Init

old_scripts/convert_man.py

@@ -0,0 +1,48 @@
+import os
+import subprocess
+from pathlib import Path
+
+ROCKY_VERSION = "8.10"
+MAN_PATH = f"./export/{ROCKY_VERSION}/"
+HTML_BASE_PATH = f"./html_data2/{ROCKY_VERSION}/"
+
+def process_file(file):
+    rpm_name = file.parts[3]
+    man_context = file.parts[7]
+    man_filename = file.name.replace('.gz', '').rsplit('.', 1)[0]
+
+    output_folder = Path(HTML_BASE_PATH) / rpm_name / man_context
+    output_folder.mkdir(parents=True, exist_ok=True)
+
+    print(man_filename)
+
+    try:
+        html_content = subprocess.check_output(
+            f'zcat "{file}" | mandoc -T html -O fragment 2>/tmp/mandoc_error.log | python3 ./apply_template.py --rpm_name "{rpm_name}" --file_name "{man_filename}"',
+            shell=True,
+            text=True
+        )
+    except subprocess.CalledProcessError:
+        print(f"Error processing file: {file}")
+        with open('/tmp/mandoc_error.log', 'r') as error_log:
+            print(error_log.read())
+        return
+
+    title = ""
+    for line in html_content.splitlines():
+        if '<h1>NAME</h1>' in line:
+            title = line.split('<p>')[1].split('</p>')[0].strip()
+            break
+    title = title or man_filename
+
+    if html_content:
+        with open(output_folder / f"{man_filename}.html", 'w') as f:
+            f.write(html_content)
+
+def main():
+    for root, _, files in os.walk(MAN_PATH):
+        for file in files:
+            process_file(Path(root) / file)
+
+if __name__ == "__main__":
+    main()