ncs_mdes 0.10.0 → 0.11.0

data/documents/3.2/extract_disposition_codes.rb

@@ -0,0 +1,58 @@
+#!/usr/bin/env ruby
+
+##
+# A script to generate disposition_codes.yml from the MDES 3.2
+# spreadsheet's Dispositions tab.
+#
+# This script requires the 'roo' gem, which is not included in
+# ncs_mdes's gemspec because it has a huge number of dependencies and
+# is not needed at runtime.
+
+require 'rubygems'
+require 'roo'
+require 'yaml'
+
+MDES_XLSX = ARGV.first or fail 'Please provide the path to the MDES spreadsheet'
+TARGET = File.expand_path('../disposition_codes.yml', __FILE__)
+SHEET_NAME = 'Dispositions'
+
+MAPPED_COLUMNS = {
+  'A' => 'final_category',
+  'B' => 'sub_category',
+  'C' => 'disposition'
+}
+
+def normalize_whitespace(s)
+  s.strip.gsub(/\s+/, " ")
+end
+
+book = Excelx.new(MDES_XLSX)
+
+# This is the array of hashes that will eventually be serialized to
+# disposition_codes.yml.
+dispositions = []
+
+current_event = nil
+current_category_code = nil
+1.upto(book.last_row(SHEET_NAME)) do |row_number|
+  a, b = %w(A B).collect { |col| book.cell(row_number, col, SHEET_NAME) }
+  if a =~ /Category\s+(\d)\s+\((.*?)\)\s+Disposition\s+Codes/
+    current_event = normalize_whitespace $2
+    current_category_code = $1.to_i
+    puts "Collecting for category #{current_event} (#{current_category_code})"
+  elsif b =~ /\S/ && a !~ /FINAL/
+    disposition_hash = MAPPED_COLUMNS.inject({}) do |h, (col, key)|
+      h[key] = normalize_whitespace(book.cell(row_number, col, SHEET_NAME)); h
+    end
+    disposition_hash['event'] = current_event
+    disposition_hash['category_code'] = current_category_code
+    disposition_hash['interim_code'], disposition_hash['final_code'] =
+      normalize_whitespace(book.cell(row_number, 'D', SHEET_NAME)).split('/')
+
+    dispositions << disposition_hash
+  end
+end
+
+File.open(TARGET, 'w') do |f|
+  f.write dispositions.sort_by { |h| [h['category_code'], h['final_code']] }.to_yaml
+end

data/documents/3.2/heuristic_overrides.yml

@@ -0,0 +1,681 @@
+foreign_keys:
+  birth_visit:
+    new_address_id: address
+    mail_address_id: address
+    new_address_b_id: address
+
+  birth_visit_2:
+    new_address_id: address
+    mail_address_id: address
+    new_address_b_id: address
+
+  birth_visit_3:
+    cwork_address_id: address
+    work_address_id: address
+
+  birth_visit_4:
+    cwork_address_id: address
+    work_address_id: address
+
+  birth_visit_baby_name:
+    bv_id: birth_visit
+
+  birth_visit_baby_name_2:
+    bv_id: birth_visit_2
+
+  birth_visit_decorate_room:
+    bv_id: birth_visit
+
+  birth_visit_decorate_room_2:
+    bv_id: birth_visit_2
+
+  birth_visit_li:
+    new_address_id: address
+    mail_address_id: address
+    new_address_b_id: address
+
+  birth_visit_li_2:
+    cwork_address_id: address
+    work_address_id: address
+
+  birth_visit_renovate_room:
+    bv_id: birth_visit
+
+  birth_visit_renovate_room_2:
+    bv_id: birth_visit_2
+
+  bitsea_saq:
+    r_p_id: participant
+
+  breast_milk_saq:
+    r_p_id: participant
+
+  bsi_saq:
+    r_p_id: participant
+
+  child_anthro:
+    r_p_id: participant
+
+  child_blood:
+    r_p_id: participant
+    # According to the corresponding instrument, equip_id is manually entered
+    # and not a reference to another table.
+    equip_id: false
+
+  child_bp:
+    r_p_id: participant
+
+  child_saliva:
+    r_p_id: participant
+
+  child_saliva_saq:
+    r_p_id: participant
+
+  child_urine:
+    r_p_id: participant
+
+  core_quest:
+    r_p_id: participant
+
+  core_quest_child_care:
+    r_p_id: participant
+    c_address_id: address
+
+  core_quest_occupation:
+    r_p_id: participant
+    work_address_id: address
+
+  drf_therm_verification:
+    equip_id: env_equipment
+
+  eighteen_mth_mother:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  eighteen_mth_mother_2:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  eighteen_mth_mother_3:
+    r_p_id: participant
+
+  eighteen_mth_mother_cond:
+    eighteen_mth_habits_id: eighteen_mth_mother_habits
+
+  eighteen_mth_mother_lice_2:
+    eighteen_mth_mother_id: eighteen_mth_mother_2
+
+  eighteen_mth_mother_saq:
+    asq18_address_id: address
+    r_p_id: participant
+
+  eighteen_mth_mother_saq_2:
+    r_p_id: participant
+    asq18_address_id: address
+    home_phone_id: telephone
+
+  env_equipment_prob_log:
+    equip_id: env_equipment
+    staff_id_reviewer: staff
+
+  event:
+    participant_id: participant
+
+  fourteen_mth_asq_saq:
+    r_p_id: participant
+    asq14_address_id: address
+    home_phone_id: telephone
+
+  household_enumeration_hidden_du:
+    hdu_address_id: address
+
+  incident:
+    inc_staff_reporter_id: staff
+    inc_staff_supervisor_id: staff
+    inc_recip_is_participant: participant
+    inc_recip_is_du: dwelling_unit
+    inc_recip_is_staff: staff
+    inc_recip_is_family: person
+    inc_recip_is_acquaintance: person
+    inc_contact_person: person
+
+  itsp_saq:
+    r_p_id: participant
+
+  m_chat_saq:
+    r_p_id: participant
+
+  multi_mode:
+    r_p_id: participant
+
+  nine_mth_mother:
+    r_p_id: participant
+
+  nine_mth_mother_2:
+    r_p_id: participant
+
+  non_interview_respondent:
+    r_p_id: participant
+
+  participant_consent:
+    person_who_consented_id: person
+    person_wthdrw_consent_id: person
+
+  participant_consent_reconsideration:
+    r_p_id: participant
+
+  participant_verif:
+    r_p_id: participant
+    c_address_id: address
+    pa_phone_id: telephone
+    s_address_id: address
+    sa_phone_id: telephone
+
+  participant_vis_consent:
+    vis_person_who_consented_id: person
+
+  participant_rvis:
+    rvis_person: person
+
+  pb_recruitment:
+    address_pb_id: address
+
+  pb_recruitment_2:
+    address_pb_id: address
+
+  pb_recruitment_info_source:
+    pb_recruitment_id: pb_recruitment
+
+  pb_recruitment_info_source_2:
+    pb_recruitment_id: pb_recruitment_2
+
+  pb_recruitment_prov_source:
+    pb_recruitment_id: pb_recruitment
+
+  pb_recruitment_prov_source_2:
+    pb_recruitment_id: pb_recruitment_2
+
+  pb_recruitment_prov_svc:
+    pb_recruitment_id: pb_recruitment
+
+  pb_recruitment_prov_svc_2:
+    pb_recruitment_id: pb_recruitment_2
+
+  pb_samp_frame:
+    practice_address_id: address
+    contact_address_id: address
+    contact_email_id: email
+    contact_phone_num_id: telephone
+    contact_address_larger_id: address
+    contact_email_larger_id: email
+    contact_phone_num_larger_id: telephone
+
+  pb_samp_frame_2:
+    practice_address_id: address
+    contact_address_id: address
+    contact_email_id: email
+    contact_phone_num_id: telephone
+    contact_address_larger_id: address
+    contact_email_larger_id: email
+    contact_phone_num_larger_id: telephone
+
+  pbs_elig_screener:
+    r_phone_1_id: telephone
+    r_phone_2_id: telephone
+    r_email_id: email
+
+  pbs_elig_screener_hosp:
+    r_p_id: participant
+    r_phone_1_id: telephone
+    r_phone_2_id: telephone
+    r_email_id: email
+
+  pbs_elig_screener_pr_office:
+    pr_address_id: address
+
+  pbs_elig_screener_hosp_pr_office:
+    pr_address_id: address
+
+  pbs_participant_verif:
+    r_p_id: participant
+
+  person:
+    new_address_id: address
+
+  prec_therm_cert:
+    equip_id: env_equipment
+
+  pre_preg:
+    c_addr1_id: address
+    c_addr2_id: address
+
+  preg_screen_eh:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_eh_2:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_eh_know_ncs:
+    ps_eh_id: preg_screen_eh
+
+  preg_screen_eh_know_ncs_2:
+    ps_eh_id: preg_screen_eh_2
+
+  preg_screen_eh_race:
+    ps_eh_id: preg_screen_eh
+
+  preg_screen_eh_race_2:
+    ps_eh_id: preg_screen_eh_2
+
+  preg_screen_hi:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_hi_2:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_hi_know_ncs:
+    ps_hi_id: preg_screen_hi
+
+  preg_screen_hi_know_ncs_2:
+    ps_hi_id: preg_screen_hi_2
+
+  preg_screen_hi_race:
+    ps_hi_id: preg_screen_hi
+
+  preg_screen_hi_race_2:
+    ps_hi_id: preg_screen_hi_2
+
+  preg_screen_pb:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_pb_2:
+    mail_address_id: address
+    new_address_id: address
+
+  preg_screen_pb_know_ncs:
+    ps_pb_id: preg_screen_pb
+
+  preg_screen_pb_know_ncs_2:
+    ps_pb_id: preg_screen_pb_2
+
+  preg_screen_pb_race:
+    ps_pb_id: preg_screen_pb
+
+  preg_screen_pb_race_2:
+    ps_pb_id: preg_screen_pb_2
+
+  preg_visit_1:
+    b_address_id: address
+    c_addr1_id: address
+    c_addr2_id: address
+
+  preg_visit_1_2:
+    b_address_id: address
+    c_addr1_id: address
+    c_addr2_id: address
+
+  preg_visit_1_3:
+    b_address_id: address
+    work_address_id: address
+
+  preg_visit_1_commute:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_commute_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_cool:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_cool_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_diagnose_2:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_diagnose_2_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_heat2:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_heat2_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_local_trav:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_local_trav_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_nonenglish2_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_pdecorate_room:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_pdecorate_room_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_pet_type:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_pet_type_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_prenovate2_room:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_prenovate_room:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_prenovate_room_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_room_mold:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_room_mold_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_1_saq_2:
+    f_addr_id: address
+
+  preg_visit_1_saq_3:
+    f_addr_id: address
+
+  preg_visit_1_saq_4:
+    f_addr_id: address
+    f_phone_id: telephone
+
+  preg_visit_1_sp_race:
+    pv1_id: preg_visit_1
+
+  preg_visit_1_sp_race_2:
+    pv1_id: preg_visit_1_2
+
+  preg_visit_2:
+    b_address_id: address
+    c_addr1_id: address
+    c_addr2_id: address
+
+  preg_visit_2_2:
+    b_address_id: address
+    c_addr1_id: address
+    c_addr2_id: address
+
+  preg_visit_2_3:
+    b_address_id: address
+    work_address_id: address
+    cwork_address_id: address
+
+  preg_visit_2_cool:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_cool_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_2_diagnose_2:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_diagnose_2_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_2_heat2:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_heat2_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_2_pdecorate2_room:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_pdecorate2_room_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_2_prenovate_room:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_prenovate_room_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_2_room_mold:
+    pv2_id: preg_visit_2
+
+  preg_visit_2_room_mold_2:
+    pv2_id: preg_visit_2_2
+
+  preg_visit_li:
+    b_address_id: address
+
+  preg_visit_li_2:
+    b_address_id: address
+
+  preg_visit_li_cool:
+    pv_li_id: preg_visit_li
+
+  preg_visit_li_cool_2:
+    pv_li_id: preg_visit_li_2
+
+  reconsideration_ins:
+    r_p_id: participant
+
+  ref_freezer_verification:
+    equip_id: env_equipment
+
+  sample_receipt_store:
+    equip_id: env_equipment
+
+  sample_shipping:
+    staff_id_track: staff
+
+  six_mth_mother:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  six_mth_mother_2:
+    r_p_id: participant
+
+  six_mth_saq_formula_type:
+    six_mth_saq_id: six_mth_saq
+
+  six_mth_saq_2:
+    r_p_id: participant
+
+  six_mth_saq_3:
+    r_p_id: participant
+
+  six_mth_saq_4:
+    r_p_id: participant
+
+  six_mth_saq_formula_type_2:
+    six_mth_saq_id: six_mth_saq_2
+
+  six_mth_saq_supp:
+    six_mth_saq_id: six_mth_saq
+
+  six_mth_saq_supp_2:
+    six_mth_saq_id: six_mth_saq_2
+
+  six_mth_saq_water:
+    six_mth_saq_id: six_mth_saq
+
+  six_mth_saq_water_2:
+    six_mth_saq_id: six_mth_saq_2
+
+  sixteen_mth_asq_saq:
+    r_p_id: participant
+    asq16_address_id: address
+    home_phone_id: telephone
+
+  spec_blood:
+    # According to the corresponding instrument, equip_id is manually entered
+    # and not a reference to another table.
+    equip_id: false
+
+  spec_blood_2:
+    # According to the corresponding instrument, equip_id is manually entered
+    # and not a reference to another table.
+    equip_id: false
+
+  spec_cord_blood_specimen:
+    spec_cord_blood_id: spec_cord_blood
+
+  spec_cord_blood_specimen_2:
+    spec_cord_blood_id: spec_cord_blood_2
+
+  spec_receipt:
+    # Insufficient info to determine what this is supposed to link to
+    storage_container_id: false
+    equip_id: spec_equipment
+
+  spec_shipping:
+    # Insufficient info to determine what this is supposed to link to
+    shipper_id: false
+
+  spec_storage:
+    equip_id: spec_equipment
+
+  staff_exp_mngmnt_tasks:
+    staff_weekly_expense_id: staff_weekly_expense
+
+  staff_exp_data_cllctn_tasks:
+    staff_weekly_expense_id: staff_weekly_expense
+
+  thirty_month_interview:
+    r_p_id: participant
+
+  thirty_mth_asq_saq:
+    r_p_id: participant
+    asq30_address_id: address
+    home_phone_id: telephone
+
+  three_mth_mother:
+    r_p_id: participant
+
+  three_mth_mother_2:
+    r_p_id: participant
+
+  three_mth_mother_3:
+    r_p_id: participant
+
+  tracing_int:
+    r_p_id: participant
+    new_address_id: address
+    home_phone_id: telephone
+    cell_phone_id: telephone
+    c_addr1_id: address
+    contact_phone_id: telephone
+    contact_phone_2_1_id: telephone
+    c_addr2_id: address
+    contact_phone_2_id: telephone
+    contact_phone_2_2_id: telephone
+    c_addr3_id: address
+    contact_phone_3_id: telephone
+    contact_phone_2_3_id: telephone
+
+  trh_meter_calibration:
+    equip_id: spec_equipment
+
+  twelve_mth_mother:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  twelve_mth_mother_2:
+    r_p_id: participant
+    c_addr1_id: address
+    c_addr2_id: address
+
+  twelve_mth_mother_3:
+    r_p_id: participant
+
+  twelve_mth_saq_2:
+    r_p_id: participant
+
+  twelve_mth_saq_3:
+    r_p_id: participant
+
+  twelve_mth_saq_formula_brand:
+    twelve_mth_saq_id: twelve_mth_saq
+
+  twelve_mth_saq_formula_brand_2:
+    twelve_mth_saq_id: twelve_mth_saq_2
+
+  twelve_mth_saq_formula_type:
+    twelve_mth_saq_id: twelve_mth_saq
+
+  twelve_mth_saq_formula_type_2:
+    twelve_mth_saq_id: twelve_mth_saq_2
+
+  twelve_mth_saq_supplement:
+    twelve_mth_saq_id: twelve_mth_saq
+
+  twelve_mth_saq_supplement_2:
+    twelve_mth_saq_id: twelve_mth_saq_2
+
+  twelve_mth_saq_water:
+    twelve_mth_saq_id: twelve_mth_saq
+
+  twelve_mth_saq_water_2:
+    twelve_mth_saq_id: twelve_mth_saq_2
+
+  twenty_four_mth_mother:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  twenty_four_mth_mother_2:
+    c_addr1_id: address
+    c_addr2_id: address
+    r_p_id: participant
+
+  twenty_four_mth_mother_3:
+    r_p_id: participant
+    work_address_id: address
+    work_address_a_id: address
+
+  twenty_four_mth_mother_otc:
+    twenty_four_mth_mother_id: twenty_four_mth_mother
+
+  twenty_four_mth_mother_prescr:
+    twenty_four_mth_mother_id: twenty_four_mth_mother
+
+  twenty_four_mth_mother_suppl:
+    twenty_four_mth_mother_id: twenty_four_mth_mother
+
+  twenty_four_mth_saq:
+    asq24_address_id: address
+    r_p_id: participant
+
+  twenty_four_mth_saq_2:
+    r_p_id: participant
+    asq24_address_id: address
+    home_phone_id: telephone
+
+  twenty_mth_asq_saq:
+    r_p_id: participant
+    asq20_address_id: address
+    home_phone_id: telephone
+
+  twenty_seven_mth_asq_saq:
+    r_p_id: participant
+    asq27_address_id: address
+    home_phone_id: telephone
+
+  twenty_two_mth_asq_saq:
+    r_p_id: participant
+    asq22_address_id: address
+    home_phone_id: telephone
+
+  validation_ins_2:
+    r_p_id: participant
+    phone_nbr_id: telephone