|
| 1 | +""" |
| 2 | + This script is used to automatically add new vectorstore classes to the newrelic-python-agent. |
| 3 | + To run this script, start from the root of the newrelic-python-agent repository and run: |
| 4 | + `python tests/mlmodel_langchain/new_vectorstore_adder.py` |
| 5 | + This will generate the necessary code to instrument the new vectorstore classes in the local |
| 6 | + copy of the newrelic-python-agent repository. |
| 7 | +""" |
| 8 | + |
| 9 | +import os |
| 10 | + |
| 11 | +from langchain_community import vectorstores |
| 12 | + |
| 13 | +from newrelic.hooks.mlmodel_langchain import VECTORSTORE_CLASSES |
| 14 | + |
| 15 | +dir_path = os.path.dirname(os.path.realpath(__file__)) |
| 16 | +test_dir = os.path.abspath(os.path.join(dir_path, os.pardir)) |
| 17 | +REPO_PATH = os.path.abspath(os.path.join(test_dir, os.pardir)) |
| 18 | + |
| 19 | + |
| 20 | +def add_to_config(directory, instrumented_class=None): |
| 21 | + # Only implement this if there is not an instrumented class within the directory already. |
| 22 | + if instrumented_class: |
| 23 | + return |
| 24 | + |
| 25 | + with open(f"{REPO_PATH}/newrelic/config.py", "r+") as file: |
| 26 | + text = file.read() |
| 27 | + text = text.replace( |
| 28 | + "VectorStores with similarity_search method", |
| 29 | + "VectorStores with similarity_search method\n " |
| 30 | + + "_process_module_definition(\n " |
| 31 | + + f'"{directory}",\n ' |
| 32 | + + '"newrelic.hooks.mlmodel_langchain",\n ' |
| 33 | + + '"instrument_langchain_vectorstore_similarity_search",\n ' |
| 34 | + + ")\n", |
| 35 | + 1, |
| 36 | + ) |
| 37 | + file.seek(0) |
| 38 | + file.write(text) |
| 39 | + |
| 40 | + |
| 41 | +def add_to_hooks(class_name, directory, instrumented_class=None): |
| 42 | + with open(f"{REPO_PATH}/newrelic/hooks/mlmodel_langchain.py", "r+") as file: |
| 43 | + text = file.read() |
| 44 | + |
| 45 | + # The directory does not exist yet. Add the new directory and class name to the beginning of the dictionary |
| 46 | + if not instrumented_class: |
| 47 | + text = text.replace( |
| 48 | + "VECTORSTORE_CLASSES = {", "VECTORSTORE_CLASSES = {\n " + f'"{directory}": "{class_name}",', 1 |
| 49 | + ) |
| 50 | + |
| 51 | + # The directory exists, and there are multiple instrumented classes in it. Append to the list. |
| 52 | + elif isinstance(instrumented_class, list): |
| 53 | + original_list = str(instrumented_class).replace("'", '"') |
| 54 | + instrumented_class.append(class_name) |
| 55 | + instrumented_class = str(instrumented_class).replace("'", '"') |
| 56 | + text = text.replace( |
| 57 | + f'"{directory}": {original_list}', f'"{directory}": {instrumented_class}' # TODO: NOT WORKING |
| 58 | + ) |
| 59 | + |
| 60 | + # The directory exists, but it only has one class. We need to convert this to a list. |
| 61 | + else: |
| 62 | + text = text.replace(f'"{instrumented_class}"', f'["{instrumented_class}", "{class_name}"]', 1) |
| 63 | + |
| 64 | + file.seek(0) |
| 65 | + file.write(text) |
| 66 | + |
| 67 | + |
| 68 | +def main(): |
| 69 | + _test_vectorstore_modules_instrumented_ignored_classes = set( |
| 70 | + [ |
| 71 | + "VectorStore", # Base class |
| 72 | + "Zilliz", # Inherited from Milvus, which we are already instrumenting. |
| 73 | + ] |
| 74 | + ) |
| 75 | + |
| 76 | + vector_store_class_directory = vectorstores._module_lookup |
| 77 | + for class_name, directory in vector_store_class_directory.items(): |
| 78 | + class_ = getattr(vectorstores, class_name) |
| 79 | + instrumented_class = VECTORSTORE_CLASSES.get(directory, None) |
| 80 | + |
| 81 | + if ( |
| 82 | + not hasattr(class_, "similarity_search") |
| 83 | + or class_name in _test_vectorstore_modules_instrumented_ignored_classes |
| 84 | + ): |
| 85 | + continue |
| 86 | + |
| 87 | + if not instrumented_class or class_name not in instrumented_class: |
| 88 | + if class_name in vector_store_class_directory: |
| 89 | + uninstrumented_directory = vector_store_class_directory[class_name] |
| 90 | + |
| 91 | + # Add in newrelic/config.py if there is not an instrumented directory |
| 92 | + # Otherwise, config already exists, so no need to duplicate it. |
| 93 | + add_to_config(uninstrumented_directory, instrumented_class) |
| 94 | + |
| 95 | + # Add in newrelic/hooks/mlmodel_langchain.py |
| 96 | + add_to_hooks(class_name, uninstrumented_directory, instrumented_class) |
| 97 | + |
| 98 | + |
| 99 | +if __name__ == "__main__": |
| 100 | + main() |
![mergify[bot]](https://avatars.githubusercontent.com/in/10562?v=4&size=40){kind=avatar}
0 commit comments