Unify Cloudflare and GitHub Actions build scripts (#137)

* Unify Cloudflare and GitHub Actions build scripts

* Make markdown dependency optional

> Have the build_colophon.py try/except import the markdown module and if it is not available fall back to just copying in the text

Refs #137 (comment)

* Revert "Make markdown dependency optional"

This reverts commit ac29359.

I'm going to have it work on Cloudflare instead.

Author: simonw

.github/workflows/pages.yml

@@ -36,14 +36,11 @@ jobs:
         run: |
           python -m pip install markdown llm llm-anthropic>=0.20
 
-      - name: Generate links and build colophon
+      - name: Build site
         env:
           ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
-        run: |
-          python gather_links.py
-          python write_docs.py
-          python build_colophon.py
-          python build_index.py
+          GENERATE_LLM_DOCS: "1"
+        run: sh build.sh
 
       - name: Find modified docs stems
         id: docs
@@ -70,24 +67,6 @@ jobs:
           git commit -m "Generated docs: ${{ steps.docs.outputs.stems }}" || exit 0
           git push
 
-      - name: Inject footer.js into HTML files
-        run: |
-          # Get the git hash of the last commit that touched footer.js
-          FOOTER_HASH=$(git log -1 --format="%H" -- footer.js)
-          FOOTER_SHORT_HASH=$(echo "$FOOTER_HASH" | cut -c1-8)
-
-          # Insert footer.js script tag into all root-level .html files except index.html
-          for file in *.html; do
-            if [ -f "$file" ] && [ "$file" != "index.html" ]; then
-              # Check if footer.js is not already included
-              if ! grep -q 'src="footer.js' "$file"; then
-                # Insert script tag before the LAST </body> tag only
-                # Using tac to reverse, replace first match, then reverse back
-                tac "$file" | sed '0,/<\/body>/s|</body>|<script src="footer.js?'"${FOOTER_SHORT_HASH}"'"></script>\n</body>|' | tac > "$file.tmp" && mv "$file.tmp" "$file"
-              fi
-            fi
-          done
-
       - name: Setup Pages
         uses: actions/configure-pages@v4
 

build.sh

@@ -1,24 +1,41 @@
 #!/bin/bash
 set -e
 
-# Mae sure we have the full git history for finding commit dates:
+# Make sure we have the full git history for finding commit dates:
 git fetch --unshallow
 
-echo "=== Building tools.simonwillison.net for Cloudflare Pages ==="
+echo "=== Building tools.simonwillison.net ==="
 
-# Install Python dependencies
-echo "Installing Python dependencies..."
-pip install --quiet markdown
-
-# Run Python build scripts (but NOT write_docs.py which generates LLM descriptions)
 echo "Gathering links and metadata..."
 python gather_links.py
 
+# Only generate LLM summaries if GENERATE_LLM_DOCS is set
+if [ "$GENERATE_LLM_DOCS" = "1" ]; then
+    echo "Generating LLM documentation..."
+    python write_docs.py
+fi
+
 echo "Building colophon page..."
 python build_colophon.py
 
-# Convert README.md to index.html using Python's markdown library
-echo "Converting README.md to index.html..."
+echo "Building index page..."
 python build_index.py
 
+echo "Injecting footer.js into HTML files..."
+# Get the git hash of the last commit that touched footer.js
+FOOTER_HASH=$(git log -1 --format="%H" -- footer.js)
+FOOTER_SHORT_HASH=$(echo "$FOOTER_HASH" | cut -c1-8)
+
+# Insert footer.js script tag into all root-level .html files except index.html
+for file in *.html; do
+  if [ -f "$file" ] && [ "$file" != "index.html" ]; then
+    # Check if footer.js is not already included
+    if ! grep -q 'src="footer.js' "$file"; then
+      # Insert script tag before the LAST </body> tag only
+      # Using tac to reverse, replace first match, then reverse back
+      tac "$file" | sed '0,/<\/body>/s|</body>|<script src="footer.js?'"${FOOTER_SHORT_HASH}"'"></script>\n</body>|' | tac > "$file.tmp" && mv "$file.tmp" "$file"
+    fi
+  fi
+done
+
 echo "=== Build complete! ==="