From 5db16a08a9726b67b5bb96fd17f813fa1b74f568 Mon Sep 17 00:00:00 2001
From: Volpeon <git@volpeon.ink>
Date: Tue, 5 Jan 2021 10:58:04 +0100
Subject: Moved all metadata processing from the build script into a Pandoc Lua
 filter

---
 scripts/build_content.sh    | 200 ++++++++-------------------------------
 scripts/metadata_filter.lua | 225 ++++++++++++++++++++++++++++++++++++++++++++
 scripts/metadata_tpl.json   |   7 +-
 3 files changed, 267 insertions(+), 165 deletions(-)
 create mode 100644 scripts/metadata_filter.lua

(limited to 'scripts')

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
diff --git a/scripts/metadata_filter.lua b/scripts/metadata_filter.lua
new file mode 100644
index 0000000..f31bc1f
--- /dev/null
+++ b/scripts/metadata_filter.lua
@@ -0,0 +1,225 @@
+function format_date(date)
+    if date == nil then
+        return date
+    end
+
+    date = pandoc.utils.normalize_date(pandoc.utils.stringify(date))
+    local year, month, day = date:match("(%d%d%d%d)-(%d%d)-(%d%d)")
+    if year == nil then
+        return nil
+    end
+
+    local time = os.time({
+        year = tonumber(year),
+        month = tonumber(month),
+        day = tonumber(day)
+    })
+    return pandoc.MetaMap({
+        yyyy_mm_dd = pandoc.MetaString(os.date("%F", time)),
+        yyyy = pandoc.MetaString(os.date("%Y", time)),
+        mm_dd = pandoc.MetaString(os.date("%m-%d", time)),
+        rfc3339 = pandoc.MetaString(os.date("%FT%T+00:00", time))
+    })
+end
+
+function table_to_list(t, cmp)
+    local l = pandoc.MetaList({})
+
+    for key, value in pairs(t) do
+        l:insert(pandoc.MetaMap({
+            key = key,
+            value = value
+        }))
+    end
+
+    l:sort(cmp or function(i1, i2)
+        return i1.key < i2.key
+    end)
+
+    return l
+end
+
+function splitstr(input, sep)
+    sep = sep or "%s"
+    local t = {}
+    for str in input:gmatch("([^" .. sep .. "]+)") do
+        table.insert(t, str)
+    end
+    return t
+end
+
+function relative_to(dir, target)
+    dir = splitstr(dir, "/")
+    target = splitstr(target, "/")
+
+    local prefix = true
+    local path = ""
+
+    for i = 1, math.min(#dir, #target) do
+        local t = target[i]
+        if prefix then
+            if dir[i] ~= t then
+                prefix = false
+                path = "../" .. t
+            end
+        else
+            path = "../" .. path .. "/" .. t
+        end
+    end
+
+    if #dir < #target then
+        for i = #dir + 1, #target do
+            path = path .. (path == "" and "" or "/") .. target[i]
+        end
+    elseif #dir > #target then
+        for i = #target + 1, #dir do
+            path = "../" .. path
+        end
+    end
+
+    return path
+end
+
+function resolve_url(page_type, site_url, content_dir, base_dir, cur_file)
+    if page_type == "page" then
+        cur_file = cur_file:gsub("%.md$", ".html")
+    elseif page_type == "feed" then
+        cur_file = cur_file:gsub("%.md$", ".xml")
+    end
+
+    local abs = cur_file:gsub("^" .. content_dir, ""):gsub("/index.html$", "/")
+    local rel = relative_to(base_dir, cur_file):gsub("/index.html$", "/")
+
+    return pandoc.MetaMap({
+        abs = pandoc.MetaString(abs),
+        rel = pandoc.MetaString(rel),
+        full = pandoc.MetaString(site_url .. abs)
+    })
+end
+
+function resolve_section(abs_url)
+    local section = abs_url:match("^/(.-)[/.]") or "index"
+    return pandoc.MetaMap({
+        id = pandoc.MetaString(section),
+        ["is_" .. section] = pandoc.MetaBool(true)
+    })
+end
+
+function organize_subpages(site_url, content_dir, base_dir, pages, categories)
+    local categories_data = pandoc.MetaList({})
+
+    pages:sort(function(p1, p2)
+        if p1.date and p2.date then
+            return pandoc.utils.stringify(p1.date.yyyy_mm_dd) > pandoc.utils.stringify(p2.date.yyyy_mm_dd)
+        elseif p2.date then
+            return true
+        elseif p1.date then
+            return false
+        else
+            return pandoc.utils.stringify(p1.title) < pandoc.utils.stringify(p2.title)
+        end
+    end)
+
+    local pages_data = pandoc.MetaMap({
+        all = pages,
+        by_year = pandoc.MetaList({}),
+        last_update = nil
+    })
+
+    if pages then
+        local pages_by_year_map = {}
+        local categories_map = {}
+
+        for i = 1, #pages do
+            local page = pages[i]
+
+            if page.date then
+                local yyyy = pandoc.utils.stringify(page.date.yyyy)
+                local pages_by_yyyy = pages_by_year_map[yyyy]
+
+                if not pages_by_yyyy then
+                    pages_by_yyyy = pandoc.MetaList(pandoc.List())
+                    pages_by_year_map[yyyy] = pages_by_yyyy
+                end
+
+                pages_by_yyyy:insert(page)
+            end
+
+            if page.category and categories then
+                local category = pandoc.utils.stringify(page.category)
+
+                if categories[category] then
+                    local current_category = categories_map[category]
+
+                    if not current_category then
+                        current_category = pandoc.MetaMap({
+                            name = pandoc.MetaString(categories[category]),
+                            count = 0
+                        })
+                        categories_map[category] = current_category
+                    end
+
+                    current_category.count = current_category.count + 1
+                end
+            end
+        end
+
+        pages_data.by_year = table_to_list(pages_by_year_map, function(i1, i2)
+            return i1.key > i2.key
+        end)
+
+        categories_data = table_to_list(categories_map)
+
+        for _, item in ipairs(categories_data) do
+            item.value.count = pandoc.MetaString(("%d"):format(item.value.count))
+        end
+
+        if #pages_data.by_year ~= 0 then
+            pages_data.last_update = pages_data.by_year[1].value[1].last_update
+        end
+    end
+
+    return pages_data, categories_data
+end
+
+function Meta(meta)
+    meta.content_dir = meta.content_dir:gsub("/$", "")
+    meta.site.url = pandoc.utils.stringify(meta.site.url):gsub("/$", "")
+    meta.base_dir = meta.base_file:gsub("^(.*)/.-$", "%1")
+    meta.type = meta.type or "page"
+
+    meta.date = format_date(meta.date)
+    if meta.last_update ~= nil then
+        meta.last_update = format_date(meta.last_update)
+    else
+        meta.last_update = meta.date
+    end
+
+    if meta.type == "feed" then
+        meta.page = pandoc.MetaMap({
+            url = resolve_url("page", meta.site.url, meta.content_dir, meta.base_dir, meta.file)
+        })
+    end
+
+    meta.url = resolve_url(meta.type, meta.site.url, meta.content_dir, meta.base_dir, meta.file)
+    meta.section = resolve_section(meta.url.abs)
+    meta.categories = meta.categories[meta.section.id]
+
+    if meta.menus and meta.menus.main then
+        for i = 1, #meta.menus.main do
+            local item = meta.menus.main[i]
+            item.active = pandoc.MetaBool(pandoc.utils.stringify(item.id) == meta.section.id)
+        end
+    end
+
+    if meta.pages then
+        local pages, categories = organize_subpages(meta.site.url, meta.content_dir, meta.base_dir, meta.pages,
+                                      meta.categories)
+        meta.pages = pages
+        meta.categories = categories
+    else
+        meta.categories = nil
+    end
+
+    return meta
+end
diff --git a/scripts/metadata_tpl.json b/scripts/metadata_tpl.json
index 79eabe1..fb36f0f 100755
--- a/scripts/metadata_tpl.json
+++ b/scripts/metadata_tpl.json
@@ -1,6 +1 @@
-{
-    "title": "$title$",
-    "date": "$date$",
-    "category": "$category$",
-    "create_feed": "$create_feed$"
-}
+$meta-json$
-- 
cgit v1.2.3-54-g00ecf