final

Signed-off-by: Yang Wang <[email protected]>

Author: yangw-dev

.ci/scripts/benchmark_tooling/README.md

@@ -13,9 +13,9 @@ Install dependencies:
 pip install -r requirements.txt
 ```
 
-Run with default output (CLI):
+Run with csv output (CLI):
 ```bash
-python3 .ci/scripts/benchmark_tooling/get_benchmark_analysis_data.py --startTime "2025-06-11T00:00:00" --endTime "2025-06-17T18:00:00"
+python3 .ci/scripts/benchmark_tooling/get_benchmark_analysis_data.py --startTime "2025-06-11T00:00:00" --endTime "2025-06-17T18:00:00" --outputType "csv"
 ```
 
 Additional options:
@@ -24,6 +24,23 @@ Additional options:
 - `--outputType excel --outputDir "{YOUR_LOCAL_DIRECTORY}"`: Generate Excel file with multiple sheets (`res_private.xlsx` and `res_public.xlsx`)
 - `--outputType csv --outputDir "{YOUR_LOCAL_DIRECTORY}"`: Generate CSV files in folders (`private` and `public`)
 
+you can then call methods in common.py to convert the file date back to df version
+```python3
+import logging
+logging.basicConfig(level=logging.INFO)
+from common.py import
+
+# assume the folder private for csv is in cunrrent directory
+folder_path = './private'
+res = read_all_csv_with_metadata(folder_path)
+logging.info(res)
+
+# assume the excel file for private device is in cunrrent directory
+folder_path = "./private.xlsx"
+res = read_excel_with_json_header(folder_path)
+logging.info(res)
+```
+
 ### Python API Usage
 
 To use the benchmark fetcher in your own scripts:
@@ -33,7 +50,7 @@ import ExecutorchBenchmarkFetcher from benchmark_tooling.get_benchmark_analysis_
 fetcher = ExecutorchBenchmarkFetcher()
 # Must call run first
 fetcher.run()
-private, public = fetcher.to_df()
+res = fetcher.
 ```
 
 ## analyze_benchmark_stability.py
@@ -50,7 +67,6 @@ python .ci/scripts/benchmark_tooling/analyze_benchmark_stability.py \
     Benchmark\ Dataset\ with\ Private\ AWS\ Devices.xlsx \
     --reference_file Benchmark\ Dataset\ with\ Public\ AWS\ Devices.xlsx
 ```
-
 ## Run unittest
 ```
 cd execuTorch/

.ci/scripts/benchmark_tooling/common.py

@@ -0,0 +1,39 @@
+import json
+import os
+
+import pandas as pd
+
+
+def read_excel_with_json_header(path: str):
+    # Read all sheets into a dict of DataFrames, without altering
+    all_sheets = pd.read_excel(path, sheet_name=None, header=None, engine="openpyxl")
+
+    results = []
+    for sheet, df in all_sheets.items():
+        # Extract JSON string from A1 (row 0, col 0)
+        json_str = df.iat[0, 0]
+        meta = json.loads(json_str) if isinstance(json_str, str) else {}
+
+        # The actual data starts from the next row; treat row 1 as header
+        df_data = pd.read_excel(path, sheet_name=sheet, skiprows=1, engine="openpyxl")
+        results.append({"groupInfo": meta, "df": df_data})
+    print(f"successfully fetched {len(results)} sheets from {path}")
+    return results
+
+
+def read_all_csv_with_metadata(folder_path: str):
+    results = []  # {filename: {"meta": dict, "df": DataFrame}}
+    for fname in os.listdir(folder_path):
+        if not fname.lower().endswith(".csv"):
+            continue
+        path = os.path.join(folder_path, fname)
+        with open(path, "r", encoding="utf-8") as f:
+            first_line = f.readline().strip()
+        try:
+            meta = json.loads(first_line)
+        except json.JSONDecodeError:
+            meta = {}
+        df = pd.read_csv(path, skiprows=1)
+        results.append({"groupInfo": meta, "df": df})
+    print(f"successfully fetched {len(results)} sheets from {folder_path}")
+    return results

.ci/scripts/benchmark_tooling/requirements.txt

@@ -4,3 +4,4 @@ pandas>=2.3.0
 openpyxl
 tabulate
 matplotlib
+openpyxl