tests: fix notebook tests

Author: baptiste-olivier

docs/sdk/tutorials/llm_dynamic.md

@@ -147,7 +147,7 @@ Now, we'll generate a conversation by providing a prompt.
 
 ```python
 conversation = kili.llm.create_conversation(
-    project_id=project_id, prompt="Give me Schrödinger equation."
+    project_id=project_id, initial_prompt="Give me Schrödinger equation."
 )
 ```
 

recipes/frame_dicom_data.ipynb

@@ -42,7 +42,6 @@
    "source": [
     "import os\n",
     "import shutil\n",
-    "import subprocess\n",
     "\n",
     "import requests\n",
     "import tqdm"
@@ -165,7 +164,10 @@
     "for asset_key in sorted_files.keys():\n",
     "    print(asset_key)\n",
     "    im = read_dcm_image(sorted_files[asset_key][20])\n",
-    "    im.save(f\"./recipes/img/frame_dicom_data_{asset_key}.png\")"
+    "    im_directory = os.path.dirname(os.path.abspath(\"\"))\n",
+    "    img_name = f\"frame_dicom_data_{asset_key}.png\"\n",
+    "    save_location = os.path.join(im_directory, \"img\", img_name)\n",
+    "    im.save(img_name)"
    ]
   },
   {
@@ -316,56 +318,15 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Finally, let's import the volumes using `appendManyToDataset` (see [link](https://staging.cloud.kili-technology.com/docs/python-graphql-api/python-api/#append_many_to_dataset)). The key argument is `json_content_array`, which is a list of list of strings. Each element is the list of urls or paths pointing to images of the volume considered.\n",
-    " - Let's host these images locally to demonstrate how we would do it with cloud URLs for example :"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "subprocess.Popen(\n",
-    "    f\"python -m http.server 8001 --directory {ASSET_ROOT}\",\n",
-    "    shell=True,\n",
-    "    stdin=None,\n",
-    "    stdout=None,\n",
-    "    stderr=None,\n",
-    "    close_fds=True,\n",
-    ")\n",
-    "ROOT_URL = \"http://localhost:8001/\""
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def files_to_urls(files):\n",
-    "    return list(map(lambda file: ROOT_URL + file.split(\"TCGA-LUAD\")[1], files))"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "kili.append_many_to_dataset(\n",
-    "    project_id=project_id,\n",
-    "    external_id_array=list(sorted_images.keys()),\n",
-    "    json_content_array=list(map(files_to_urls, sorted_images.values())),\n",
-    ")"
+    "Finally, let's import the volumes using `appendManyToDataset` (see [link](https://staging.cloud.kili-technology.com/docs/python-graphql-api/python-api/#append_many_to_dataset)). The key argument is `json_content_array`, which is a list of list of strings. Each element is the list of urls or paths pointing to images of the volume considered."
    ]
   },
   {
    "attachments": {},
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Or, as mentionned, you can simply provide the paths to your images, and call the function like below : "
+    "You can simply provide the paths to your images, and call the function like below : "
    ]
   },
   {
@@ -405,7 +366,7 @@
    "source": [
     "ds_size = kili.count_assets(project_id=project_id)\n",
     "print(ds_size)\n",
-    "assert ds_size == 6"
+    "assert ds_size == 3"
    ]
   },
   {

recipes/img/assets_inserted.png

@@ -237,7 +237,7 @@
    "outputs": [],
    "source": [
     "conversation = kili.llm.create_conversation(\n",
-    "    project_id=project_id, prompt=\"Give me Schrödinger equation.\"\n",
+    "    project_id=project_id, initial_prompt=\"Give me Schrödinger equation.\"\n",
     ")"
    ]
   },

recipes/img/frame_annotation.png

@@ -295,9 +295,6 @@ def _scale_normalized_vertices_image_video_annotation(
             " the resolution of your asset.`"
         )
 
-    width = asset["resolution"]["width"] if "resolution" in asset else 0
-    height = asset["resolution"]["height"] if "resolution" in asset else 0
-
     # bbox, segmentation, polygons
     if "boundingPoly" in annotation and normalized_vertices:
         annotation["boundingPoly"] = [
@@ -310,6 +307,9 @@ def _scale_normalized_vertices_image_video_annotation(
         ]
         return
 
+    width = asset["resolution"]["width"] if "resolution" in asset else 0
+    height = asset["resolution"]["height"] if "resolution" in asset else 0
+
     if "boundingPoly" in annotation and not normalized_vertices:
         annotation["boundingPoly"] = [
             {

recipes/llm_dynamic.ipynb

@@ -1,6 +1,7 @@
 """Module for managing bucket's signed urls."""
 
 import itertools
+import os
 from typing import List, Union
 from urllib.parse import parse_qs, urlparse
 
@@ -67,5 +68,15 @@ def clean_signed_url(url: str, endpoint: str) -> str:
     """Return a cleaned signed url for frame upload."""
     query = urlparse(url).query
     id_param = parse_qs(query)["id"][0]
-    base_path = endpoint.replace("/graphql", "/files").replace("http://", "https://")
+    # Check if Kili is using http or https
+    kili_path = os.getenv(
+        "KILI_API_ENDPOINT", "https://cloud.kili-technology.com/api/label/v2/graphql"
+    )
+    is_using_http = kili_path.startswith("http://")
+    if is_using_http:
+        base_path = endpoint.replace("https://", "http://")
+    else:
+        base_path = endpoint.replace("http://", "https://")
+
+    base_path = endpoint.replace("/graphql", "/files")
     return f"{base_path}?id={id_param}"