Fix and enhance handling of Bluesky facet links

core/include/fastsm/models/status.hpp

@@ -20,6 +20,13 @@ struct StatusFilterMatch {
     bool hide = false;  // filter_action: true = "hide", false = "warn"
 };
 
+// A semantically linked span in a post's plain text. Bluesky carries these as
+// rich-text facets, where the visible label may differ from the destination.
+struct StatusTextLink {
+    std::string text;
+    std::string url;
+};
+
 // A post/status. Boosts and quotes reference another Status via shared_ptr to
 // break the recursion (the Swift core uses a boxed indirect enum).
 struct Status {
@@ -51,6 +58,7 @@ struct Status {
     std::shared_ptr<Status> quote;  // the quoted status
     std::vector<MediaAttachment> media_attachments;
     std::vector<Mention> mentions;
+    std::vector<StatusTextLink> text_links;
     std::vector<std::string> tags; // hashtag names in this post (no '#'), Mastodon
     std::optional<Visibility> visibility; // Mastodon only
     std::optional<std::string> spoiler_text;

core/src/platform/bluesky/bluesky_map.cpp

@@ -78,7 +78,7 @@ std::shared_ptr<Status> map_quote_record(const json& rec) {
     return std::make_shared<Status>(std::move(q));
 }
 
-// Populate mentions/tags from a post record's richtext `facets` (byte-range
+// Populate links/mentions/tags from a post record's richtext `facets` (byte-range
 // annotations over the plain text). Mention facets carry the DID; the handle is
 // the sliced display text ("@handle"). Tag facets carry the bare tag.
 void map_facets(const json& record, Status& s) {
@@ -113,6 +113,10 @@ void map_facets(const json& record, Status& s) {
                     tag = slice.substr(1);
                 if (!tag.empty())
                     s.tags.push_back(std::move(tag));
+            } else if (type == "app.bsky.richtext.facet#link") {
+                std::string uri = str(feat, "uri");
+                if (!uri.empty())
+                    s.text_links.push_back({slice, std::move(uri)});
             }
         }
     }
@@ -262,6 +266,7 @@ Notification map_notification(const json& j) {
         if (const json* record = obj(j, "record")) {
             s.text = str(*record, "text");
             s.created_at = util::parse_iso8601(str(*record, "createdAt")).value_or(n.created_at);
+            map_facets(*record, s);
         }
         if (s.created_at == 0)
             s.created_at = n.created_at;

core/src/presentation/status_presenter.cpp

@@ -1,5 +1,6 @@
 #include "fastsm/presentation/status_presenter.hpp"
 
+#include <algorithm>
 #include <cctype>
 #include <cstring>
 #include <string>
@@ -538,8 +539,11 @@ std::vector<PostLink> post_links(const Status& status) {
     // Links embedded in the text. A Mastodon post carries HTML in `content`; a
     // Bluesky post has none, so fall back to scanning its plain `text`.
     std::vector<std::pair<std::string, std::string>> text_links;
+    for (const auto& link : s.text_links)
+        text_links.push_back({link.text, link.url});
+    const size_t facet_link_count = text_links.size();
     anchors(s.content, text_links);
-    if (text_links.empty()) {
+    if (text_links.size() == facet_link_count) {
         std::vector<std::string> urls;
         find_urls_in_text(s.text, urls);
         for (const auto& u : urls)
@@ -573,13 +577,20 @@ std::vector<PostLink> post_links(const Status& status) {
 std::vector<std::string> post_text_link_urls(const Status& status) {
     const Status& s = status.display_status(); // unwrap a boost
     std::vector<std::pair<std::string, std::string>> text_links;
+    for (const auto& link : s.text_links)
+        text_links.push_back({link.text, link.url});
+    const size_t facet_link_count = text_links.size();
     anchors(s.content, text_links); // HTML anchors (skips @mention / #hashtag)
     std::vector<std::string> out;
-    if (!text_links.empty()) {
-        for (const auto& [text, url] : text_links)
+    if (text_links.size() == facet_link_count) {
+        std::vector<std::string> literal_urls;
+        find_urls_in_text(s.text, literal_urls);
+        for (const auto& url : literal_urls)
+            text_links.push_back({std::string{}, url});
+    }
+    for (const auto& [text, url] : text_links) {
+        if (std::find(out.begin(), out.end(), url) == out.end())
             out.push_back(url);
-    } else {
-        find_urls_in_text(s.text, out); // Bluesky: URLs live in plain text
     }
     return out;
 }

core/src/store/timeline_codec.cpp

@@ -10,7 +10,8 @@ namespace {
 // cleanly (a magic mismatch -> empty) instead of being read with a mismatched
 // reader. v2 added Status::url. v6 added Notification group_key + notifications_count.
 // v7 added Status::filtered + tags. v9 added Bluesky reply-parent metadata.
-constexpr char kMagic[4] = {'F', 'S', 'C', 'A'};
+// v11 added Status::text_links.
+constexpr char kMagic[4] = {'F', 'S', 'C', 'B'};
 // Guard against runaway recursion if a file is ever corrupt/misaligned: boost/
 // quote nesting is shallow in practice.
 constexpr int kMaxStatusDepth = 24;
@@ -289,6 +290,11 @@ void write_status(Writer& w, const Status& s) {
     w.u32(static_cast<std::uint32_t>(s.tags.size()));
     for (const auto& t : s.tags)
         w.str(t); // else "Follow hashtag" pre-fills blank for cached posts
+    w.u32(static_cast<std::uint32_t>(s.text_links.size()));
+    for (const auto& link : s.text_links) {
+        w.str(link.text);
+        w.str(link.url);
+    }
 }
 
 Status read_status(Reader& r, int depth = 0) {
@@ -359,6 +365,9 @@ Status read_status(Reader& r, int depth = 0) {
     const std::uint32_t tag_n = r.u32();
     for (std::uint32_t i = 0; i < tag_n && r.ok; ++i)
         s.tags.push_back(r.str());
+    const std::uint32_t link_n = r.u32();
+    for (std::uint32_t i = 0; i < link_n && r.ok; ++i)
+        s.text_links.push_back({r.str(), r.str()});
     return s;
 }
 

docs/changelog.txt

@@ -7,6 +7,7 @@ FastSMRW changelog
 - New: send someone a direct message from their profile, from the post menu, or with the D key — it opens a new post already addressed to them and set to go only to them (Mastodon).
 - Fixed: on the Mac and iOS, the cursor now starts at the end of a post you are editing instead of in front of it.
 - Fixed: the ages shown on posts (like "2m" or "1h") now keep counting up while you sit on a timeline, instead of freezing until something new arrives.
+- Fixed: Open Links now finds Bluesky links hidden behind descriptive text, not just web addresses written out in the post.
 
 0.5.5
 -----

tests/test_bluesky_map.cpp

@@ -142,43 +142,64 @@ void test_bluesky_notification_mapping() {
     CHECK_EQ(like.account.display_name, std::string("Dana"));
     CHECK(like.status == nullptr); // like/repost carry no incoming post
 
-    // A reply notification reads as a Mention and carries the incoming post text.
+    // A reply notification reads as a Mention and carries the incoming post text
+    // and rich-text link facets.
     const char* kReply = R"JSON({
       "uri": "at://did:plc:eve/app.bsky.feed.post/r1", "cid": "c2",
       "author": { "did": "did:plc:eve", "handle": "eve.test", "displayName": "Eve" },
       "reason": "reply",
-      "record": { "text": "@me hi there", "createdAt": "2024-06-28T13:00:00.000Z" },
+      "record": {
+        "text": "@me read this",
+        "createdAt": "2024-06-28T13:00:00.000Z",
+        "facets": [
+          { "index": { "byteStart": 9, "byteEnd": 13 },
+            "features": [ { "$type": "app.bsky.richtext.facet#link", "uri": "https://example.com/notification" } ] }
+        ]
+      },
       "indexedAt": "2024-06-28T13:00:01.000Z"
     })JSON";
     const Notification reply = bluesky::map_notification(json::parse(kReply));
     CHECK(reply.type == Notification::Kind::Mention); // reply/quote read as mention
     CHECK(reply.status != nullptr);
     if (reply.status) {
-        CHECK_EQ(reply.status->text, std::string("@me hi there"));
+        CHECK_EQ(reply.status->text, std::string("@me read this"));
         CHECK_EQ(reply.status->id, std::string("at://did:plc:eve/app.bsky.feed.post/r1"));
+        CHECK_EQ(reply.status->text_links.size(), size_t(1));
+        if (!reply.status->text_links.empty()) {
+            CHECK_EQ(reply.status->text_links[0].text, std::string("this"));
+            CHECK_EQ(reply.status->text_links[0].url,
+                     std::string("https://example.com/notification"));
+        }
     }
 }
 
 void test_bluesky_facet_mapping() {
-    // A post whose record carries mention + tag facets populates mentions/tags.
+    // A post whose record carries link + mention + tag facets preserves all three.
     const char* kFaceted = R"JSON({
       "post": {
         "uri": "at://did:plc:x/app.bsky.feed.post/2", "cid": "cid2",
         "author": { "did": "did:plc:x", "handle": "bob.test", "displayName": "Bob" },
         "record": {
-          "text": "hi @alice.bsky.social #a11y",
+          "text": "😀 read this hi @alice.bsky.social #a11y",
           "createdAt": "2024-06-28T10:00:00Z",
           "facets": [
-            { "index": { "byteStart": 3, "byteEnd": 21 },
+            { "index": { "byteStart": 10, "byteEnd": 14 },
+              "features": [ { "$type": "app.bsky.richtext.facet#link", "uri": "https://example.com/article" } ] },
+            { "index": { "byteStart": 18, "byteEnd": 36 },
               "features": [ { "$type": "app.bsky.richtext.facet#mention", "did": "did:plc:alice" } ] },
-            { "index": { "byteStart": 22, "byteEnd": 27 },
+            { "index": { "byteStart": 37, "byteEnd": 42 },
               "features": [ { "$type": "app.bsky.richtext.facet#tag", "tag": "a11y" } ] }
           ]
         },
         "likeCount": 0, "repostCount": 0, "replyCount": 0
       }
     })JSON";
     const Status s = bluesky::map_feed_item(json::parse(kFaceted));
+    CHECK_EQ(s.text_links.size(), size_t(1));
+    if (!s.text_links.empty()) {
+        CHECK_EQ(s.text_links[0].text, std::string("this"));
+        CHECK_EQ(s.text_links[0].url, std::string("https://example.com/article"));
+    }
     CHECK_EQ(s.mentions.size(), size_t(1));
     if (!s.mentions.empty()) {
         CHECK_EQ(s.mentions[0].id, std::string("did:plc:alice"));

tests/test_models.cpp

@@ -31,6 +31,7 @@ static Status sample_inner() {
     s.platform = Platform::Mastodon;
     s.tags.push_back("cats");
     s.tags.push_back("welcome");
+    s.text_links.push_back({"welcome", "https://example.com/article"});
     s.filtered.push_back({"Spoilers", true}); // a "hide" filter match
     s.filtered.push_back({"Politics", false}); // a "warn" filter match
 
@@ -85,6 +86,9 @@ void test_status_roundtrip() {
     CHECK(back.reblog->filtered[1].hide == false);
     CHECK_EQ(back.reblog->tags.size(), size_t(2));
     CHECK_EQ(back.reblog->tags[1], std::string("welcome"));
+    CHECK_EQ(back.reblog->text_links.size(), size_t(1));
+    CHECK_EQ(back.reblog->text_links[0].text, std::string("welcome"));
+    CHECK_EQ(back.reblog->text_links[0].url, std::string("https://example.com/article"));
     CHECK(back.reblog->visibility.value() == Visibility::Unlisted);
     CHECK_EQ(back.reblog->media_attachments.size(), size_t(1));
     CHECK(back.reblog->media_attachments[0].type == MediaAttachment::Kind::Image);

tests/test_presentation.cpp

@@ -314,6 +314,22 @@ void test_post_links() {
     CHECK_EQ(links[3].url, std::string("https://x.social/@me/123"));
     CHECK_EQ(links[3].title, std::string("Open this post in browser"));
 
+    // Mastodon HTML is authoritative: URL-looking anchor text must not become a
+    // second destination distinct from the anchor's href.
+    Status deceptive_anchor;
+    deceptive_anchor.content =
+        "<a href=\"https://actual.example\">https://displayed.example</a>";
+    deceptive_anchor.text = "https://displayed.example";
+    const std::vector<present::PostLink> anchor_links = present::post_links(deceptive_anchor);
+    CHECK_EQ(anchor_links.size(), static_cast<size_t>(1));
+    if (!anchor_links.empty())
+        CHECK_EQ(anchor_links[0].url, std::string("https://actual.example"));
+    const std::vector<std::string> anchor_urls =
+        present::post_text_link_urls(deceptive_anchor);
+    CHECK_EQ(anchor_urls.size(), static_cast<size_t>(1));
+    if (!anchor_urls.empty())
+        CHECK_EQ(anchor_urls[0], std::string("https://actual.example"));
+
     // A boost unwraps to the boosted post's links.
     Status boost;
     boost.reblog = std::make_shared<Status>(s);
@@ -328,6 +344,23 @@ void test_post_links() {
     CHECK_EQ(bl[0].url, std::string("https://bsky.example/x"));
     CHECK_EQ(bl[1].title, std::string("Open this post in browser"));
 
+    // Bluesky facets can link ordinary display text whose destination is not
+    // visible in the post body.
+    Status faceted;
+    faceted.text = "read this article, https://other.example/page and https://example.com/article";
+    faceted.text_links.push_back({"this article", "https://example.com/article"});
+    std::vector<present::PostLink> fl = present::post_links(faceted);
+    CHECK_EQ(fl.size(), static_cast<size_t>(2));
+    CHECK_EQ(fl[0].title, std::string("this article"));
+    CHECK_EQ(fl[0].url, std::string("https://example.com/article"));
+    CHECK_EQ(fl[1].url, std::string("https://other.example/page"));
+    const std::vector<std::string> facet_urls = present::post_text_link_urls(faceted);
+    CHECK_EQ(facet_urls.size(), static_cast<size_t>(2));
+    if (facet_urls.size() == 2) {
+        CHECK_EQ(facet_urls[0], std::string("https://example.com/article"));
+        CHECK_EQ(facet_urls[1], std::string("https://other.example/page"));
+    }
+
     // A post with only its own URL still offers that one link.
     Status plain;
     plain.text = "just some text, no links here.";