FEATURE: Allows CSV file result to be attached in automated PMs (#318)

This commit adds an optional setting that allows to attach query results
in CSV format as a file to PMs sent by Data Explorer's automation
scripts.

meta topic: https://meta.discourse.org/t/turn-data-explorer-query-results-into-csv-to-attach-to-discourse-automated-emails/267529

Co-authored-by: Drenmi <drenmi@gmail.com>

Author: Lhcfl

app/controllers/discourse_data_explorer/query_controller.rb

@@ -179,49 +179,27 @@ def run
 
         render json: { success: false, errors: [err_msg] }, status: 422
       else
-        pg_result = result[:pg_result]
-        cols = pg_result.fields
+        content_disposition =
+          "attachment; filename=#{query.slug}@#{Slug.for(Discourse.current_hostname, "discourse")}-#{Date.today}.dcqresult"
+
         respond_to do |format|
           format.json do
-            if params[:download]
-              response.headers[
-                "Content-Disposition"
-              ] = "attachment; filename=#{query.slug}@#{Slug.for(Discourse.current_hostname, "discourse")}-#{Date.today}.dcqresult.json"
-            end
-            json = {
-              success: true,
-              errors: [],
-              duration: (result[:duration_secs].to_f * 1000).round(1),
-              result_count: pg_result.values.length || 0,
-              params: query_params,
-              columns: cols,
-              default_limit: SiteSetting.data_explorer_query_result_limit,
-            }
-            json[:explain] = result[:explain] if opts[:explain]
-
-            if !params[:download]
-              relations, colrender = DataExplorer.add_extra_data(pg_result)
-              json[:relations] = relations
-              json[:colrender] = colrender
-            end
-
-            json[:rows] = pg_result.values
-
-            render json: json
+            response.headers["Content-Disposition"] = "#{content_disposition}.json" if params[
+              :download
+            ]
+
+            render json:
+                     ResultFormatConverter.convert(
+                       :json,
+                       result,
+                       query_params:,
+                       download: params[:download],
+                     )
           end
           format.csv do
-            response.headers[
-              "Content-Disposition"
-            ] = "attachment; filename=#{query.slug}@#{Slug.for(Discourse.current_hostname, "discourse")}-#{Date.today}.dcqresult.csv"
-
-            require "csv"
-            text =
-              CSV.generate do |csv|
-                csv << cols
-                pg_result.values.each { |row| csv << row }
-              end
-
-            render plain: text
+            response.headers["Content-Disposition"] = "#{content_disposition}.csv"
+
+            render plain: ResultFormatConverter.convert(:csv, result)
           end
         end
       end

config/locales/client.en.yml

@@ -116,3 +116,5 @@ en:
               label: Data Explorer Query parameters
             skip_empty:
               label: Skip sending PM if there are no results
+            attach_csv:
+              label: Attach the CSV file to the PM

lib/report_generator.rb

@@ -9,13 +9,13 @@ def self.generate(query_id, query_params, recipients, opts = {})
       recipients = filter_recipients_by_query_access(recipients, query)
       params = params_to_hash(query_params)
 
-      result = DataExplorer.run_query(query, params)[:pg_result]
+      result = DataExplorer.run_query(query, params)
       query.update!(last_run_at: Time.now)
 
-      return [] if opts[:skip_empty] && result.values.empty?
-      table = ResultToMarkdown.convert(result)
+      return [] if opts[:skip_empty] && result[:pg_result].values.empty?
+      table = ResultToMarkdown.convert(result[:pg_result])
 
-      build_report_pms(query, table, recipients)
+      build_report_pms(query, table, recipients, attach_csv: opts[:attach_csv], result:)
     end
 
     private
@@ -40,8 +40,20 @@ def self.params_to_hash(query_params)
       params_hash
     end
 
-    def self.build_report_pms(query, table = "", targets = [])
+    def self.build_report_pms(query, table = "", targets = [], attach_csv: false, result: nil)
       pms = []
+      upload =
+        if attach_csv
+          tmp_filename =
+            "#{query.slug}@#{Slug.for(Discourse.current_hostname, "discourse")}-#{Date.today}.dcqresult.csv"
+          tmp = Tempfile.new(tmp_filename)
+          tmp.write(ResultFormatConverter.convert(:csv, result))
+          tmp.rewind
+          UploadCreator.new(tmp, tmp_filename, type: "csv_export").create_for(
+            Discourse.system_user.id,
+          )
+        end
+
       targets.each do |target|
         name = target[0]
         pm_type = "target_#{target[1]}s"
@@ -53,6 +65,9 @@ def self.build_report_pms(query, table = "", targets = [])
           "Query Name:\n#{query.name}\n\nHere are the results:\n#{table}\n\n" +
           "<a href='#{Discourse.base_url}/admin/plugins/explorer?id=#{query.id}'>View query in Data Explorer</a>\n\n" +
           "Report created at #{Time.zone.now.strftime("%Y-%m-%d at %H:%M:%S")} (#{Time.zone.name})"
+        if upload
+          pm["raw"] << "\n\nAppendix: [#{upload.original_filename}|attachment](#{upload.short_url})"
+        end
         pms << pm
       end
       pms

lib/result_format_converter.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+module ::DiscourseDataExplorer
+  class ResultFormatConverter
+    def self.convert(file_type, result, opts = {})
+      self.new(result, opts).send("to_#{file_type}")
+    end
+
+    def initialize(result, opts)
+      @result = result
+      @opts = opts
+    end
+
+    private
+
+    attr_reader :result
+    attr_reader :opts
+
+    def pg_result
+      @pg_result ||= @result[:pg_result]
+    end
+
+    def cols
+      @cols ||= pg_result.fields
+    end
+
+    def to_csv
+      require "csv"
+      CSV.generate do |csv|
+        csv << cols
+        pg_result.values.each { |row| csv << row }
+      end
+    end
+
+    def to_json
+      json = {
+        success: true,
+        errors: [],
+        duration: (result[:duration_secs].to_f * 1000).round(1),
+        result_count: pg_result.values.length || 0,
+        params: opts[:query_params],
+        columns: cols,
+        default_limit: SiteSetting.data_explorer_query_result_limit,
+      }
+      json[:explain] = result[:explain] if opts[:explain]
+
+      if !opts[:download]
+        relations, colrender = DataExplorer.add_extra_data(pg_result)
+        json[:relations] = relations
+        json[:colrender] = colrender
+      end
+
+      json[:rows] = pg_result.values
+
+      json
+    end
+
+    #TODO: we can move ResultToMarkdown here
+  end
+end

plugin.rb

@@ -79,6 +79,7 @@ module ::DiscourseDataExplorer
 
   require_relative "lib/report_generator"
   require_relative "lib/result_to_markdown"
+  require_relative "lib/result_format_converter"
   reloadable_patch do
     if defined?(DiscourseAutomation)
       add_automation_scriptable("recurring_data_explorer_result_pm") do
@@ -90,6 +91,7 @@ module ::DiscourseDataExplorer
         field :query_id, component: :choices, required: true, extra: { content: queries }
         field :query_params, component: :"key-value", accepts_placeholders: true
         field :skip_empty, component: :boolean
+        field :attach_csv, component: :boolean
 
         version 1
         triggerables [:recurring]
@@ -99,6 +101,7 @@ module ::DiscourseDataExplorer
           query_id = fields.dig("query_id", "value")
           query_params = fields.dig("query_params", "value") || {}
           skip_empty = fields.dig("skip_empty", "value") || false
+          attach_csv = fields.dig("attach_csv", "value") || false
 
           unless SiteSetting.data_explorer_enabled
             Rails.logger.warn "#{DiscourseDataExplorer::PLUGIN_NAME} - plugin must be enabled to run automation #{automation.id}"
@@ -111,7 +114,7 @@ module ::DiscourseDataExplorer
           end
 
           DiscourseDataExplorer::ReportGenerator
-            .generate(query_id, query_params, recipients, { skip_empty: })
+            .generate(query_id, query_params, recipients, { skip_empty:, attach_csv: })
             .each do |pm|
               begin
                 utils.send_pm(pm, automation_id: automation.id, prefers_encrypt: false)

spec/report_generator_spec.rb

@@ -130,5 +130,25 @@
       expect(result[1]["target_group_names"]).to eq([group.name])
       expect(result[2]["target_emails"]).to eq(["john@doe.com"])
     end
+
+    it "works with attached csv file" do
+      SiteSetting.personal_message_enabled_groups = group.id
+      DiscourseDataExplorer::ResultToMarkdown.expects(:convert).returns("le table")
+      freeze_time
+
+      result =
+        described_class.generate(query.id, query_params, [user.username], { attach_csv: true })
+
+      filename =
+        "#{query.slug}@#{Slug.for(Discourse.current_hostname, "discourse")}-#{Date.today}.dcqresult.csv"
+
+      expect(result[0]["raw"]).to include(
+        "Hi #{user.username}, your data explorer report is ready.\n\n" +
+          "Query Name:\n#{query.name}\n\nHere are the results:\nle table\n\n" +
+          "<a href='#{Discourse.base_url}/admin/plugins/explorer?id=#{query.id}'>View query in Data Explorer</a>\n\n" +
+          "Report created at #{Time.zone.now.strftime("%Y-%m-%d at %H:%M:%S")} (#{Time.zone.name})\n\n" +
+          "Appendix: [#{filename}|attachment](upload://",
+      )
+    end
   end
 end

spec/result_format_converter_spec.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+describe DiscourseDataExplorer::ResultFormatConverter do
+  fab!(:user)
+  fab!(:post)
+  fab!(:query) { DiscourseDataExplorer::Query.find(-1) }
+
+  let(:query_params) { [{ from_days_ago: 0 }, { duration_days: 15 }] }
+  let(:query_result) { DiscourseDataExplorer::DataExplorer.run_query(query, query_params) }
+
+  before { SiteSetting.data_explorer_enabled = true }
+
+  describe ".convert" do
+    context "for csv files" do
+      it "format results as a csv table with headers and columns" do
+        result = described_class.convert(:csv, query_result)
+
+        table = <<~CSV
+          liker_user_id,liked_user_id,count
+        CSV
+
+        expect(result).to include(table)
+      end
+    end
+
+    context "for json files" do
+      it "format results as a json file" do
+        result = described_class.convert(:json, query_result, { query_params: })
+
+        expect(result[:columns]).to contain_exactly("liker_user_id", "liked_user_id", "count")
+        expect(result[:params]).to eq(query_params)
+      end
+    end
+  end
+end