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/update_timestamps.py

@@ -0,0 +1,57 @@
+import re
+import sys
+import os
+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", "--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)))
+
+    date_command = ["date", "-d", "@" + timestamp, "+%I:%M %p, %B %d, %Y"]
+    return subprocess.check_output(date_command).decode().strip()
+
+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)
+    timestamp_line = f'{"# " if file_path.endswith("py") else ""}**Updated:** *{timestamp}*\n'
+    timestamp_pattern = r'\*\*Updated:\*\*\s\**\d{1,2}:\d{2} [AP]M, \w+ \d{1,2}, \d{4}\*'
+
+    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 +1,16 @@
+#!/bin/bash
+
+SOURCEDIR=$1
+
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null && pwd )"
+
+directories=("$SOURCEDIR/beginner_source" "$SOURCEDIR/intermediate_source" "$SOURCEDIR/advanced_source")
+
+for dir in "${directories[@]}"; do
+    # Process .py and .rst files in the current directory
+    for file in "$dir"/*.{py,rst}; do
+        if [ -f "$file" ]; then
+            python "$DIR/update_timestamps.py" "$file"
+        fi
+    done
+done