',
message='https://github.com/stedolan/jq')
}
def requires(self):
return {
'input': DOAJIntermediateSchema(date=self.date),
'jq': Executable(name='jq', message='http://git.io/NYpfTw'),
}
def requires(self):
return {'dump': GutenbergDump(date=self.date),
'apps': Executable(name='marctotsv',
message='https://github.com/miku/gomarckit')}
def requires(self):
return [Executable(name='wget'), Executable(name='bunzip2')]
def requires(self):
"""
The setup of siskin should install this automatically.
"""
return Executable(name='ia',
message='https://pypi.org/project/internetarchive/')
def requires(self):
return [Executable(name='ia')]
def requires(self):
return {
'span': Executable(name='span-import',
message='http://git.io/vI8NV'),
'file': CrossrefUniqItems(begin=self.begin, date=self.date)
}
def requires(self):
""" Outsource the download. """
return Executable(name='wget')
def requires(self):
return Executable(name='metha-sync',
message='https://github.com/miku/metha')
def test_executable(self):
task = Executable(name='ls')
self.assertTrue(task.complete())
task = Executable(name='veryunlikely123')
self.assertFalse(task.complete())
def requires(self):
return Executable(name='srufetch',
message='https://github.com/miku/srufetch')
def test_executable(self):
task = Executable(name='ls')
self.assertTrue(task.complete())
task = Executable(name='veryunlikely123')
self.assertFalse(task.complete())