def main():
armor = Armor(start_date_in_current_exhibition=Date(1, 1, 2001))
vase = Vase(start_date_in_current_exhibition=Date(2, 1, 2003))
crown = Crown(start_date_in_current_exhibition=Date(5, 3, 2007))
painting = Painting(start_date_in_current_exhibition=Date(6, 9, 2011))
list_of_exhibits = [painting, vase, crown, armor]
manager = ExhibitManager(list_of_exhibits)
manager.find_by_theme(Topic.ANCIENT_ROME)
manager.print_exhibit_list_info()
manager.sort_by_age()
manager.print_exhibit_list_info()
manager.sort_by_date_in_current_exhibition()
manager.print_exhibit_list_info()
def get_date(properties): # TODO : refactor this
if len(properties) == 0:
return None
data_property = properties[0]
if not 'datavalue' in data_property['mainsnak']:
return None
precision = data_property['mainsnak']['datavalue']['value']['precision']
time = data_property['mainsnak']['datavalue']['value']['time']
if precision < 9:
return None
return Date(time[1:5])
def __init__(self,
name="some_exhibit",
portable=True,
century_of_creation=0,
country_of_creation="Italy",
size=3,
popularity=InterestedPeople.TWO,
theme=Topic.ANCIENT_GREECE,
destroyed_in_persentage=20,
start_date_in_current_exhibition=Date(1, 1, 2020),
made_of_metal=True,
pattern=False):
super().__init__(name, portable, century_of_creation,
country_of_creation, size, popularity, theme,
destroyed_in_persentage,
start_date_in_current_exhibition)
self.made_of_metal = made_of_metal
self.pattern = pattern
def __init__(self,
name="some_exhibit",
portable=True,
century_of_creation=0,
country_of_creation="Italy",
size=3,
popularity=InterestedPeople.TWO,
theme=Topic.ANCIENT_GREECE,
destroyed_in_persentage=20,
start_date_in_current_exhibition=Date(2, 1, 2050),
insertion=True,
number_of_diamonds=18):
super().__init__(name, portable, century_of_creation,
country_of_creation, size, popularity, theme,
destroyed_in_persentage,
start_date_in_current_exhibition)
self.insertion = insertion
self.number_of_diamonds = number_of_diamonds
def __init__(self,
name="some_exhibit",
portable=True,
century_of_creation=12,
country_of_creation="Italy",
size=3,
popularity=InterestedPeople.TWO,
theme=Topic.ANCIENT_ROME,
destroyed_in_persentage=20,
start_date_in_current_exhibition=Date(4, 1, 2007),
material="metal",
painted=True):
super().__init__(name, portable, century_of_creation,
country_of_creation, size, popularity, theme,
destroyed_in_persentage,
start_date_in_current_exhibition)
self.material = material
self.painted = painted
def __init__(self,
name="some_exhibit",
portable=True,
century_of_creation=0,
country_of_creation="Italy",
size=3,
popularity=InterestedPeople.TWO,
theme=Topic.ANCIENT_ROME,
destroyed_in_persentage=20,
start_date_in_current_exhibition=Date(3, 1, 2007),
painting_style=Style.ABSTRACT_ART,
frame=True):
super().__init__(name, portable, century_of_creation,
country_of_creation, size, popularity, theme,
destroyed_in_persentage,
start_date_in_current_exhibition)
self.painting_style = painting_style
self.frame = frame
def __init__(self,
name="some_exhibit",
portable=True,
century_of_creation=0,
country_of_creation="Italy",
size=3,
popularity=InterestedPeople.TWO,
theme=Topic.ANCIENT_ROME,
destroyed_in_persentage=20,
start_date_in_current_exhibition=Date()):
self.name = name
self.portable = portable
self.century_of_creation = century_of_creation
self.country_of_creation = country_of_creation
self.size = size
self.popularity = popularity
self.theme = theme
self.destroyed_in_persentage = destroyed_in_persentage
self.start_date_in_current_exhibition = start_date_in_current_exhibition
def saveInmateProfile(browser):
inmate = Inmate() # inmate profile
record = InmateRecord() # inmate current record
facility = Facility()
# find inmate ID, will go in active record
time.sleep(2)
element = browser.find_elements_by_class_name("section_data")
backgroundPersonInfo = BeautifulSoup(element[1].get_attribute('innerHTML'), 'html.parser').find("tbody").find("tbody")
personInfoCells = backgroundPersonInfo.find_all("tr")
for ind in range(len(personInfoCells)):
cell = personInfoCells[ind]
if not isinstance(cell, NavigableString):
txt = " ".join(cell.text.strip().split())
if "Name" in txt:
fullName = txt.replace(",", "").split(" ")
lastName = fullName[1]
firstName = fullName[2]
middleName = fullName[-1] if len(fullName) == 4 else None
inmate.name = Name(firstName, middleName, lastName)
elif "Age" in txt:
inmate.age = txt.split(" ")[-1]
inmate.DOB = Date(inmate.age, None, None, True)
elif "Gender" in txt:
inmate.sex = txt.split(" ")[-1]
elif "Ethnicity" in txt:
inmate.race = txt.split(" ")[-1]
elif "Hair Color" in txt:
inmate.hairColor = txt.split(" ")[-1]
elif "Eye Color" in txt:
inmate.eyeColor = txt.split(" ")[-1]
elif "Height" in txt:
inmate.height = txt.split(" ")[-2] + txt.split(" ")[-1]
elif "Weight" in txt:
inmate.weight = txt.split(" ")[-1]
backgroundPersonPrisonInfo = backgroundPersonInfo.find_next("tbody")
personPrisonInfoCells = backgroundPersonPrisonInfo.find_all("tr")
for ind in range(len(personPrisonInfoCells)):
cell = personPrisonInfoCells[ind]
if not isinstance(cell, NavigableString):
txt = " ".join(cell.text.strip().split())
if "DOC Number" in txt:
# inmate's id given by Colorado Department of Corrections
record.recordNumber = txt.split(" ")[-1]
elif "Est. Parole Eligibility Date" in txt:
dateSplit = txt.split(" ")[-1].split("/")
if len(dateSplit) > 1:
record.paroleEligibilityDate = Date(dateSplit[-1], dateSplit[0], dateSplit[1])
elif "Next Parole Hearing Date" in txt:
dateSplit = txt.split(":")
if len(dateSplit) > 1:
record.nextParoleHearingDate = dateSplit[-1].strip()
elif "Est. Sentence Discharge Date" in txt:
dateSplit = txt.split(" ")[-1].split("/")
if len(dateSplit) > 1:
record.estReleaseDate = Date(dateSplit[-1], dateSplit[0], dateSplit[1])
elif "Current Facility Assignment" in txt:
facility.name = txt.split(":")[-1].strip()
# saves profile to the database
writeToDB(inmate)
browser.find_element_by_id("btn_search_txt").click()
return inmate.name.first + " " + inmate.name.last
def saveInmateProfile(soup, browser):
inmate = Inmate() # inmate profile
record = InmateRecord() # inmate current record
record.state = "GA"
facility = Facility()
inmateID = soup.find(
'h5', text=re.compile('GDC ID:.*')).get_text().strip().split()[
-1] # find inmate ID, will go in active record
# find inmate name
name = soup.find(
'h4', text=re.compile('NAME:.*')).get_text().strip('NAME:').strip()
# finds text that includes "NAME:", for example; regex
lastName, firstNames = name.split(
',') # only if "last, first" order; otherwise vice versa
firstNames = firstNames.split()
firstName = firstNames[0]
middleName = "" if len(firstNames[1:]) == 0 else " ".join(firstNames[1:])
inmate.name = Name(firstName, middleName, lastName)
# find way to parse and collect rest of info; will vary with different crawlers
info = soup.findAll('p')
caseNumbers = [
x.text.strip('CASE NO:').strip() for x in soup.findAll('h7')
]
recordCounter = 0
for row in info:
if row and 'class="offender"' in str(row):
data = row.text.split('\n')
data = map(str.strip, data)
data = list(filter(None, data))
# past convictions
if any("SENTENCE LENGTH" in s for s in data):
for info in data:
try:
entry, value = info.split(': ')
except:
continue
if entry == 'OFFENSE':
past_record = InmateRecord()
past_record.inmateNumber = inmateID
past_record.status = RecordStatus.INACTIVE
past_record.state = "GA"
past_record.offense = value
elif entry == 'CRIME COMMIT DATE':
past_record.sentenceDate = Date(value)
elif entry == 'SENTENCE LENGTH':
past_record.estReleaseDate = past_record.sentenceDate
try:
(paramY, paramM, paramD) = [
int(x.strip().split()[0])
for x in value.split(",")
]
except (ValueError):
(paramY, paramM, paramD) = (None, None, None)
past_record.estReleaseDate.addTime(
paramY, paramM, paramD)
past_record.estReleaseDate.estimated = True
past_record.recordNumber = caseNumbers[recordCounter]
inmate.addRecord(past_record)
recordCounter += 1
# information about current conviction
else:
for trait in data:
try:
entry, value = trait.split(': ')
except:
continue
if entry == 'RACE':
inmate.race = value
elif entry == 'GENDER':
inmate.sex = value
elif entry == 'HEIGHT':
inmate.height = value
elif entry == 'EYE COLOR':
inmate.eyeColor = value
elif entry == 'YOB':
inmate.DOB = Date(value)
elif entry == 'WEIGHT':
inmate.weight = value
elif entry == 'HAIR COLOR':
inmate.hairColor = value
elif entry == 'MOST RECENT INSTITUTION':
facility.name = value
facility.state = record.state
facility.getGeneratedID()
record.addFacility(facility)
elif entry == 'MAX POSSIBLE RELEASE DATE':
record.maxReleaseDate = Date(value)
record.status = RecordStatus.ACTIVE
record.recordNumber = caseNumbers[0]
record.status = RecordStatus.ACTIVE
record.inmateNumber = inmateID
inmate.addRecord(record)
# saves profile to the database
writeToDB(inmate)
# click back button
browser.find_element_by_xpath(
"//a[text()=' Return to previous screen']").click()
return name
def getDict(self):
if (not isinstance(self.admissionDate, Date)
and self.admissionDate is not None):
self.admissionDate = Date(self.admissionDate.year,
self.admissionDate.month,
self.admissionDate.day)
if (not isinstance(self.sentenceDate, Date)
and self.sentenceDate is not None):
self.sentenceDate = Date(self.sentenceDate.year,
self.sentenceDate.month,
self.sentenceDate.day)
if (not isinstance(self.maxReleaseDate, Date)
and self.maxReleaseDate is not None):
self.maxReleaseDate = Date(self.maxReleaseDate.year,
self.maxReleaseDate.month,
self.maxReleaseDate.day)
if (not isinstance(self.admissionDate, Date)
and self.estReleaseDate is not None):
self.estReleaseDate = Date(self.estReleaseDate.year,
self.estReleaseDate.month,
self.estReleaseDate.day)
return {
"_id":
self.generatedID if self.generatedID is not None else
generate_record_id(self.state, self.recordNumber), # db key!
"inmateID":
self.inmateId, # db key!
"recordNumber":
self.recordNumber,
"admissionDate":
self.admissionDate.getDict()
if self.admissionDate is not None else None,
"sentenceDate":
self.sentenceDate.getDict()
if self.sentenceDate is not None else None,
"maxReleaseDate":
self.maxReleaseDate.getDict()
if self.maxReleaseDate is not None else None,
"estReleaseDate":
self.estReleaseDate.getDict()
if self.estReleaseDate is not None else None,
"paroleEligibilityDate":
self.paroleEligibilityDate,
"nextParoleHearingDate":
self.nextParoleHearingDate,
"facilityID":
self.facility.getGeneratedID(), # db key!
"state":
self.state,
"county":
self.county,
"bondAmt":
self.bondAmt,
"status":
self.status.value,
"supervisionStatus":
self.currentSupervisionStatus,
"offense":
self.offense,
}
class InmateRecord:
status = RecordStatus
def __init__(self):
self.generatedID = ""
self.recordNumber = None # db key! This is the value (usually 6 characters) that is used by every database
self.inmateId = None # optional number that may be found on some databases
self.admissionDate = None
self.sentenceDate = None
self.maxReleaseDate = None
self.estReleaseDate = None
self.paroleEligibilityDate = None
self.nextParoleHearingDate = None
self.facility = None
self.state = None
self.county = None
self.bondAmt = None
self.status = RecordStatus.UNDEFINED
self.currentSupervisionStatus = None
self.offense = ""
def __str__(self):
return str(self.getDict())
def __setattr__(self, name, value):
if name == 'status' and not isinstance(value, RecordStatus):
raise TypeError('Status must be of type RecordStatus')
super().__setattr__(name, value)
def getGeneratedID(self):
if (len(self.generatedID) == 0):
self.generatedID = generate_record_id(self.state,
self.recordNumber)
return self.generatedID
def getDict(self):
if (not isinstance(self.admissionDate, Date)
and self.admissionDate is not None):
self.admissionDate = Date(self.admissionDate.year,
self.admissionDate.month,
self.admissionDate.day)
if (not isinstance(self.sentenceDate, Date)
and self.sentenceDate is not None):
self.sentenceDate = Date(self.sentenceDate.year,
self.sentenceDate.month,
self.sentenceDate.day)
if (not isinstance(self.maxReleaseDate, Date)
and self.maxReleaseDate is not None):
self.maxReleaseDate = Date(self.maxReleaseDate.year,
self.maxReleaseDate.month,
self.maxReleaseDate.day)
if (not isinstance(self.admissionDate, Date)
and self.estReleaseDate is not None):
self.estReleaseDate = Date(self.estReleaseDate.year,
self.estReleaseDate.month,
self.estReleaseDate.day)
return {
"_id":
self.generatedID if self.generatedID is not None else
generate_record_id(self.state, self.recordNumber), # db key!
"inmateID":
self.inmateId, # db key!
"recordNumber":
self.recordNumber,
"admissionDate":
self.admissionDate.getDict()
if self.admissionDate is not None else None,
"sentenceDate":
self.sentenceDate.getDict()
if self.sentenceDate is not None else None,
"maxReleaseDate":
self.maxReleaseDate.getDict()
if self.maxReleaseDate is not None else None,
"estReleaseDate":
self.estReleaseDate.getDict()
if self.estReleaseDate is not None else None,
"paroleEligibilityDate":
self.paroleEligibilityDate,
"nextParoleHearingDate":
self.nextParoleHearingDate,
"facilityID":
self.facility.getGeneratedID(), # db key!
"state":
self.state,
"county":
self.county,
"bondAmt":
self.bondAmt,
"status":
self.status.value,
"supervisionStatus":
self.currentSupervisionStatus,
"offense":
self.offense,
}
def addFacility(self, facility):
self.facility = facility
def saveInmateProfile(soup, browser):
inmate = Inmate() # inmate profile
record = InmateRecord() # inmate current record
record.state = "Alabama"
facility = Facility() # current facility info
# inmate id specific to alabama
inmateID = soup.find(id="MainContent_DetailsView2_Label2").get_text()
# find name element
name = soup.find(id="MainContent_DetailsView2_Label1").get_text()
lastName, firstNames = name.split(',')
firstNames = firstNames.split()
firstName = firstNames[0]
middleName = "" if len(firstNames[1:]) == 0 else " ".join(firstNames[1:])
# add name to inmate object
inmate.name = Name(firstName, middleName, lastName)
# get info off profile pages
info = soup.findAll('table')
# facility names
facility.name = info[0].find(
id="MainContent_DetailsView2_Label3").get_text()
facility.state = record.state
# record.facilityID = facility.generatedID
# recordCounter = 0
for i in range(1, len(info)):
count = 0
for row in info[i]:
# skip first dummy row
if count != 1:
count += 1
else:
currentRecord = False
for trait in row:
if i == 1:
try:
lines = trait.text.split("\n")
traits = lines[1].split(":")
except AttributeError:
continue
if traits[0] == "Race":
inmate.race = traits[1]
elif traits[0] == "Sex":
inmate.sex = traits[1]
elif traits[0] == "Hair Color":
inmate.hairColor = traits[1]
elif traits[0] == "Eye Color":
inmate.eyeColor = traits[1]
elif traits[0] == "Height":
inmate.height = traits[1]
elif traits[0] == "Birth Year":
inmate.DOB = Date(traits[1], None, None, True)
elif traits[0] == "Sex":
inmate.sex = traits[1]
elif i == 2:
# print('NEW ROW')
# print(trait)
try:
lines = trait.text.split("\n")
traits = lines[1].split(" ")
print(traits)
except AttributeError:
continue
# save profile to the database
writeToDB(inmate)
browser.set_page_load_timeout(10)
# need way to get back to original search list results
# EX: browser.find_element_by_xpath("//a[text()=' Return to previous screen']").click()
return name
# TESTS:
# baseCrawler("Adams", "")