feat: expose url field for link/appmsg messages (#18)

* feat: expose url field for link/appmsg messages

Extract <url> from appmsg XML in type-49 messages and append it as
a 'url' field in history/search output. The field is omitted when
the message has no valid URL (non-link types, empty, non-http).

* fix: normalize appmsg urls across query outputs

---------

Co-authored-by: tsinghu <tsinghu@tencent.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>

src/daemon/query.rs

@@ -549,15 +549,20 @@ fn query_messages(
         let content = decompress_message(&content_bytes, ct);
         let sender = sender_label(real_sender_id, &content, is_group, chat_username, &id2u, names_map, group_nicknames);
         let text = fmt_content(local_id, local_type, &content, is_group);
+        let url = appmsg_url_for_message(local_type, &content);
 
-        result.push(json!({
+        let mut msg = json!({
             "timestamp": ts,
             "time": fmt_time(ts, "%Y-%m-%d %H:%M"),
             "sender": sender,
             "content": text,
             "type": fmt_type(local_type),
             "local_id": local_id,
-        }));
+        });
+        if let Some(u) = url {
+            msg["url"] = serde_json::Value::String(u);
+        }
+        result.push(msg);
     }
     Ok(result)
 }
@@ -636,15 +641,20 @@ fn search_in_table(
         if search_decoded_content && !matches_search_text(&content, &text, keyword, &keyword_lower) {
             continue;
         }
+        let url = appmsg_url_for_message(local_type, &content);
 
-        result.push(json!({
+        let mut msg = json!({
             "timestamp": ts,
             "time": fmt_time(ts, "%Y-%m-%d %H:%M"),
             "chat": "",
             "sender": sender,
             "content": text,
             "type": fmt_type(local_type),
-        }));
+        });
+        if let Some(u) = url {
+            msg["url"] = serde_json::Value::String(u);
+        }
+        result.push(msg);
         if search_decoded_content && result.len() >= limit {
             break;
         }
@@ -1273,6 +1283,37 @@ fn extract_xml_text(xml: &str, tag: &str) -> Option<String> {
     Some(xml[content_start..content_start + end].trim().to_string())
 }
 
+fn appmsg_url_for_message(local_type: i64, content: &str) -> Option<String> {
+    if (local_type as u64 & 0xFFFFFFFF) != 49 {
+        return None;
+    }
+    extract_appmsg_url(content)
+}
+
+fn strip_xml_cdata(s: &str) -> &str {
+    s.strip_prefix("<![CDATA[")
+        .and_then(|inner| inner.strip_suffix("]]>"))
+        .unwrap_or(s)
+}
+
+/// 从 appmsg XML 中提取链接 URL（优先取 <url>，fallback 到 <url1>）
+fn extract_appmsg_url(text: &str) -> Option<String> {
+    let xml = strip_group_prefix(text);
+    if !xml.contains("<appmsg") {
+        return None;
+    }
+    if extract_xml_text(&xml, "type").as_deref() == Some("57") {
+        return None;
+    }
+    let url = extract_xml_text(&xml, "url")
+        .or_else(|| extract_xml_text(&xml, "url1"))
+        .map(|s| unescape_html(strip_xml_cdata(&s)))?;
+    if url.is_empty() || !(url.starts_with("http://") || url.starts_with("https://")) {
+        return None;
+    }
+    Some(url)
+}
+
 fn extract_xml_attr(xml: &str, tag: &str, attr: &str) -> Option<String> {
     let open = format!("<{}", tag);
     let start = xml.find(&open)?;
@@ -1906,7 +1947,8 @@ pub async fn q_new_messages(
                     let content = decompress_message(&content_bytes, ct);
                     let sender = sender_label(real_sender_id, &content, is_group, &uname2, &id2u, &names_map, &group_nicknames2);
                     let text = fmt_content(local_id, local_type, &content, is_group);
-                    result.push(json!({
+                    let url = appmsg_url_for_message(local_type, &content);
+                    let mut msg = json!({
                         "chat": display2,
                         "username": uname2,
                         "is_group": is_group,
@@ -1916,7 +1958,11 @@ pub async fn q_new_messages(
                         "sender": sender,
                         "content": text,
                         "type": fmt_type(local_type),
-                    }));
+                    });
+                    if let Some(u) = url {
+                        msg["url"] = serde_json::Value::String(u);
+                    }
+                    result.push(msg);
                 }
                 Ok::<_, anyhow::Error>(result)
             }).await {
@@ -2926,6 +2972,71 @@ mod sns_tests {
         assert_eq!(escape_like_pattern(""), "");
     }
 
+    #[test]
+    fn extract_appmsg_url_unescapes_html_entities() {
+        let xml = concat!(
+            "<appmsg>",
+            "<type>5</type>",
+            "<url>https://mp.weixin.qq.com/s?__biz=MzI4&amp;mid=2247&amp;idx=1</url>",
+            "</appmsg>"
+        );
+        assert_eq!(
+            extract_appmsg_url(xml).as_deref(),
+            Some("https://mp.weixin.qq.com/s?__biz=MzI4&mid=2247&idx=1")
+        );
+    }
+
+    #[test]
+    fn extract_appmsg_url_strips_group_prefix_and_cdata() {
+        let xml = concat!(
+            "wxid_sender:\n",
+            "<appmsg>",
+            "<type>5</type>",
+            "<url><![CDATA[https://example.com/x?a=1&b=2]]></url>",
+            "</appmsg>"
+        );
+        assert_eq!(
+            extract_appmsg_url(xml).as_deref(),
+            Some("https://example.com/x?a=1&b=2")
+        );
+    }
+
+    #[test]
+    fn extract_appmsg_url_falls_back_to_url1() {
+        let xml = concat!(
+            "<appmsg>",
+            "<type>5</type>",
+            "<url1>https://example.com/fallback</url1>",
+            "</appmsg>"
+        );
+        assert_eq!(
+            extract_appmsg_url(xml).as_deref(),
+            Some("https://example.com/fallback")
+        );
+    }
+
+    #[test]
+    fn extract_appmsg_url_ignores_non_http_values() {
+        let xml = concat!(
+            "<appmsg>",
+            "<type>5</type>",
+            "<url>weixin://bizmsgmenu?msgmenucontent=foo</url>",
+            "</appmsg>"
+        );
+        assert_eq!(extract_appmsg_url(xml), None);
+    }
+
+    #[test]
+    fn extract_appmsg_url_ignores_refermsg() {
+        let xml = concat!(
+            "<appmsg>",
+            "<type>57</type>",
+            "<url>https://example.com/nested</url>",
+            "</appmsg>"
+        );
+        assert_eq!(extract_appmsg_url(xml), None);
+    }
+
     fn media_object(value: &Value) -> &serde_json::Map<String, Value> {
         value.as_object().expect("media entry should be an object")
     }