Merge pull request #28 from biosimulations/show-saved-plots

show saved plots from simulators in notebook

Author: jcschaff

biosim_client/verify/dataset_comparison.py

@@ -1,8 +1,15 @@
+from io import BytesIO
 from typing import TypeAlias
+from zipfile import ZIP_DEFLATED, ZipFile
 
+import fitz  # type: ignore
 import matplotlib.pyplot as plt
 import numpy as np
+import PIL
+import urllib3
 from IPython.display import Markdown, display
+from PIL.Image import Image
+from PIL.PngImagePlugin import PngImageFile
 
 from biosim_client.api.biosim.models.biosim_simulation_run import BiosimSimulationRun
 from biosim_client.api.biosim.models.biosimulator_version import BiosimulatorVersion
@@ -229,6 +236,81 @@ def _create_markdown_report(self) -> str:
                 simulator_markdown += f"- no match for {run_link_map[run_group[0].id]} (closest score {min_score})\n"
         return simulator_markdown
 
+    def show_saved_plots(self, max_columns: int = 3, width_in: float = 10) -> None:
+        images: list[Image] = []
+        image_run_ids = []
+        image_not_found_run_ids = []
+        sim_version_names = self.simulator_version_names
+        http = urllib3.PoolManager()
+        for run_id in self.run_ids:
+            pdf_images: list[PngImageFile] = self._retrieve_pdf_as_image(run_id=run_id, http=http)
+            if len(pdf_images) == 0:
+                image_not_found_run_ids.append(run_id)
+            elif len(pdf_images) == 1:
+                images.append(pdf_images[0])
+                image_run_ids.append(run_id)
+            else:
+                # concatenate all images in pdf_images into a single image
+                width = max([image.width for image in pdf_images])
+                height = sum([image.height for image in pdf_images])
+                new_image = PIL.Image.new("RGB", (width, height))
+                y_offset = 0
+                for image in pdf_images:
+                    new_image.paste(image, (0, y_offset))
+                    y_offset += image.height
+                images.append(new_image)
+                # images.append(ImageFile(new_image.tobytes("png")))
+                image_run_ids.append(run_id)
+
+        if len(images) == 0:
+            display(f"no saved plots for runs {image_not_found_run_ids}")  # type: ignore
+            return
+
+        # determine number of columns and rows, start with max_columns and reduce to two columns
+        # until the number of images is divisible (or almost divisible) by number of columns
+        ncols = min(len(images), max_columns)
+        while len(images) % ncols != 0 and len(images) % ncols != (ncols - 1) and ncols > 2:
+            ncols -= 1
+        nrows = len(images) // ncols + 1
+
+        width_pixels = max([image.width for image in images]) * ncols
+        height_pixels = max([image.height for image in images]) * nrows
+        height_in = width_in * height_pixels / width_pixels
+        figsize = (width_in, height_in)
+        fig, ax = plt.subplots(ncols=ncols, nrows=nrows, figsize=figsize)
+        ax = ax.flatten()
+        for i in range(nrows * ncols):
+            if i >= len(images):
+                ax[i].axis("off")
+                ax[i].set_title("")
+                ax[i].set_visible(False)
+            else:
+                ax[i].axis("off")
+                ax[i].set_title(f"{sim_version_names[i]}")
+                ax[i].imshow(images[i])
+
+        plt.tight_layout()
+        plt.show()
+
+    def _retrieve_pdf_as_image(self, run_id: str, http: urllib3.PoolManager) -> list[PngImageFile]:
+        pdf_url = f"https://api.biosimulations.org/results/{run_id}/download"
+        images: list[PngImageFile] = []
+        head_response = http.request("HEAD", pdf_url)
+        if head_response.status == 200:
+            response = http.request("GET", pdf_url)
+            with BytesIO(response.data) as zip_buffer, ZipFile(zip_buffer, "a", ZIP_DEFLATED, False) as zip_file:
+                file_names = zip_file.namelist()
+                for file_name in file_names:
+                    if file_name.endswith(".pdf") and file_name.lower().index("plot") != -1:
+                        with zip_file.open(file_name) as pdf_file:
+                            content = pdf_file.read()
+                            with fitz.open(stream=content, filetype="pdf") as doc:
+                                for page in doc:
+                                    pix: fitz.Pixmap = page.get_pixmap()
+                                    png_content = pix.tobytes("png")
+                                    images.append(PIL.Image.open(BytesIO(png_content)))  # type: ignore
+        return images
+
 
 def _extract_dataset_attr(
     dataset_name: str, attr_key: str, hdf5_file: BiosimHDF5File | SimDataHDF5File

biosim_client/verify/models.py

@@ -27,7 +27,17 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "75fa30ae7e01483c",
+   "id": "1cec8049-5ca4-4b45-b68c-773cc7fc8a11",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "run.show_saved_plots(max_columns=4, width_in=10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "82c8576f-030b-4e31-ba95-7f93cb4a5637",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -45,6 +55,7 @@
     "omex_file = \"/Users/jimschaff/Documents/workspace/biosim-client/tests/fixtures/data/BIOMD0000000010_tellurium_Negative_feedback_and_ultrasen.omex\"\n",
     "run_2: VerifyResults = BiosimClient().compare_omex(omex_path=omex_file, simulators=simulators)\n",
     "run_2.show_report()\n",
+    "run_2.show_saved_plots()\n",
     "run_2.show_dataset_heatmaps()"
    ]
   },