def main(app_name):
app_path = join(curdir, app_name)
xml_path = join(app_path, 'view.xml')
views = parser.parse_xml(xml_path)
try:
globals.app = __import__(app_name + '.extensions', fromlist=[app_name])
except ImportError as e:
stderr.write(e.message + '\n')
return
main_window = application.Application(views[0], views[1])
globals.views = main_window.extensions
main_window.start(views[2])
def test_parser():
xml = """<root>
<object>
<field1>foo</field1>
<field1>bar</field1>
<field2>baz</field2>
<field3>
goat
<nested1>rabbit</nested1>
<nested2>llama</nested2>
herd
</field3>
</object>
<thing>
<field>basic</field>
</thing>
</root>"""
nodes = []
def emit(node, arg1, arg2):
nodes.append(node)
assert arg1 == 'arg1'
assert arg2 == 'arg2'
assert parse_xml(StringIO(xml), emit, 'arg1', 'arg2') == 2
assert len(nodes) == 2
assert nodes[0].name == 'object'
assert len(nodes[0].fields) == 4
assert nodes[0].fields[0].name == 'field1'
assert nodes[0].fields[0].value() == 'foo'
assert nodes[0].fields[1].name == 'field1'
assert nodes[0].fields[1].value() == 'bar'
assert nodes[0].fields[2].name == 'field2'
assert nodes[0].fields[2].value() == 'baz'
assert nodes[0].fields[3].name == 'field3'
assert nodes[0].fields[3].value() == 'goatherd'
assert len(nodes[0].fields[3].fields) == 2
assert nodes[0].fields[3].fields[0].name == 'nested1'
assert nodes[0].fields[3].fields[0].value() == 'rabbit'
assert nodes[0].fields[3].fields[1].name == 'nested2'
assert nodes[0].fields[3].fields[1].value() == 'llama'
assert nodes[1].name == 'thing'
assert len(nodes[1].fields) == 1
assert nodes[1].fields[0].name == 'field'
assert nodes[1].fields[0].value() == 'basic'
def main(argv):
# Get parameters from user input
accountKey = argv[0]
target_precision = float(argv[1])
query = argv[2].lower().split()
precision = 0.0
# create a new search engine instance
search_engine = search.SearchEngine(accountKey)
while True:
# search this query on Bing
xml = search_engine.search(query)
# print out search query parameters on interminal
interface.print_parameters(search_engine, target_precision)
# parse the xml search results to Siteinfo objects
result = parser.parse_xml(xml)
# if there are less than 10 results, break and quit
if len(result) < 10:
break
# interact with user and let user mark relevence of each result
array_score, precision = interface.interact(result)
# if precision is 0, break and quit
if precision == 0.0:
print 'Below desired precision, but can no longer augment the query'
break
interface.print_transcript(precision, query, target_precision)
# if precision is equal or larger than target precision, break and quit
if precision >= target_precision:
break
# if precision is less than target precision, expand and reoder query and search again
else:
new_terms, new_query = search.query_expand(result, array_score, query)
interface.print_new_terms(new_terms)
if new_query == query:
break
else:
query = new_query
#self.accountKey = 'KtKgk8Mo5p6/rJE0FnlmA8qKVi1F7kS3OQbxik1ZnCg'
self.accountKey = key
self.url = None
self.query = None
def search(self, siteUrl, query):
#bingBaseUrl = 'https://api.datamarket.azure.com/Bing/Search/Web?Query=%27'
bingBaseUrl = 'https://api.datamarket.azure.com/Data.ashx/Bing/SearchWeb/v1/Composite?Query=%27site%3a'
self.url = bingBaseUrl + siteUrl +'%20'+ '%20'.join(query) + "%27&$top=10&$format=Atom"
self.query = query
accountKeyEnc = base64.b64encode(self.accountKey + ':' + self.accountKey)
headers = {'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8','Authorization': 'Basic ' + accountKeyEnc}
req = urllib2.Request(self.url, headers = headers)
response = urllib2.urlopen(req)
return response
def getKey(self):
return self.accountKey
def getUrl(self):
return self.url
def getQuery(self):
return self.query
if __name__ == '__main__':
s = SearchEngine('KtKgk8Mo5p6/rJE0FnlmA8qKVi1F7kS3OQbxik1ZnCg')
response = s.search('afifa.com',['premiership'])
parser.parse_xml(response)
# concatenate faa file contents to temp.faa
cat_faa_cmd = "cat faa/*.faa > faa/temp.faa"
os.system(cat_faa_cmd)
# make a blast db
blast_cmd1 = "makeblastdb -parse_seqids -in faa/temp.faa -dbtype prot -out " + project_name
os.system(blast_cmd1)
blast_cmd2 = "diamond makedb -p N --in faa/temp.faa -d " + project_name
os.system(blast_cmd2)
# remove the concatenated fna files
remove_cmd = "rm faa/temp.faa && rm faa/temp1.faa"
os.system(remove_cmd)
faa_list = []
for entry in os.scandir('faa'):
if entry.is_file() and (entry.name[0] !=
".") and (entry.name.split(".")[-1]) == "faa":
faa_list.append(entry.name)
for faa_file in faa_list:
xml_name = "xml/" + faa_file.split(".")[0] + ".xml"
tab_name = "tab/" + faa_file.split(".")[0] + ".tab"
dict_name = project_name + ".dict"
blast_command = "diamond blastp --max-target-seqs 0 --outfmt 5 --query-cover 90 --subject-cover 90 --query faa/" +\
faa_file + " --db " + project_name + " --out " + xml_name
os.system(blast_command)
parser.parse_xml(xml_name, tab_name, dict_name, pct_id)
os.system("rm " + xml_name)
