feat: check and update notebook links

.cloud-build/UpdateNotebookLinks.py

@@ -0,0 +1,86 @@
+
+from nbconvert.preprocessors import Preprocessor
+
+import argparse
+import nbformat
+import os
+import re
+
+
+class UpdateNotebookLinks(Preprocessor):
+
+  def __init__(self, repo_name):
+    self._repo_name = repo_name
+    self._regex = {
+        "colab": "https://colab\\.research\\.google\\.com/github/GoogleCloudPlatform/.*\\.ipynb",
+        "github": "https://github\\.com/GoogleCloudPlatform/.*\\.ipynb"
+    }
+    self._prefix = {
+        "colab": "https://colab.research.google.com/github/GoogleCloudPlatform/",
+        "github": "https://github.com/GoogleCloudPlatform/"
+    }
+
+  def generate_link(self, notebook_file_path, type):
+    return self._prefix[type] + self._repo_name + "/blob/master" + notebook_file_path
+
+  def search_link(self, text, type):
+    return re.search(self._regex[type], text)
+
+  def replace_link(self, text, notebook_file_path, type):
+    return re.sub(self._regex[type], self.generate_link(notebook_file_path, type), text)
+
+  def update_text(self, original_text, notebook_file_path):
+    updated_text = self.replace_link(original_text, notebook_file_path, 'colab')
+    updated_text = self.replace_link(updated_text, notebook_file_path, 'github')
+    return updated_text
+
+  def preprocess(self, notebook, notebook_file_path):
+    updated = False
+    for cell in notebook.cells:
+      if cell.cell_type == 'markdown' and (
+          self.search_link(cell.source, 'colab') or self.search_link(cell.source, 'github')):
+        cell.source = self.update_text(cell.source, notebook_file_path)
+        updated = True
+        break
+    return notebook, updated
+
+def main(args):
+
+  notebook_file_paths = list()
+  for root, dirs, files in os.walk(args.repo_path):
+    for file in files:
+      if file.endswith(".ipynb"):
+        notebook_file_paths.append(os.path.join(root.replace(args.repo_path, ''), file))
+
+  for notebook_file_path in notebook_file_paths:
+    with open(args.repo_path + notebook_file_path) as f:
+      nb = nbformat.read(f, as_version=4)
+
+    update_notebook_links_preprocessor = UpdateNotebookLinks(repo_name=args.repo_name)
+    nb, updated = update_notebook_links_preprocessor.preprocess(nb, notebook_file_path)
+
+    if updated:
+      with open(args.repo_path + notebook_file_path, mode="w", encoding="utf-8") as f:
+        nbformat.write(nb, f)
+
+  return
+
+parser = argparse.ArgumentParser(description="Run changed notebooks.")
+
+parser.add_argument(
+    "--repo-path",
+    type=str,
+    default=os.getcwd(),
+    help="The local path containing the folders of notebooks that should be updated.",
+)
+parser.add_argument(
+    "--repo-name",
+    type=str,
+    default='vertex-ai-samples',
+    help="The repository name used to generate the notebook links.",
+)
+
+args = parser.parse_args()
+
+if __name__ == '__main__':
+  main(args)

UpdateLinks.sh

@@ -0,0 +1,45 @@
+#!/bin/bash
+
+export search_dir=community-content
+
+notebook_filenames=$(find $search_dir -type f -name "*.ipynb")
+
+export repo_name=vertex-ai-samples
+
+export search_links=https:.*\.ipynb
+
+export search_console=https://console\.cloud\.google\.com/ai/platform/notebooks/deploy-notebook?download_url=.*\.ipynb
+export search_colab=https://colab\.research\.google\.com/github/GoogleCloudPlatform/.*/blob/master/.*\.ipynb
+export search_github=https://github\.com/GoogleCloudPlatform/.*/blob/master/.*\.ipynb
+
+export console_prefix=https://console.cloud.google.com/ai/platform/notebooks/deploy-notebook?download_url=https://github.com/GoogleCloudPlatform
+export colab_prefix=https://colab.research.google.com/github/GoogleCloudPlatform
+export github_prefix=https://github.com/GoogleCloudPlatform
+
+for notebook_filename in $notebook_filenames
+  do
+
+    echo "Notebook: $notebook_filename"
+
+    echo "        Existing Linkes"
+    grep $search_links $notebook_filename
+    grep $search_console $notebook_filename
+    grep $search_colab $notebook_filename
+    grep $search_github $notebook_filename
+
+    export replace_console="$console_prefix/$repo_name/raw/master/$notebook_filename"
+    export replace_colab="$colab_prefix/$repo_name/blob/master/$notebook_filename"
+    export replace_github="$github_prefix/$repo_name/blob/master/$notebook_filename"
+
+    echo "        New Linkes"
+    echo "replace_console:  $replace_console"
+    echo "replace_colab:    $replace_colab"
+    echo "replace_github:   $replace_github"
+
+    sed -i -e s,$search_console,$replace_console,g $notebook_filename
+    sed -i -e s,$search_colab,$replace_colab,g $notebook_filename
+    sed -i -e s,$search_github,$replace_github,g $notebook_filename
+
+  done
+
+find $search_dir -name '*.ipynb-e' -delete