[Other] Use symbolic link pointed to TensorRT library (#1461)

* Add force rpath

* update to dev-6

* remove force-rpath

* remove nvinfer so

* Add symbol link creation when import fastdeploy first

* Add dst

* update symbol link creation

* Add logging

* logging -> logger

* update tips

* Set logging level

* Add --force-rpath

python/fastdeploy/__init__.py

@@ -16,6 +16,32 @@ import logging
 import os
 import sys
 
+# Create a symbol link to tensorrt library.
+trt_directory = os.path.join(
+    os.path.dirname(os.path.abspath(__file__)),
+    "libs/third_libs/tensorrt/lib/")
+if os.name != "nt" and os.path.exists(trt_directory):
+    logging.basicConfig(level=logging.INFO)
+    for trt_lib in [
+            "libnvcaffe_parser.so", "libnvinfer_plugin.so", "libnvinfer.so",
+            "libnvonnxparser.so", "libnvparsers.so"
+    ]:
+        dst = os.path.join(trt_directory, trt_lib)
+        src = os.path.join(trt_directory, trt_lib + ".8")
+        if not os.path.exists(dst):
+            try:
+                os.symlink(src, dst)
+                logging.info(
+                    f"Create a symbolic link pointing to {src} named {dst}.")
+            except OSError as e:
+                logging.warning(
+                    f"Failed to create a symbolic link pointing to {src} by an unprivileged user. "
+                    "It may failed when you use Paddle TensorRT backend. "
+                    "Please use administator privilege to import fastdeploy at first time."
+                )
+                break
+    logging.basicConfig(level=logging.NOTSET)
+
 # Note(zhoushunjie): Fix the import order of paddle and fastdeploy library.
 # This solution will be removed it when the confilct of paddle and
 # fastdeploy is fixed.