codepuncher · March 26, 2021 15:07 · Mar 26, 2021 · Mar 26, 2021
diff --git a/fix_post_urls.sh b/fix_post_urls.sh
@@ -0,0 +1,175 @@
+#!/usr/bin/env bash
+
+# set -x # debug mode - shows all command output
+
+# Get the full path to the wp program.
+wp=$(which wp)
+
+# Get the full path to the curl program
+curl=$(which curl)
+
+# Example path for remote server (e.g. Kinsta).
+path="$HOME/public/current/web/wp"
+
+echo "Begin fixing Post URLs $(date)"
+
+# Generate print numbers 3-9. We don't want 1 or 2 because they are irrelevant.
+site_ids=$(printf '%s,' {3..8}; echo 9)
+site_urls=$("$wp" site list --field=domain --site__in="$site_ids" --path="$path")
+
+# An array for storing the URLs for a search-replace.
+declare -A SR_URLS
+
+for site_url in $site_urls; do
+  read -rp "Do you wish to skip site $site_url? [y/n] " do_skip_site
+  if [[ $do_skip_site =~ ^[Yy]$ ]]; then
+    continue
+  fi
+
+  # Output directory
+  output_dir="fix_post_urls_output/$site_url"
+  if [ -d "$output_dir" ]; then
+    rm -rf "$output_dir"
+  else
+    mkdir -p "$output_dir"
+  fi
+
+  # Clear the search replace array from previous loop.
+  SR_URLS=()
+
+  echo "$site_url"
+
+  post_urls_file="$site_url-post-urls.csv"
+
+  # If a redirect file exists, delete it first.
+  if [ -f "$output_dir/$post_urls_file" ]; then
+    rm "$output_dir/$post_urls_file"
+  fi
+
+  current_permastruct=$("$wp" option get permalink_structure --url="$site_url" --path="$path")
+  old_path=${current_permastruct//\/%postname%\//}
+
+  # Do not include a leading or trailing slash.
+  new_path='about-us/news'
+
+  # Get all URLs for Posts with Post Type of "post".
+  post_urls=$("$wp" post list --post_type=post --post_status=publish --field=url --url="$site_url" --path="$path")
+
+  # Generate a CSV for redirects.
+  for post_url in $post_urls; do
+    # Create the new URL.
+    new_url=${post_url/${site_url}${old_path}/${site_url}/${new_path}}
+    if [ "$post_url" = "$new_url" ]; then
+      echo "$post_url == $new_url"
+      echo 'Old/current URL is the same as the new URL. Skipping'
+      printf '\n'
+      continue 2
+    fi
+
+    # This is the row in the redirect import file.
+    # source URL,target URL,regex,HTTP code
+    echo "$post_url,$new_url,0,301" >> "$output_dir/$post_urls_file"
+
+    SR_URLS["$post_url"]="$new_url"
+  done
+  echo "Redirects generated for $site_url"
+
+  # Change rewrite rule / permalink structure.
+  new_permastruct="/$new_path/%postname%/"
+  read -rp "Do you wish to change permalink structure from $current_permastruct to $new_permastruct? [y/n] " do_rewrite_structure
+  if [[ $do_rewrite_structure =~ ^[Yy]$ ]]; then
+    res=$("$wp" rewrite structure "$new_permastruct" --url="$site_url" --path="$path")
+
+    echo "$res"
+
+    if [ ! "$res" ]; then
+      exit 1
+    fi
+  else
+    echo "Skipping rewrite structure for $site_url"
+  fi
+
+  # Perform wp search-replace.
+  read -rp 'Do you wish to perform a search-replace? [y/n] ' do_replacement
+  if [[ $do_replacement =~ ^[Yy]$ ]]; then
+    echo "Performing search-replace for $site_url"
+
+    for source_url in "${!SR_URLS[@]}"; do 
+      target_url="${SR_URLS[$source_url]}"
+      res=$("$wp" search-replace \
+       "$source_url" \
+       "$target_url" \
+       --url="$site_url" \
+       --path="$path" \
+       --network \
+       --skip-columns=guid)
+
+      if [ ! "$res" ]; then
+        echo "Failed to replace $source_url with $target_url"
+        exit 1
+      fi
+    done
+  else
+    echo "Skipping search-replace for $site_url"
+  fi
+
+  # Import new redirects from file.
+  redirect_group_id=10
+  read -rp "Do you wish to import new redirects for $site_url? [y/n] " do_redirect_import
+  if [[ $do_redirect_import =~ ^[Yy]$ ]]; then
+    res=$("$wp" redirection import "$output_dir/$post_urls_file" --format=csv --group="$redirect_group_id" --url="$site_url" --path="$path")
+
+    if [ ! "$res" ]; then
+      exit 1;
+    fi
+  else
+    echo "Skipping redirect import for $site_url"
+  fi
+
+  # Check if redirects are successful 301.
+  read -rp "Do you wish to validate the new redirects for $site_url? [y/n] " do_redirect_check
+  if [[ $do_redirect_check =~ ^[Yy]$ ]]; then
+    # If a redirect file exists, delete it first.
+    redirect_no_301_file="$site_url-redirects-no-301s.csv"
+    redirect_incorrect_301_file="$site_url-redirects-incorrect-301s.csv"
+    redirect_target_404_file="$site_url-redirects-target-404s.csv"
+    redirect_correct_301_file="$site_url-redirects-correct-301s.csv"
+    for file in $redirect_no_301_file $redirect_incorrect_301_file $redirect_target_404_file $redirect_correct_301_file; do
+      if [ -f "$output_dir/$file" ]; then
+        rm "$output_dir/$file" && echo "Deleted $file"
+      fi
+    done
+
+    for source_url in "${!SR_URLS[@]}"; do
+      echo "Performing a check for $source_url"
+      target_url="${SR_URLS[$source_url]}"
+      # -I "Perform HEAD request, -s "Silent".
+      CURL_SOURCE=$("$curl" -I -s "$source_url")
+      CURL_TARGET=$("$curl" -I -s "$target_url")
+      source_is_301=$(echo "$CURL_SOURCE" | grep -iE '^HTTP\/2 301')
+      source_redirects_to_target=$(echo "$CURL_SOURCE" | grep -iE "^Location: $target_url")
+      target_is_404=$(echo "$CURL_TARGET" | grep -iE '^HTTP\/2 404')
+
+      echo "$CURL_SOURCE"
+      echo "$CURL_TARGET"
+
+      if [ -z "$source_is_301" ]; then
+        echo "$source_url" >> "$output_dir/$redirect_no_301_file"
+        echo "$source_url does not have a 301 redirect"
+      elif [ -z "$source_redirects_to_target" ]; then
+        echo "$source_url" >> "$output_dir/$redirect_incorrect_301_file"
+        echo "$source_url does not redirect to $target_url"
+      elif [ -n "$target_is_404" ]; then
+        echo "$target_url" >> "$output_dir/$redirect_target_404_file"
+        echo "$target_url is a 404"
+      else
+        echo "$source_url,$target_url" >> "$output_dir/$redirect_correct_301_file"
+        echo "$source_url successfully redirects to $target_url"
+      fi
+    done
+  fi
+
+  printf '\n'
+done
+
+echo "End fixing Post URLs $(date)"