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 "lsup.namespace"
local term = require "lsup.term"
local triple = require "lsup.triple"
local graph = require "lsup.graph"

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

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

-- Compile all templates 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()


-- HTML generator module.
local M = {
    res_dir = pkar.config.htmlgen.out_dir .. "/res",
    asset_dir = pkar.config.htmlgen.out_dir .. "/assets",
    media_dir = 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 generate_dres(s, mconf)
    local gr = graph.new(pkar.store, term.DEFAULT_CTX)

    local dmd = {}
    local rel = {}
    local children = {}
    -- Metadata
    local attrs = gr:connections(s, term.LINK_OUTBOUND)
    --require "debugger"()
    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.
            local p = term.new_iriref_ns("pas:next")
            local dc_title = term.new_iriref_ns("dc:title")
            local tn_p = term.new_iriref_ns("pas:thumbnail")
            for o in pairs(ots) do
                -- Loop through all first children.
                local node_uri = o
                logger:debug("local node_uri", node_uri.data)
                local ll = {}
                while node_uri do
                    -- Loop trough all next nodes for each first child.
                    local el_gr = graph.get(node_uri, pkar.store)
                    table.insert(ll, {
                        uri = node_uri,
                        label = gr:attr(node_uri, dc_title)[1],
                        tn = gr:attr(node_uri, tn_p)[1],
                    })
                    local next_attr = gr:attr(node_uri, p)
                    node_uri = next(next_attr)  -- There can only be one "next"
                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.
            dmd[fname] = {label = p_label, uri = fname}
            -- TODO differentiate term types
            for o in pairs(ots) do table.insert(dmd[fname], o.data) end
            table.sort(dmd[fname])
        end
        ::skip::
    end
    table.sort(dmd)
    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 = ((dmd["dc:title"] or NT)[1] or NT) or s.data,
        mconf = mconf,
        uri = s,
        dmd = dmd,
        rel = rel,
        children = children,
        ls_next = ls_next,
        breadcrumbs = get_breadcrumbs(mconf),
        deliverable = deliverable,
    })

    local res_id = s.data:gsub(nsm.get_ns("par"), "")
    local ofh = assert(io.open(string.format(
            "%s/%s.html", M.res_dir, res_id), "w"))
    ofh:write(out_html)
    ofh:close()
end


local function generate_ores(s, mconf)
    local gr = graph.new(pkar.store, term.DEFAULT_CTX)

    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))

    --require "debugger"()
    -- Transform and move media assets.
    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"}
    local dest_fname = plpath.basename(res_path[1])
    if txconf.ext then
        dest_fname = plpath.splitext(dest_fname) .. txconf.ext
    end
    dest = M.asset_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("Deliverable: ", dest)

    out_html = ores_tpl({
        site_title = pkar.config.site.title or pkar.default_title,
        fname = plpath.basename(techmd["pas:sourcePath"][1]),
        mconf = mconf,
        uri = s,
        techmd = techmd,
        rel = rel,
        ls_next = ls_next,
        breadcrumbs = get_breadcrumbs(mconf),
        deliverable = deliverable,
    })

    local res_id = s.data:gsub(nsm.get_ns("par"), "")
    local ofh = assert(io.open(string.format(
            "%s/%s.html", M.res_dir, res_id), "w"))
    ofh:write(out_html)
    ofh:close()
end


M.generate_resource = function(s)
    local gr = graph.new(pkar.store, term.DEFAULT_CTX)

    local res_type
    local type_attr = gr:attr(
        s, term.new_iriref_ns("pas:contentType"))
    res_type = next(type_attr).data
    local mconf = model.models[res_type]

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


M.generate_resources = function()
    dir.rmtree(M.res_dir)
    dir.makepath(M.res_dir)
    dir.rmtree(M.asset_dir)
    dir.makepath(M.asset_dir)
    dir.rmtree(M.media_dir)
    dir.makepath(M.media_dir)
    local gr = graph.new(pkar.store, term.DEFAULT_CTX)
    local subjects = gr:unique_terms(triple.POS_S)

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


M.generate_idx = function()
    local obj_idx = {}
    local gr = graph.new(pkar.store, term.DEFAULT_CTX)
    -- 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 s_label = nsm.denormalize_uri(s.data)
        local titles = gr:attr(s, term.new_iriref_ns("dc:title"))
        local obj = {title = next(titles)}
        if obj.title then obj_idx[s_label] = obj end
    end

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

    local idx_path = 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()
end


return M