class EcoLexCrawl(object):
def __init__(self, gApi = None):
if gApi != None:
self.gApi = defineApi(gApi)
else:
self.gApi = gApi
while self.gApi == None:
self.gApi = defineApi()
self.firstTime()
self.search = EcoLexSearch()
self.entriesAdded = 0
self.enterNewResults()
def firstTime(self):
"""Prompts user about creating a new data base."""
first = raw_input("Is this the first time you run EcoLex Crawler? 1. 'Yes', 2. 'No' >> " )
if first == '1':
self.firsttime = True
else:
self.firsttime = False
def enterResults(self,links):
"""Pass results to database"""
for link in links:
newentry = EcoLexResult(link,self.gApi)
addEntry(newentry,self.firsttime)
if self.firsttime == True:
self.firsttime = False
self.entriesAdded = self.entriesAdded + 1
def enterNewResults(self):
"""Checks whether scraped entries are in database already and passes new ones to database"""
links = self.search.entrylinks
if not self.firsttime:
for link in links:
start = link.rfind('id=') + 3
end = link.rfind('&',start)
ecolexID = link[start:end]
session = startSession()
check = session.query(addEcoLexEntry.EcoLexEntry).filter(addEcoLexEntry.EcoLexEntry.ecolex_id == ecolexID ).count()
session.commit()
if check == 1:
links.remove(link)
else:
pass
self.enterResults(links)
def printReport(self):
"""Prints a report of the specified search."""
self.search.printSearchInput()
print "This specification led to %i new entries in your data base." % self.entriesAdded
def __init__(self, gApi = None):
if gApi != None:
self.gApi = defineApi(gApi)
else:
self.gApi = gApi
while self.gApi == None:
self.gApi = defineApi()
self.firstTime()
self.search = EcoLexSearch()
self.entriesAdded = 0
self.enterNewResults()
