langfuse.Sampler

 1import hashlib
 2import logging
 3
 4
 5log = logging.getLogger("langfuse")
 6
 7
 8class Sampler:
 9    sample_rate: float
10
11    def __init__(self, sample_rate: float):
12        self.sample_rate = sample_rate
13
14    def sample_event(self, event: dict):
15        # need to get trace_id from a given event
16        # returns true if
17
18        if "type" in event and "body" in event:
19            event_type = event["type"]
20
21            if event_type == "sdk-log":
22                return True
23
24            trace_id = None
25
26            if event_type == "trace-create" and "id" in event["body"]:
27                trace_id = event["body"]["id"]
28            elif "trace_id" in event["body"]:
29                trace_id = event["body"]["trace_id"]
30            elif "traceId" in event["body"]:
31                trace_id = event["body"]["traceId"]
32            else:
33                log.error("Unexpected event format: No trace id found in event")
34                return True
35
36            return self.deterministic_sample(trace_id, self.sample_rate)
37
38        else:
39            log.error("Unexpected event format: No trace id found in event")
40            return True
41
42    def deterministic_sample(self, trace_id: str, sample_rate: float):
43        """Determins if an event should be sampled based on the trace_id and sample_rate. Event will be sent to server if True"""
44        log.debug(
45            f"Applying deterministic sampling to trace_id: {trace_id} with rate {sample_rate}"
46        )
47
48        # Use SHA-256 to hash the trace_id
49        hash_object = hashlib.sha256(trace_id.encode())
50        # Get the hexadecimal representation of the hash
51        hash_hex = hash_object.hexdigest()
52
53        # Take the first 8 characters of the hex digest and convert to integer
54        hash_int = int(hash_hex[:8], 16)
55
56        # Normalize the integer to a float in the range [0, 1)
57        normalized_hash = hash_int / 0xFFFFFFFF
58
59        result = normalized_hash < sample_rate
60
61        if not result:
62            log.debug(
63                f"event with trace_id: {trace_id} and rate {sample_rate} was sampled and not sent to the server"
64            )
65
66        return result