Merge pull request #127 from anarkiwi/path

Image inference outputs image path, and handles multiline JSON from torchserve.

lib/image_inference_impl.cc

@@ -326,7 +326,8 @@ void image_inference_impl::output_image_() {
   std::stringstream ss("", std::ios_base::app | std::ios_base::out);
   ss << "{"
      << "\"ts\": " << host_now_str_(output_item.ts)
-     << ", \"rx_freq\": " << output_item.rx_freq;
+     << ", \"rx_freq\": " << output_item.rx_freq << ", \"image_path\": \""
+     << full_image_file_png << "\"";
   // TODO: synchronous requests for testing. Should be parallel.
   if (host_.size() && port_.size()) {
     boost::asio::io_context ioc;
@@ -357,7 +358,9 @@ void image_inference_impl::output_image_() {
       ss << ", \"error\": \"" << ex.what() << "\"";
     }
   }
-  ss << "}" << std::endl;
+  // double new line to faciliate json parsing, since prediction may contain new
+  // lines.
+  ss << "}\n" << std::endl;
   const std::string s = ss.str();
   out_buf_.insert(out_buf_.end(), s.begin(), s.end());
   delete_output_();

python/iqtlabs/qa_image_inference.py

@@ -241,8 +241,8 @@ def simulate_torchserve(self, port, model_name, result):
 
         # nosemgrep:github.workflows.config.useless-inner-function
         @app.route(f"/predictions/{model_name}", methods=["POST"])
-        def predictions_test():  
-            return json.dumps(result), 200
+        def predictions_test():
+            return json.dumps(result, indent=2), 200
 
         try:
             app.run(host="127.0.0.1", port=11001)
@@ -252,7 +252,7 @@ def predictions_test():
     def test_instance(self):
         port = 11001
         model_name = "testmodel"
-        predictions_result = {"modulation": 999}
+        predictions_result = {"modulation": [{"conf": 0.9, "xywh": [1, 2, 3, 4]}]}
         if self.pid == 0:
             self.simulate_torchserve(port, model_name, predictions_result)
             return
@@ -312,9 +312,15 @@ def test_instance(self):
                 self.assertEqual(imghdr.what(image_file), "png")
             self.assertTrue(os.stat(test_file).st_size)
             with open(test_file) as f:
-                for line in f.readlines():
-                    result = json.loads(line)
-                    self.assertEqual(result["predictions"], predictions_result)
+                content = f.read()
+            json_raw_all = content.split("\n\n")
+            self.assertTrue(json_raw_all)
+            for json_raw in json_raw_all:
+                if not json_raw:
+                    continue
+                result = json.loads(json_raw)
+                self.assertTrue(os.path.exists(result["image_path"]))
+                self.assertEqual(result["predictions"], predictions_result)
 
 
 if __name__ == "__main__":