From 621587c1b1c84a57e1724fc436248e6356ad1dd6 Mon Sep 17 00:00:00 2001
From: Mason Ballengee <masaball@iu.edu>
Date: Mon, 16 Sep 2024 15:05:12 -0400
Subject: [PATCH] Bring in sharepoint code

Refer to https://github.com/samvera/browse-everything/pull/418 for
what is being built off of.
---
 SharePoint.md                                 |  20 ++
 lib/browse_everything.rb                      |   8 +
 .../auth/sharepoint/session.rb                | 157 ++++++++++++++
 lib/browse_everything/driver/sharepoint.rb    | 198 ++++++++++++++++++
 4 files changed, 383 insertions(+)
 create mode 100644 SharePoint.md
 create mode 100644 lib/browse_everything/auth/sharepoint/session.rb
 create mode 100644 lib/browse_everything/driver/sharepoint.rb

diff --git a/SharePoint.md b/SharePoint.md
new file mode 100644
index 00000000..7bf71b64
--- /dev/null
+++ b/SharePoint.md
@@ -0,0 +1,20 @@
+# Sharepoint Provider
+
+This provider will allow browse-everything to access a _specific_ SharePoint location
+
+First register an application on azure to give access to the relevant location
+
+https://learn.microsoft.com/en-us/graph/auth-v2-service?tabs=http (steps 1,2 and 3)
+
+To us the sharepoint provider add the following to config/browse_everything_providers.yml
+
+```
+sharepoint:
+  client_id: [MyAppClientID]
+  client_secret: [MyAppClientSecret]
+  tenant_id: [MyAzuerTenantID]
+  grant_type: client_credentials
+  scope: https://graph.microsoft.com/.default
+  domain: mydomain.sharepoint.com
+  site_name: [MySiteName]
+```
\ No newline at end of file
diff --git a/lib/browse_everything.rb b/lib/browse_everything.rb
index a50dcf11..16091d61 100644
--- a/lib/browse_everything.rb
+++ b/lib/browse_everything.rb
@@ -16,6 +16,8 @@ module Driver
     autoload :Box,         'browse_everything/driver/box'
     autoload :GoogleDrive, 'browse_everything/driver/google_drive'
     autoload :S3,          'browse_everything/driver/s3'
+    autoload :Sharepoint,  'browse_everything/driver/sharepoint'
+
 
     # Access the sorter set for the base driver class
     # @return [Proc]
@@ -39,6 +41,12 @@ module Google
     end
   end
 
+  module Auth
+    module Sharepoint
+      autoload :Session,            'browse_everything/auth/sharepoint/session'
+    end
+  end
+
   class InitializationError < RuntimeError; end
   class ConfigurationError < StandardError; end
   class NotImplementedError < StandardError; end
diff --git a/lib/browse_everything/auth/sharepoint/session.rb b/lib/browse_everything/auth/sharepoint/session.rb
new file mode 100644
index 00000000..f2a1ad68
--- /dev/null
+++ b/lib/browse_everything/auth/sharepoint/session.rb
@@ -0,0 +1,157 @@
+require 'oauth2'
+
+# BrowseEverything OAuth2 session for 
+# Sharepoint provider
+module BrowseEverything
+  module Auth
+    module Sharepoint
+      class Session
+
+        OAUTH2_URLS = { 
+          :site => 'https://login.microsoftonline.com', 
+        }
+#          :scope => "https://graph.microsoft.com/.default" 
+
+        def initialize(opts={})
+
+          @config = BrowseEverything.config['sharepoint']
+
+          if opts[:client_id]
+            @oauth2_client = OAuth2::Client.new(opts[:client_id], opts[:client_secret],{:authorize_url => authorize_url, :token_url => token_url, :scope => scope}.merge!(OAUTH2_URLS.dup))
+            @access_token = OAuth2::AccessToken.new(@oauth2_client, opts[:access_token]) if opts[:access_token]
+            @access_token = get_access_token if opts[:access_token].blank?
+            @refresh_token = opts[:refresh_token] if @config[:grant_type] == 'authorization_code'
+#        @as_user = opts[:as_user]
+          end
+        end
+        
+        def authorize_url
+          @config['tenant_id']+"/oauth2/v2.0/authorize"
+        end
+
+        def token_url
+          @config['tenant_id']+"/oauth2/v2.0/token"
+        end
+
+        def scope
+          @config['scope']
+        end
+
+#        def authorize_url(redirect_uri, state=nil)
+#          opts = { :redirect_uri => redirect_uri }
+#          opts[:state] = state if state
+#
+#          @oauth2_client.auth_code.authorize_url(opts)
+#        end
+
+        def get_access_token(code=nil)
+
+          if @config[:grant_type] == 'client_credentials'
+             @access_token ||= @oauth2_client.client_credentials.get_token({:scope => @config[:scope]})
+          else
+            # assume authorization_code grant_type..?
+            @access_token ||= @oauth2_client.auth_code.get_token(code)
+          end
+        end
+
+        def refresh_token(refresh_token)
+          refresh_access_token_obj = OAuth2::AccessToken.new(@oauth2_client, @access_token.token, {'refresh_token' => refresh_token})
+          @access_token = refresh_access_token_obj.refresh!
+        end
+
+        def build_auth_header
+          "BoxAuth api_key=#{@api_key}&auth_token=#{@auth_token}"
+        end
+
+        def get(url, raw=false)
+          uri = URI.parse(url)
+          request = Net::HTTP::Get.new( uri.request_uri )
+          resp = request( uri, request, raw )
+        end
+
+        def delete(url, raw=false)
+          uri = URI.parse(url)
+          request = Net::HTTP::Delete.new( uri.request_uri )
+          resp = request( uri, request, raw )
+        end
+
+        def request(uri, request, raw=false, retries=0)
+
+          http = Net::HTTP.new(uri.host, uri.port)
+          http.use_ssl = true
+          #http.set_debug_output($stdout)
+
+          if @access_token
+            request.add_field('Authorization', "Bearer #{@access_token.token}")
+          else
+            request.add_field('Authorization', build_auth_header)
+          end
+
+
+          request.add_field('As-User', "#{@as_user}") if @as_user
+
+          response = http.request(request)
+
+          if response.is_a? Net::HTTPNotFound
+            raise RubyBox::ObjectNotFound
+          end
+
+          # Got unauthorized (401) status, try to refresh the token
+          if response.code.to_i == 401 and @refresh_token and retries == 0
+            refresh_token(@refresh_token)
+            return request(uri, request, raw, retries + 1)
+          end
+
+          sleep(@backoff) # try not to excessively hammer API.
+
+          handle_errors( response, raw )
+        end
+
+        def do_stream(url, opts)
+          params = {
+            :content_length_proc => opts[:content_length_proc],
+            :progress_proc => opts[:progress_proc]
+          }
+
+          if @access_token
+            params['Authorization'] = "Bearer #{@access_token.token}"
+          else
+            params['Authorization'] = build_auth_header
+          end
+
+          params['As-User'] = @as_user if @as_user
+
+          open(url, params)
+        end
+
+        def handle_errors( response, raw )
+          status = response.code.to_i
+          body = response.body
+          begin
+            parsed_body = JSON.parse(body)
+          rescue
+            msg = body.nil? || body.empty? ? "no data returned" : body
+            parsed_body = { "message" =>  msg }
+          end
+
+          # status is used to determine whether
+          # we need to refresh the access token.
+          parsed_body["status"] = status
+
+          case status / 100
+          when 3
+            # 302 Found. We should return the url
+            parsed_body["location"] = response["Location"] if status == 302
+          when 4
+            raise(RubyBox::ItemNameInUse.new(parsed_body, status, body), parsed_body["message"]) if parsed_body["code"] == "item_name_in_use"
+            raise(RubyBox::AuthError.new(parsed_body, status, body), parsed_body["message"]) if parsed_body["code"] == "unauthorized" || status == 401
+            raise(RubyBox::RequestError.new(parsed_body, status, body), parsed_body["message"])
+          when 5
+            raise(RubyBox::ServerError.new(parsed_body, status, body), parsed_body["message"])
+          end
+          raw ? body : parsed_body
+        end
+      end
+    end
+  end
+end
\ No newline at end of file
diff --git a/lib/browse_everything/driver/sharepoint.rb b/lib/browse_everything/driver/sharepoint.rb
new file mode 100644
index 00000000..f8cb42aa
--- /dev/null
+++ b/lib/browse_everything/driver/sharepoint.rb
@@ -0,0 +1,198 @@
+# frozen_string_literal: true
+
+require_relative 'authentication_factory'
+
+module BrowseEverything
+  module Driver
+    # Driver for accessing the MS-Graph API (https://learn.microsoft.com/en-us/graph/overview)
+    class Sharepoint < Base
+
+      class << self
+        attr_accessor :authentication_klass
+
+        def default_authentication_klass
+          BrowseEverything::Auth::Sharepoint::Session
+        end
+      end
+
+      # Constructor
+      # @param config_values [Hash] configuration for the driver
+      def initialize(config_values)
+        self.class.authentication_klass ||= self.class.default_authentication_klass
+        super(config_values)
+      end
+
+      def icon
+        'cloud'
+      end
+
+      # Validates the configuration for the Sharepoint provider
+      def validate_config
+        raise InitializationError, 'Sharepoint driver requires a :client_id argument' unless config[:client_id]
+        raise InitializationError, 'Sharepoint driver requires a :client_secret argument' unless config[:client_secret]
+        raise InitializationError, 'Sharepoint driver requires a :domain argument' unless config[:domain]
+        raise InitializationError, 'Sharepoint driver requires a :site_name argument' unless config[:site_name]
+      end
+
+      # Retrieves the file entry objects for a given path to MS-graph drive resource
+      # @param [String] id of the file or folder
+      # @return [Array<BrowseEverything::FileEntry>]
+      def contents(id = '')
+        sharepoint_session
+        folder = id.empty? ? drives : items_by_id(id)
+        values = []
+
+        folder.each do |f|
+          values << directory_entry(f)
+        end
+        @entries = values.compact
+
+        @sorter.call(@entries)
+      end
+
+      # Not used as we currently only deal with Client Credentials flow
+      # @return [String]
+      # Authorization url that is used to request the initial access code from Sharepoint/Onedrive/365/etc
+      def auth_link(*_args)
+        Addressable::URI.parse("https://login.microsoftonline.com/kingsfund.org.uk/oauth2/v2.0/authorize")
+      end
+
+      # @return [Boolean]
+      def authorized?
+        unless @token.present?
+          authorize!
+        end
+        @token.present?
+      end
+
+      def authorize!
+        # TODO handle other authentication strategies (other than client_credentials)
+        register_access_token(sharepoint_session.get_access_token)
+      end
+
+      # @param [String] id of the file on MS graph drive
+      # @return [Array<String, Hash>]
+      def link_for(id)
+         file = items_by_id(id)
+         extras = {file_name: file['name'], file_size: file['size'].to_i}
+         [download_url(file), extras]
+      end
+
+
+      private
+
+      def token_expired?
+        return true if expiration_time.nil?
+        Time.now.to_i > expiration_time
+      end
+
+
+      def session
+        AuthenticationFactory.new(
+          self.class.authentication_klass,
+          client_id: config[:client_id],
+          client_secret: config[:client_secret],
+          access_token: sharepoint_token,
+          domain: config[:domain],
+          site_name: config[:site_name],
+        )
+      end
+
+      def authenticate
+        session.authenticate
+      end
+
+      # If there is an active session, {@token} will be set by {BrowseEverythingController} using data stored in the
+      # session. 
+      #
+      # @param [OAuth2::AccessToken] access_token
+      def register_access_token(access_token)
+        @token = {
+          'token' => access_token.token,
+          'expires_at' => access_token.expires_at
+        }
+      end
+
+      def sharepoint_token
+        return unless @token
+        @token.fetch('token', nil)
+      end
+
+      def expiration_time
+        return unless @token
+        @token.fetch('expires_at', nil).to_i
+      end
+
+      # Constructs a BrowseEverything::FileEntry object for a Sharepoint file
+      # resource
+      # @param file [String] ID to the file resource
+      # @return [BrowseEverything::File]
+      def directory_entry(file)
+        BrowseEverything::FileEntry.new(make_path(file), [key, make_path(file)].join(':'), file['name'], file['size'] ? file['size'] : nil, Date.parse(file['lastModifiedDateTime']), folder?(file))
+      end
+
+      # Derives a path from item (file or folder or drive) metadata 
+      # that can be used in subsequent items_by_id calls
+      def make_path(file)
+        if file['parentReference'].present? 
+          folder?(file) ? "#{file['parentReference']['driveId']}/items/#{file['id']}/children" : "#{file['parentReference']['driveId']}/items/#{file['id']}"
+        else 
+          "#{file['id']}/root/children"
+        end
+      end
+
+      def folder?(file)
+        !file['file'].present?
+      end
+
+     ##################################################################
+     # The below are all candidates to go its own sharepoint api module
+     # or some such
+     ##################################################################
+
+      def sharepoint_request(sharepoint_uri)
+        sharepoint_client
+        @auth = "Bearer "+sharepoint_token
+
+        uri = URI.parse(sharepoint_uri)
+        http = Net::HTTP.new(uri.host, uri.port)
+        http.use_ssl = true if uri.scheme == 'https'
+
+        response = http.start do
+          request = Net::HTTP::Get.new(uri.request_uri,{'Authorization' => @auth})
+          http.request(request)
+        end
+        JSON.parse(response.body)
+      end
+
+      def site_id
+        @site_id ||= sharepoint_request("https://graph.microsoft.com/v1.0/sites/#{config[:domain]}:/sites/#{config[:site_name]}/")['id']
+      end
+
+      def drives
+       @drives ||= sharepoint_request("https://graph.microsoft.com/v1.0/sites/#{site_id}/drives")['value']
+     end
+
+     def items_by_id(id)
+       item = sharepoint_request("https://graph.microsoft.com/v1.0/sites/#{site_id}/drives/#{id}")
+       item['value'].present? ? item['value'] : item
+     end
+
+     def download_url(file)
+       file['@microsoft.graph.downloadUrl']
+     end
+
+    def sharepoint_client
+      if token_expired?
+        session = sharepoint_session
+        register_access_token(sharepoint_session.get_access_token)
+      end
+    end
+
+    def sharepoint_session
+      authenticate
+     end
+
+    end
+  end
+end
\ No newline at end of file