def getMatchingKnowledgeArea(candidate):
# find matching knowledge area and return Id
for knowledgearea in knowledgeAreas:
if knowledgearea.getText().strip() == str(candidate).strip():
return knowledgearea
return KnowledgeArea()
def ExtractKnowledgeAreas(firebase, document, knowledgeAreas):
# find the knowledge areas in the document and write them to a database table
global firstPositionOfCKA
firstPositionOfCKA = 0
for i, p in enumerate(document.paragraphs):
pText = p.text.strip()
lastPos = len(pText)
kaText = ""
isKnowledgeArea, kaText = ParagraphIsKnowledgeArea(
document, p, i, lastPos, pText)
if isKnowledgeArea == 'true' and kaText:
knowledgeArea = KnowledgeArea()
knowledgeArea.setText(kaText)
knowledgeAreas.append(knowledgeArea)
diff = (i - firstPositionOfCKA)
if firstPositionOfCKA > 0 and diff > 120:
break
for knowledgeArea in knowledgeAreas:
newKnowledgeArea = {'Content': knowledgeArea.getText()}
result = firebase.post('KnowledgeArea', newKnowledgeArea)
knowledgeArea.setId(result.get("name"))
def FindKnowledgeAreaId(KnowledgeAreaText: str) -> str:
"""
Find existing KnowledgeArea in the global collection.
If found, get the Id.
:param newKnowledgeArea: a KnowledgeArea object
:return string representation of the Id
"""
global knowledgeAreas
for KnowledgeArea in knowledgeAreas:
if KnowledgeArea.getText().strip().lower() == KnowledgeAreaText.strip(
).lower():
return KnowledgeArea.getId()
return ""
def GetKnowledgeAreas(firebase: firebase) -> None:
"""
Get KnowledgeAreas from the database and
insert them into the global collection.
:param firebase: a firebase connection
"""
global knowledgeAreas
obj_key_list = []
result = firebase.get('/knowledgearea', None)
if result is None:
return
for i in result.keys():
obj_key_list.append(i)
for i in obj_key_list:
knowledgeArea = KnowledgeArea()
knowledgeArea.setText(result[i]['content'])
knowledgeArea.setId(i)
knowledgeAreas.append(knowledgeArea)
def GetExistingKnowledgeAreas(firebase: firebase, document: Document) -> None:
"""
Get existing KnowledgeAreas from the database and
insert them into the global collection.
:param firebase: a firebase connection
:param document: a Python-Docx Document object
"""
global existingKnowledgeAreas
obj_key_list = []
result = firebase.get('/knowledgearea', None)
if result is None:
return
for i in result.keys():
obj_key_list.append(i)
for i in obj_key_list:
knowledgeArea = KnowledgeArea()
knowledgeArea.setText(result[i]['content'])
knowledgeArea.setId(i)
existingKnowledgeAreas.append(knowledgeArea)
def getMatchingKnowledgeArea(candidate: str) -> KnowledgeArea:
"""
Find matching knowledge area and return the Knowledge Area as a string.
:param candidate: a string representing an extracted paragraph that
we want to check to see if it matches a Knowledge Area.
"""
for knowledgearea in knowledgeAreas:
if knowledgearea.getText().strip() == str(candidate).strip():
return knowledgearea
return KnowledgeArea()
def ExtractKnowledgeAreas(firebase: firebase, document: Document, knowledgeAreas: List[KnowledgeArea]) -> None:
"""
Find the KnowledgeAreas in the document and
insert them into a database table.
:param firebase: a firebase connection
:param document: a Python-Docx Document object
:param knowledgeAreas: a List of KnowledgeArea objects
"""
global firstPositionOfCKA
firstPositionOfCKA = 0
for i, p in enumerate(document.paragraphs):
pText = p.text.strip()
lastPos = len(pText)
kaText = ""
isKnowledgeArea, kaText = ParagraphIsKnowledgeArea(document, p, i, lastPos, pText)
if isKnowledgeArea and kaText:
knowledgeArea = KnowledgeArea()
knowledgeArea.setText(kaText)
knowledgeAreas.append(knowledgeArea)
diff = (i - firstPositionOfCKA)
if firstPositionOfCKA > 0 and diff > 120:
break
for knowledgeArea in knowledgeAreas:
newKnowledgeArea = {
'content': knowledgeArea.getText()
}
existingKey = FindExistingKnowledgeAreaId(knowledgeArea)
if existingKey == "":
result = firebase.post('knowledgearea', newKnowledgeArea)
knowledgeArea.setId(result.get("name"))
else:
knowledgeArea.setId(existingKey)
def writeCourseRows(ws: xlsxwriter.worksheet,
knowledgeArea: KnowledgeArea) -> None:
"""
Write the rows in the courses worksheet for all courses under
the knowledgearea.
:param ws: Excel Worksheet Object
:param knowledgeArea: KnowledgeArea
"""
global courses
global coursesWsRow
col = 0
knowledgeAreaId = knowledgeArea.getId()
for course in courses:
if str(course.getKnowledgeAreaId()) == str(knowledgeAreaId):
ws.write(coursesWsRow, col, knowledgeArea.getText())
col += 1
courseId = course.getId()
ws.write(coursesWsRow, col, courseId)
col += 1
ws.write(coursesWsRow, col, course.getTitle())
col += 2
ws.write(coursesWsRow, col, course.getDescription())
col += 2
ws.write(coursesWsRow, col, course.getInstructor())
col += 2
ws.write(coursesWsRow, col, course.getFee())
coursesWsRow += 1
col = 0
def FindExistingKnowledgeAreaId(newKnowledgeArea: KnowledgeArea) -> str:
"""
Find existing KnowledgeArea in the global collection.
If found, get the Id.
:param newKnowledgeArea: a KnowledgeArea object
:return string representation of the Id
"""
global existingKnowledgeAreas
for existingKnowledgeArea in existingKnowledgeAreas:
if existingKnowledgeArea.getText() == newKnowledgeArea.getText():
return existingKnowledgeArea.getId()
return ""
from textblob import TextBlob
from firebase import firebase
from catalog import Catalog
from knowledgearea import KnowledgeArea
from course import Course
from learningobjective import LearningObjective
from learningobjective_course import LearningObjective_course
from paragraph import Paragraph
import constant
import time
existingCatalogRecords = []
existingKnowledgeAreas = []
firstPositionOfCKA = 0
currentKnowledgeArea = KnowledgeArea()
courseNameStyles = [constant.STYLE_NORMAL, constant.STYLE_HEADING, constant.STYLE_HEADING7]
firebase = firebase.FirebaseApplication('https://scholacity-org.firebaseio.com/')
document = Document('../Catalogs/Fall2020_LLCatalog.docx')
documentName = "Fall2020_LLCatalog.docx"
documentSemester = "Fall"
documentYear = "2020"
# document = Document('Spring2020_LeisureLearningCatalogFULL.docx')
# documentName = "Spring2020_LeisureLearningCatalogFULL.docx"
# documentSemester = "Spring"
# documentYear = "2020"
def writeWorksheet(ws: xlsxwriter.worksheet,
knowledgeArea: KnowledgeArea) -> None:
"""
Iterate through the global collection of KnowledgeAreas
and write them to a worksheet in an Excel workbook.
:param ws: Excel Worksheet Object
:param knowledgeArea: KnowledgeArea Object
"""
global courses
global learningObjectives
row = 0
col = 0
ws.write(row, col, 'CourseId')
col += 1
ws.write(row, col, 'CourseTitle - As Extracted')
col += 1
ws.write(row, col, 'learningObjectiveId')
col += 1
ws.write(row, col, 'learningObjective - As Extracted')
col += 1
ws.write(row, col, 'learningObjective - As Revised')
row += 1
col = 0
knowledgeAreaId = knowledgeArea.getId()
for course in courses:
#print("Course {}, {} ".format(course.getTitle(), course.getKnowledgeAreaId()))
if str(course.getKnowledgeAreaId()) == str(knowledgeAreaId):
courseId = course.getId()
if course.getDescription() == "":
ws.write(row, col, courseId)
col += 1
ws.write(row, col, course.getTitle())
row += 1
col = 0
# write another row for the course to allow for 2 learning outcomes
ws.write(row, col, courseId)
col += 1
ws.write(row, col, course.getTitle())
row += 1
col = 0
continue
else:
for learningObjective in learningObjectives:
#print("LearningObjective: {}, {} ".format(learningObjective.getText(), learningObjective.getCourseId()))
if str(learningObjective.getCourseId()) == str(courseId):
ws.write(row, col, courseId)
col += 1
ws.write(row, col, course.getTitle())
col += 1
ws.write(row, col, learningObjective.getId())
col += 1
ws.write(row, col, learningObjective.getText())
row += 1
col = 0
continue