Rank competitors by leveraging Google's autocomplete function while searching for "<target> vs"

competitive-landscape.py

import argparse
import requests
from tqdm import tqdm
from pprint import pprint
from itertools import islice
from urllib.parse import quote
from functools import lru_cache
from collections import Counter, OrderedDict


@lru_cache
def autocomplete(request):
    response = requests.get(
        f"http://google.com/complete/search?output=toolbar&client=chrome&q={quote(request)}"
    )
    return [item[len(request) + 1:] for item in response.json()[1]
            ], response.json()[4]['google:suggestrelevance']


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('target', type=str)
    parser.add_argument('--top_k', type=int)
    parser.add_argument('--iterations', type=int)
    args = parser.parse_args()
    top_k = args.top_k or 10
    iterations = args.iterations or 10
    items = Counter({args.target: 0})

    for iteration in tqdm(range(iterations)):
        for keyword in items.copy():
            try:
                request = f"{keyword} vs"
                keywords, scores = autocomplete(request)
                new_items = Counter(dict(zip(keywords, scores)))
                items = Counter(dict((items + new_items).most_common(top_k)))
            except:
                continue

    pprint(items)

e.g.
python competitive-landscape.py mlflow

Counter({'airflow': 68241,
         'kubeflow': 41095,
         'mlflow': 36177,
         'sagemaker': 24768,
         'argo': 21076,
         'prefect': 16668,
         'weights and biases': 12518,
         'luigi': 11286,
         'scale and polish': 11277,
         'metaflow': 11259})