Improved subpage organization code in metadata filter

2 files changed, 82 insertions, 73 deletions

diff --git a/scripts/build_content.sh b/scripts/build_content.sh
index c88ea86..0b2ce2e 100755
--- a/scripts/build_content.sh
+++ b/scripts/build_content.sh
@@ -48,7 +48,7 @@ get_metadata() {
 get_content() {
     pandoc "$2" \
         -f markdown \
-        -t plain \
+        -t markdown \
         --metadata content_dir="$CONTENT_DIR" \
         --metadata base_file="$1" \
         --metadata file="$2" \
@@ -131,13 +131,11 @@ handle () {
                 --template "${TEMPLATES_DIR}feed.xml" \
                 -o "${target_file%.html}.xml" \
                 --metadata content_dir="$CONTENT_DIR" \
-                --metadata base_file="$1" \
                 --metadata file="$1" \
-                --metadata type=feed \
+                --metadata page_type=feed \
                 --metadata-file metadata/metadata.yaml \
                 --metadata-file "$meta_file" \
-                --lua-filter scripts/metadata_filter.lua \
-                $FILTERS
+                --lua-filter scripts/metadata_filter.lua
         fi
 
         echo -e "\033[0;90m[COMPILE ]\033[0m     Creating page"
@@ -149,13 +147,24 @@ handle () {
             --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
 
+        # echo "$(pandoc "$1" \
+        #     -f markdown \
+        #     -t plain \
+        #     --no-highlight \
+        #     --template scripts/metadata_tpl.json \
+        #     --metadata content_dir="$CONTENT_DIR" \
+        #     --metadata file="$1" \
+        #     --metadata-file metadata/metadata.yaml \
+        #     --metadata-file "$meta_file" \
+        #     --lua-filter scripts/metadata_filter.lua \
+        #     $FILTERS)"
+
         rm "$meta_file"
 
         # echo "$included_metadata"
diff --git a/scripts/metadata_filter.lua b/scripts/metadata_filter.lua
index f31bc1f..d6379b9 100644
--- a/scripts/metadata_filter.lua
+++ b/scripts/metadata_filter.lua
@@ -23,7 +23,7 @@ function format_date(date)
 end
 
 function table_to_list(t, cmp)
-    local l = pandoc.MetaList({})
+    local l = pandoc.List()
 
     for key, value in pairs(t) do
         l:insert(pandoc.MetaMap({
@@ -39,6 +39,32 @@ function table_to_list(t, cmp)
     return l
 end
 
+function group_by(l, field, insert)
+    insert = insert or function(group, _, item)
+        if not group then
+            group = l:new({item})
+            return group
+        end
+        group:insert(item)
+    end
+
+    local groups = {}
+
+    for i = 1, #l do
+        local item = l[i]
+        local f = field(item)
+
+        if f ~= nil then
+            local out = insert(groups[f], f, item)
+            if out ~= nil then
+                groups[f] = out
+            end
+        end
+    end
+
+    return groups
+end
+
 function splitstr(input, sep)
     sep = sep or "%s"
     local t = {}
@@ -106,77 +132,50 @@ function resolve_section(abs_url)
 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
+    local grouped_pages = group_by(pages, function(p)
+        return not p.date
     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
+    local pages_undated = grouped_pages[true] or pandoc.MetaList({})
+    pages_undated:sort(function(p1, p2)
+        return pandoc.utils.stringify(p1.title) < pandoc.utils.stringify(p2.title)
+    end)
 
-                    current_category.count = current_category.count + 1
-                end
-            end
-        end
+    local pages_dated = grouped_pages[false] or pandoc.MetaList({})
+    pages_dated:sort(function(p1, p2)
+        return pandoc.utils.stringify(p1.date.yyyy_mm_dd) > pandoc.utils.stringify(p2.date.yyyy_mm_dd)
+    end)
 
-        pages_data.by_year = table_to_list(pages_by_year_map, function(i1, i2)
-            return i1.key > i2.key
-        end)
+    local pages_by_year = group_by(pages_dated, function(p)
+        return pandoc.utils.stringify(p.date.yyyy)
+    end)
+    pages_by_year = pandoc.MetaList(table_to_list(pages_by_year, function(i1, i2)
+        return i1.key > i2.key
+    end))
 
-        categories_data = table_to_list(categories_map)
+    local pages_data = pandoc.MetaMap({
+        all_dated = pages_dated,
+        all_undated = pages_undated,
+        by_year = pages_by_year,
+        last_update = #pages_dated ~= 0 and pages_dated[1].last_update
+    })
 
-        for _, item in ipairs(categories_data) do
-            item.value.count = pandoc.MetaString(("%d"):format(item.value.count))
+    local categories_data = group_by(pages_dated, function(p)
+        return p.category and pandoc.utils.stringify(p.category)
+    end, function(stats, category)
+        if not stats then
+            return {
+                name = pandoc.MetaString(categories[category] or category),
+                count = 1
+            }
+        else
+            stats.count = stats.count + 1
         end
+    end)
+    categories_data = pandoc.MetaList(table_to_list(categories_data))
 
-        if #pages_data.by_year ~= 0 then
-            pages_data.last_update = pages_data.by_year[1].value[1].last_update
-        end
+    for i = 1, #categories_data do
+        categories_data[i].value.count = pandoc.MetaString(("%d"):format(categories_data[i].value.count))
     end
 
     return pages_data, categories_data
@@ -185,8 +184,9 @@ end
 function Meta(meta)
     meta.content_dir = meta.content_dir:gsub("/$", "")
     meta.site.url = pandoc.utils.stringify(meta.site.url):gsub("/$", "")
+    meta.base_file = meta.base_file or meta.file
     meta.base_dir = meta.base_file:gsub("^(.*)/.-$", "%1")
-    meta.type = meta.type or "page"
+    meta.page_type = meta.page_type or "page"
 
     meta.date = format_date(meta.date)
     if meta.last_update ~= nil then
@@ -195,13 +195,13 @@ function Meta(meta)
         meta.last_update = meta.date
     end
 
-    if meta.type == "feed" then
+    if meta.page_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.url = resolve_url(meta.page_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]