From bd9ea7a88dc44f92516315aa1e997a7c1b0e616d Mon Sep 17 00:00:00 2001
From: Wim <wim@42.be>
Date: Sat, 9 Jun 2018 14:35:02 +0200
Subject: [PATCH] Add MediaDownloadBlacklist option. Closes #442

---
 bridge/config/config.go  |  3 ++-
 bridge/helper/helper.go  | 14 ++++++++++++++
 matterbridge.toml.sample |  8 +++++++-
 3 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/bridge/config/config.go b/bridge/config/config.go
index 351e6cad..0fa76b52 100644
--- a/bridge/config/config.go
+++ b/bridge/config/config.go
@@ -74,10 +74,11 @@ type Protocol struct {
 	Jid                    string // xmpp
 	Label                  string // all protocols
 	Login                  string // mattermost, matrix
+	MediaDownloadBlackList []string
+	MediaDownloadPath      string // Basically MediaServerUpload, but instead of uploading it, just write it to a file on the same server.
 	MediaDownloadSize      int    // all protocols
 	MediaServerDownload    string
 	MediaServerUpload      string
-	MediaDownloadPath      string     // Basically MediaServerUpload, but instead of uploading it, just write it to a file on the same server.
 	MessageDelay           int        // IRC, time in millisecond to wait between messages
 	MessageFormat          string     // telegram
 	MessageLength          int        // IRC, max length of a message allowed
diff --git a/bridge/helper/helper.go b/bridge/helper/helper.go
index a3f4255c..13d94ede 100644
--- a/bridge/helper/helper.go
+++ b/bridge/helper/helper.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"regexp"
 	"strings"
 	"time"
 
@@ -73,6 +74,19 @@ func GetAvatar(av map[string]string, userid string, general *config.Protocol) st
 }
 
 func HandleDownloadSize(flog *log.Entry, msg *config.Message, name string, size int64, general *config.Protocol) error {
+	// check blacklist here
+	for _, entry := range general.MediaDownloadBlackList {
+		if entry != "" {
+			re, err := regexp.Compile(entry)
+			if err != nil {
+				flog.Errorf("incorrect regexp %s for %s", entry, msg.Account)
+				continue
+			}
+			if re.MatchString(name) {
+				return fmt.Errorf("Matching blacklist %s. Not downloading %s", entry, name)
+			}
+		}
+	}
 	flog.Debugf("Trying to download %#v with size %#v", name, size)
 	if int(size) > general.MediaDownloadSize {
 		msg.Event = config.EVENT_FILE_FAILURE_SIZE
diff --git a/matterbridge.toml.sample b/matterbridge.toml.sample
index 1bc4f3b4..2a77bca1 100644
--- a/matterbridge.toml.sample
+++ b/matterbridge.toml.sample
@@ -1331,9 +1331,15 @@ MediaServerDownload="https://youserver.com/download"
 #It will only download from bridges that don't have public links available, which are for the moment
 #slack, telegram, matrix and mattermost
 #
-#Optional (default 1000000 (1 megabyte))
+#OPTIONAL (default 1000000 (1 megabyte))
 MediaDownloadSize=1000000
 
+#MediaDownloadBlacklist allows you to blacklist specific files from being downloaded.
+#Filenames matching these regexp will not be download/uploaded to the mediaserver
+#You can use regex for this, see https://regex-golang.appspot.com/assets/html/index.html for more regex info
+#OPTIONAL (default empty)
+MediaDownloadBlacklist=[".html$",".htm$"]
+
 ###################################################################
 #Gateway configuration
 ###################################################################