-- media-microdata Soupault Plugin
-- SPDX-FileCopyrightText: 2026 toastal <https://toast.al/contact>
-- SPDX-License-Identifier: MPL-2.0
--
-- Applies Schema.org media markup on /media/ only. The page holds one
-- section per medium (audio-casts/talks/blogs), each with h3 item
-- subsections carrying flat field lists:
--   article#Article       → mainEntity ItemList
--   h3 item section       → itemListElement + AudioObject/VideoObject/Article
--                           (medium → type from custom_options.media_types)
--   h3                    → name
--   Date dd               → datePublished
--   Host dd               → publisher Organization stub
--   Guest/Speaker dd      → actor (on the peers-plugin Person spans)
--   Author dd             → author (ditto)
--   Listen/Watch/Read dd  → url (copied link href)

Plugin.require_version("5.0.0")

function url_path()
	local url = page_url
	url = Regex.replace(url, "^[a-z]+://[^/]+", "")
	if url == "" then
		url = "/"
	elseif not String.starts_with(url, "/") then
		url = "/" .. url
	end
	return url
end

function hidden_meta(itemprop, content, parent)
	local elem = HTML.create_element("meta")
	HTML.set_attribute(elem, "itemprop", itemprop)
	HTML.set_attribute(elem, "content", content)
	HTML.append_child(parent, elem)
	return elem
end

function hidden_link(itemprop, href, parent)
	local elem = HTML.create_element("link")
	HTML.set_attribute(elem, "itemprop", itemprop)
	HTML.set_attribute(elem, "href", href)
	HTML.append_child(parent, elem)
	return elem
end

function field_text(dd)
	return String.trim(HTML.strip_tags(dd))
end

path = url_path()

if path == "/media" or path == "/media/" then
	main = HTML.select_one(page, "main#Main")
	if not HTML.is_element(main) then
		Plugin.fail(format("No main#Main found in “%s”, skipping media markup", page_file))
	end

	article = HTML.select_one(page, "article#Article")
	if not HTML.is_element(article) then
		Plugin.fail(format("No article#Article found in “%s”, cannot apply ItemList scope", page_file))
	end
	HTML.set_attribute(article, "itemscope", "")
	HTML.set_attribute(article, "itemprop", "mainEntity")
	HTML.set_attribute(article, "itemtype", "https://schema.org/ItemList")

	h1 = HTML.select_one(main, "h1[itemprop=\"headline\"]")
	if not HTML.is_element(h1) then
		h1 = HTML.select_one(main, "h1")
	end
	if HTML.is_element(h1) then
		HTML.set_attribute(h1, "itemprop", "name")
		headline = HTML.create_element("meta")
		HTML.set_attribute(headline, "itemprop", "headline")
		HTML.set_attribute(headline, "content", String.trim(HTML.strip_tags(h1)))
		HTML.prepend_child(main, headline)
	else
		Log.warning(format("No h1 found in “%s”, skipping WebPage.headline compensation", page_file))
	end

	types = soupault_config["custom_options"]["media_types"]
	if not Value.is_table(types) then
		Plugin.fail("custom_options.media_types is missing: cannot type media items")
	end

	mediums = HTML.select(article, "section#Content > section")
	if Table.is_empty(mediums) then
		Log.warning(format("No media medium sections found in “%s”", page_file))
	end
	local idx = 1
	while mediums[idx] do
		local medium = mediums[idx]
		local medium_id = HTML.get_attribute(medium, "id")
		local typename = nil
		if Value.is_string(medium_id) and medium_id ~= "" then
			typename = types[medium_id]
		end
		if not Value.is_string(typename) or typename == "" then
			Log.warning(format("Unknown media medium “%s”, skipping its items", Value.repr(medium_id)))
		else
			local itemtype = "https://schema.org/" .. typename
			local items = HTML.select(medium, "section")
			if Table.is_empty(items) then
				Log.warning(format("No media items found in medium “%s”", Value.repr(medium_id)))
			end
			local jdx = 1
			while items[jdx] do
				local item = items[jdx]
				local heading = HTML.select_one(item, "h3")
				if not HTML.is_element(heading) then
					Log.warning("Media item without h3, skipping it")
				else
					HTML.set_attribute(item, "itemscope", "")
					HTML.set_attribute(item, "itemprop", "itemListElement")
					HTML.set_attribute(item, "itemtype", itemtype)
					HTML.set_attribute(heading, "itemprop", "name headline")

					local dts = HTML.select(item, "dl.field-list dt")
					local dds = HTML.select(item, "dl.field-list dd")
					local kdx = 1
					while dts[kdx] do
						local key = field_text(dts[kdx])
						local dd = dds[kdx]
						if Value.is_nil(dd) or not HTML.is_element(dd) then
							Log.warning(format("Media field “%s” without value, skipping it", key))
						elseif key == "Date:" then
							hidden_meta("datePublished", field_text(dd), item)
						elseif key == "Host:" then
							local org = HTML.create_element("span")
							HTML.set_attribute(org, "itemprop", "publisher")
							HTML.set_attribute(org, "itemscope", "")
							HTML.set_attribute(org, "itemtype", "https://schema.org/Organization")
							hidden_meta("name", field_text(dd), org)
							HTML.append_child(item, org)
						elseif key == "Guest:" or key == "Speaker:" then
							local persons = HTML.select(dd, "span.peer")
							if Table.is_empty(persons) then
								Log.warning(format("Media field “%s” without person, skipping actor", key))
							end
							local pdx = 1
							while persons[pdx] do
								HTML.set_attribute(persons[pdx], "itemprop", "actor")
								pdx = pdx + 1
							end
						elseif key == "Author:" or key == "Writer:" then
							local persons = HTML.select(dd, "span.peer")
							if Table.is_empty(persons) then
								Log.warning(format("Media field “%s” without person, skipping author", key))
							end
							local pdx = 1
							while persons[pdx] do
								HTML.set_attribute(persons[pdx], "itemprop", "author")
								pdx = pdx + 1
							end
						elseif key == "Listen:" or key == "Watch:" or key == "Read:" then
							local link = HTML.select_one(dd, "a[href]")
							if HTML.is_element(link) then
								hidden_link("url", HTML.get_attribute(link, "href"), item)
							else
								Log.warning(format("Media field “%s” without link, skipping url", key))
							end
						else
							Log.warning(format("Unrecognized media field “%s”, skipping it", key))
						end
						kdx = kdx + 1
					end
				end
				jdx = jdx + 1
			end
		end
		idx = idx + 1
	end
end