Greptimedb script for Ramer–Douglas–Peucker algorithm

rdp.py

import math


@coprocessor(
    returns=["lon", "lat"],
)
def simplify(**params) -> (vector[f64], vector[f64]):
    from greptime import query
    results = query().sql(params["sql"])
    lon = results[0]
    lat = results[1]
    e = float(params["e"])
    line = list(zip(lon, lat))
    simplified_line = rdp(line, e)
    result = list(zip(*simplified_line))
    return list(result[0]), list(result[1])


def rdp(line, epsilon):
    start_idx = 0
    end_idx = len(line) - 1
    max_dist = 0  # var to store furthest point dist
    max_id = 0  # var to store furthest point index

    for i in range(start_idx + 1, end_idx):
        d = perpendicular_distance(line[i], line[start_idx], line[end_idx])
        if d > max_dist:
            max_dist = d  # overwrite max distance
            max_id = i  # overwrite max index

    if max_dist > epsilon:
        l = rdp(line[start_idx : max_id + 1], epsilon)
        r = rdp(line[max_id:], epsilon)

        results = l[:-1]
        results.extend(r)
        return results
    else:
        return [line[0], line[end_idx]]


def perpendicular_distance(point, start, end):
    m = (end[1] - start[1]) / (end[0] - start[0])
    b = start[1] - m * start[0]  # plug in either start or end
    a = m
    b = -1
    c = b
    d = abs(a * point[0] + b * point[1] + c) / math.sqrt(
        a**2 + b**2
    )  # formula line - point distance

    return d

https://en.wikipedia.org/wiki/Ramer%E2%80%93Douglas%E2%80%93Peucker_algorithm

1. Save the script as rdp.py
2. Submit to greptimedb: curl --data-binary "@rdp.py" -XPOST "http://localhost:4000/v1/scripts?name=rdp&db=public"
3. Call the script: curl -v -XPOST "http://localhost:4000/v1/run-script?db=public&name=rdp&sql=select%20lon%2Clat%20from%20waypoints%3B&e=4", arguments explained:

• sql: a sql statement that will return two columns lon and lat
• e: the epsilon in RDP algorithm