Init

old_scripts/convert_man.sh

@@ -0,0 +1,46 @@
+#! /bin/bash
+
+ROCKY_VERSION=8.10
+MAN_PATH=./export/${ROCKY_VERSION}/
+LOCAL_MAN_PATH=
+HTML_BASE_PATH=./html_data/${ROCKY_VERSION}/
+
+process_file() {
+    local file=$1
+
+    local rpm_name
+    rpm_name=$(echo "$file" | cut -d'/' -f 4)
+    local man_context
+    man_context=$(echo "$file" | cut -d'/' -f 8)
+    local man_filename
+    man_filename=$(echo "$file" | awk -F'/' '{print $NF}' | sed -e 's/.gz//g' -e 's/\.[0-9]*$//g')
+
+    local output_folder="${HTML_BASE_PATH}/${rpm_name}/${man_context}/"
+
+    echo "$man_filename"
+
+    mkdir -p "${output_folder}"
+
+    # Try to convert the file and capture any errors
+    # if ! html_content=$(zcat "$file" | groff -Thtml -P-D/dev/null -man 2>/tmp/groff_error.log | pandoc -f html -t html 2>/tmp/pandoc_error.log); then
+    if ! html_content=$(zcat "$file" | mandoc -T html -O fragment 2>/tmp/mandoc_error.log | python3 ./apply_template.py  --rpm_name "$rpm_name" --file_name "$man_filename"); then
+        echo "Error processing file: $file"
+        cat /tmp/pandoc_error.log
+        return
+    fi
+    
+    local title
+    title=$(echo "$html_content" | sed -n 's/.*<h1>NAME<\/h1>\s*<p>\(.*\)<\/p>/\1/p' | sed 's/<[^>]*>//g')
+    [ -z "$title" ] && title="$man_filename"
+
+    # Check if html_content is empty
+    if [ -n "$html_content" ]; then
+        echo -e "$html_content" > "${output_folder}${man_filename}.html"
+        # echo -e "---\ntitle: \"$title\"\n---\n$html_content" > "${output_folder}${man_filename}.html"
+    fi
+}
+
+export -f process_file
+export HTML_BASE_PATH
+
+find "$MAN_PATH" -type f | parallel --will-cite process_file