Advanced csv import

This is the documentation of the advanced csv import command advanced_csv_import. For the documentation of the less complicated bulk-upload command bulk-add-officers see Using the bulk upload feature. bulk-add-officers accepts one csv containing information about the officer, including badge-number, jobs and salary and makes decisions on whether to update rows in the database or create new entries based on the existing data.

The advanced csv upload is for the most part a way to copy data for one department into the database with as little as possible logic added on. So the tables provided in csv form represent the data that will be inside the sql tables after running the command. (With a few exceptions for many-to-many relationships and auxiliary models like location and license plates)

Before you start

CSV uploads should always be tested locally or in other non-production environments and it is strongly recommended to have the database backed up before running the command. The command is designed to fail early and will only commit the changes if it didn’t encounter any problems. The command however is pretty powerful and can therefore lead to data loss and inconsistencies if the provided csv files are not prepared correctly.

Explanation of the command

/usr/src/app/OpenOversight$ flask advanced-csv-import --help
Usage: flask advanced-csv-import [OPTIONS] DEPARTMENT_NAME

Add or update officers, assignments, salaries, links and incidents from
csv files in the department DEPARTMENT_NAME.

The csv files are treated as the source of truth. Existing entries might
be overwritten as a result, backing up the database and running the
command locally first is highly recommended.

See the documentation before running the command.

Options:
--officers-csv PATH
--assignments-csv PATH
--salaries-csv PATH
--links-csv PATH
--incidents-csv PATH
--force-create          Only for development/testing!
--help                  Show this message and exit.

The command expects one mandatory argument the department name. This is to reduce the chance of making changes to the wrong department by mixing up files. Then there are 5 options to include paths to officers, assignments, salaries, incidents and links csv files. Finally there is a --force-create flag that allows to delete and overwrite existing entries in the database. This is only supposed to be used in non-production environments and to allow replication of the data of another (in most cases production) instance to local environments to test the import command locally first. More details on that flag at the end of the document: Local development flag --force-create.

General overview of the csv import:

The following lists the header fields that each csv can contain. If the csv includes any other fields, the command will fail. However the fields are not case-sensitive and spaces are treated as _. So Officer ID can be used instead of officer_id.

All optional fields can be left blank and will be inserted as NULL or empty string as appropriate. Warning: When updating a record a field that is left blank might overwrite an existing record. This can only be prevented by not including the column in the csv at all.

Formats

  • date: The date should be provided in YYYY-MM-DD format.
  • time: Time should be provided in HH:MM:SS 24h-format in the respective timezone.

The id field

Each csv corresponds to a table in the OpenOversight database. And each csv file has to include id as a field in the table. That field has one primary purpose: If the field is blank, it is assumed that that row is a new entry. If the field contains a number however, it is assumed that a record with that particular id already exists in the database and the record will be updated according to the provided fields. Finally in the case of officers and incidents there is a third option where the id field can contain a string that starts with #. This also indicates a new record, but that new record can be referenced in other provided tables. (for example as the officer_id in the salaries csv)

Officers csv

  • Required: id, department_name
  • Optional: last_name, first_name, middle_initial, suffix, race, gender, employment_date, birth_year, unique_internal_identifier
  • Ignored: badge_number, job_title, most_recent_salary (Unused but command will not fail when field is present)

Details:

  • department_name Name of department exactly as in the server database. This needs to match the department name provided with the command.
  • unique_internal_identifier A string or number that can be used to uniquely identify the officer, in departments in which the badge number stays with the officer using that number is fine. Can and should be left blank if no such number is available.
  • first_name & last_name Will be inserted into the database as is.
  • middle_initial Usually up to one character, but can be more.
  • suffix Choice of Jr, Sr, II, III, IV, V.
  • gender Choice of M, F, Other.
  • race One of BLACK, WHITE, ASIAN, HISPANIC, NATIVE AMERICAN, PACIFIC ISLANDER, Other.
  • employment_date Date representing the start of employment with this department.
  • birth_year Integer representing the birth year of the officer.

Assignments csv

  • Required: id, officer_id, job_title
  • Optional: badge_number, unit_id, start_date, resign_date

Details:

  • officer_id Number referring to id of existing officer or string starting with # referring to a newly created officer in the provided officers csv.
  • badge_number Any string that represents the star or badge number of the officer. In some departments this number changes with the assignment.
  • job_title The job title, needs to be created for that department.
  • unit_id Id of existing unit within the department.
  • start_date Start date of this assignment.
  • resign_date End date of this assignment.

Salaries csv

  • Required: id, officer_id, salary, year
  • Optional: overtime_pay, is_fiscal_year

Details:

  • officer_id Integer referring to id of existing officer or string starting with # referring to a newly created officer in the provided officers csv
  • salary Number representing the officer’s salary in the given year.
  • year Integer, the year this salary information refers to.
  • overtime_pay Number representing the amount of overtime payment for offer in given year.
  • is_fiscal_year Boolean value, indicating whether the provided year refers to calendar year or fiscal year. The values true, t, yes and y are treated as “yes, the salary is for the fiscal year”, all others (including blank) as “no”

Incidents csv

  • Required: id, department_name
  • Optional: date, time, report_number, description, street_name, cross_street1, cross_street2, city, state, zip_code, creator_id, last_updated_id, officer_ids, license_plates

Details:

  • department_name Name of department exactly as in the server database. This needs to match the department name provided with the command.
  • date Date of the incident
  • time Time of the incident
  • report_number String representing any kind of number assigned to complaints or incidents by the police department.
  • description Text description of the incident.
  • street_name Name of the street the incident occurred, but should not include the street number.
  • cross_street1, cross_street2 The two closest intersecting streets.
  • city, state, zip_code State needs to be in 2 letter abbreviated notation.
  • creator_id, last_updated_id Id of existing user shown as responsible for adding this entry.
  • officer_ids Ids of officers involved in the incident, separated by |.
    • Each individual id can either be an integer referring to an existing officer or a string starting with # referring to a newly created officer.
    • Example: 123|#C1|1627 for three officers, one with id 123, one with 1627 and one whose record was created in the officers csv and whose id-field was the string #C1.
  • license_plates all license plates involved in the incident. If there is more than one, they can be separated with a |.
    • Each license plate consists of the license plate number and optionally a state in abbreviated form separated by an underscore _.
    • Example: ABC123_IL|B991 for one license plate with number ABC123 from Illinois and one with number B991 and no associated state.

Examples

Example csvs can be found in the repository under OpenOversight/tests/test_csvs.

Local development flag --force-create

This flag chances the behavior when an integer is provided as id. Instead of updating an existing record, a new record will be created and assigned the given id. If a record with that id already exists in the database, it will be deleted before the new record is created.

This functionality is intended to be used to import csv files downloaded from OpenOversight download page to get a local copy of the production data for one department in the local development database.