Skip to content
fyrnheim

Primitives

Primitives are reusable Python functions that replace SQL snippets. Hashing, date operations, categorization — import and compose them instead of copy-pasting SQL.

Usage

Section titled “Usage”
from fyrnheim.primitives import hash_email, date_trunc_month
from fyrnheim import ComputedColumn


ComputedColumn(name="email_hash", expression=hash_email("email"))
ComputedColumn(name="signup_month", expression=date_trunc_month("created_at"))

Primitives generate Ibis expressions, so they compile to the correct SQL for every backend (DuckDB, BigQuery, ClickHouse, etc.).

Available Primitives

Section titled “Available Primitives”

Hashing

Section titled “Hashing”
  • hash_email(column) — SHA-256 hash of a lowered, trimmed email
  • hash_id(column) — SHA-256 hash of a column value
  • hash_md5(column) — MD5 hash of a column value
  • hash_sha256(column) — SHA-256 hash of a column value
  • concat_hash(*columns) — Concatenate columns and hash the result

Dates

Section titled “Dates”
  • date_trunc_month(column) — Truncate a date to the first of the month
  • date_trunc_quarter(column) — Truncate a date to the first of the quarter
  • date_trunc_year(column) — Truncate a date to the first of the year
  • days_since(column) — Number of days between a date column and today
  • date_diff_days(start, end) — Number of days between two date columns
  • extract_year(column) — Extract year from a date
  • extract_month(column) — Extract month from a date
  • extract_day(column) — Extract day from a date
  • earliest_date(*columns) — The earliest (minimum) date across columns
  • latest_date(*columns) — The latest (maximum) date across columns

Categorization

Section titled “Categorization”
  • categorize(column, mapping) — Map column values to categories via a dictionary
  • categorize_contains(column, mapping) — Categorize by substring matching
  • lifecycle_flag(column, states) — Produce a boolean flag based on column value
  • boolean_to_int(column) — Convert a boolean column to 0/1

JSON

Section titled “JSON”
  • to_json_struct(column) — Parse a JSON string column into a struct
  • json_extract_scalar(column, path) — Extract a scalar value from JSON
  • json_value(column, path) — Extract a value from JSON

Aggregations

Section titled “Aggregations”
  • sum_(column) — Sum of a column
  • count_(column) — Count of a column
  • count_distinct(column) — Count of distinct values
  • avg_(column) — Average of a column
  • min_(column) — Minimum value
  • max_(column) — Maximum value
  • row_number_by(partition, order) — Row number within a partition
  • cumulative_sum(column) — Cumulative sum
  • lag_value(column) — Previous row’s value
  • lead_value(column) — Next row’s value
  • first_value(column) — First value in a window
  • last_value(column) — Last value in a window
  • any_value(column) — Any non-null value

Strings

Section titled “Strings”
  • extract_email_domain(column) — Extract the domain from an email address
  • is_personal_email_domain(column) — Check if an email domain is a personal provider
  • account_id_from_domain(column) — Derive an account ID from an email domain

Time

Section titled “Time”
  • parse_iso8601_duration(column) — Parse an ISO 8601 duration string