ltc-code 0.1.81__tar.gz → 0.1.83__tar.gz

{ltc_code-0.1.81 → ltc_code-0.1.83}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ltc-code
-Version: 0.1.81
+Version: 0.1.83
 Summary: Add your description here
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown

{ltc_code-0.1.81 → ltc_code-0.1.83}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ltc-code"
-version = "0.1.81"
+version = "0.1.83"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.9"

ltc_code-0.1.83/src/ltc_code/kipp_nj.py

@@ -0,0 +1,283 @@
+# info snap select
+infosnap = infosnap.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("school_year"),
+    pl.col("cmo_name"),
+    pl.col("data_type"),
+    pl.col("ren_num"),
+
+    # --------------------------------------------------
+    # Student name
+    # --------------------------------------------------
+    pl.col("First Name").alias("fname"),
+    pl.col("stu_MiddleName").alias("mname"),
+    pl.col("Last Name").alias("lname"),
+
+    # --------------------------------------------------
+    # DOB
+    # --------------------------------------------------
+    # Ambiguous / duplicate DOB fields; keep raw
+    pl.col("DateOfBirth"),
+    pl.col("stu_DoB"),
+    pl.col("stu_DobCheck"),
+
+    # --------------------------------------------------
+    # School / campus
+    # --------------------------------------------------
+    pl.col("School").alias("school_name"),
+
+    # Ambiguous duplicate/alternate campus fields; keep raw
+    pl.col("stu_Campus"),
+    pl.col("Campus"),
+
+    # --------------------------------------------------
+    # Grades
+    # --------------------------------------------------
+    # Grade only exists in 2010 and can be missing
+    pl.col("Grade"),
+
+    pl.col("Enroll Grade").alias("enrollment_grade"),
+
+    # Ambiguous / alternate enrollment-grade fields; keep raw
+    pl.col("stu_EnrollGrade"),
+    pl.col("stu_GradeLastYear"),
+
+    # --------------------------------------------------
+    # IDs
+    # --------------------------------------------------
+    pl.col("Snapcode"),
+    pl.col("ExternalStudentID").alias("student_id"),
+
+    # --------------------------------------------------
+    # Status / enrollment status
+    # --------------------------------------------------
+    pl.col("Status"),
+    pl.col("Enroll Status").alias("enroll_status"),
+    pl.col("stu_Status"),
+
+    # --------------------------------------------------
+    # Dates / timestamps
+    # --------------------------------------------------
+    pl.col("Imported").alias("imported_date"),
+
+    # Started is after imported date; may proxy registration started/submitted
+    pl.col("Started"),
+
+    # Notification/contact dates
+    pl.col("NotificationDate"),
+    pl.col("LetterHistory"),
+
+    # --------------------------------------------------
+    # Address / geography
+    # --------------------------------------------------
+    pl.col("p1_Address").alias("address1"),
+    pl.col("p1_Address2").alias("address2"),
+    pl.col("p1_City").alias("city"),
+    pl.col("p1_Zip").alias("zip"),
+    pl.col("p1_NewarkRes"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    pl.col("p1_FirstName").alias("p1_fname"),
+    pl.col("p1_LastName").alias("p1_lname"),
+    pl.col("p_Agree1"),
+
+    # --------------------------------------------------
+    # Sibling variables
+    # --------------------------------------------------
+    pl.col("sib1_FirstName").alias("sib1_fname"),
+    pl.col("sib1_LastName").alias("sib1_lname"),
+    pl.col("sib1_Grade").alias("sib1_grade"),
+    pl.col("stu_NumberOfSibs"),
+
+    # --------------------------------------------------
+    # Prior school / preschool
+    # --------------------------------------------------
+    pl.col("stu_PrevSchool").alias("prior_school"),
+    pl.col("stu_PreSchool"),
+
+    # --------------------------------------------------
+    # Language / student attributes
+    # --------------------------------------------------
+    pl.col("stu_Language1").alias("preferred_language"),
+
+    # Possible SPED / IEP fields
+    pl.col("stu_IEPStatus").alias("sped"),
+    pl.col("stu_NeedsIEP"),
+    pl.col("stu_AdaptiveSkillsIssue"),
+    pl.col("stu_GrossMotorSkillsIssue"),
+
+    # Health / documentation fields
+    pl.col("stu_ImmunizationReceived"),
+    pl.col("stu_BirthCityNJ"),
+    pl.col("stu_BirthCityNotNJ"),
+
+    # --------------------------------------------------
+    # Notification validation / communication
+    # --------------------------------------------------
+    pl.col("EmailNotificationValidationStatus"),
+    pl.col("LetterNotificationValidationStatus"),
+    pl.col("NotificationDate"),
+
+    # --------------------------------------------------
+    # Call / follow-up fields
+    # --------------------------------------------------
+    pl.col("Call_Date"),
+    pl.col("Call_HomeVisit"),
+    pl.col("Call_HomeVisitStatus"),
+    pl.col("Call_Meeting_Date"),
+    pl.col("Call_Notes"),
+    pl.col("Call_OutcomeDecl"),
+    pl.col("Follow-up"),
+    pl.col("Outcome"),
+
+    # --------------------------------------------------
+    # Notes / admin flags
+    # --------------------------------------------------
+    pl.col("Tags"),
+    pl.col("Notes"),
+    pl.col("stu_OptInAgreement"),
+)
+
+
+# --- survey monkey select ---
+
+survey_monkey = survey_monkey.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("school_year"),
+    pl.col("cmo_name"),
+    pl.col("data_type"),
+    pl.col("ren_num"),
+
+    # --------------------------------------------------
+    # SurveyMonkey / application IDs
+    # --------------------------------------------------
+    pl.col("Respondent ID").alias("application_id"),
+    pl.col("Collector ID"),
+
+    # --------------------------------------------------
+    # Dates
+    # --------------------------------------------------
+    # Probably application/submission date
+    pl.col("End Date").alias("application_date"),
+
+    # --------------------------------------------------
+    # Student name / DOB
+    # --------------------------------------------------
+    pl.col("fname"),
+    pl.col("lname"),
+    pl.col("dob"),
+
+    # --------------------------------------------------
+    # Grade
+    # --------------------------------------------------
+    # Missing in Spark Academy file
+    pl.col("entry_grade"),
+
+    # Spark-only fields that may help infer entry grade
+    pl.col("daycare"),
+    pl.col("Headstartprek"),
+    pl.col("otherprek"),
+
+    # Many “Other” fields may indicate kindergarten/1st/2nd/etc.; keep raw for grade imputation
+    pl.col("Other"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    pl.col("p_fname"),
+    pl.col("p_lname"),
+    pl.col("p_sig"),
+
+    # --------------------------------------------------
+    # Prior school / prior setting
+    # --------------------------------------------------
+    pl.col("prior_school_type"),
+    pl.col("prior_school"),
+
+    # --------------------------------------------------
+    # Sibling priority
+    # --------------------------------------------------
+    pl.col("sib_enroll_school"),
+    pl.col("sib_fname"),
+    pl.col("sib_lname"),
+    pl.col("Sib_school").alias("sib_school"),
+
+    # Only exists in 2011-2012 CLOSED file
+    pl.col("sib_concur_school"),
+
+    # TEAM/Rise/NCA-specific sibling campus fields from 2009 import
+    pl.col("sib_TEAM_campus"),
+    pl.col("sib_Rise_Academy_Campus"),
+    pl.col("sib_Newark_Collegiate_Academy_campus"),
+
+    # --------------------------------------------------
+    # Address / geography
+    # --------------------------------------------------
+    pl.col("address1"),
+    pl.col("address2"),
+    pl.col("city"),
+    pl.col("zip"),
+
+    # NJ residency field; null may imply non-NJ / missing residency status
+    pl.col("state_nj_residents_only"),
+)
+
+
+# --- kipp newark enrollment ---
+
+enrollment = enrollment.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("academic_year").alias("school_year"),
+
+    # --------------------------------------------------
+    # Student IDs
+    # --------------------------------------------------
+    pl.col("Student_number").alias("student_id"),
+
+    # Alternate network-specific identifiers; keep raw
+    pl.col("newark_enrollment_number"),
+    pl.col("infosnap_id"),
+
+    # --------------------------------------------------
+    # Student name
+    # --------------------------------------------------
+    pl.col("first_name").alias("fname"),
+    pl.col("middle_name").alias("mname"),
+    pl.col("last_name").alias("lname"),
+
+    # --------------------------------------------------
+    # Grade / enrollment
+    # --------------------------------------------------
+    pl.col("grade_level").alias("grade"),
+
+    # Years enrolled in TEAM/KIPP network
+    pl.col("year_in_network"),
+
+    # Enrollment entry date
+    pl.col("entrydate"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    # Full names; can split later if desired
+    pl.col("Mother"),
+    pl.col("Father"),
+
+    # --------------------------------------------------
+    # Address
+    # --------------------------------------------------
+    pl.col("street").alias("address1"),
+    pl.col("city"),
+    pl.col("Zip").alias("zip"),
+)
+
+
+

ltc_code-0.1.83/src/ltc_code/kipp_tx.py

@@ -0,0 +1,581 @@
+# KIPP HOUSTON LOTTO
+df = df.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("school_year"),
+    pl.col("cmo_name"),
+    pl.col("data_type"),
+    pl.col("ren_num"),
+
+    # --------------------------------------------------
+    # Student name
+    # --------------------------------------------------
+    pl.col("First Name").alias("fname"),
+    pl.col("stu_MiddleName").alias("mname"),
+    pl.col("Last Name").alias("lname"),
+
+    # Ambiguous / duplicate DOB fields; keep raw
+    pl.col("Date Of Birth"),
+    pl.col("stu_DOB"),
+
+    # --------------------------------------------------
+    # Grades
+    # --------------------------------------------------
+    pl.col("Apply Grade").alias("entry_grade"),
+    pl.col("Grade").alias("grade"),
+
+    # Current grade, not necessarily application grade
+    pl.col("Stu_CurrGrade"),
+
+    # --------------------------------------------------
+    # Application IDs / student IDs
+    # --------------------------------------------------
+    pl.col("SubmissionRecordID").alias("application_id"),
+    pl.col("ExternalStudentID").alias("student_id"),
+    pl.col("LotteryFamilyCode").alias("family_id"),
+    pl.col("Confirmation ID"),
+    pl.col("ActionSessionID"),
+
+    # --------------------------------------------------
+    # Application dates / admin timestamps
+    # --------------------------------------------------
+    pl.col("Submitted").alias("application_date"),
+
+    # Ambiguous: lottery-specific app date
+    pl.col("LotterySubmissionDate"),
+
+    # Ambiguous: unclear how this differs from Submitted
+    pl.col("Started"),
+
+    pl.col("ApprovalDate"),
+    pl.col("DeliveryDate").alias("last_update_date"),
+    pl.col("stu_RegistrationDate").alias("registration_complete_date"),
+    pl.col("p_Date").alias("parent_signature_date"),
+
+    # --------------------------------------------------
+    # Lottery / status
+    # --------------------------------------------------
+    pl.col("LotteryStatus").alias("lottery_status"),
+    pl.col("LotteryPosition").alias("lottery_number"),
+    pl.col("LotteryPriority").alias("priority_group"),
+
+    # Ambiguous alt lottery number field
+    pl.col("stu_NumberPickedInLottery"),
+
+    pl.col("Status").alias("application_status"),
+    pl.col("ApprovalHistory"),
+    pl.col("DeliveryHistory"),
+    pl.col("SWMMatchStatus"),
+
+    # --------------------------------------------------
+    # Offer / enrollment proxies
+    # --------------------------------------------------
+    pl.col("stu_OfferingSpot").alias("offer"),
+    pl.col("Accepted School1"),
+    pl.col("Accepted School2"),
+    pl.col("Accepted School3"),
+    pl.col("Accepted School4"),
+
+    # Ambiguous post-offer/enrollment process variables
+    pl.col("stu_SignedCte"),
+    pl.col("stu_HomeVisit"),
+    pl.col("stu_HSPacketAgree"),
+    pl.col("stu_ExportToFE"),
+
+    # --------------------------------------------------
+    # School applying to / school attended
+    # --------------------------------------------------
+    pl.col("School Applying To").alias("school_name"),
+
+    # Ambiguous school-applied-to alternatives; keep raw
+    pl.col("hid_ApplyingSchoolName"),
+    pl.col("lottery_ApplyingToWhichSchool"),
+    pl.col("Applied School"),
+    pl.col("School"),
+
+    # Ambiguous: current/prior/next-year school fields
+    pl.col("Stu_CurrSchool"),
+    pl.col("stu_CurrentlyAttendingSchool"),
+    pl.col("stu_SchoolNextYear"),
+    pl.col("stu_School"),
+    pl.col("stu_School2"),
+
+    # Zoned / feeder school fields
+    pl.col("Zoned School").alias("feeder_school"),
+    pl.col("hid_ZonedSchoolName"),
+    pl.col("lottery_ZonedForWhichSchools"),
+    pl.col("zone_school_id"),
+    pl.col("zone_school_address"),
+    pl.col("zone_school_address2"),
+    pl.col("zone_school_grades"),
+    pl.col("zone_school_grades2"),
+
+    # --------------------------------------------------
+    # Address / geography
+    # --------------------------------------------------
+    pl.col("Stu_address").alias("address1"),
+    pl.col("Stu_address2").alias("address2"),
+    pl.col("Stu_address3"),
+    pl.col("Stu_city").alias("city"),
+    pl.lit("TX").alias("state"),
+
+    # Duplicate/alternate zip fields; keep raw
+    pl.col("stu_Zip"),
+    pl.col("Zip"),
+    pl.col("stu_ZipOther"),
+
+    pl.col("stu_CurrSchoolDistrict"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    pl.col("p1_FirstName").alias("p1_fname"),
+    pl.col("p1_LastName").alias("p1_lname"),
+    pl.col("p2_FirstName").alias("p2_fname"),
+    pl.col("p2_LastName").alias("p2_lname"),
+    pl.col("p_Sig").alias("p_name"),
+
+    pl.col("Stu_kippsterparent").alias("p_employed"),
+
+    # Ambiguous / alternate parent employment fields
+    pl.col("Parent Employee Priority"),
+    pl.col("stu_KIPPParentName"),
+    pl.col("stu_KIPPsterParentSchool"),
+
+    # --------------------------------------------------
+    # Sibling priority / sibling names
+    # --------------------------------------------------
+    pl.col("Stu_kippsibling").alias("sib_enroll_cmo"),
+    pl.col("Stu_kippsiblingfirstname").alias("sib_fname"),
+    pl.col("Stu_kippsiblinglastname").alias("sib_lname"),
+    pl.col("Stu_kippsiblingschool").alias("sib_school"),
+
+    pl.col("sib1_FirstName").alias("sib1_fname"),
+    pl.col("sib1_LastName").alias("sib1_lname"),
+    pl.col("sib2_FirstName").alias("sib2_fname"),
+    pl.col("sib2_LastName").alias("sib2_lname"),
+    pl.col("sib3_FirstName").alias("sib3_fname"),
+    pl.col("sib3_LastName").alias("sib3_lname"),
+
+    # Ambiguous / alternate sibling priority fields
+    pl.col("Sibling Priority"),
+    pl.col("stu_KIPPSiblingName"),
+    pl.col("stu_KIPPSiblingAlum").alias("sib_grad_cmo"),
+    pl.col("stu_KIPPSiblingAlumFirstName"),
+    pl.col("stu_KIPPSiblingLastName"),
+    pl.col("stu_KIPPSiblingAlumSchool"),
+    pl.col("stu_KIPPSiblingAlumGradYear"),
+
+    # --------------------------------------------------
+    # Priority / geography priority
+    # --------------------------------------------------
+    pl.col("Priority Tier"),
+    pl.col("Stu_priorityzone"),
+    pl.col("nozonedschool"),
+    pl.col("outofdistrict"),
+    pl.col("stu_RequestingZonedSchool"),
+    pl.col("lottery_HasAffiliationToKIPP"),
+    pl.col("stu_CurrKipp"),
+
+    # --------------------------------------------------
+    # Late app / submission type
+    # --------------------------------------------------
+    pl.col("stu_Deadline").alias("late_app"),
+    pl.col("stu_SubmissionType").alias("application_source"),
+
+    # --------------------------------------------------
+    # Language / demographics / student attributes
+    # --------------------------------------------------
+    pl.col("stu_Language").alias("preferred_language"),
+    pl.col("stu_LanguageOther"),
+    pl.col("stu_HomeLanguage"),
+    pl.col("Stu_age"),
+    pl.col("stu_QualifyPK"),
+    pl.col("stu_BirthCertificate"),
+    pl.col("stu_LunchAppResults"),
+    pl.col("stu_ResidencyQuestionaire"),
+    pl.col("stu_SpecNeedQuestionnaire").alias("sped"),
+    pl.col("stu_TXCharter"),
+
+    # --------------------------------------------------
+    # School application flags
+    # --------------------------------------------------
+    pl.col("stu_SHARP"),
+    pl.col("stu_Sharpstown"),
+    pl.col("stu_SHINE"),
+    pl.col("stu_KMS"),
+    pl.col("stu_KHHS"),
+    pl.col("stu_Unity"),
+    pl.col("stu_WestHoustonAcademy"),
+    pl.col("stu_3D"),
+    pl.col("stu_BellaireMiddle"),
+    pl.col("stu_BellairePrimary"),
+    pl.col("stu_CONNECTMiddle"),
+    pl.col("stu_CONNEXTPrimary"),
+    pl.col("stu_Coastal"),
+    pl.col("stu_Courage"),
+    pl.col("stu_DREAM"),
+    pl.col("stu_Explore"),
+    pl.col("stu_Generations"),
+    pl.col("stu_Intrepid"),
+    pl.col("stu_Polaris"),
+    pl.col("stu_Legacy"),
+    pl.col("stu_Liberation"),
+    pl.col("stu_NortheastCollegePrep"),
+    pl.col("stu_PEACE"),
+    pl.col("stu_Spirit"),
+    pl.col("stu_Sunnyside"),
+    pl.col("stu_Voyage"),
+    pl.col("stu_ZENITH"),
+
+    # --------------------------------------------------
+    # Notes / admin follow-up fields
+    # --------------------------------------------------
+    pl.col("Tags"),
+    pl.col("Notes"),
+    pl.col("Notes86"),
+    pl.col("Notes18"),
+    pl.col("Notes34"),
+    pl.col("stu_NotesComments"),
+    pl.col("stu_WD"),
+    pl.col("stu_PostLotteryParentCommunicationDate"),
+    pl.col("stu_CloseWaitlistDate"),
+    pl.col("stu_CloseWaitlistForSchool"),
+    pl.col("stu_ApplyOtherSchools"),
+    pl.col("stu_ConsiderOtherKipp"),
+    pl.col("stu_OtherSchool1"),
+    pl.col("stu_OtherSchool2"),
+    pl.col("stu_OtherSchool3"),
+    pl.col("stu_OtherSchool4"),
+    pl.col("stu_OtherSchool5"),
+)
+
+
+
+
+# Kipp houston registration data
+registration = registration.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("school_year"),
+    pl.col("cmo_name"),
+    pl.col("data_type"),
+    pl.col("ren_num"),
+
+    # --------------------------------------------------
+    # Student name
+    # --------------------------------------------------
+    pl.col("First Name").alias("fname"),
+    pl.col("stu_MiddleName").alias("mname"),
+    pl.col("Last Name").alias("lname"),
+
+    # Ambiguous / duplicate student-name fields; keep raw
+    pl.col("stu_FirstName"),
+    pl.col("stu_LastName"),
+    pl.col("stu_Suffix").alias("suffix"),
+
+    # --------------------------------------------------
+    # DOB
+    # --------------------------------------------------
+    # Ambiguous / duplicate DOB fields; keep raw
+    pl.col("DateOfBirth"),
+    pl.col("stu_DoB"),
+
+    # --------------------------------------------------
+    # Grades
+    # --------------------------------------------------
+    pl.col("Grade").alias("grade"),
+    pl.col("stu_Grade").alias("enrollment_grade"),
+
+    # --------------------------------------------------
+    # School / enrollment school
+    # --------------------------------------------------
+    pl.col("School").alias("enrollment_school_name"),
+
+    # Ambiguous duplicate school field; keep raw
+    pl.col("stu_School"),
+
+    # --------------------------------------------------
+    # Enrollment / registration status
+    # --------------------------------------------------
+    pl.col("Enroll Status").alias("enroll_status"),
+    pl.col("stu_Status"),
+    pl.col("Status").alias("registration_status"),
+
+    # --------------------------------------------------
+    # IDs
+    # --------------------------------------------------
+    pl.col("ID"),
+    pl.col("SubmissionRecordID").alias("application_id"),
+    pl.col("stu_ID").alias("student_id"),
+    pl.col("SnapCode"),
+    pl.col("FamilyMemberID"),
+    pl.col("FamilyID").alias("family_id"),
+    pl.col("ActionSessionID"),
+    pl.col("stu_TSDSLocalID"),
+
+    # --------------------------------------------------
+    # Registration / admin dates
+    # --------------------------------------------------
+    pl.col("Submitted").alias("registration_submit_date"),
+
+    # Ambiguous: starts around June; maybe registration/enrollment start date
+    pl.col("Started"),
+
+    pl.col("ApprovalDate"),
+    pl.col("DeliveryDate").alias("last_update_date"),
+    pl.col("p_Date").alias("parent_signature_date"),
+
+    # Only exists 2017-2018; unclear exact meaning
+    pl.col("stu_Date"),
+
+    # --------------------------------------------------
+    # Lottery fields included in registration data
+    # --------------------------------------------------
+    # Mostly non-useful constants in registration files, but keep for checks
+    pl.col("LotteryStatus"),
+    pl.col("LotteryPosition").alias("lottery_number"),
+    pl.col("LotteryPriority").alias("priority_group"),
+    pl.col("LotterySubmissionDate"),
+
+    # --------------------------------------------------
+    # Prior / zoned school
+    # --------------------------------------------------
+    pl.col("stu_ZonedDistrict"),
+    pl.col("stu_ZonedSchool").alias("feeder_school"),
+    pl.col("stu_PrevKIPP"),
+    pl.col("stu_PrevKIPPYear"),
+
+    # Prior school name; other prev-school fields mostly broad missing per notes
+    pl.col("stu_PrevSch1Name").alias("prior_school"),
+
+    # --------------------------------------------------
+    # Address / geography
+    # --------------------------------------------------
+    pl.col("stu_Addrnum"),
+    pl.col("stu_AddrStreet"),
+    pl.col("stu_AddrCity").alias("city"),
+    pl.col("stu_AddrZip").alias("zip"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    pl.col("p1_FirstName").alias("p1_fname"),
+    pl.col("p1_LastName").alias("p1_lname"),
+    pl.col("p2_FirstName").alias("p2_fname"),
+    pl.col("p2_LastName").alias("p2_lname"),
+
+    # Parent volunteer fields, not employment priority
+    pl.col("p1_Volunteer"),
+    pl.col("p2_Volunteer"),
+
+    pl.col("p_Agree"),
+    pl.col("p_Sig").alias("p_name"),
+
+    # --------------------------------------------------
+    # Sibling variables
+    # --------------------------------------------------
+    # Count/number of KIPP siblings, not binary
+    pl.col("stu_KIPPSibling"),
+
+    pl.col("stu_KIPPSib1Fname").alias("sib1_fname"),
+    pl.col("stu_KIPPSib1Lname").alias("sib1_lname"),
+    pl.col("stu_KIPPSib1Grade").alias("sib1_grade"),
+    pl.col("stu_KIPPSib1Sch").alias("sib_school"),
+
+    # --------------------------------------------------
+    # Language / student attributes
+    # --------------------------------------------------
+    pl.col("Language").alias("preferred_language"),
+    pl.col("stu_SNAPTANF"),
+    pl.col("stu_SNAPTANFNumber"),
+
+    # --------------------------------------------------
+    # Returning student indicators
+    # --------------------------------------------------
+    # Only 2018; useful for separating new vs returning registrations
+    pl.col("stu_ReturningNo"),
+    pl.col("stu_ReturningYesNo"),
+)
+
+
+# other files
+app = app.select(
+    # --------------------------------------------------
+    # Metadata
+    # --------------------------------------------------
+    pl.col("school_year"),
+    pl.col("cmo_name"),
+    pl.col("data_type"),
+    pl.col("ren_num"),
+
+    # --------------------------------------------------
+    # Student IDs / application IDs
+    # --------------------------------------------------
+    pl.col("Application ID").alias("application_id"),
+    pl.col("Application_ID"),
+    pl.col("Applicant_ID"),
+    pl.col("Account ID").alias("account_id"),
+    pl.col("Account_ID"),
+    pl.col("SM_Student_Id"),
+    pl.col("SIS_Student_Id"),
+    pl.col("SIS ID"),
+    pl.col("Student_ID_Schoolmint"),
+    pl.col("Student_ID_Local").alias("student_id"),
+    pl.col("Student_ID_State"),
+
+    # --------------------------------------------------
+    # Student name
+    # --------------------------------------------------
+    pl.col("Students First Name").alias("fname"),
+    pl.col("Student Middle Name").alias("mname"),
+    pl.col("Students Last Name").alias("lname"),
+
+    # Duplicate/alternate name fields from KIPP Texas-style exporter; keep raw
+    pl.col("First_Name"),
+    pl.col("Middle_Name"),
+    pl.col("Last_Name"),
+
+    # --------------------------------------------------
+    # DOB
+    # --------------------------------------------------
+    pl.col("Student Birth Date").alias("dob"),
+
+    # Duplicate/alternate DOB from KIPP Texas-style exporter; keep raw
+    pl.col("Birth_Date"),
+
+    # --------------------------------------------------
+    # School applying to
+    # --------------------------------------------------
+    pl.col("School Applying to").alias("school_name"),
+    pl.col("School_Name"),
+    pl.col("School_Id"),
+
+    # --------------------------------------------------
+    # Grades
+    # --------------------------------------------------
+    pl.col("Grade Applying To").alias("entry_grade"),
+    pl.col("Current Grade"),
+
+    # KIPP Texas-style exporter fields; keep raw if ambiguous/alternate
+    pl.col("Grade"),
+    pl.col("Current_Grade_Level"),
+
+    # --------------------------------------------------
+    # Application / enrollment period
+    # --------------------------------------------------
+    pl.col("Enrollment Period"),
+    pl.col("Academic_Year"),
+
+    # --------------------------------------------------
+    # Application status
+    # --------------------------------------------------
+    pl.col("Application Status").alias("application_status"),
+    pl.col("Status"),
+    pl.col("Public_Status"),
+
+    # Status updated timestamp
+    pl.col("Status_Timestamp").alias("status_timestamp"),
+
+    # --------------------------------------------------
+    # Application type / rank
+    # --------------------------------------------------
+    pl.col("Application Type").alias("application_type"),
+    pl.col("Application_Type"),
+    pl.col("Applicant_Rank").alias("rank"),
+
+    # --------------------------------------------------
+    # Dates
+    # --------------------------------------------------
+    pl.col("Submission Date").alias("application_date"),
+    pl.col("Submission_Date"),
+    pl.col("Created_At"),
+
+    pl.col("Offered Date").alias("offer_date"),
+    pl.col("Offered_Date"),
+
+    pl.col("Accepted Date").alias("offer_accept_date"),
+    pl.col("Accepted_Date"),
+
+    pl.col("Registration Completed Date").alias("registration_complete_date"),
+    pl.col("Registration_Completed_Date"),
+
+    # Ambiguous: completion vs verification gap needs checking
+    pl.col("Registration_Verified_Date"),
+
+    # --------------------------------------------------
+    # Lottery / waitlist
+    # --------------------------------------------------
+    pl.col("Waitlist Number").alias("waitlist_number"),
+    pl.col("Waitlist_Number"),
+
+    # Ambiguous: compare global vs non-global lottery numbers
+    pl.col("Lottery_Global_Number"),
+    pl.col("Lottery_Number"),
+    pl.col("Lottery_Tickets"),
+
+    # --------------------------------------------------
+    # Parent / guardian
+    # --------------------------------------------------
+    pl.col("Primary Guardian First Name").alias("p1_fname"),
+    pl.col("Primary Guardian Last Name").alias("p1_lname"),
+    pl.col("Secondary Guardian First Name").alias("p2_fname"),
+    pl.col("Secondary Guardian Last Name").alias("p2_lname"),
+
+    # Parent/guardian employed by organization
+    pl.col("Is one of the parents or guardians employed by this organization?").alias("p_employed"),
+
+    # --------------------------------------------------
+    # Address / geography
+    # --------------------------------------------------
+    pl.col("StudentAddress: Street 1").alias("address1"),
+    pl.col("StudentAddress: Street 2").alias("address2"),
+    pl.col("StudentAddress: City").alias("city"),
+    pl.col("StudentAddress: ZIP").alias("zip"),
+    pl.col("District"),
+    pl.col("District_Name"),
+
+    # --------------------------------------------------
+    # Current / prior school
+    # --------------------------------------------------
+    pl.col("Current School").alias("prior_school"),
+    pl.col("Current_School_Name"),
+
+    # --------------------------------------------------
+    # Language
+    # --------------------------------------------------
+    pl.col("Account Preferred Language").alias("preferred_language"),
+
+    # --------------------------------------------------
+    # Admin/source fields
+    # --------------------------------------------------
+    pl.col("admin_info_source"),
+    pl.col("Homeless Shelter"),
+
+    # --------------------------------------------------
+    # Priority fields
+    # --------------------------------------------------
+    # Broad priority text/code fields; keep both because they differ by exporter
+    pl.col("Priorities"),
+    pl.col("Priority_List"),
+
+    # --------------------------------------------------
+    # Sibling priority fields
+    # --------------------------------------------------
+    pl.col(
+        "Does the student have a brother or sister currently attending a school within this organization?"
+    ).alias("sib_enroll_cmo"),
+
+    pl.col(
+        "Does the student have a brother or sister who has graduated from a school within this organization?"
+    ).alias("sib_grad_cmo"),
+
+    pl.col("Sibling 1 Name").alias("sib1_name"),
+    pl.col("Sibling 1 Grade").alias("sib1_grade"),
+    pl.col("Sibling 1 School").alias("sib1_school"),
+
+    pl.col("Sibling 2 Name").alias("sib2_name"),
+    pl.col("Sibling 2 School").alias("sib2_school"),
+)