add unit tests for edge cases and errors in responseStreamer

Author: ccreutzi

+llms/+stream/responseStreamer.m

@@ -25,11 +25,16 @@
             end
         end
     end
-    
+
     methods
         function [len,stop] = putData(this, data)
             [len,stop] = this.putData@matlab.net.http.io.BinaryConsumer(data);
-            
+            stop = doPutData(this, data, stop);
+        end
+    end
+
+    methods (Access=?tresponseStreamer)
+        function stop = doPutData(this, data, stop)
             % Extract out the response text from the message
             str = native2unicode(data','UTF-8');
             str = this.Incomplete + string(str);
@@ -88,8 +93,13 @@
                         end
                     else
                         txt = json.message.content;
-                        this.StreamFun(txt);
-                        this.ResponseText = [this.ResponseText txt];
+                        if strlength(txt) > 0
+                            this.StreamFun(txt);
+                            this.ResponseText = [this.ResponseText txt];
+                        end
+                        if isfield(json,"done")
+                            stop = json.done;
+                        end
                     end
                 end
             end

tests/tresponseStreamer.m

@@ -0,0 +1,73 @@
+classdef tresponseStreamer < matlab.unittest.TestCase
+% Tests for llms.stream.reponseStreamer
+%
+% This test file contains unit tests, with a specific focus on edge cases that
+% are hard to trigger in end-to-end tests.
+
+%   Copyright 2024 The MathWorks, Inc.
+
+    methods (Test)
+        function singleResponse(testCase)
+            s = tracingStreamer;
+            inp = 'data: {"choices":[{"content_filter_results":{},"delta":{"content":"foo","role":"assistant"}}]}';
+            inp = [inp newline 'data: [DONE]'];
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyTrue(s.doPutData(inp,false));
+            testCase.verifyEqual(s.StreamFun(),"foo");
+        end
+
+        function skipEmpty(testCase)
+            s = tracingStreamer;
+            inp = [...
+                'data: {"choices":[{"content_filter_results":{},"delta":{"content":"foo","role":"assistant"}}]}' newline ...
+                'data: {"choices":[]}' newline ...
+                'data: [DONE]'];
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyTrue(s.doPutData(inp,false));
+            testCase.verifyEqual(s.StreamFun(),"foo");
+        end
+
+        function splitResponse(testCase)
+            % it can happen that the server sends packets split in the
+            % middle of a JSON object. Hard to trigger on purpose.
+            s = tracingStreamer;
+            inp = 'data: {"choices":[{"content_filter_results":{},"delta":{"content":"foo","role":"assistant"}}]}';
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyFalse(s.doPutData(inp(1:42),false));
+            testCase.verifyFalse(s.doPutData(inp(43:end),false));
+            testCase.verifyEqual(s.StreamFun(),"foo");
+        end
+
+        function ollamaFormat(testCase)
+            s = tracingStreamer;
+            inp = '{"model":"mistral","created_at":"2024-06-07T07:43:30.658793Z","message":{"role":"assistant","content":" Hello"},"done":false}';
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyFalse(s.doPutData(inp,false));
+            inp = '{"model":"mistral","created_at":"2024-06-07T07:43:30.658793Z","message":{"role":"assistant","content":" World"},"done":true}';
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyTrue(s.doPutData(inp,false));
+            testCase.verifyEqual(s.StreamFun(),[" Hello"," World"]);
+        end
+
+        function badJSON(testCase)
+            s = tracingStreamer;
+            inp = 'data: {"choices":[{"content_filter_results":{};"delta":{"content":"foo","role":"assistant"}}]}';
+            inp = [inp newline inp];
+            inp = unicode2native(inp,"UTF-8").';
+            testCase.verifyError(@() s.doPutData(inp,false),'llms:stream:responseStreamer:InvalidInput');
+            testCase.verifyEmpty(s.StreamFun());
+        end
+    end
+end
+
+function s = tracingStreamer
+    data = strings(1, 0);
+    function seen = sf(str)
+        % Append streamed text to an empty string array of length 1
+        if nargin > 0
+            data = [data, str];
+        end
+        seen = data;
+    end
+    s = llms.stream.responseStreamer(@sf);
+end