pocket_archive/src/html_generator.lua

local datafile = require "datafile"
local dir = require "pl.dir"
local etlua = require "etlua"
local plpath = require "pl.path"
local pp = require "pl.pretty"

local nsm = require "volksdata.namespace"
local term = require "volksdata.term"
local triple = require "volksdata.triple"
local graph = require "volksdata.graph"

local pkar = require "pocket_archive"
local logger = pkar.logger
local model = require "pocket_archive.model"
local transformers = require "pocket_archive.transformers"

local dbg = require "debugger"


-- "nil" table - for missing key fallback in chaining.
local NT = {}

-- Default store graph to search all triples.
local gr

-- HTML templates. Compile them only once.
-- TODO Add override for user-maintained templates.
local fh, idx_tpl, dres_tpl, ores_tpl
fh = datafile.open("templates/index.html")
idx_tpl = assert(etlua.compile(fh:read("a")))
fh:close()

fh = datafile.open("templates/dres.html")
dres_tpl = assert(etlua.compile(fh:read("a")))
fh:close()

fh = datafile.open("templates/ores.html")
ores_tpl = assert(etlua.compile(fh:read("a")))
fh:close()

fh = datafile.open("templates/header.html")
header_tpl = assert(etlua.compile(fh:read("a")))
fh:close()


-- HTML generator module.
local M = {
    res_dir = plpath.join(pkar.config.htmlgen.out_dir, "res"),
    asset_dir = plpath.join(pkar.config.htmlgen.out_dir, "assets"),
    media_dir = plpath.join(pkar.config.htmlgen.out_dir, "media"),
}


local function get_breadcrumbs(mconf)
    -- Breadcrumbs, from top class to current class.
    -- Also verify if it's a File subclass.
    local breadcrumbs = {}
    for i = 1, #mconf.lineage do
        breadcrumbs[i] = {
            mconf.lineage[i],
            model.models[mconf.lineage[i]].label
        }
    end

    return breadcrumbs
end


local function get_tn_url(s)
    if gr:attr(s, pkar.RDF_TYPE)[pkar.FILE_T.hash] then
        -- The subject is a file.
        tn_fname = (s.data:gsub(pkar.PAR_NS, "") .. ".jpg")  -- FIXME do not hardcode.
        return plpath.join(
                M.media_dir, tn_fname:sub(1, 2), tn_fname:sub(3, 4), tn_fname)
    end

    -- Recurse through all first children until one with a thumbnail, or a
    -- leaf without children, is found.
    local first_child
    _, first_child = next(gr:attr(s, pkar.FIRST_P))
    if first_child then return get_tn_url(first_child) end
end


local function generate_dres(s, mconf)
    local dmd = {}
    local rel = {}
    local children = {}
    local title
    -- Metadata
    local attrs = gr:connections(s, term.LINK_OUTBOUND)
    for p, ots in pairs(attrs) do
        local fname = nsm.denormalize_uri(p.data)
        p_label = ((mconf.properties or NT)[fname] or NT).label
        -- RDF types are shown in in breadcrumbs.
        if fname == "rdf:type" then goto skip
        elseif ((mconf.properties or NT)[fname] or NT).type == "rel" then
            -- Relationship.
            rel[fname] = {label = p_label, uri = fname}
            for _, o in pairs(ots) do table.insert(dmd[fname], o.data) end
        elseif fname == "pas:first" then
            -- Build a linked list for every first found.
            for _, o in pairs(ots) do
                -- Loop through all first children.
                local child_s = o
                logger:debug("local child_s: ", child_s.data)
                local ll = {}

                -- Fallback labels.
                local label
                _, label = next(gr:attr(child_s, pkar.DC_TITLE_P))
                if label then label = label.data
                else
                    _, label = next(gr:attr(child_s, pkar.PATH_P))
                    if label then label = plpath.basename(label.data)
                        else label = child_s.data end
                end

                while child_s do
                    -- Loop trough all next nodes for each first child.
                    require "debugger".assert(get_tn_url(child_s))
                    table.insert(ll, {
                        href = pkar.gen_pairtree(
                                "/res", child_s.data, ".html", true),
                        label = label,
                        tn = get_tn_url(child_s):gsub(M.media_dir, "/media/tn"),
                    })
                    logger:debug("Child label for ", child_s.data, ": ", ll[#ll].label or "nil")
                    -- There can only be one "next"
                    _, child_s = next(gr:attr(child_s, pkar.NEXT_P))
                end
                table.insert(children, ll)
            end
        elseif fname == "pas:next" then
            -- Sibling.
            for _, o in pairs(ots) do ls_next = o.data break end
        else
            -- Descriptive metadata.
            local attr = {label = p_label, uri = fname}
            -- TODO differentiate term types
            for _, o in pairs(ots) do table.insert(attr, o.data) end
            table.sort(attr)
            if p == pkar.DC_TITLE_P then title = attr[1] end
            table.insert(dmd, attr)
        end
        ::skip::
    end
    table.sort(
        dmd, function(a, b)
            return ((a.label or a.uri) < (b.label or b.uri))
        end
    )
    table.sort(rel)
    table.sort(children)
    logger:debug("Lineage:", pp.write(mconf.lineage))
    logger:debug("DMD:", pp.write(dmd))
    logger:debug("REL:", pp.write(rel))
    logger:debug("Children:", pp.write(children))
    logger:debug("Breadcrumbs:", pp.write(get_breadcrumbs(mconf)))

    out_html = dres_tpl({
        site_title = pkar.config.site.title or pkar.default_title,
        title = title or s.data,
        header_tpl = header_tpl,
        mconf = mconf,
        uri = s,
        dmd = dmd,
        rel = rel,
        children = children,
        ls_next = ls_next,
        breadcrumbs = get_breadcrumbs(mconf),
        rdf_href = pkar.gen_pairtree("/res", s.data, ".ttl", true),
    })

    local res_path = pkar.gen_pairtree(M.res_dir, s.data, ".html")
    local ofh = assert(io.open(res_path, "w"))
    ofh:write(out_html)
    ofh:close()

   return true
end


local function generate_ores(s, mconf)
    local techmd = {}
    local rel = {}
    -- Metadata
    local attrs = gr:connections(s, term.LINK_OUTBOUND)
    for p, ots in pairs(attrs) do
        local fname = nsm.denormalize_uri(p.data)
        p_label = ((mconf.properties or NT)[fname] or NT).label
        -- RDF types are shown in in breadcrumbs.
        if fname == "rdf:type" then goto skip
        elseif ((mconf.properties or NT)[fname] or NT).type == "rel" then
            -- Relationship.
            rel[fname] = {label = p_label, uri = fname}
            for _, o in pairs(ots) do table.insert(techmd[fname], o.data) end
        elseif fname == "pas:next" then
            -- Sibling.
            for _, o in pairs(ots) do ls_next = o.data break end
        else
            -- Descriptive metadata.
            techmd[fname] = {label = p_label, uri = fname}
            -- TODO differentiate term types
            for _, o in pairs(ots) do table.insert(techmd[fname], o.data) end
            table.sort(techmd[fname])
        end
        ::skip::
    end
    table.sort(techmd)
    table.sort(rel)
    logger:debug("Lineage:", pp.write(mconf.lineage))
    logger:debug("Breadcrumbs:", pp.write(get_breadcrumbs(mconf)))
    logger:debug("techmd:", pp.write(techmd))
    logger:debug("REL:", pp.write(rel))

    -- Transform and move media assets.
    local dest_fname, dest_dir, dest  -- Reused for thumbnail.
    logger:info("Transforming resource file.")
    local res_path = techmd["pas:path"]
    if not res_path then error("No file path for File resource!") end
    local txconf = (mconf.transformers or NT).deliverable or {fn = "copy"}
    -- Set file name to resource ID + source extension.
    dest_fname = (
            s.data:gsub(pkar.PAR_NS, "") ..
            (txconf.ext or plpath.extension(res_path[1])))
    dest_dir = plpath.join(
            M.media_dir, dest_fname:sub(1, 2), dest_fname:sub(3, 4))
    dir.makepath(dest_dir)
    dest = plpath.join(dest_dir, dest_fname)
    assert(transformers[txconf.fn](
            res_path[1], dest, table.unpack(txconf or NT)))
    local deliverable = dest:gsub(pkar.config.htmlgen.out_dir, "")
    logger:info("Access file: ", deliverable)

    -- Thumbnail.
    local tn
    txconf = (mconf.transformers or NT).thumbnail
    if txconf then
        if txconf.ext then
            dest_fname = plpath.splitext(dest_fname) .. txconf.ext
        end
        dest_dir = plpath.join(
                M.media_dir, "tn", dest_fname:sub(1, 2), dest_fname:sub(3, 4))
        dir.makepath(dest_dir)
        dest = plpath.join(dest_dir, dest_fname)
        assert(transformers[txconf.fn](
                res_path[1], dest, table.unpack(txconf or NT)))
        tn = dest:gsub(M.media_dir, "/media/tn")
        logger:info("Thumbnail: ", tn)
    end

    out_html = ores_tpl({
        site_title = pkar.config.site.title or pkar.default_title,
        fname = plpath.basename(techmd["pas:sourcePath"][1]),
        header_tpl = header_tpl,
        mconf = mconf,
        uri = s,
        techmd = techmd,
        rel = rel,
        ls_next = ls_next,
        breadcrumbs = get_breadcrumbs(mconf),
        deliverable = deliverable,
        thumbnail = tn,
        rdf_href = pkar.gen_pairtree("/res", s.data, ".ttl", true),
    })

    local res_path = pkar.gen_pairtree(M.res_dir, s.data, ".html")
    local ofh = assert(io.open(res_path, "w"))
    ofh:write(out_html)
    ofh:close()

   return true
end


M.get_graph = function(s)
    out_gr = graph.new(nil, s.data)
    gr:copy(out_gr, s)

    return out_gr
end


M.generate_resource = function(s)
    local res_type
    _, res_type = next(gr:attr(s, pkar.CONTENT_TYPE_P))
    local mconf = model.models[res_type.data]

    -- Generate RDF/Turtle doc.
    local res_gr = M.get_graph(s)
    logger:debug("Serializing graph: ", s.data)
    local res_path = pkar.gen_pairtree(M.res_dir, s.data, ".ttl")
    local ofh = assert(io.open(res_path, "w"))
    ofh:write(res_gr:encode("ttl"))
    ofh:close()

    -- Generate HTML doc.
    if mconf.types["pas:File"] then return generate_ores(s, mconf)
    else return generate_dres(s, mconf) end
end


M.generate_resources = function()
    local subjects = gr:unique_terms(triple.POS_S)

    -- TODO parallelize
    for _, s in pairs(subjects) do assert(M.generate_resource(s)) end

    return true
end


M.generate_idx = function()
    local obj_idx = {}
    -- Get all subject of type: Artifact.
    s_ts = gr:term_set(
        pkar.RDF_TYPE, triple.POS_P,
        term.new_iriref_ns("pas:Artifact"), triple.POS_O
    )
    for _, s in pairs(s_ts) do
        local title, created
        _, title = next(gr:attr(s, pkar.DC_TITLE_P))
        _, created = next(gr:attr(s, pkar.DC_CREATED_P))

        local obj = {
            href = pkar.gen_pairtree("/res", s.data, ".html", true),
            title = title,
            created = created.data,
            tn = get_tn_url(s):gsub(M.media_dir, "/media/tn"),
        }
        table.insert(obj_idx, obj)
    end
    table.sort(obj_idx, function(a, b) return a.created < b.created end)

    logger:debug(pp.write(obj_idx))
    out_html = idx_tpl({
        title = pkar.config.site.title or pkar.default_title,
        header_tpl = header_tpl,
        nsm = nsm,
        obj_idx = obj_idx,
    })

    local idx_path = plpath.join(pkar.config.htmlgen.out_dir, "index.html")
    local ofh = assert(io.open(idx_path, "w"))

    logger:debug("Writing info at ", idx_path)
    ofh:write(out_html)
    ofh:close()

    return true
end


M.generate_site = function()
    if plpath.isdir(M.res_dir) then dir.rmtree(M.res_dir) end
    dir.makepath(M.res_dir)
    if plpath.isdir(M.asset_dir) then dir.rmtree(M.asset_dir) end
    dir.makepath(M.asset_dir)
    if plpath.isdir(M.media_dir) then dir.rmtree(M.media_dir) end
    dir.makepath(plpath.join(M.media_dir, "tn"))

    gr = graph.new(pkar.store, term.DEFAULT_CTX)

    assert(M.generate_resources())
    assert(M.generate_idx())
    dir.clonetree("templates/assets", plpath.dirname(M.asset_dir), dir.copyfile)
end


return M