v0.10 Add support for SQLite DB and CSV file generation for mapping of old to new usernames

gossamer_forums.rb

@@ -1,9 +1,16 @@
+# gossamer threads migration-import code
+# v0.10
+
 require 'mysql2'
 require 'open-uri'
 require 'net/http'
 require 'tempfile'
 require 'sqlite3'
 
+require 'digest'
+require 'fileutils'
+require 'csv'
+
 require File.expand_path("../../../config/environment", __FILE__)
 require_relative 'base'
 
@@ -25,11 +32,11 @@ class GossamerForumsImporter < ImportScripts::Base
 
 #    # Create a mapping of old Gossamer user IDs to new Discourse user IDs
 #    @user_id_map = {}
-      initialize_sqlite_user_id_db
+      initialize_sqlite_id_name_url_db
   end
 
-  def initialize_sqlite_user_id_db
+  def initialize_sqlite_id_name_url_db
-    @db = SQLite3::Database.new 'user_id_map.db'
+    @db = SQLite3::Database.new 'id_name_url_map.db'
     @db.execute <<-SQL
       CREATE TABLE IF NOT EXISTS user_id_map (
         old_user_id INTEGER PRIMARY KEY,
@@ -42,6 +49,15 @@ class GossamerForumsImporter < ImportScripts::Base
         new_category_id INTEGER
       );
     SQL
+    @db.execute <<-SQL
+      CREATE TABLE IF NOT EXISTS username_map (
+        id INTEGER PRIMARY KEY,
+        old_username TEXT,
+        new_username TEXT,
+        email TEXT,
+        real_name TEXT
+      );
+    SQL
   end
 
   def insert_user_id_mapping(old_user_id, new_user_id)
@@ -60,41 +76,61 @@ class GossamerForumsImporter < ImportScripts::Base
     @db.get_first_value "SELECT new_category_id FROM category_id_map WHERE old_category_id = ?", old_category_id
   end
 
+  def insert_username_mapping(old_username, new_username, email, real_name)
+    @db.execute "INSERT INTO username_map (old_username, new_username, email, real_name) VALUES (?, ?, ?, ?)", old_username, new_username, email, real_name
+  end
+
+  # Define a method to export the username mapping table to a CSV file
+  def export_username_mapping_to_csv(filename)
+    CSV.open(filename, 'wb') do |csv|
+      # Add headers
+      csv << ['Old Username', 'New Username', 'Email', 'Full Name']
+      
+      # Fetch data from the database
+      @db.execute("SELECT old_username, new_username, email, real_name FROM username_map") do |row|
+        csv << row
+      end
+    end
+  end
+
+
   # Execute an SQL query on the Gossamer Forums database
   def execute_query(query)
     @mysql_client.query(query, as: :hash)
   end
 
   # Sanitize the username to meet Discourse's requirements
-  def sanitize_username(username, email, name)
+  def sanitize_username(original_username, email, real_name)
-    original_username = username
+#     original_username = username
-    sanitized = username.gsub(/[^a-zA-Z0-9._-]/, '_')
+    sanitized_username = username.gsub(/[^a-zA-Z0-9._-]/, '_')
-    sanitized = "#{sanitized}." if sanitized.length < 2 # Allow two-character usernames
+    sanitized_username = "#{sanitized_username}." if sanitized_username.length < 2 # Allow two-character usernames
-    sanitized = sanitized[0, 20] if sanitized.length > 20
+    sanitized_username = sanitized_username[0, 20] if sanitized_username.length > 20
-    original_sanitized = sanitized
+    firststep_sanitized = sanitized_username
 
-    existing_user = User.find_by(username: sanitized)
+    existing_user = User.find_by(username: sanitized_username)
     
     if existing_user
       if existing_user.email.downcase == email.downcase && existing_user.name == name
-        return sanitized
+        return sanitized_username
       else
         counter = 1
-        while User.exists?(username: sanitized)
+        while User.exists?(username: sanitized_username)
-          sanitized = "#{original_sanitized}_#{counter}"
+          sanitized_username = "#{firststep_sanitized}_#{counter}"
-          sanitized = sanitized[0, 20] if sanitized.length > 20
+          sanitized_username = sanitized_username[0, 20] if sanitized_username.length > 20
           counter += 1
         end
       end
     end
 
-    if original_username != sanitized
+    if original_username != sanitized_username
-      puts "Sanitized username: '#{original_username}' --> '#{sanitized}'"
+      # The Discourse username is not the same as the Gossamer Forums username
+      puts "Sanitized username: '#{original_username}' --> '#{sanitized_username}'"
+      insert_username_mapping(original_username, sanitized_username, email, real_name)
 #    else
-#      puts "UNsanitized username:  '#{original_username}' --> '#{sanitized}'"
+#      puts "UNsanitized username:  '#{original_username}' --> '#{sanitized_username}'"
     end
 
-    sanitized
+    sanitized_username
   end
 
 
@@ -214,14 +250,14 @@ end
     end
 
     # For each user, add user ID mapping to SQLite now that we know what the Discourse user ID is, ... and append user bio and import user files
-    users.each do |user|
+    users.each do |discourse_user|
-      discourse_username = sanitize_username(user[:username], user[:email], user[:name])
+      # discourse_username = sanitize_username(user[:username], user[:email], user[:name])
-      discourse_user = User.find_by(username: discourse_username)
+      # discourse_user = User.find_by(username: discourse_username)
 
-      if discourse_user.nil?
+      # if discourse_user.nil?
-        puts "User #{user[:username]} --> #{discourse_username} not found in Discourse. Skipping file import."
+      #   puts "User #{user[:username]} --> #{discourse_username} not found in Discourse. Skipping file import."
-        next
+      #   next
-      end
+      # end
 
 #      # Store the user ID mapping
 #      @user_id_map[user[:id]] = discourse_user.id
@@ -556,6 +592,7 @@ end
     RateLimiter.disable
     puts "Starting Gossamer Forums import..."
     # import_users
+    # export_username_mapping_to_csv
     # import_categories
     # import_topics_and_posts
     import_personal_messages