Tutorial timestamps

.jenkins/build.sh

@@ -56,6 +56,9 @@ if [[ "${JOB_TYPE}" == "worker" ]]; then
   # Files to run must be accessible to subprocessed (at least to `download_data.py`)
   export FILES_TO_RUN
 
+  # Step 2.1: Add timestamps to .py and .rst files in source directories
+  bash $DIR/update_timestamps_batch.sh
+
   # Step 3: Run `make docs` to generate HTML files and static files for these tutorialis
   pip3 install -e git+https://github.com/pytorch/pytorch_sphinx_theme.git#egg=pytorch_sphinx_theme
   make docs
@@ -118,6 +121,10 @@ if [[ "${JOB_TYPE}" == "worker" ]]; then
   7z a worker_${WORKER_ID}.7z docs
   awsv2 s3 cp worker_${WORKER_ID}.7z s3://${BUCKET_NAME}/${COMMIT_ID}/worker_${WORKER_ID}.7z
 elif [[ "${JOB_TYPE}" == "manager" ]]; then
+
+  # Step 0.9: Add timestamps to .py and .rst files in source directories
+  bash $DIR/update_timestamps_batch.sh
+
   # Step 1: Generate no-plot HTML pages for all tutorials
   pip3 install -e git+https://github.com/pytorch/pytorch_sphinx_theme.git#egg=pytorch_sphinx_theme
   make html-noplot

.jenkins/get_sphinx_filenames.py

@@ -1,4 +1,3 @@
-from pathlib import Path
 from typing import List
 
 from get_files_to_run import get_all_files
@@ -11,3 +10,6 @@ def get_files_for_sphinx() -> List[str]:
 
 
 SPHINX_SHOULD_RUN = "|".join(get_files_for_sphinx())
+
+if __name__ == "__main__":
+    print(SPHINX_SHOULD_RUN)

.jenkins/update_timestamps.py

@@ -0,0 +1,74 @@
+import re
+import sys
+import os
+import datetime
+import subprocess
+
+
+def get_last_commit_timestamp_for_file(file_path: str) -> str:
+    """Get the last commit timestamp for a file.
+
+    Args:
+        file_path (str): Path to file
+
+    Returns:
+        str: Last committed timestamp string
+    """
+
+    git_command = [
+        "git",
+        "log",
+        "-1",
+        "--date=format:%B %d, %Y",
+        "--format=%at",
+        "--",
+        file_path,
+    ]
+    timestamp = subprocess.check_output(git_command).decode().strip()
+
+    if not timestamp:
+        # If there is no git commit history, use last modified date
+        timestamp = str(int(os.path.getmtime(file_path)))
+
+    dt = datetime.datetime.fromtimestamp(int(timestamp), tz=datetime.timezone.utc)
+    return dt.strftime("%I:%M %p, %B %d, %Y")
+
+
+def update_timestamp(file_path: str):
+    """Adds a timestamp of the most recent time the file was edited.
+
+    Args:
+        file_path (str): Path to file
+    """
+    with open(file_path, "r") as file:
+        lines = file.readlines()
+
+    # Get current timestamp
+    timestamp = get_last_commit_timestamp_for_file(file_path)
+    if not timestamp:
+        return
+
+    timestamp_line = (
+        f'{"# " if file_path.endswith("py") else ""}_Last Updated: {timestamp}_\n'
+    )
+    timestamp_pattern = r"_Last Updated:\s.+_"
+
+    if not lines:
+        lines = [timestamp_line]
+    else:
+        i = len(lines) - 1
+        while i > 0 and not lines[i].strip():
+            i -= 1
+
+        if re.search(timestamp_pattern, lines[i]):
+            lines[i] = timestamp_line
+        else:
+            lines.append("\n\n" + timestamp_line)
+
+    # Write updated lines back to file
+    with open(file_path, "w") as file:
+        file.writelines(lines)
+
+
+file_path = sys.argv[1]
+update_timestamp(file_path)

.jenkins/update_timestamps_batch.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null && pwd )"
+
+# Check if SPHINX_SHOULD_RUN is already set
+if [ -z "$SPHINX_SHOULD_RUN" ]; then
+    # If not set, retrieve it using get_sphinx_filenames.py and export it
+    SPHINX_SHOULD_RUN=$(python "$DIR/get_sphinx_filenames.py")
+    export SPHINX_SHOULD_RUN
+fi
+
+# Convert the pipe-separated filenames into an array
+IFS='|' read -r -a file_array <<< "$SPHINX_SHOULD_RUN"
+
+# Loop through each file and update timestamps if it exists
+for file in "${file_array[@]}"; do
+    file="$DIR/../$file"
+    if [ -f "$file" ]; then
+        python "$DIR/update_timestamps.py" "$file"
+    fi
+done