Improve handling of Instant View anchors.

GitOrigin-RevId: 9293ea4c1c7f4767c7f91421b3719e00d836470c

td/generate/scheme/td_api.tl

@@ -843,8 +843,14 @@ richTextPhoneNumber text:RichText phone_number:string = RichText;
 //@height Height of a bounding box in which the image should be shown; 0 if unknown
 richTextIcon document:document width:int32 height:int32 = RichText;
 
-//@description A rich text anchor @text Text @name Anchor name
+//@description A rich text reference of a text on the same web page @text The text @reference_text The text to show on click
-richTextAnchor text:RichText name:string = RichText;
+richTextReference text:RichText reference_text:RichText = RichText;
+
+//@description An anchor @name Anchor name
+richTextAnchor name:string = RichText;
+
+//@description A link to an anchor on the same web page @text The link text @name The anchor name
+richTextAnchorLink text:RichText name:string = RichText;
 
 //@description A concatenation of rich texts @texts Texts
 richTexts texts:vector<RichText> = RichText;
@@ -980,10 +986,9 @@ pageBlockMap location:location zoom:int32 width:int32 height:int32 caption:pageB
 
 //@description Describes an instant view page for a web page @page_blocks Content of the web page
 //@version Version of the instant view, currently can be 1 or 2
-//@url Instant view URL; may be different from WebPage.url and must be used for the correct anchors handling
 //@is_rtl True, if the instant view must be shown from right to left
 //@is_full True, if the instant view contains the full page. A network request might be needed to get the full web page instant view
-webPageInstantView page_blocks:vector<PageBlock> version:int32 url:string is_rtl:Bool is_full:Bool = WebPageInstantView;
+webPageInstantView page_blocks:vector<PageBlock> version:int32 is_rtl:Bool is_full:Bool = WebPageInstantView;
 
 
 //@description Describes a web page preview

td/telegram/WebPageBlock.cpp

@@ -38,8 +38,15 @@
 
 namespace td {
 
+class RichText;
+
 struct GetWebPageBlockObjectContext {
-  Td *td_;
+  Td *td_ = nullptr;
+  Slice base_url_;
+
+  bool is_first_pass_ = true;
+  bool has_anchor_urls_ = false;
+  std::unordered_map<Slice, const RichText *, SliceHash> anchors_;  // anchor -> text
 };
 
 static vector<td_api::object_ptr<td_api::PageBlock>> get_page_block_objects(
@@ -49,8 +56,6 @@ static vector<td_api::object_ptr<td_api::PageBlock>> get_page_block_objects(
   });
 }
 
-namespace {
-
 class RichText {
   static vector<td_api::object_ptr<td_api::RichText>> get_rich_text_objects(const vector<RichText> &rich_texts,
                                                                             GetWebPageBlockObjectContext *context) {
@@ -112,6 +117,23 @@ class RichText {
       case RichText::Type::Fixed:
         return make_tl_object<td_api::richTextFixed>(texts[0].get_rich_text_object(context));
       case RichText::Type::Url:
+        if (!context->base_url_.empty() && begins_with(content, context->base_url_) &&
+            content.size() > context->base_url_.size() + 1 && content[context->base_url_.size()] == '#') {
+          if (context->is_first_pass_) {
+            context->has_anchor_urls_ = true;
+          } else {
+            auto anchor = Slice(content).substr(context->base_url_.size() + 1);
+            auto it = context->anchors_.find(anchor);
+            if (it != context->anchors_.end()) {
+              if (it->second == nullptr) {
+                return make_tl_object<td_api::richTextAnchorLink>(texts[0].get_rich_text_object(context), anchor.str());
+              } else {
+                return make_tl_object<td_api::richTextReference>(texts[0].get_rich_text_object(context),
+                                                                 it->second->get_rich_text_object(context));
+              }
+            }
+          }
+        }
         return make_tl_object<td_api::richTextUrl>(texts[0].get_rich_text_object(context), content,
                                                    web_page_id.is_valid());
       case RichText::Type::EmailAddress:
@@ -134,7 +156,13 @@ class RichText {
             context->td_->documents_manager_->get_document_object(document_file_id), width, height);
       }
       case RichText::Type::Anchor:
-        return make_tl_object<td_api::richTextAnchor>(texts[0].get_rich_text_object(context), content);
+        if (context->is_first_pass_) {
+          context->anchors_.emplace(Slice(content), texts[0].empty() ? nullptr : &texts[0]);
+        }
+        if (texts[0].empty()) {
+          return make_tl_object<td_api::richTextAnchor>(content);
+        }
+        return texts[0].get_rich_text_object(context);
     }
     UNREACHABLE();
     return nullptr;
@@ -177,6 +205,8 @@ class RichText {
   }
 };
 
+namespace {
+
 class WebPageBlockCaption {
  public:
   RichText text;
@@ -740,6 +770,9 @@ class WebPageBlockAnchor : public WebPageBlock {
   }
 
   td_api::object_ptr<td_api::PageBlock> get_page_block_object(Context *context) const override {
+    if (context->is_first_pass_) {
+      context->anchors_.emplace(name, nullptr);
+    }
     return make_tl_object<td_api::pageBlockAnchor>(name);
   }
 
@@ -2348,9 +2381,16 @@ vector<unique_ptr<WebPageBlock>> get_web_page_blocks(
 }
 
 vector<td_api::object_ptr<td_api::PageBlock>> get_page_block_objects(
-    const vector<unique_ptr<WebPageBlock>> &page_blocks, Td *td) {
+    const vector<unique_ptr<WebPageBlock>> &page_blocks, Td *td, Slice base_url) {
   GetWebPageBlockObjectContext context;
   context.td_ = td;
+  context.base_url_ = base_url;
+  auto blocks = get_page_block_objects(page_blocks, &context);
+  if (context.anchors_.empty() || !context.has_anchor_urls_) {
+    return blocks;
+  }
+
+  context.is_first_pass_ = false;
   return get_page_block_objects(page_blocks, &context);
 }
 

td/telegram/WebPageBlock.h

@@ -14,6 +14,7 @@
 #include "td/telegram/Photo.h"
 
 #include "td/utils/common.h"
+#include "td/utils/Slice.h"
 
 #include <unordered_map>
 
@@ -103,6 +104,6 @@ vector<unique_ptr<WebPageBlock>> get_web_page_blocks(
     const std::unordered_map<int64, FileId> &videos, const std::unordered_map<int64, FileId> &voice_notes);
 
 vector<td_api::object_ptr<td_api::PageBlock>> get_page_block_objects(
-    const vector<unique_ptr<WebPageBlock>> &page_blocks, Td *td);
+    const vector<unique_ptr<WebPageBlock>> &page_blocks, Td *td, Slice base_url);
 
 }  // namespace td

td/telegram/WebPagesManager.cpp

@@ -1226,9 +1226,9 @@ tl_object_ptr<td_api::webPageInstantView> WebPagesManager::get_web_page_instant_
     LOG(ERROR) << "Trying to get not loaded web page instant view";
     return nullptr;
   }
-  return make_tl_object<td_api::webPageInstantView>(get_page_block_objects(web_page_instant_view->page_blocks, td_),
+  return td_api::make_object<td_api::webPageInstantView>(
-                                                    web_page_instant_view->is_v2 ? 2 : 1, web_page_instant_view->url,
+      get_page_block_objects(web_page_instant_view->page_blocks, td_, web_page_instant_view->url),
-                                                    web_page_instant_view->is_rtl, web_page_instant_view->is_full);
+      web_page_instant_view->is_v2 ? 2 : 1, web_page_instant_view->is_rtl, web_page_instant_view->is_full);
 }
 
 void WebPagesManager::on_web_page_changed(WebPageId web_page_id, bool have_web_page) {