Fix pipeline config and agent tools hashing for telemetry (#4508)

haystack/agents/base.py

@@ -191,7 +191,8 @@ class Agent:
         See haystack/telemetry.py::send_event
         """
         try:
-            self.hash = md5(" ".join([tool.pipeline_or_node.__class__.__name__ for tool in self.tools.values()]))
+            tool_names = " ".join([tool.pipeline_or_node.__class__.__name__ for tool in self.tools.values()])
+            self.hash = md5(tool_names.encode()).hexdigest()
         except Exception as exc:
             logger.debug("Telemetry exception: %s", str(exc))
             self.hash = "[an exception occurred during hashing]"

haystack/pipelines/base.py

@@ -438,7 +438,7 @@ class Pipeline:
             for comp in config_to_hash["components"]:
                 del comp["name"]
             config_hash = json.dumps(config_to_hash, default=str)
-            self.config_hash = md5(config_hash)
+            self.config_hash = md5(config_hash.encode()).hexdigest()
         except Exception as exc:
             logger.debug("Telemetry exception: %s", str(exc))
             self.config_hash = "[an exception occurred during hashing]"

test/agents/test_agent.py

@@ -3,6 +3,8 @@ import os
 import re
 from typing import Tuple
 
+from test.conftest import MockRetriever, MockPromptNode
+from unittest import mock
 import pytest
 
 from haystack import BaseComponent, Answer
@@ -11,7 +13,6 @@ from haystack.agents.base import Tool
 from haystack.errors import AgentError
 from haystack.nodes import PromptModel, PromptNode, PromptTemplate
 from haystack.pipelines import ExtractiveQAPipeline, DocumentSearchPipeline, BaseStandardPipeline
-from test.conftest import MockRetriever, MockPromptNode
 
 
 @pytest.mark.unit
@@ -335,3 +336,30 @@ def test_agent_run_batch(reader, retriever_with_docs, document_store_with_docs):
     # TODO Replace Count tool once more tools are implemented so that we do not need to account for off-by-one errors
     assert any(digit in results["answers"][0][0].answer for digit in ["5", "6", "five", "six"])
     assert any(digit in results["answers"][1][0].answer for digit in ["5", "6", "five", "six"])
+
+
+@pytest.mark.unit
+def test_update_hash():
+    agent = Agent(prompt_node=mock.Mock(), prompt_template=mock.Mock())
+    assert agent.hash == "d41d8cd98f00b204e9800998ecf8427e"
+    agent.add_tool(
+        Tool(
+            name="Search",
+            pipeline_or_node=mock.Mock(),
+            description="useful for when you need to answer "
+            "questions about where people live. You "
+            "should ask targeted questions",
+            output_variable="answers",
+        )
+    )
+    assert agent.hash == "d41d8cd98f00b204e9800998ecf8427e"
+    agent.add_tool(
+        Tool(
+            name="Count",
+            pipeline_or_node=mock.Mock(),
+            description="useful for when you need to count how many characters are in a word. Ask only with a single word.",
+        )
+    )
+    assert agent.hash == "d41d8cd98f00b204e9800998ecf8427e"
+    agent.update_hash()
+    assert agent.hash == "5ac8eca2f92c9545adcce3682b80d4c5"

test/pipelines/test_pipeline.py

@@ -2,7 +2,6 @@ import ssl
 import json
 import platform
 import sys
-import datetime
 from typing import Tuple
 from copy import deepcopy
 from unittest import mock
@@ -2048,3 +2047,31 @@ def test_fix_to_pipeline_execution_when_join_follows_join():
     res = pipeline.run(query="Alpha Beta Gamma Delta")
     documents = res["documents"]
     assert len(documents) == 4  # all four documents should be found
+
+
+@pytest.mark.unit
+def test_update_config_hash():
+    fake_configs = {
+        "version": "ignore",
+        "components": [
+            {
+                "name": "MyReader",
+                "type": "FARMReader",
+                "params": {"no_ans_boost": -10, "model_name_or_path": "deepset/roberta-base-squad2"},
+            }
+        ],
+        "pipelines": [
+            {
+                "name": "my_query_pipeline",
+                "nodes": [
+                    {"name": "MyRetriever", "inputs": ["Query"]},
+                    {"name": "MyReader", "inputs": ["MyRetriever"]},
+                ],
+            }
+        ],
+    }
+    with mock.patch("haystack.pipelines.base.Pipeline.get_config", return_value=fake_configs):
+        test_pipeline = Pipeline()
+        assert test_pipeline.config_hash == None
+        test_pipeline.update_config_hash()
+        assert test_pipeline.config_hash == "a30d3273de0d70e63e8cd91d915255b3"