From 98c304df17f5cd264763002b9b8e5006a9bf41fc Mon Sep 17 00:00:00 2001
From: traut <sergey@polzunov.com>
Date: Thu, 2 Jan 2025 00:28:28 +0100
Subject: [PATCH] Add `use_browser_user_agent` field

---
 docs/plugins/builtin/data-sources/rss.md |  7 +++++++
 docs/plugins/plugins.json                |  3 ++-
 internal/builtin/data_rss.go             | 25 ++++++++++++++++++++----
 3 files changed, 30 insertions(+), 5 deletions(-)

diff --git a/docs/plugins/builtin/data-sources/rss.md b/docs/plugins/builtin/data-sources/rss.md
index ac8d0529..86c9bd16 100644
--- a/docs/plugins/builtin/data-sources/rss.md
+++ b/docs/plugins/builtin/data-sources/rss.md
@@ -58,6 +58,13 @@ data rss {
   # Default value:
   fill_in_content = false
 
+  # If the data source should pretend to be a browser while fetching the feed and the feed items.
+  # If set to "false", the default user-agent value "blackstork-rss/0.0.1" will be used.
+  #
+  # Optional bool.
+  # Default value:
+  use_browser_user_agent = false
+
   # Maximum number of items to fill the content in per feed.
   #
   # Optional number.
diff --git a/docs/plugins/plugins.json b/docs/plugins/plugins.json
index e95779c0..b73a1ce4 100644
--- a/docs/plugins/plugins.json
+++ b/docs/plugins/plugins.json
@@ -123,7 +123,8 @@
           "fill_in_content",
           "fill_in_max_items",
           "only_items_after_time",
-          "url"
+          "url",
+          "use_browser_user_agent"
         ]
       },
       {
diff --git a/internal/builtin/data_rss.go b/internal/builtin/data_rss.go
index 24e812e0..79df793f 100644
--- a/internal/builtin/data_rss.go
+++ b/internal/builtin/data_rss.go
@@ -28,6 +28,7 @@ import (
 
 const (
 	defaultRequestTimeout = 30 * time.Second
+	defaultUserAgent      = "blackstork-rss/0.0.1"
 )
 
 // https://techblog.willshouse.com/2012/01/03/most-common-user-agents/
@@ -69,6 +70,16 @@ func makeRSSDataSource() *plugin.DataSource {
 						If the full content should be added when it's not present in the feed items.
 					`,
 				},
+				{
+					Name:        "use_browser_user_agent",
+					Type:        cty.Bool,
+					DefaultVal:  cty.BoolVal(false),
+					Constraints: constraint.NonNull,
+					Doc: fmt.Sprintf(`
+						If the data source should pretend to be a browser while fetching the feed and the feed items.
+						If set to "false", the default user-agent value "%s" will be used.
+					`, defaultUserAgent),
+				},
 				{
 					Name:         "fill_in_max_items",
 					Type:         cty.Number,
@@ -150,7 +161,7 @@ func filterItems(ctx context.Context, feed *gofeed.Feed, from time.Time) *gofeed
 	return feed
 }
 
-func fetchFeedItems(ctx context.Context, feed *gofeed.Feed, itemsCap int) *gofeed.Feed {
+func fetchFeedItems(ctx context.Context, feed *gofeed.Feed, userAgent string, itemsCap int) *gofeed.Feed {
 	log := slog.Default()
 	log = log.With("feed_url", feed.Link, "items_cap", itemsCap)
 	log.InfoContext(ctx, "Fetching content for the items in the feed")
@@ -192,7 +203,7 @@ func fetchFeedItems(ctx context.Context, feed *gofeed.Feed, itemsCap int) *gofee
 				_log.ErrorContext(ctx, "Error while creating a HTTP request for a feed item link", "err", err)
 				return
 			}
-			req.Header.Set("User-Agent", getRandUserAgent())
+			req.Header.Set("User-Agent", userAgent)
 
 			resp, err := client.Do(req)
 			if err != nil {
@@ -226,14 +237,20 @@ func fetchRSSData(ctx context.Context, params *plugin.RetrieveDataParams) (plugi
 	log := slog.Default()
 
 	fp := gofeed.NewParser()
-	fp.UserAgent = getRandUserAgent()
 
 	url := params.Args.GetAttrVal("url").AsString()
 
 	fillInContent := params.Args.GetAttrVal("fill_in_content").True()
+	useBrowserUserAgent := params.Args.GetAttrVal("use_browser_user_agent").True()
 	fillInMaxItems, _ := params.Args.GetAttrVal("fill_in_max_items").AsBigFloat().Int64()
 	onlyItemsAfterTimeAttr := params.Args.GetAttrVal("only_items_after_time")
 
+	userAgent := defaultUserAgent
+	if useBrowserUserAgent {
+		userAgent = getRandUserAgent()
+	}
+	fp.UserAgent = userAgent
+
 	basicAuth := params.Args.Blocks.GetFirstMatching("basic_auth")
 	if basicAuth != nil {
 		fp.AuthConfig = &gofeed.Auth{
@@ -283,7 +300,7 @@ func fetchRSSData(ctx context.Context, params *plugin.RetrieveDataParams) (plugi
 	}
 
 	if fillInContent {
-		feed = fetchFeedItems(ctx, feed, int(fillInMaxItems))
+		feed = fetchFeedItems(ctx, feed, userAgent, int(fillInMaxItems))
 		log.InfoContext(ctx, "The content for the feed items downloaded")
 	}