Reduce sitemap size

scripts/build-docc.sh

@@ -55,16 +55,16 @@ if test "$BUILD_SYMBOLS" == 1; then
 fi
 
 # Build documentation
-mkdir -p $OUTPUT_PATH
-rm -rf ${OUTPUT_PATH:?}/*
+mkdir -p "$OUTPUT_PATH"
+rm -rf "${OUTPUT_PATH:?}"/*
 $DOCC convert Hummingbird.docc \
     --transform-for-static-hosting \
     --hosting-base-path /"$HUMMINGBIRD_VERSION" \
     --fallback-display-name Hummingbird \
     --fallback-bundle-identifier com.opticalaberration.hummingbird \
     --fallback-bundle-version 1 \
     --additional-symbol-graph-dir $HB_SG_FOLDER \
-    --output-path $OUTPUT_PATH \
+    --output-path "$OUTPUT_PATH" \
     --hosting-base-path /hummingbird-docs/"$HUMMINGBIRD_VERSION"
 # copy root files template to docs file
 rsync -trv scripts/docsTemplate/* $BASE_OUTPUT_PATH

scripts/build-sitemap.sh

@@ -16,39 +16,42 @@ BASE_PATH=docs/hummingbird-docs/
 
 pushd "$BASE_PATH"
 
-FILES=$(find $DOCUMENTATION_PATH -name 'index.html' -print)
+FILES=$(find "$DOCUMENTATION_PATH" -name 'index.html' -print)
 
 cat > sitemap.xml << EOF
 <?xml version="1.0" encoding="UTF-8"?>
 <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
 EOF
 
-# Calculate the minimum number of slashes in a file
-minSlashes=100
-for FILE in $FILES; do
-    slashCount=$(echo $FILE | text.count "/")
-    minSlashes=$(($minSlashes>$slashCount ? $slashCount : $minSlashes))
-done
+slashCount=$(echo "$DOCUMENTATION_PATH" | text.count "/")
+minSlashes=$(($slashCount + 2))
 
 for FILE in $FILES; do
-    URL=$(echo ${FILE%index.html} | sed 's/:/%3A/g')
+    URL=$(echo "${FILE%index.html}" | sed 's/:/%3A/g')
     # work out priority, by counting slashes in filename, subtract the minimum
     # number of slashes, calculate 2 to the power of the resulting number and
     # invert it
-    slashCount=$(echo $FILE | text.count "/")
+    slashCount=$(echo "$FILE" | text.count "/")
     slashCount=$(($slashCount - $minSlashes))
     slashCount=$(($slashCount>0 ? $slashCount : 0))
-    inv_priority=$((2 ** $slashCount))
-    priority=$(bc <<< "scale=2; 1/$inv_priority" )
-
-    DATE=$(date -r $FILE "+%Y-%m-%d")
-    cat >> sitemap.xml << EOF
+    # only include pages which are modules or top level types
+    if [ "$slashCount" -lt 2 ]; then
+        # Give Hummingbird page higher priority than anything else
+        if [[ "$URL" == */hummingbird/ ]]; then
+            priority=1.0
+        else
+            inv_priority=$((2 ** $slashCount))
+            priority=$(bc <<< "scale=2; 0.8/$inv_priority" )
+        fi
+        DATE=$(date -r "$FILE" "+%Y-%m-%d")
+        cat >> sitemap.xml << EOF
   <url>
     <loc>$URL_PREFIX/$URL</loc>
     <lastmod>$DATE</lastmod>
     <priority>$priority</priority>
   </url>
 EOF
+    fi
 done   
 cat >> sitemap.xml << EOF
 </urlset>