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

1 files changed, 225 insertions, 0 deletions

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