redisbench-admin 0.11.72__tar.gz → 0.11.73__tar.gz

{redisbench_admin-0.11.72 → redisbench_admin-0.11.73}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: redisbench-admin
-Version: 0.11.72
+Version: 0.11.73
 Summary: Redis benchmark run helper. A wrapper around Redis and Redis Modules benchmark tools ( ftsb_redisearch, memtier_benchmark, redis-benchmark, aibench, etc... ).
 License-File: LICENSE
 Author: filipecosta90

{redisbench_admin-0.11.72 → redisbench_admin-0.11.73}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "redisbench-admin"
-version = "0.11.72"
+version = "0.11.73"
 description = "Redis benchmark run helper. A wrapper around Redis and Redis Modules benchmark tools ( ftsb_redisearch, memtier_benchmark, redis-benchmark, aibench, etc... )."
 authors = ["filipecosta90 <filipecosta.90@gmail.com>","Redis Performance Group <performance@redis.com>"]
 readme = "README.md"

{redisbench_admin-0.11.72 → redisbench_admin-0.11.73}/redisbench_admin/run/common.py

@@ -38,6 +38,7 @@ from redisbench_admin.run.ycsb.ycsb import (
     prepare_ycsb_benchmark_command,
     prepare_go_ycsb_benchmark_command,
 )
+from redisbench_admin.utils.utils import get_remote_input_file_from_url
 from redisbench_admin.run_remote.args import OVERRIDE_MODULES
 from redisbench_admin.run_remote.remote_helpers import (
     extract_module_semver_from_info_modules_cmd,
@@ -66,6 +67,38 @@ PERFORMANCE_GH_TOKEN = os.getenv("PERFORMANCE_GH_TOKEN", None)
 REDIS_BINARY = os.getenv("REDIS_BINARY", "redis-server")
 
 
+def extract_input_file_url_from_parameters(entry, benchmark_tool):
+    """
+    Extract the input file URL from the parameters entry.
+    Different tools use different parameter names:
+    - ftsb_*: uses "input" parameter
+    - tsbs_*, aibench_*: uses "file" parameter
+
+    Args:
+        entry: The benchmark config entry containing "parameters"
+        benchmark_tool: The benchmark tool name
+
+    Returns:
+        The file URL if found, None otherwise
+    """
+    if "parameters" not in entry:
+        return None
+
+    # Determine which parameter name to look for based on the tool
+    if "ftsb_" in benchmark_tool:
+        param_name = "input"
+    elif "tsbs_" in benchmark_tool or "aibench_" in benchmark_tool:
+        param_name = "file"
+    else:
+        return None
+
+    for param in entry["parameters"]:
+        if param_name in param:
+            return param[param_name]
+
+    return None
+
+
 def prepare_benchmark_parameters(
     benchmark_config,
     benchmark_tool,
@@ -88,6 +121,10 @@ def prepare_benchmark_parameters(
     if type(benchmark_config[config_key]) == list:
         for entry in benchmark_config[config_key]:
             if "parameters" in entry:
+                # Extract input file URL from parameters
+                input_file_url = extract_input_file_url_from_parameters(
+                    entry, benchmark_tool
+                )
                 command_arr, command_str = prepare_benchmark_parameters_specif_tooling(
                     benchmark_tool,
                     cluster_api_enabled,
@@ -104,10 +141,13 @@ def prepare_benchmark_parameters(
                     private_key,
                     client_ssh_port,
                     redis_password,
+                    input_file_url,
                 )
     # v0.4 spec
     elif type(benchmark_config[config_key]) == dict:
         entry = benchmark_config[config_key]
+        # Extract input file URL from parameters
+        input_file_url = extract_input_file_url_from_parameters(entry, benchmark_tool)
         command_arr, command_str = prepare_benchmark_parameters_specif_tooling(
             benchmark_tool,
             cluster_api_enabled,
@@ -124,6 +164,7 @@ def prepare_benchmark_parameters(
             private_key,
             client_ssh_port,
             redis_password,
+            input_file_url,
         )
     printed_command_str = command_str
     printed_command_arr = command_arr
@@ -154,6 +195,7 @@ def prepare_benchmark_parameters_specif_tooling(
     private_key,
     client_ssh_port,
     redis_password=None,
+    input_file_url=None,
 ):
     if "redis-benchmark" in benchmark_tool:
         command_arr, command_str = prepare_redis_benchmark_command(
@@ -207,7 +249,7 @@ def prepare_benchmark_parameters_specif_tooling(
         input_data_file = None
         if isremote is True:
             benchmark_tool = "/tmp/{}".format(benchmark_tool)
-            input_data_file = "/tmp/input.data"
+            input_data_file = get_remote_input_file_from_url(input_file_url)
         (
             command_arr,
             command_str,
@@ -260,7 +302,7 @@ def prepare_benchmark_parameters_specif_tooling(
         input_data_file = None
         if isremote is True:
             benchmark_tool = "/tmp/{}".format(benchmark_tool)
-            input_data_file = "/tmp/input.data"
+            input_data_file = get_remote_input_file_from_url(input_file_url)
         (
             command_arr,
             command_str,
@@ -280,7 +322,7 @@ def prepare_benchmark_parameters_specif_tooling(
         input_data_file = None
         if isremote is True:
             benchmark_tool = "/tmp/{}".format(benchmark_tool)
-            input_data_file = "/tmp/input.data"
+            input_data_file = get_remote_input_file_from_url(input_file_url)
         (
             command_arr,
             command_str,

{redisbench_admin-0.11.72 → redisbench_admin-0.11.73}/redisbench_admin/run_remote/remote_helpers.py

@@ -28,6 +28,7 @@ from redisbench_admin.utils.remote import (
     extract_redisgraph_version_from_resultdict,
 )
 from redisbench_admin.utils.results import post_process_benchmark_results
+from redisbench_admin.utils.utils import get_remote_input_file_from_url
 
 
 def absoluteFilePaths(directory):
@@ -166,7 +167,7 @@ def remote_tool_pre_bench_step(
                 queries_file_link, remote_tool_link, tool_link
             )
         )
-        remote_input_file = "/tmp/input.data"
+        remote_input_file = get_remote_input_file_from_url(queries_file_link)
         setup_remote_benchmark_tool_requirements_ftsb(
             client_public_ip,
             username,
@@ -191,7 +192,7 @@ def remote_tool_pre_bench_step(
             "redisearch",
             "go-ycsb",
         )
-        remote_input_file = "/tmp/input.data"
+        remote_input_file = get_remote_input_file_from_url(queries_file_link)
         setup_remote_benchmark_tool_requirements_tsbs(
             client_public_ip,
             username,
@@ -211,7 +212,7 @@ def remote_tool_pre_bench_step(
         ) = extract_remote_tool_extra_links(
             benchmark_config, benchmark_tool, config_key, os_str, arch_str
         )
-        remote_input_file = "/tmp/input.data"
+        remote_input_file = get_remote_input_file_from_url(queries_file_link)
         setup_remote_benchmark_tool_requirements_tsbs(
             client_public_ip,
             username,
@@ -228,7 +229,7 @@ def remote_tool_pre_bench_step(
             remote_tool_link,
             tool_link,
         ) = extract_aibench_extra_links(benchmark_config, benchmark_tool)
-        remote_input_file = "/tmp/input.data"
+        remote_input_file = get_remote_input_file_from_url(queries_file_link)
         setup_remote_benchmark_tool_requirements_tsbs(
             client_public_ip,
             username,

{redisbench_admin-0.11.72 → redisbench_admin-0.11.73}/redisbench_admin/utils/utils.py

@@ -11,10 +11,12 @@ import logging
 import operator
 import os
 import os.path
+import re
 import tarfile
 import time
 from functools import reduce
-from urllib.parse import quote_plus
+from pathlib import PurePosixPath
+from urllib.parse import quote_plus, urlparse, unquote
 from zipfile import ZipFile
 
 import boto3
@@ -484,6 +486,60 @@ def make_dashboard_callback(
     )
 
 
+def get_remote_input_file_from_url(url):
+    """
+    Generate a unique remote input file path based on the file name from the URL.
+    This ensures each input file has its own path and avoids reuse issues between benchmark runs.
+
+    Example:
+        url = "https://s3.amazonaws.com/.../AND_QUERY.CSV"
+        returns "/tmp/input-AND_QUERY.data"
+
+    Args:
+        url: The URL or path to the input file
+
+    Returns:
+        A unique remote file path like /tmp/input-{file_name}.data,
+        or "/tmp/input.data" as fallback if url is None or empty
+    """
+    default_path = PurePosixPath("/tmp/input.data")
+
+    if url is None or url == "":
+        return str(default_path)
+
+    # Parse the URL to extract the path component
+    # This handles both http(s):// URLs and s3:// URIs
+    parsed = urlparse(url)
+    url_path = parsed.path if parsed.path else url
+
+    # Use PurePosixPath to extract the file name (works for both URLs and paths)
+    path = PurePosixPath(url_path)
+    file_name = path.name
+
+    # If no file name could be extracted, return default
+    if not file_name:
+        return str(default_path)
+
+    # Decode URL-encoded characters (e.g., %3A -> :)
+    file_name = unquote(file_name)
+
+    # Remove extension using pathlib's stem property
+    base_name = PurePosixPath(file_name).stem
+
+    # If stem is empty (e.g., hidden file like ".data"), use the full name
+    if not base_name:
+        base_name = file_name
+
+    # Sanitize the base name: replace characters that might cause issues in file paths
+    # Keep only alphanumeric, underscore, hyphen, and dot
+    base_name = re.sub(r"[^a-zA-Z0-9_\-.]", "_", base_name)
+
+    # Construct the final path using PurePosixPath for proper path joining
+    result_path = PurePosixPath("/tmp") / f"input-{base_name}.data"
+
+    return str(result_path)
+
+
 EC2_REGION = os.getenv("AWS_DEFAULT_REGION", None)
 EC2_SECRET_KEY = os.getenv("AWS_SECRET_ACCESS_KEY", None)
 EC2_PRIVATE_PEM = os.getenv("EC2_PRIVATE_PEM", None)