|
| 1 | +"""Unit tests configuration module.""" |
| 2 | + |
| 3 | +import json |
| 4 | +import os |
| 5 | +import re |
| 6 | +from typing import Any, Mapping, MutableMapping |
| 7 | + |
| 8 | +import pytest |
| 9 | +import vertexai |
| 10 | +import yaml |
| 11 | +from google.auth.credentials import AnonymousCredentials |
| 12 | +from vcr import VCR |
| 13 | +from vcr.record_mode import RecordMode |
| 14 | +from vcr.request import Request |
| 15 | + |
| 16 | +from opentelemetry.instrumentation.vertexai import VertexAIInstrumentor |
| 17 | +from opentelemetry.instrumentation.vertexai.utils import ( |
| 18 | + OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT, |
| 19 | +) |
| 20 | +from opentelemetry.sdk._events import EventLoggerProvider |
| 21 | +from opentelemetry.sdk._logs import LoggerProvider |
| 22 | +from opentelemetry.sdk._logs.export import ( |
| 23 | + InMemoryLogExporter, |
| 24 | + SimpleLogRecordProcessor, |
| 25 | +) |
| 26 | +from opentelemetry.sdk.metrics import ( |
| 27 | + MeterProvider, |
| 28 | +) |
| 29 | +from opentelemetry.sdk.metrics.export import ( |
| 30 | + InMemoryMetricReader, |
| 31 | +) |
| 32 | +from opentelemetry.sdk.trace import TracerProvider |
| 33 | +from opentelemetry.sdk.trace.export import SimpleSpanProcessor |
| 34 | +from opentelemetry.sdk.trace.export.in_memory_span_exporter import ( |
| 35 | + InMemorySpanExporter, |
| 36 | +) |
| 37 | + |
| 38 | +FAKE_PROJECT = "fake-project" |
| 39 | + |
| 40 | + |
| 41 | +@pytest.fixture(scope="function", name="span_exporter") |
| 42 | +def fixture_span_exporter(): |
| 43 | + exporter = InMemorySpanExporter() |
| 44 | + yield exporter |
| 45 | + |
| 46 | + |
| 47 | +@pytest.fixture(scope="function", name="log_exporter") |
| 48 | +def fixture_log_exporter(): |
| 49 | + exporter = InMemoryLogExporter() |
| 50 | + yield exporter |
| 51 | + |
| 52 | + |
| 53 | +@pytest.fixture(scope="function", name="metric_reader") |
| 54 | +def fixture_metric_reader(): |
| 55 | + exporter = InMemoryMetricReader() |
| 56 | + yield exporter |
| 57 | + |
| 58 | + |
| 59 | +@pytest.fixture(scope="function", name="tracer_provider") |
| 60 | +def fixture_tracer_provider(span_exporter): |
| 61 | + provider = TracerProvider() |
| 62 | + provider.add_span_processor(SimpleSpanProcessor(span_exporter)) |
| 63 | + return provider |
| 64 | + |
| 65 | + |
| 66 | +@pytest.fixture(scope="function", name="event_logger_provider") |
| 67 | +def fixture_event_logger_provider(log_exporter): |
| 68 | + provider = LoggerProvider() |
| 69 | + provider.add_log_record_processor(SimpleLogRecordProcessor(log_exporter)) |
| 70 | + event_logger_provider = EventLoggerProvider(provider) |
| 71 | + |
| 72 | + return event_logger_provider |
| 73 | + |
| 74 | + |
| 75 | +@pytest.fixture(scope="function", name="meter_provider") |
| 76 | +def fixture_meter_provider(metric_reader): |
| 77 | + return MeterProvider( |
| 78 | + metric_readers=[metric_reader], |
| 79 | + ) |
| 80 | + |
| 81 | + |
| 82 | +@pytest.fixture(autouse=True) |
| 83 | +def vertexai_init(vcr: VCR) -> None: |
| 84 | + # When not recording (in CI), don't do any auth. That prevents trying to read application |
| 85 | + # default credentials from the filesystem or metadata server and oauth token exchange. This |
| 86 | + # is not the interesting part of our instrumentation to test. |
| 87 | + credentials = None |
| 88 | + project = None |
| 89 | + if vcr.record_mode == RecordMode.NONE: |
| 90 | + credentials = AnonymousCredentials() |
| 91 | + project = FAKE_PROJECT |
| 92 | + vertexai.init( |
| 93 | + api_transport="rest", credentials=credentials, project=project |
| 94 | + ) |
| 95 | + |
| 96 | + |
| 97 | +@pytest.fixture |
| 98 | +def instrument_no_content( |
| 99 | + tracer_provider, event_logger_provider, meter_provider |
| 100 | +): |
| 101 | + os.environ.update( |
| 102 | + {OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT: "False"} |
| 103 | + ) |
| 104 | + |
| 105 | + instrumentor = VertexAIInstrumentor() |
| 106 | + instrumentor.instrument( |
| 107 | + tracer_provider=tracer_provider, |
| 108 | + event_logger_provider=event_logger_provider, |
| 109 | + meter_provider=meter_provider, |
| 110 | + ) |
| 111 | + |
| 112 | + yield instrumentor |
| 113 | + os.environ.pop(OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT, None) |
| 114 | + instrumentor.uninstrument() |
| 115 | + |
| 116 | + |
| 117 | +@pytest.fixture |
| 118 | +def instrument_with_content( |
| 119 | + tracer_provider, event_logger_provider, meter_provider |
| 120 | +): |
| 121 | + os.environ.update( |
| 122 | + {OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT: "True"} |
| 123 | + ) |
| 124 | + instrumentor = VertexAIInstrumentor() |
| 125 | + instrumentor.instrument( |
| 126 | + tracer_provider=tracer_provider, |
| 127 | + event_logger_provider=event_logger_provider, |
| 128 | + meter_provider=meter_provider, |
| 129 | + ) |
| 130 | + |
| 131 | + yield instrumentor |
| 132 | + os.environ.pop(OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT, None) |
| 133 | + instrumentor.uninstrument() |
| 134 | + |
| 135 | + |
| 136 | +@pytest.fixture(scope="module") |
| 137 | +def vcr_config(): |
| 138 | + filter_header_regexes = [ |
| 139 | + r"X-.*", |
| 140 | + "Server", |
| 141 | + "Date", |
| 142 | + "Expires", |
| 143 | + "Authorization", |
| 144 | + ] |
| 145 | + |
| 146 | + def filter_headers(headers: Mapping[str, str]) -> Mapping[str, str]: |
| 147 | + return { |
| 148 | + key: val |
| 149 | + for key, val in headers.items() |
| 150 | + if not any( |
| 151 | + re.match(filter_re, key, re.IGNORECASE) |
| 152 | + for filter_re in filter_header_regexes |
| 153 | + ) |
| 154 | + } |
| 155 | + |
| 156 | + def before_record_cb(request: Request): |
| 157 | + request.headers = filter_headers(request.headers) |
| 158 | + request.uri = re.sub( |
| 159 | + r"/projects/[^/]+/", "/projects/fake-project/", request.uri |
| 160 | + ) |
| 161 | + return request |
| 162 | + |
| 163 | + def before_response_cb(response: MutableMapping[str, Any]): |
| 164 | + response["headers"] = filter_headers(response["headers"]) |
| 165 | + return response |
| 166 | + |
| 167 | + return { |
| 168 | + "decode_compressed_response": True, |
| 169 | + "before_record_request": before_record_cb, |
| 170 | + "before_record_response": before_response_cb, |
| 171 | + "ignore_hosts": ["oauth2.googleapis.com"], |
| 172 | + } |
| 173 | + |
| 174 | + |
| 175 | +class LiteralBlockScalar(str): |
| 176 | + """Formats the string as a literal block scalar, preserving whitespace and |
| 177 | + without interpreting escape characters""" |
| 178 | + |
| 179 | + |
| 180 | +def literal_block_scalar_presenter(dumper, data): |
| 181 | + """Represents a scalar string as a literal block, via '|' syntax""" |
| 182 | + return dumper.represent_scalar("tag:yaml.org,2002:str", data, style="|") |
| 183 | + |
| 184 | + |
| 185 | +yaml.add_representer(LiteralBlockScalar, literal_block_scalar_presenter) |
| 186 | + |
| 187 | + |
| 188 | +def process_string_value(string_value): |
| 189 | + """Pretty-prints JSON or returns long strings as a LiteralBlockScalar""" |
| 190 | + try: |
| 191 | + json_data = json.loads(string_value) |
| 192 | + return LiteralBlockScalar(json.dumps(json_data, indent=2)) |
| 193 | + except (ValueError, TypeError): |
| 194 | + if len(string_value) > 80: |
| 195 | + return LiteralBlockScalar(string_value) |
| 196 | + return string_value |
| 197 | + |
| 198 | + |
| 199 | +def convert_body_to_literal(data): |
| 200 | + """Searches the data for body strings, attempting to pretty-print JSON""" |
| 201 | + if isinstance(data, dict): |
| 202 | + for key, value in data.items(): |
| 203 | + # Handle response body case (e.g., response.body.string) |
| 204 | + if key == "body" and isinstance(value, dict) and "string" in value: |
| 205 | + value["string"] = process_string_value(value["string"]) |
| 206 | + |
| 207 | + # Handle request body case (e.g., request.body) |
| 208 | + elif key == "body" and isinstance(value, str): |
| 209 | + data[key] = process_string_value(value) |
| 210 | + |
| 211 | + else: |
| 212 | + convert_body_to_literal(value) |
| 213 | + |
| 214 | + elif isinstance(data, list): |
| 215 | + for idx, choice in enumerate(data): |
| 216 | + data[idx] = convert_body_to_literal(choice) |
| 217 | + |
| 218 | + return data |
| 219 | + |
| 220 | + |
| 221 | +class PrettyPrintJSONBody: |
| 222 | + """This makes request and response body recordings more readable.""" |
| 223 | + |
| 224 | + @staticmethod |
| 225 | + def serialize(cassette_dict): |
| 226 | + cassette_dict = convert_body_to_literal(cassette_dict) |
| 227 | + return yaml.dump( |
| 228 | + cassette_dict, default_flow_style=False, allow_unicode=True |
| 229 | + ) |
| 230 | + |
| 231 | + @staticmethod |
| 232 | + def deserialize(cassette_string): |
| 233 | + return yaml.load(cassette_string, Loader=yaml.Loader) |
| 234 | + |
| 235 | + |
| 236 | +@pytest.fixture(scope="module", autouse=True) |
| 237 | +def fixture_vcr(vcr): |
| 238 | + vcr.register_serializer("yaml", PrettyPrintJSONBody) |
| 239 | + return vcr |
0 commit comments