example_search_in_transcript.py

import lh3.api
from lh3.api import *

import re

content_range_pattern = re.compile(r"chats (\d+)-(\d+)\/(\d+)")

def extract_content_range(content_range):
    matches = content_range_pattern.match(content_range)
    begin = matches.group(1)
    end = matches.group(2)
    total = matches.group(3)
    return (begin, end, total)
    
def search_chats(client, query, chat_range):
    begin, end = chat_range
    _, x_api_version = lh3.api._API.versions.get("v4")
    headers = {
        "Content-Type": "application/json",
        "Range": "chats {begin}-{end}".format(begin=begin, end=end),
        "X-Api-Version": x_api_version,
    }

    request = getattr(client.api().session, "post")
    response = request(
        client.api()._api("v4", "/chat/_search"), headers=headers, json=query
    )
    chats = client.api()._maybe_json(response)
    content_range = extract_content_range(response.headers["Content-Range"])
    return chats, content_range
    
def query_for_search_transcript_with_this_keyword(in_transcript, results=(0, 500)):
    query = {
        "query": {
            "transcript": [in_transcript],
        },
        "sort": [{"started": "descending"}],
    }
    #import pdb; pdb.set_trace()
    client = Client()
    chats, content_range = search_chats(client, query, chat_range=results)
    return chats