Support input type payload for timing strategy

genai-perf/genai_perf/inputs/input_constants.py

@@ -36,6 +36,7 @@ class ModelSelectionStrategy(Enum):
 class PromptSource(Enum):
     SYNTHETIC = auto()
     FILE = auto()
+    PAYLOAD = auto()
 
 
 class OutputFormat(Enum):

genai-perf/genai_perf/parser.py

@@ -351,17 +351,29 @@ def parse_goodput(values):
 
 
 def _infer_prompt_source(args: argparse.Namespace) -> argparse.Namespace:
-
     args.synthetic_input_files = None
+    args.payload_input_file = None
 
     if args.input_file:
-        if str(args.input_file).startswith("synthetic:"):
+        input_file_str = str(args.input_file)
+        if input_file_str.startswith("synthetic:"):
             args.prompt_source = ic.PromptSource.SYNTHETIC
-            synthetic_input_files_str = str(args.input_file).split(":", 1)[1]
+            synthetic_input_files_str = input_file_str.split(":", 1)[1]
             args.synthetic_input_files = synthetic_input_files_str.split(",")
             logger.debug(
                 f"Input source is synthetic data: {args.synthetic_input_files}"
             )
+        elif input_file_str.startswith("payload:"):
+            args.prompt_source = ic.PromptSource.PAYLOAD
+            payload_input_file_str = input_file_str.split(":", 1)[1]
+            if not payload_input_file_str:
+                raise ValueError(
+                    f"Invalid payload input: '{input_file_str}' is missing the file path"
+                )
+            args.payload_input_file = payload_input_file_str.split(",")
+            logger.debug(
+                f"Input source is a payload file with timing information in the following path: {args.payload_input_file}"
+            )
         else:
             args.prompt_source = ic.PromptSource.FILE
             logger.debug(f"Input source is the following path: {args.input_file}")
@@ -384,7 +396,7 @@ def _convert_str_to_enum_entry(args, option, enum):
 
 
 def file_or_directory(value: str) -> Path:
-    if value.startswith("synthetic:"):
+    if value.startswith("synthetic:") or value.startswith("payload:"):
         return Path(value)
     else:
         path = Path(value)
@@ -459,12 +471,17 @@ def _add_input_args(parser):
         required=False,
         help="The input file or directory containing the content to use for "
         "profiling. Each line should be a JSON object with a 'text' or "
-        "'image' field 'in JSONL format. Example: {\"text\":"
-        ' "Your prompt here"}\'. To use synthetic files for a converter that '
+        "'image' field in JSONL format. Example: {\"text\": "
+        '"Your prompt here"}. To use synthetic files for a converter that '
         "needs multiple files, prefix the path with 'synthetic:', followed "
         "by a comma-separated list of filenames. The synthetic filenames "
         "should not have extensions. For example, "
-        "'synthetic:queries,passages'. ",
+        "'synthetic:queries,passages'. For payload data, prefix the path with 'payload:', "
+        "followed by a JSON string representing a payload object. The payload should "
+        "contain fields such as 'timestamp', 'input_length', 'output_length', "
+        "and you can optionally add 'text_input', 'session_id', 'hash_ids', and 'priority'. "
+        'Example: \'payload:{"timestamp": 123.45, "input_length": 10, "output_length": 12, '
+        '"session_id": 1, "priority": 5, "text_input": "Your prompt here"}\'.',
     )
 
     input_group.add_argument(

genai-perf/genai_perf/wrapper.py

@@ -92,6 +92,7 @@ def build_cmd(args: Namespace, extra_args: Optional[List[str]] = None) -> List[s
             "output_tokens_mean",
             "output_tokens_mean_deterministic",
             "output_tokens_stddev",
+            "payload_input_file",
             "prompt_source",
             "random_seed",
             "request_rate",

genai-perf/tests/test_cli.py

@@ -785,25 +785,55 @@ def test_goodput_args_warning(self, monkeypatch, args, expected_error):
         assert str(exc_info.value) == expected_error
 
     @pytest.mark.parametrize(
-        "args, expected_prompt_source",
+        "args, expected_prompt_source, expected_payload_input_file, expect_error",
         [
-            ([], PromptSource.SYNTHETIC),
-            (["--input-file", "prompt.txt"], PromptSource.FILE),
+            ([], PromptSource.SYNTHETIC, None, False),
+            (["--input-file", "prompt.txt"], PromptSource.FILE, None, False),
             (
                 ["--input-file", "prompt.txt", "--synthetic-input-tokens-mean", "10"],
                 PromptSource.FILE,
+                None,
+                False,
+            ),
+            (
+                ["--input-file", "payload:test.jsonl"],
+                PromptSource.PAYLOAD,
+                ["test.jsonl"],
+                False,
             ),
+            (["--input-file", "payload:"], PromptSource.PAYLOAD, [], True),
+            (
+                ["--input-file", "synthetic:test.jsonl"],
+                PromptSource.SYNTHETIC,
+                None,
+                False,
+            ),
+            (["--input-file", "invalidinput"], PromptSource.FILE, None, False),
         ],
     )
     def test_inferred_prompt_source(
-        self, monkeypatch, mocker, args, expected_prompt_source
+        self,
+        monkeypatch,
+        mocker,
+        args,
+        expected_prompt_source,
+        expected_payload_input_file,
+        expect_error,
     ):
         mocker.patch.object(Path, "is_file", return_value=True)
         combined_args = ["genai-perf", "profile", "--model", "test_model"] + args
         monkeypatch.setattr("sys.argv", combined_args)
-        args, _ = parser.parse_args()
 
-        assert args.prompt_source == expected_prompt_source
+        if expect_error:
+            with pytest.raises(ValueError):
+                parser.parse_args()
+        else:
+            args, _ = parser.parse_args()
+
+            assert args.prompt_source == expected_prompt_source
+
+            if expected_payload_input_file is not None:
+                assert args.payload_input_file == expected_payload_input_file
 
     @pytest.mark.parametrize(
         "args",

genai-perf/tests/test_json_exporter.py

@@ -227,6 +227,7 @@ class TestJsonExporter:
           "output_tokens_mean": -1,
           "output_tokens_mean_deterministic": false,
           "output_tokens_stddev": 0,
+          "payload_input_file": null,
           "random_seed": 0,
           "request_count": 0,
           "synthetic_input_files": null,