Method findMatchingClusterForTokens

pkg/drain/drain.go:135–179

(tokens []string)

}

func (d *Drain) findMatchingClusterForTokens(tokens []string) *LogCluster {
	tokenCount := len(tokens)

	// at first level, children are grouped by token (word) count
	curNode, ok := d.rootNode.keyToChildNode[strconv.Itoa(tokenCount)]
	if !ok {
		// no template with same token count yet
		return nil
	}

	// we always end the token list with an <END> token, so <2 tokens is a
	// special case for an empty input string. In this case, we return the
	// single cluster in that group.
	if tokenCount < 2 {
		return d.idToCluster.Get(curNode.clusterIDs[0])
	}

	// otherwise, we need to find the leaf node for this log.
	curNodeDepth := 1
	for _, token := range tokens {
		// at max depth
		if curNodeDepth >= d.maxNodeDepth {
			break
		}

		// this is last token
		if curNodeDepth == tokenCount {
			break
		}

		keyToChildNode := curNode.keyToChildNode
		curNode, ok = keyToChildNode[d.config.ParamString]
		if !ok { // no wildcard node, try exact match
			curNode, ok = keyToChildNode[token]
		}
		if !ok { // no existing path
			return nil
		}
		curNodeDepth++
	}

	// get best match among all clusters with same prefix, or None if no match is above sim_th
	cluster := d.findBestClusterForTokens(curNode.clusterIDs, tokens)
	return cluster
}

// findBestClusterForTokens finds the best match for a log message (represented
// as tokens) versus a list of clusters. A Match is considered better if it has