Tool calls support from mainline (#723)

* Tool calls support from mainline

* update cmake

* revert api for /completions

* Fix broken thinking process for gpt-oss

* add missing args and fix webui bugs

* add missing args and fix webui bugs2

* Fix reasoning format error

* add usage

* change default post_sampling_probs to true

* add back generated_text

* Remove server endpoints tests

* add log

* Chat fixes

* Remove logs

* webui: revert extra handling of thinking process

---------

Co-authored-by: firecoperana <firecoperana>
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>

examples/server/webui/src/utils/misc.ts

@@ -80,13 +80,22 @@ export function normalizeMsgsForAPI(messages: Readonly<Message[]>) {
  * recommended for DeepsSeek-R1, filter out content between <think> and </think> tags
  */
 export function filterThoughtFromMsgs(messages: APIMessage[]) {
+  console.debug({ messages });
   return messages.map((msg) => {
+    if (msg.role !== 'assistant') {
+      return msg;
+    }
+    // assistant message is always a string
+    const contentStr = msg.content as string;
     return {
       role: msg.role,
       content:
         msg.role === 'assistant'
-          ? msg.content.split('</think>').at(-1)!.trim()
-          : msg.content,
+          ? contentStr
+              .split(/<\/think>|<\|end\|>/)
+              .at(-1)!
+              .trim()
+          : contentStr,
     } as APIMessage;
   });
 }