Moved all metadata processing from the build script into a Pandoc Lua filter

1 files changed, 41 insertions, 159 deletions

diff --git a/scripts/build_content.sh b/scripts/build_content.sh
index c640b12..c88ea86 100755
--- a/scripts/build_content.sh
+++ b/scripts/build_content.sh
@@ -23,8 +23,6 @@ get_filters() {
 
 FILTERS=$(get_filters)
 
-CATEGORIES=$(cat "$METADATA_DIR/categories.json")
-
 get_target_filename() {
     filename=$1
     filename=${filename#"$CONTENT_DIR"}
@@ -34,43 +32,29 @@ get_target_filename() {
     echo "$OUTPUT_DIR$filename"
 }
 
-get_url_rel() {
-    basedir=$1
-    filename=$2
-    ext=$3
-    filename=$(realpath --relative-to="$basedir" "$filename")
-    filename="${filename%.md}.$ext"
-    if [ "${filename##*/}" = "index.html" ]; then
-        filename=${filename%index.html}
-    fi
-    echo "$filename"
-}
-
-get_url_abs() {
-    echo "/$(get_url_rel "$CONTENT_DIR" "$1" "$2")"
-}
-
 get_metadata() {
-    pandoc "$1" \
+    pandoc "$2" \
         -f markdown \
         -t plain \
+        --metadata content_dir="$CONTENT_DIR" \
+        --metadata base_file="$1" \
+        --metadata file="$2" \
+        --metadata-file metadata/metadata.yaml \
         --no-highlight \
-        --template scripts/metadata_tpl.json
+        --template scripts/metadata_tpl.json \
+        --lua-filter scripts/metadata_filter.lua
 }
 
 get_content() {
-    pandoc "$1" \
+    pandoc "$2" \
         -f markdown \
         -t plain \
-        --no-highlight
-}
-
-get_section() {
-    filename=$1
-    filename=${filename#"$CONTENT_DIR"}
-    filename=${filename%.md}
-    filename=$(echo "$filename" | cut -f1 -d /)
-    echo "$filename"
+        --metadata content_dir="$CONTENT_DIR" \
+        --metadata base_file="$1" \
+        --metadata file="$2" \
+        --metadata-file metadata/metadata.yaml \
+        --no-highlight \
+        --lua-filter scripts/metadata_filter.lua
 }
 
 get_subpages_basedir() {
@@ -82,7 +66,7 @@ get_subpages_basedir() {
     echo "$filename"
 }
 
-get_subpages_meta() {
+get_subpages_metadata() {
     basedir=$(get_subpages_basedir "$1")
     child_pages=()
 
@@ -97,98 +81,25 @@ get_subpages_meta() {
             -print0)
     fi
 
-    pages=$(echo "{}" | jq ". + { \
-        all: [], \
-        by_year: {}, \
-    }")
-
-    categories="{}"
+    pages="[]"
 
     if [ ${#child_pages[@]} -ne 0 ]; then
-        #echo -e "\033[0;90m[////////]\033[0m Child pages:"
-
         for file in "${child_pages[@]}"; do
-            #echo -e "\033[0;90m[////////]\033[0m   -  $file"
-
-            metadata=$(get_metadata "$file")
-
-            section=$(get_section "$file")
-
-            url_rel=$(get_url_rel "$basedir" "$file" "html")
-            url_abs=$(get_url_abs "$file" "html")
-            url_full="${SITE%/}$url_abs"
-
-            date=$(echo "$metadata" | jq -r .date)
-            date_rfc3339=""
-
-            if [ "$date" != "" ]; then
-                date_rfc3339=$(date -d "$date" --rfc-3339=seconds | sed 's/ /T/')
-            fi
-
-            last_update=$(echo "$metadata" | jq -r .last_update)
-            last_update_rfc3339=""
-
-            if [ "$last_update" != "" ]; then
-                last_update=$date
-            fi
-
-            if [ "$last_update" != "" ]; then
-                last_update_rfc3339=$(date -d "$last_update" --rfc-3339=seconds | sed 's/ /T/')
-            fi
-
-            category=$(echo "$metadata" | jq -r .category)
-            category_name=""
-
-            if [ "$section" != "" ] && [ "$category" != "" ]; then
-                category_name=$(echo "$CATEGORIES" | jq -r ".$section.$category")
-            fi
-
-            content=$(get_content "$file")
-
-            metadata=$(echo "$metadata" | jq ". + { \
-                content: \"$(echo "$content" | sed -z 's/\\/\\\\/g;s/\n/\\n/g;s/"/\"/g')\", \
-                date: { \
-                    yyyy_mm_dd: \"$date\", \
-                    rfc3339: \"$date_rfc3339\" \
-                }, \
-                last_update: { \
-                    yyyy_mm_dd: \"$last_update\", \
-                    rfc3339: \"$last_update_rfc3339\" \
-                }, \
-                url: { \
-                    rel: \"$url_rel\", \
-                    abs: \"$url_abs\", \
-                    full: \"$url_full\" \
-                } \
-            }")
-
-            pages=$(echo "$pages" | jq ".all += [ $metadata ]")
-
-            if [ "$date" != "" ]; then
-                date_year=$(date -d "$date" +%Y)
-                #date_month=$(date -d "$date" +%m)
-                #date_day=$(date -d "$date" +%d)
-
-                pages=$(echo "$pages" | jq ".by_year.\"$date_year\" += [ $metadata ]")
-            fi
-
-            if [ "$category" != "" ]; then
-                categories=$(echo "$categories" | jq ".$category = { name: \"$category_name\", count: (.$category.count + 1) } ")
-            fi
+            metadata=$(get_metadata "$1" "$file" | jq "{ \
+                url: .url, \
+                author: .author, \
+                title: .title, \
+                date: .date, \
+                last_update: .last_update, \
+                category: .category, \
+                content: \"$(get_content "$1" "$file" | sed -z 's/\\/\\\\/g;s/\n/\\n/g;s/"/\"/g')\" \
+            } | del(.[] | nulls)")
+
+            pages=$(echo "$pages" | jq ". += [ $metadata ]")
         done
     fi
 
-    pages=$(echo "$pages" \
-        | jq ".all |= (sort_by(.date) | reverse) \
-            | .by_year |= (to_entries | sort_by(.key) | reverse \
-                | .[].value |= (sort_by(.date) | reverse))")
-
-    categories=$(echo "$categories" | jq ". | to_entries | { by_id: (. | sort_by(.key)), by_count: (. | sort_by(.value.count) | reverse) }")
-
-    echo "{}" | jq ". + { \
-        pages: $pages, \
-        categories: $categories
-    }"
+    echo "{ \"pages\": $pages }"
 }
 
 handle () {
@@ -201,47 +112,10 @@ handle () {
 
         echo -e "\033[0;90m[COMPILE ]\033[0m     Getting metadata"
 
-        included_metadata=$(get_metadata "$1")
+        included_metadata=$(get_metadata "$1" "$1")
+        added_metadata=$(get_subpages_metadata "$1")
 
         create_feed=$(echo "$included_metadata" | jq -r .create_feed)
-        
-        url_abs=$(get_url_abs "$1" "html")
-        url_full="${SITE%/}$url_abs"
-
-        section=$(get_section "$1")
-
-        is_home=false
-        if [ "$target_file" = "${OUTPUT_DIR}index.html" ]; then
-            is_home=true
-        fi
-
-        added_metadata=$(echo "$(get_subpages_meta "$1")" | jq ". + { \
-            url: { \
-                abs: \"$url_abs\", \
-                full: \"$url_full\" \
-            }, \
-            section: { \
-                id: \"$section\", \
-                is_$section: true \
-            }, \
-            is_home: $is_home \
-        }")
-
-        if [ "$create_feed" = "true" ]; then
-            feed_url_abs="$(get_url_abs "$1" "xml")"
-            feed_url_full="${SITE%/}$feed_url_abs"
-            last_update=$(echo "$added_metadata" | jq -r ".pages.all[0].last_update")
-
-            added_metadata=$(echo "$added_metadata" | jq ". + { \
-                feed: { \
-                    last_update: $last_update, \
-                    url: { \
-                        abs: \"$feed_url_abs\", \
-                        full: \"$feed_url_full\" \
-                    } \
-                } \
-            }")
-        fi
 
         meta_file=$(mktemp)
 
@@ -256,8 +130,13 @@ handle () {
                 --no-highlight \
                 --template "${TEMPLATES_DIR}feed.xml" \
                 -o "${target_file%.html}.xml" \
+                --metadata content_dir="$CONTENT_DIR" \
+                --metadata base_file="$1" \
+                --metadata file="$1" \
+                --metadata type=feed \
                 --metadata-file metadata/metadata.yaml \
                 --metadata-file "$meta_file" \
+                --lua-filter scripts/metadata_filter.lua \
                 $FILTERS
         fi
 
@@ -269,15 +148,18 @@ handle () {
             --no-highlight \
             --template "${TEMPLATES_DIR}base.html" \
             -o "$target_file" \
+            --metadata content_dir="$CONTENT_DIR" \
+            --metadata base_file="$1" \
+            --metadata file="$1" \
             --metadata-file metadata/metadata.yaml \
             --metadata-file "$meta_file" \
+            --lua-filter scripts/metadata_filter.lua \
             $FILTERS
 
         rm "$meta_file"
 
-        # echo $(pandoc "$1" \
-        #     -f markdown \
-        #     -t json)
+        # echo "$included_metadata"
+        # echo "$added_metadata"
 
         echo -e "\033[0;90m[COMPILE ]\033[0m     Done"
     else