Merge pull request #923 from cornell-dti/automate_email

Eric / Implemented Script For Automating Email Outreach Information (In Progress)

.gitignore

@@ -48,3 +48,6 @@ yarn-error.log*
 
 .idea
 .DS_Store
+
+# ignore virtual environment
+venv/

src/scripts/classes.csv

@@ -0,0 +1,187 @@
+Course,Professor,Emails,Has course/professor use QMI before
+AEP 1100,Chris Xu,[email protected],-1
+AEP 2550,Or Katz,[email protected],-1
+AEP 3100,Peter McMahon,[email protected],-1
+AEP 3200,Ankit Disa,[email protected],-1
+AEP 3550,Valla Fatemi,[email protected],-1
+AEP 3620,Kenji Yasuda,[email protected],-1
+AEP 4340,Lisa Wickham,[email protected],-1
+AEP 4400,Jeffrey Moses,[email protected],-1
+BIOMG 1290,Charles Aquadro,[email protected],-1
+BIOMG 1350,"Chris Fromme, Kelly Liu","[email protected], [email protected]",-1
+BIOMG 2800,"Kristina Blake-Hodek, Maria Serrano","[email protected], [email protected]",-1
+BIOMG 3300,Kevin Siegenthaler,[email protected],-1
+BIOMG 3350,"Stephen Jesch, Frank Pugh","[email protected], [email protected]",-1
+BIOMG 4320,Chun Han,[email protected],-1
+BIOMG 4400,Stephen Jesch,[email protected],-1
+BIOMG 4610,Mariana Wolfner,[email protected],-1
+BIOMG 4880,Zeribe Nwosu,[email protected],-1
+CHEM 2070,Julie Laudenschlager,[email protected],-1
+CHEM 2080,Robert DiStasio,[email protected],-1
+CHEM 2090,Youn Jue Bae,[email protected],-1
+CHEM 3580,"Geoffrey Coates, Brett Fors","[email protected], [email protected]",-1
+CHEM 3880,Barbara Baird,[email protected],-1
+CHEM 4100,Peter Wolczanski,[email protected],-1
+CHEM 4810,Fernando Escobedo,[email protected],-1
+CHEME 2200,T. Michael Duncan,[email protected],-1
+CHEME 3230,Sarah Hormozi,[email protected],-1
+CHEME 3320,Shuwen Yue,[email protected],-1
+CHEME 3900,Sijin Li,[email protected],-1
+CHEME 4020,Ilana Brito,[email protected],-1
+CHEME 4620,"Brian Bauer, William Cleary, Marsha Kowal, Jeffrey Varner, Alex Woltornist","[email protected], [email protected], [email protected], [email protected], [email protected]",-1
+CS 1110,"Michael Clarkson, Lillian Lee","[email protected], [email protected]",0
+CS 1112,Daisy Fan,[email protected],2
+CS 1340,"Jon Kleinberg, Karen Levy","[email protected], [email protected]",-1
+CS 1700,Haym Hirsh,[email protected],-1
+CS 2110,"Matthew Eichhorn, Curran Muhlberger","[email protected], [email protected]",2
+CS 2800,"Joe Halpern, Eva Tardos","[email protected], [email protected]",0
+CS 3110,Anshuman Mohan,[email protected],0
+CS 3410,"Zachary Susag, Hakim Weatherspoon","[email protected], [email protected]",2
+CS 3420,Nils Napp,[email protected],-1
+CS 3700,Kevin Ellis,[email protected],-1
+CS 3780,"Tushaar Gangavarapu, Karthik Sridharan","[email protected], [email protected]",-1
+CS 4300,Cristian Danescu-Niculescu-Mizil,[email protected],2
+CS 4410,Robbert VanRenesse,[email protected],0
+CS 4450,Rachit Agarwal,[email protected],-1
+CS 4670,"Bharath Hariharan, Wei-Chiu Ma","[email protected], [email protected]",2
+CS 4740,"Claire Cardie, Tanya Goyal","[email protected], [email protected]",-1
+CS 4744,Mats Rooth,[email protected],-1
+CS 4756,Kuan Fang,[email protected],-1
+CS 4782,"Jennifer Sun, Kilian Weinberger","[email protected], [email protected]",-1
+CS 4789,Wen Sun,[email protected],0
+CS 4810,Dexter Kozen,[email protected],-1
+CS 4813,Nicholas Spooner,[email protected],-1
+CS 4820,Michael Kim,[email protected],2
+CS 4850,Robert Kleinberg,[email protected],-1
+ECE 2100,Carl Bernard,[email protected],-1
+ECE 2300,Zhiru Zhang,[email protected],2
+ECE 2400,Anne Bracy,[email protected],-1
+ECE 2720,Vikram Krishnamurthy,[email protected],-1
+ECE 3100,Qing Zhao,[email protected],-1
+ECE 3140,Nils Napp,[email protected],2
+ECE 3150,Amit Lal,[email protected],-1
+ECE 3200,Ziv Goldfeld,[email protected],-1
+ECE 4150,David Hysell,[email protected],-1
+ECE 4160,Elizabeth Helbling,[email protected],-1
+ECE 4210,Francesca Parise,[email protected],-1
+ECE 4271,David Delchamps,[email protected],-1
+ECE 4370,Karan Mehta,[email protected],-1
+ECE 4520,Lang Tong,[email protected],-1
+ECE 4590,R. van Dover,[email protected],-1
+ECE 4670,Aaron Wagner,[email protected],-1
+ECE 4740,Alyosha Molnar,[email protected],-1
+ECE 4760,Van Adams,[email protected],-1
+ECE 4910,Bruce Johnson,[email protected],-1
+ECON 1110,George Orlov,[email protected],-1
+ECON 1120,Fikri Pitsuwan,[email protected],-1
+ECON 2300,Arnab Basu,[email protected],-1
+ECON 3030,Philipp Kircher,[email protected],-1
+ECON 3040,Mathieu Taschereau-Dumouchel,[email protected],-1
+ECON 3110,Kevin Packard,[email protected],-1
+ECON 3120,Douglas McKee,[email protected],-1
+ECON 3140,Joerg Stoye,[email protected],-1
+ECON 3430,Stephanie Thomas,[email protected],-1
+ECON 3475,Michele Belot,[email protected],-1
+ECON 3550,Elizabeth Tennant,[email protected],-1
+ECON 3670,Brandon Tripp,[email protected],-1
+ECON 3711,John Cawley,[email protected],-1
+ECON 3770,Evan Riehl,[email protected],-1
+ECON 3860,C.-Y. Cynthia Lin Lawell,[email protected],-1
+ECON 3865,Shanjun Li,[email protected],-1
+ECON 4020,Lawrence Blume,[email protected],-1
+ECON 4210,Kristoffer Nimark,[email protected],-1
+ECON 4270,Stephen Coate,[email protected],-1
+ECON 4600,Levon Barseghyan,[email protected],-1
+ECON 4620,Adam Harris,[email protected],-1
+ECON 4660,Ted O'Donoghue,[email protected],-1
+ECON 4903,Eleonora Patacchini,[email protected],-1
+INFO 1260,"Jon Kleinberg, Karen Levy","[email protected], [email protected]",-1
+INFO 1300,Royal Westwater,[email protected],-1
+INFO 3130,Moon Duchin,[email protected],-1
+INFO 3140,Shimon Edelman,[email protected],-1
+INFO 3200,Claire Wardle,[email protected],-1
+INFO 3950,Paul Ginsparg,[email protected],-1
+INFO 4100,Rene Kizilcec,[email protected],-1
+INFO 4120,Cheng Zhang,[email protected],-1
+INFO 4152,Walker White,[email protected],-1
+INFO 4220,Cristobal Cheyre Forestier,[email protected],-1
+INFO 4250,Daniel Susser,[email protected],-1
+INFO 4260,Steven Jackson,[email protected],-1
+INFO 4300,Cristian Danescu-Niculescu-Mizil,[email protected],-1
+INFO 4310,Jeff Rzeszotarski,[email protected],-1
+INFO 4320,Francois Guimbretiere,[email protected],-1
+INFO 4360,Drew Margolin,[email protected],-1
+INFO 4450,Susan Fussell,[email protected],-1
+INFO 4940,Sharlane Cleare,[email protected],-1
+MATH 1106,Marie MacDonald,[email protected],-1
+MATH 1110,Hosea Wondo,[email protected],-1
+MATH 1120,Luke Serafin,[email protected],-1
+MATH 1340,Matthew Haulmark,[email protected],-1
+MATH 1910,Timothy Healey,[email protected],-1
+MATH 1920,Sergio Chaves,[email protected],-1
+MATH 2210,Benjamin Dozier,[email protected],-1
+MATH 2220,Justin Moore,[email protected],-1
+MATH 2240,Jason Manning,[email protected],-1
+MATH 2310,Alex Townsend,[email protected],-1
+MATH 2930,David Freund,[email protected],-1
+MATH 2940,Frans Schalekamp,[email protected],-1
+MATH 3110,Timothy Riley,[email protected],-1
+MATH 3320,Dan Barbasch,[email protected],-1
+MATH 3340,Rachel Webb,[email protected],-1
+MATH 3360,Karola Meszaros,[email protected],-1
+MATH 4180,Chun Pong Chu,[email protected],-1
+MATH 4200,John Hubbard,[email protected],-1
+MATH 4280,Alexander Vladimirsky,[email protected],-1
+MATH 4310,Andreea Iorga,[email protected],-1
+MATH 4500,James West,[email protected],-1
+MATH 4540,Xiaodong Cao,[email protected],-1
+MATH 4710,Laurent Saloff-Coste,[email protected],-1
+MATH 4740,Sungwoo Jeong,[email protected],-1
+ORIE 3120,Peter Frazier,[email protected],-1
+ORIE 3310,David Shmoys,[email protected],-1
+ORIE 3741,Soroosh Shafiee,[email protected],-1
+ORIE 4126,Eric Gentsch,[email protected],-1
+ORIE 4130,Shane Henderson,[email protected],-1
+ORIE 4152,Negin Majedi,[email protected],-1
+ORIE 4160,Paul Golz,[email protected],-1
+ORIE 4600,John Richard Callister,[email protected],-1
+ORIE 4742,Sid Banerjee,[email protected],-1
+PHYS 1102,Nam Jung Kim,[email protected],-1
+PHYS 1112,Philip Krasicky,[email protected],-1
+PHYS 1116,Csaba Csaki,[email protected],-1
+PHYS 1203,Nils Deppe,[email protected],-1
+PHYS 2208,Matthias Liepe,[email protected],-1
+PHYS 2213,Ryan Tapping,[email protected],-1
+PHYS 2214,Itai Cohen,[email protected],-1
+PHYS 2218,Liam McAllister,[email protected],-1
+PHYS 3310,"Jim Alexander, Kyle Shen","[email protected], [email protected]",-1
+PHYS 3316,Xiaomeng Liu,[email protected],-1
+PHYS 3318,Ivan Bazarov,[email protected],-1
+PHYS 4410,Anders Ryd,[email protected],-1
+PHYS 4443,Katja Nowack,[email protected],-1
+PHYS 4444,Yuval Grossman,[email protected],-1
+PHYS 4488,James Sethna,[email protected],-1
+PHYS 4688,Georg Hoffstaetter de Torquat,[email protected],-1
+STSCI 2150,Melissa Smith,[email protected],-1
+STSCI 3740,Yang Ning,[email protected],-1
+STSCI 4520,David Kent,[email protected],-1
+STSCI 4550,Younghoon Kim,[email protected],-1
+"PHYS 2217, AEP 2170",Abigail Crites,[email protected],-1
+"AEP 3630, PHYS 3360",Earl Kirkland,[email protected],-1
+"ECE 4380, AEP 4450",Francesco Monticone,[email protected],-1
+"AEP 4840, ECE 4840",David Hammer,[email protected],-1
+"BTRY 3020, STSCI 2100, STSCI 3200",Jeremy Entner,[email protected],-1
+"STSCI 3080, BTRY 3080",Florentina Bunea,[email protected],-1
+"STSCI 4090, BTRY 4090, MATH 4720",Marten Wegkamp,[email protected],-1
+"STSCI 4100, BTRY 4100",Dana Yang,[email protected],-1
+"CHEM 1570, CHEM 3570",Cynthia Kinsland,[email protected],-1
+"CS 4220, MATH 4260",Anil Damle,[email protected],-1
+"CS 4754, INFO 4410",Malte Jung,[email protected],-1
+"CS 4758, ECE 4180",Anastasia Bizyaeva,[email protected],-1
+"CS 4852, ECON 3825",Cristobal Cheyre Forestier,[email protected],-1
+"ECON 3280, ECON 4280",Milena Djourelova,[email protected],-1
+"INFO 2310, INFO 4340",Kyle Harms,[email protected],-1
+"INFO 3312, INFO 2951",Benjamin Soltoff,[email protected],-1
+"MATH 4030, MATH 3040",Anil Nerode,[email protected],-1
+"STSCI 3510, ORIE 3510",Jim Dai,[email protected],-1
+"STSCI 3110, STSCI 2110",Kevin Packard,[email protected],-1

src/scripts/course-scrape.py

@@ -0,0 +1,193 @@
+import requests
+import csv
+import json
+
+def grab_json(url):
+  return json.loads(requests.get(url).text)
+
+rosterURL = "https://classes.cornell.edu/api/2.0/config/rosters.json"
+rosters = grab_json(rosterURL)['data']['rosters']
+# Get the most recent semester from the json
+# Note that slug is sem/year (e.g SP25)
+roster = rosters[-1]['slug']
+# If the most recent semester is a WINTER or SUMMER sem, use the sem before it
+if "WI" in roster or "SU" in roster:
+  roster = rosters[-2]['slug']
+
+# Get all subjects for that semester 
+subjectsURL = "https://classes.cornell.edu/api/2.0/config/subjects.json?roster="+roster
+subjects = grab_json(subjectsURL)['data']['subjects']
+
+relevent_courses = [
+  'CS',
+  'BTRY',
+  'INFO',
+  'ORIE',
+  'ECON',
+  'PHYS',
+  'CHEM',
+  'CHEME',
+  'ECE',
+  'AEP',
+  'BIO',
+  'BIOMG',
+  'MATH',
+  'STSCI'
+]
+
+# Classes to blacklist because professors are not interested in using QMI
+blacklist = [
+   'INFO 3152',
+   'CS 3152',
+   'INFO 4400',
+   'INFO 4390'
+   'INFO 4152',
+   'CS 4152'
+]
+
+def getCourseStatus(course, professors):
+    """
+    Returns the status of the course/professors in the past semester on whether they have used QMI before
+
+    2 indicates that both the course and professor have used QMI before
+    1 indicates that professor used QMI before
+    0 indicates that course used QMI before
+    -1 indicates that neither the course nor the professor have used QMI before
+
+    :param course: The course subject and its code (e.g CS 1110)
+    :param professors: The string that consist of all the professors teaching the course this sem (e.g "Michael Clarkson, Lillian Lee" )
+    """
+    with open('../scripts/trackers/past_semester.csv', 'r') as file:
+        csvreader = csv.reader(file)
+        header = next(csvreader)
+
+        for row in csvreader:
+            if row[0] == course:
+                # List of professors in the current semester
+                current_professors = professors.split(", ")
+
+                # List of professors in the past semester
+                past_professors = row[1].split(", ")
+
+                # Checks if at least one professor in the current semester is in the past professors
+                if any(p in past_professors for p in current_professors) or professors in past_professors:
+                    # If the course previously decided to use QMI and the professor is in the past professors, return 2 (both course/prof used QMI)
+                    if row[4] == "yes":
+                        return 2
+
+                    # If the course previously decided not to use QMI and the professor is in the past professors, return the status from the past semester
+                    return row[3]
+
+                # If the professor is not in the past professors
+                else:
+                    # If the course previously decided to use QMI, return 0 (course used QMI)
+                    if row[4] == "yes":
+                        return 0
+                    # If the course previously decided not to use QMI, but the course and professor has used QMI before, return 0 (course used QMI)
+                    elif row[4] == "no" and row[3] in ["2", "0"]:
+                        return 0
+                    return -1
+    # If no match is found, return -1 after checking all rows
+    return -1
+
+classesJSON = []
+for s in subjects:
+  if s['value'] in relevent_courses:
+    classURL = "https://classes.cornell.edu/api/2.0/search/classes.json?roster="+roster+"&subject=" + s['value']
+    classesJSON.extend(grab_json(classURL)['data']['classes'])
+
+def getProfessorInfo(instructors):
+  """
+  Returns the names and emails of the professors teaching the course
+
+  :param instructors: The list of instructors teaching the course
+  """
+  accum_names = "" # To accumulate all the professor names
+  accum_emails = "" # To accumulate all the professor emails
+  for professor in instructors:
+    name = professor['firstName'] + " " + professor['lastName'] + ", "
+    email = professor['netid']  + '@cornell.edu' + ", "
+    accum_names += name 
+    accum_emails += email
+  # Remove preceding and trailing , and "
+  return accum_names.strip(', ').strip('\"'), accum_emails.strip(', ').strip('\"')
+
+# Some courses are not relevant 
+# We will filter out these courses if they contain these words
+irrelevant_courses = ["Independent", "Research", "Project", "Projects", "Academic Support", "Sem", "Seminar", "Supplement", "Honors", "Thesis"]
+
+classes = []
+for c in classesJSON:
+  instructors = c['enrollGroups'][0]['classSections'][0]['meetings'][0]['instructors']
+  names, emails = getProfessorInfo(instructors)
+  hasProfessor = names != ""
+  isUndergrad = int(c['catalogNbr']) < 5000
+  isNonResearch = all(word not in c['titleShort'] for word in irrelevant_courses)
+  # At least 3 credits
+  isCredit = c['enrollGroups'][0]['unitsMinimum'] > 2
+  # Course must be on campus and not Cornell Tech
+  isOnCampus = c['enrollGroups'][0]['classSections'][0]['locationDescr'] == "Ithaca, NY (Main Campus)"
+
+  course = c['subject'] + " " + c['catalogNbr']
+
+  # Course must not be blacklisted
+  isNotBlacklisted = course not in blacklist
+  item = {
+    'Course': course,
+    'Professor': names,
+    'Emails': emails,
+    'Has course/professor use QMI before': getCourseStatus(course, names)
+  }
+  if hasProfessor and isUndergrad and isNonResearch and isCredit and isOnCampus and isNotBlacklisted:
+    classes.append(item)
+
+# Dictionary to store professor and the courses they teach
+professor_courses = {}
+
+# Populate the dictionary
+for item in classes:
+    course = item['Course']
+    professor = item['Professor']
+    email = item['Emails']
+    status = item['Has course/professor use QMI before']
+    key = (professor, email, status)
+    if key in professor_courses:
+        # Adds the course to the set of courses the professor teaches
+        professor_courses[key].add(course)
+    else:
+        professor_courses[key] = {course}
+
+professors_with_multiple_courses = {}
+
+# Find professors teaching multiple courses
+for professor, course_list in professor_courses.items():
+    if len(course_list) > 1:
+        professors_with_multiple_courses[professor] = course_list
+
+# Output results
+if professors_with_multiple_courses:
+    for (professor_name, professor_email, status), course_list in professors_with_multiple_courses.items():
+        # Remove the individual courses from the classes list that have the same professor
+        for item in classes:
+            if item['Course'] in course_list:
+              classes.remove(item)
+        class_list = list(course_list)  # Convert set to list
+        # Append the combined version of the courses
+        classes.append({
+          'Course': ', '.join(course_list),
+          'Professor': professor_name,
+          'Emails': professor_email,
+          'Has course/professor use QMI before': status
+        })
+else:
+    print("No professors teach multiple courses.")
+
+with open('../scripts/classes.csv', 'w') as file:
+    field = ["Course", "Professor", "Emails", 'Has course/professor use QMI before']
+    writer = csv.DictWriter(file, fieldnames=field)
+
+    # Write header
+    writer.writeheader()
+
+    # Write rows to the CSV
+    writer.writerows(classes)