Mccalluc/serve notebooks plus introspection

CHANGELOG-notebook.md

@@ -0,0 +1 @@
+- Add an endpoint which serves a Jupyter notebook with the viewconf for a dataset.

context/app/api/client.py

@@ -105,7 +105,7 @@ def get_entity(self, uuid=None, hbm_id=None):
         entity = hits[0]['_source']
         return entity
 
-    def get_vitessce_conf(self, entity):
+    def get_vitessce_conf_cells(self, entity):
         # First, try "vis-lifting": Display image pyramids on their parent entity pages.
         image_pyramid_descendants = _get_image_pyramid_descendants(entity)
         if image_pyramid_descendants:
@@ -116,20 +116,19 @@ def get_vitessce_conf(self, entity):
             # about "files". Bill confirms that when the new structure comes in
             # there will be a period of backward compatibility to allow us to migrate.
             derived_entity['files'] = derived_entity['metadata']['files']
-            return self.get_vitessce_conf(derived_entity)
+            return self.get_vitessce_conf_cells(derived_entity)
 
         if 'files' not in entity or 'data_types' not in entity:
             return None
         if self.is_mock:
-            return self._get_mock_vitessce_conf()
+            return self._get_mock_vitessce_conf_cells()
 
         # Otherwise, just try to visualize the data for the entity itself:
         try:
             vc = get_view_config_class_for_data_types(
                 entity=entity, nexus_token=self.nexus_token
             )
-            conf = vc.build_vitessce_conf()
-            return conf
+            return vc.build_vitessce_conf_cells()
         except Exception:
             message = f'Building vitessce conf threw error: {traceback.format_exc()}'
             current_app.logger.error(message)

context/app/api/vitessce_confs/base_confs.py

@@ -2,6 +2,7 @@
 from pathlib import Path
 import re
 
+import nbformat
 from flask import current_app
 from vitessce import (
     VitessceConfig,
@@ -119,7 +120,8 @@ def __init__(self, entity, nexus_token, is_mock=False):
         self.image_pyramid_regex = IMAGE_PYRAMID_DIR
         super().__init__(entity, nexus_token, is_mock)
 
-    def build_vitessce_conf(self):
+    def build_vitessce_conf_cells(self):
+        cells = []
         file_paths_found = self._get_file_paths()
         found_images = get_matches(
             file_paths_found, self.image_pyramid_regex + r".*\.ome\.tiff?$",
@@ -131,8 +133,18 @@ def build_vitessce_conf(self):
             raise FileNotFoundError(message)
 
         vc = VitessceConfig(name="HuBMAP Data Portal")
+        cells.append(nbformat.v4.new_code_cell(
+            'vc = VitessceConfig(name="HuBMAP Data Portal")'
+        ))
+
         dataset = vc.add_dataset(name="Visualization Files")
+        cells.append(nbformat.v4.new_code_cell(
+            'vc.add_dataset(name="Visualization Files")'
+        ))
+
         images = []
+        cells.append(nbformat.v4.new_code_cell(f'images = []'))
+
         for img_path in found_images:
             img_url, offsets_url = self._get_img_and_offset_url(
                 img_path, self.image_pyramid_regex
@@ -142,12 +154,29 @@ def build_vitessce_conf(self):
                     img_url=img_url, offsets_url=offsets_url, name=Path(img_path).name
                 )
             )
+            cells.append(nbformat.v4.new_code_cell(
+                f"images.append(OmeTiffWrapper(img_url='{img_url}', offsets_url='{offsets_url}', name=Path('{img_path}').name))"
+            ))
+        # NOTE: If OmeTiffWrapper had a repr which reproduced the object, we could just say:
+        #    f"images = {images}"
+
         dataset = dataset.add_object(MultiImageWrapper(images))
+        cells.append(nbformat.v4.new_code_cell(
+            'dataset = dataset.add_object(MultiImageWrapper(images))'
+        ))
+
         vc = self._setup_view_config_raster(vc, dataset)
+        cells.append(nbformat.v4.new_code_cell(
+            f'vc = self._setup_view_config_raster(vc, dataset)'
+        ))
+
+        # NOTE: We could skip the serialization to JSON and then back to an object.
+        # ... but not sure how to handle the del below... does a method need to be added to the viewconf?
+
         conf = vc.to_dict()
         # Don't want to render all layers
         del conf["datasets"][0]["files"][0]["options"]["renderLayers"]
-        return conf
+        return (conf, cells)
 
 
 class ScatterplotViewConf(ViewConf):

context/app/routes_main.py

@@ -1,6 +1,7 @@
 from os.path import dirname
 from urllib.parse import urlparse
 import json
+import nbformat
 
 from flask import (Blueprint, render_template, abort, current_app,
                    session, request, redirect, url_for, Response)
@@ -95,7 +96,7 @@ def details(type, uuid):
     flask_data = {
         'endpoints': _get_endpoints(),
         'entity': entity,
-        'vitessce_conf': client.get_vitessce_conf(entity)
+        'vitessce_conf': client.get_vitessce_conf_cells(entity)[0]
     }
     return render_template(
         template,
@@ -106,25 +107,53 @@ def details(type, uuid):
     )
 
 
-@blueprint.route('/browse/<type>/<uuid>.<ext>')
-def details_ext(type, uuid, ext):
+@blueprint.route('/browse/<type>/<uuid>.json')
+def details_json(type, uuid):
     if type not in entity_types:
         abort(404)
-    if ext != 'json':
-        abort(404)
     client = _get_client()
     entity = client.get_entity(uuid)
     return entity
 
 
-@blueprint.route('/browse/<type>/<uuid>.rui.<ext>')
-def details_rui_ext(type, uuid, ext):
+@blueprint.route('/browse/<type>/<uuid>.ipynb')
+def details_notebook(type, uuid):
+    if type not in entity_types:
+        abort(404)
+    client = _get_client()
+    entity = client.get_entity(uuid)
+    vitessce_cells = client.get_vitessce_conf_cells(entity)[1]
+    if vitessce_cells is None:
+        abort(404)
+    nb = nbformat.v4.new_notebook()
+    nb['cells'] = [
+        nbformat.v4.new_markdown_cell(f"""
+Visualization for [{entity['display_doi']}]({request.base_url.replace('.ipynb','')})
+        """.strip()),
+        nbformat.v4.new_code_cell('\n'.join([
+            '!pip install vitessce==0.1.0a9',
+            '!jupyter nbextension install --py --sys-prefix vitessce',
+            '!jupyter nbextension enable --py --sys-prefix vitessce'
+        ])),
+        nbformat.v4.new_code_cell('from vitessce import VitessceConfig')
+    ] + vitessce_cells + [
+        nbformat.v4.new_code_cell('vc = VitessceConfig.from_dict(vitessce_conf)'),
+        nbformat.v4.new_code_cell('vw = vc.widget()'),
+        nbformat.v4.new_code_cell('vw')
+    ]
+    return Response(
+        response=nbformat.writes(nb),
+        headers={'Content-Disposition': f"attachment; filename={entity['display_doi']}.ipynb"},
+        mimetype='application/x-ipynb+json'
+    )
+
+
+@blueprint.route('/browse/<type>/<uuid>.rui.json')
+def details_rui_json(type, uuid):
     # Note that the API returns a blob of JSON as a string,
     # so, to return a JSON object, and not just a string, we need to decode.
     if type not in entity_types:
         abort(404)
-    if ext != 'json':
-        abort(404)
     client = _get_client()
     entity = client.get_entity(uuid)
     # For Samples...

context/app/static/js/components/error/ErrorBody/ErrorBody.jsx

@@ -56,7 +56,7 @@ function ErrorBody({ errorCode, urlPath, isAuthenticated, isGlobus401, isMainten
 
   if (errorCode === 404) {
     if (urlPath && urlPath.startsWith('/browse/')) {
-      const uuid = urlPath.split('/').pop();
+      const uuid = urlPath.split('/').pop().split('.')[0];
       if (uuid.length !== 32) {
         return (
           <>

context/requirements.txt

@@ -1,3 +1,4 @@
+nbformat==5.1.2
 Flask==1.1.2
 globus-sdk==2.0.1
 requests==2.22.0