path: root/cmd/build/build.go

package main

import (
	"bytes"
	"encoding/xml"
	"fmt"
	"io"
	"io/fs"
	"log"
	"log/slog"
	"net/url"
	"os"
	"path"
	"path/filepath"
	"slices"
	"strings"
	"time"

	"website/internal/atom"
	"website/internal/config"

	"github.com/BurntSushi/toml"
	"github.com/PuerkitoBio/goquery"
	"github.com/a-h/htmlformat"
	"github.com/adrg/frontmatter"
	"github.com/antchfx/xmlquery"
	"github.com/antchfx/xpath"
	"github.com/ardanlabs/conf/v3"
	mapset "github.com/deckarep/golang-set/v2"
	cp "github.com/otiai10/copy"
	"github.com/pkg/errors"
	"github.com/yuin/goldmark"
	"github.com/yuin/goldmark/extension"
	htmlrenderer "github.com/yuin/goldmark/renderer/html"
	"golang.org/x/net/html"
)

type PostMatter struct {
	Date        time.Time `toml:"date"`
	Description string    `toml:"description"`
	Title       string    `toml:"title"`
	Taxonomies  struct {
		Tags []string `toml:"tags"`
	} `toml:"taxonomies"`
}

type Post struct {
	Input    string
	Output   string
	Basename string
	URL      string
	Content  string
	PostMatter
}

type Tags mapset.Set[string]

func getPost(filename string) (*PostMatter, *[]byte, error) {
	matter := PostMatter{}
	content, err := os.Open(filename)
	defer content.Close()
	if err != nil {
		return nil, nil, errors.WithMessagef(err, "could not open post %s", filename)
	}
	rest, err := frontmatter.MustParse(content, &matter)
	if err != nil {
		return nil, nil, errors.WithMessagef(err, "could not parse front matter of post %s", filename)
	}

	return &matter, &rest, nil
}

func readPosts(root string, inputDir string, outputDir string) ([]Post, Tags, error) {
	tags := mapset.NewSet[string]()
	posts := []Post{}
	subdir := filepath.Join(root, inputDir)
	files, err := os.ReadDir(subdir)
	if err != nil {
		return nil, nil, errors.WithMessagef(err, "could not read post directory %s", subdir)
	}
	outputReplacer := strings.NewReplacer(root, outputDir, ".md", "/index.html")
	urlReplacer := strings.NewReplacer(root, "", ".md", "/")
	markdown := goldmark.New(
		goldmark.WithRendererOptions(
			htmlrenderer.WithUnsafe(),
		),
		goldmark.WithExtensions(
			extension.GFM,
			extension.Footnote,
			extension.Typographer,
		),
	)
	for _, f := range files {
		pathFromRoot := filepath.Join(subdir, f.Name())
		if !f.IsDir() && path.Ext(pathFromRoot) == ".md" {
			output := outputReplacer.Replace(pathFromRoot)
			url := urlReplacer.Replace(pathFromRoot)
			slog.Debug("reading post", "post", pathFromRoot)
			matter, content, err := getPost(pathFromRoot)
			if err != nil {
				return nil, nil, err
			}

			for _, tag := range matter.Taxonomies.Tags {
				tags.Add(strings.ToLower(tag))
			}

			var buf bytes.Buffer
			slog.Debug("rendering markdown in post", "post", pathFromRoot)
			if err := markdown.Convert(*content, &buf); err != nil {
				return nil, nil, errors.WithMessage(err, "could not convert markdown content")
			}
			post := Post{
				Input:      pathFromRoot,
				Output:     output,
				Basename:   filepath.Base(url),
				URL:        url,
				PostMatter: *matter,
				Content:    buf.String(),
			}

			posts = append(posts, post)
		}
	}
	slices.SortFunc(posts, func(a, b Post) int {
		return b.Date.Compare(a.Date)
	})
	return posts, tags, nil
}

func layout(filename string, config config.Config, pageTitle string) (*goquery.Document, error) {
	html, err := os.Open(filename)
	if err != nil {
		return nil, err
	}
	defer html.Close()
	css, err := os.ReadFile("templates/style.css")
	if err != nil {
		return nil, err
	}
	doc, err := goquery.NewDocumentFromReader(html)
	if err != nil {
		return nil, err
	}
	doc.Find("html").SetAttr("lang", config.DefaultLanguage)
	doc.Find("head > link[rel=alternate]").SetAttr("title", config.Title)
	doc.Find(".title").SetText(config.Title)
	doc.Find("title").Add(".p-name").SetText(pageTitle)
	doc.Find("head > style").SetHtml("\n" + string(css))
	nav := doc.Find("nav")
	navLink := doc.Find("nav a")
	nav.Empty()
	for _, link := range config.Menus["main"] {
		nav.AppendSelection(navLink.Clone().SetAttr("href", link.URL).SetText(link.Name))
	}
	return doc, nil
}

func renderHTML(doc *goquery.Document) io.Reader {
	r, w := io.Pipe()

	// TODO: return errors to main thread
	go func() {
		w.Write([]byte("<!doctype html>\n"))
		err := htmlformat.Nodes(w, []*html.Node{doc.Children().Get(0)})
		if err != nil {
			slog.Error("error rendering html", "error", err)
			w.CloseWithError(err)
			return
		}
		defer w.Close()
	}()
	return r
}

func renderPost(post Post, config config.Config) (r io.Reader, err error) {
	doc, err := layout("templates/post.html", config, post.PostMatter.Title)
	if err != nil {
		return nil, err
	}
	doc.Find(".title").AddClass("p-author h-card").SetAttr("rel", "author")
	doc.Find(".h-entry .dt-published").
		SetAttr("datetime", post.PostMatter.Date.UTC().Format(time.RFC3339)).
		SetText(
			post.PostMatter.Date.Format("2006-01-02"),
		)
	doc.Find(".h-entry .e-content").SetHtml(post.Content)
	categories := doc.Find(".h-entry .p-categories")
	tpl := categories.Find(".p-category").ParentsUntilSelection(categories)
	tpl.Remove()
	for _, tag := range post.Taxonomies.Tags {
		cat := tpl.Clone()
		cat.Find(".p-category").SetAttr("href", fmt.Sprintf("/tags/%s/", tag)).SetText("#" + tag)
		categories.AppendSelection(cat)
	}

	return renderHTML(doc), nil
}

func renderTags(tags Tags, config config.Config) (io.Reader, error) {
	doc, err := layout("templates/tags.html", config, config.Title)
	if err != nil {
		return nil, err
	}
	tagList := doc.Find(".tags")
	tpl := doc.Find(".h-feed")
	tpl.Remove()
	for _, tag := range mapset.Sorted(tags) {
		li := tpl.Clone()
		li.Find("a").SetAttr("href", fmt.Sprintf("/tags/%s/", tag)).SetText("#" + tag)
		tagList.AppendSelection(li)
	}
	return renderHTML(doc), nil
}

func renderListPage(tag string, config config.Config, posts []Post) (io.Reader, error) {
	var title string
	if len(tag) > 0 {
		title = tag
	} else {
		title = config.Title
	}
	doc, err := layout("templates/list.html", config, title)
	if err != nil {
		return nil, err
	}
	feed := doc.Find(".h-feed")
	tpl := feed.Find(".h-entry")
	tpl.Remove()

	doc.Find(".title").AddClass("p-author h-card").SetAttr("rel", "author")
	if tag == "" {
		doc.Find(".filter").Remove()
	} else {
		doc.Find(".filter").Find("h3").SetText("#" + tag)
	}

	for _, post := range posts {
		entry := tpl.Clone()
		entry.Find(".p-name").SetText(post.Title).SetAttr("href", post.URL)
		entry.Find(".dt-published").
			SetAttr("datetime", post.PostMatter.Date.UTC().Format(time.RFC3339)).
			SetText(post.PostMatter.Date.Format("2006-01-02"))
		feed.AppendSelection(entry)
	}

	return renderHTML(doc), nil
}

func renderHomepage(config config.Config, posts []Post) (io.Reader, error) {
	_, index, err := getPost("content/_index.md")
	if err != nil {
		return nil, err
	}
	doc, err := layout("templates/homepage.html", config, config.Title)
	if err != nil {
		return nil, err
	}
	doc.Find("body").AddClass("h-card")
	doc.Find(".title").AddClass("p-name u-url")
	var buf bytes.Buffer

	md := goldmark.New(goldmark.WithRendererOptions(htmlrenderer.WithUnsafe()))
	if err := md.Convert(*index, &buf); err != nil {
		return nil, err
	}
	doc.Find("#content").SetHtml(buf.String())

	feed := doc.Find(".h-feed")
	tpl := feed.Find(".h-entry")
	tpl.Remove()

	for _, post := range posts[0:3] {
		entry := tpl.Clone()
		entry.Find(".p-name").SetText(post.Title)
		entry.Find(".u-url").SetAttr("href", post.URL)
		entry.
			Find(".dt-published").
			SetAttr("datetime", post.PostMatter.Date.UTC().Format(time.RFC3339)).
			SetText(post.PostMatter.Date.Format("2006-01-02"))

		feed.AppendSelection(entry)
	}
	doc.Find(".u-email").
		SetAttr("href", fmt.Sprintf("mailto:%s", config.Email)).
		SetText(config.Email)

	elsewhere := doc.Find(".elsewhere")
	linkRelMe := elsewhere.Find(".u-url[rel=me]").ParentsUntil("ul")
	linkRelMe.Remove()

	for _, link := range config.Menus["me"] {
		el := linkRelMe.Clone()
		el.Find("a").SetAttr("href", link.URL).SetText(link.Name)
		elsewhere.AppendSelection(el)
	}

	return renderHTML(doc), nil
}

func render404(config config.Config) (io.Reader, error) {
	doc, err := layout("templates/404.html", config, "404 Not Found")
	if err != nil {
		return nil, err
	}
	return renderHTML(doc), nil
}

func renderFeed(title string, config config.Config, posts []Post, specific string) (io.Reader, error) {
	reader, err := os.Open("templates/feed.xml")
	if err != nil {
		return nil, err
	}
	defer reader.Close()
	doc, err := xmlquery.Parse(reader)
	feed := doc.SelectElement("feed")
	feed.SelectElement("title").FirstChild.Data = title
	feed.SelectElement("link").SetAttr("href", config.BaseURL.String())
	feed.SelectElement("id").FirstChild.Data = atom.MakeTagURI(config, specific)
	datetime, err := posts[0].Date.UTC().MarshalText()
	feed.SelectElement("updated").FirstChild.Data = string(datetime)
	tpl := feed.SelectElement("entry")
	xmlquery.RemoveFromTree(tpl)

	for _, post := range posts {
		fullURL, err := url.JoinPath(config.BaseURL.String(), post.URL)
		if err != nil {
			return nil, err
		}
		text, err := xml.MarshalIndent(&atom.FeedEntry{
			Title:   post.Title,
			Link:    atom.MakeLink(fullURL),
			Id:      atom.MakeTagURI(config, post.Basename),
			Updated: post.Date.UTC(),
			Summary: post.Description,
			Author:  config.Title,
			Content: atom.FeedContent{
				Content: post.Content,
				Type:    "html",
			},
		}, "  ", "    ")
		if err != nil {
			return nil, err
		}
		entry, err := xmlquery.ParseWithOptions(strings.NewReader(string(text)), xmlquery.ParserOptions{
			Decoder: &xmlquery.DecoderOptions{
				Strict:    false,
				AutoClose: xml.HTMLAutoClose,
				Entity:    xml.HTMLEntity,
			},
		})
		if err != nil {
			return nil, err
		}
		xmlquery.AddChild(feed, entry.SelectElement("entry"))
	}

	return strings.NewReader(doc.OutputXML(true)), nil
}

func renderFeedStyles() (io.Reader, error) {
	reader, err := os.Open("templates/feed-styles.xsl")
	if err != nil {
		return nil, err
	}
	defer reader.Close()
	nsMap := map[string]string{
		"xsl":   "http://www.w3.org/1999/XSL/Transform",
		"atom":  "http://www.w3.org/2005/Atom",
		"xhtml": "http://www.w3.org/1999/xhtml",
	}
	doc, err := xmlquery.Parse(reader)
	expr, err := xpath.CompileWithNS("//xhtml:style", nsMap)
	if err != nil {
		return nil, err
	}
	style := xmlquery.QuerySelector(doc, expr)
	css, err := os.ReadFile("templates/style.css")
	if err != nil {
		return nil, err
	}
	xmlquery.AddChild(style, &xmlquery.Node{
		Type: xmlquery.TextNode,
		Data: string(css),
	})
	return strings.NewReader(doc.OutputXML(true)), nil
}

func mkdirp(dirs ...string) error {
	return os.MkdirAll(path.Join(dirs...), 0755)
}

func outputToFile(output io.Reader, filename ...string) error {
	file, err := os.OpenFile(path.Join(filename...), os.O_WRONLY|os.O_CREATE, 0644)
	if err != nil {
		return errors.WithMessage(err, "could not open output file")
	}
	defer file.Close()

	if _, err := file.ReadFrom(output); err != nil {
		return errors.WithMessage(err, "could not write output file")
	}
	return nil
}

func build(outDir string, config config.Config) error {
	privateDir := path.Join(outDir, "private")
	if err := mkdirp(privateDir); err != nil {
		return errors.Errorf("could not create private directory: %w", err)
	}
	publicDir := path.Join(outDir, "public")
	if err := mkdirp(publicDir); err != nil {
		return errors.Errorf("could not create public directory: %w", err)
	}

	err := cp.Copy("static", publicDir, cp.Options{
		PreserveTimes:     true,
		PermissionControl: cp.AddPermission(0755),
	})
	if err != nil {
		log.Panic(errors.Errorf("could not copy static files: %v", err))
	}

	if err := mkdirp(publicDir, "post"); err != nil {
		return errors.WithMessage(err, "could not create post output directory")
	}
	slog.Debug("reading posts")
	posts, tags, err := readPosts("content", "post", publicDir)
	if err != nil {
		return err
	}

	for _, post := range posts {
		if err := mkdirp(publicDir, "post", post.Basename); err != nil {
			return errors.WithMessage(err, "could not create directory for post")
		}
		slog.Debug("rendering post", "post", post.Basename)
		output, err := renderPost(post, config)
		if err != nil {
			return errors.WithMessagef(err, "could not render post %s", post.Input)
		}
		if err := outputToFile(output, post.Output); err != nil {
			return err
		}
	}

	if err := mkdirp(publicDir, "tags"); err != nil {
		return errors.WithMessage(err, "could not create directory for tags")
	}
	slog.Debug("rendering tags list")
	output, err := renderTags(tags, config)
	if err != nil {
		return errors.WithMessage(err, "could not render tags")
	}
	if err := outputToFile(output, publicDir, "tags", "index.html"); err != nil {
		return err
	}

	for _, tag := range tags.ToSlice() {
		matchingPosts := []Post{}
		for _, post := range posts {
			if slices.Contains(post.Taxonomies.Tags, tag) {
				matchingPosts = append(matchingPosts, post)
			}
		}
		if err := mkdirp(publicDir, "tags", tag); err != nil {
			return errors.WithMessage(err, "could not create directory")
		}
		slog.Debug("rendering tags page", "tag", tag)
		output, err := renderListPage(tag, config, matchingPosts)
		if err != nil {
			return errors.WithMessage(err, "could not render tag page")
		}
		if err := outputToFile(output, publicDir, "tags", tag, "index.html"); err != nil {
			return err
		}

		slog.Debug("rendering tags feed", "tag", tag)
		output, err = renderFeed(fmt.Sprintf("%s - %s", config.Title, tag), config, matchingPosts, tag)
		if err != nil {
			return errors.WithMessage(err, "could not render tag feed page")
		}
		if err := outputToFile(output, publicDir, "tags", tag, "atom.xml"); err != nil {
			return err
		}
	}

	slog.Debug("rendering list page")
	listPage, err := renderListPage("", config, posts)
	if err != nil {
		return errors.WithMessage(err, "could not render list page")
	}
	if err := outputToFile(listPage, publicDir, "post", "index.html"); err != nil {
		return err
	}

	slog.Debug("rendering feed")
	feed, err := renderFeed(config.Title, config, posts, "feed")
	if err != nil {
		return errors.WithMessage(err, "could not render feed")
	}
	if err := outputToFile(feed, publicDir, "atom.xml"); err != nil {
		return err
	}

	slog.Debug("rendering feed styles")
	feedStyles, err := renderFeedStyles()
	if err != nil {
		return errors.WithMessage(err, "could not render feed styles")
	}
	if err := outputToFile(feedStyles, publicDir, "feed-styles.xsl"); err != nil {
		return err
	}

	slog.Debug("rendering homepage")
	homePage, err := renderHomepage(config, posts)
	if err != nil {
		return errors.WithMessage(err, "could not render homepage")
	}
	if err := outputToFile(homePage, publicDir, "index.html"); err != nil {
		return err
	}

	slog.Debug("rendering 404 page")
	notFound, err := render404(config)
	if err != nil {
		return errors.WithMessage(err, "could not render 404 page")
	}
	if err := outputToFile(notFound, privateDir, "404.html"); err != nil {
		return err
	}

	return nil
}

type IOConfig struct {
	Source      string `conf:"default:.,short:s"`
	Destination string `conf:"default:website,short:d"`
	BaseURL     config.URL
}

func main() {
	if os.Getenv("DEBUG") != "" {
		slog.SetLogLoggerLevel(slog.LevelDebug)
	}
	slog.Debug("starting build process")

	ioConfig := IOConfig{}
	if help, err := conf.Parse("", &ioConfig); err != nil {
		if errors.Is(err, conf.ErrHelpWanted) {
			fmt.Println(help)
			os.Exit(1)
		}
		log.Panicf("parsing I/O configuration: %v", err)
	}

	if ioConfig.Source != "." {
		err := os.Chdir(ioConfig.Source)
		if err != nil {
			log.Panic("could not change to source directory")
		}
	}

	config, err := config.GetConfig()
	if err != nil {
		log.Panic(errors.Errorf("could not get config: %v", err))
	}

	if ioConfig.BaseURL.URL != nil {
		config.BaseURL.URL, err = url.Parse(ioConfig.BaseURL.String())
		if err != nil {
			log.Panic(errors.Errorf("highly unlikely: %v", err))
		}
	}

	err = os.RemoveAll(ioConfig.Destination)
	if err != nil {
		log.Panic(errors.Errorf("could not remove public directory: %v", err))
	}

	if err := build(ioConfig.Destination, *config); err != nil {
		switch cause := errors.Cause(err).(type) {
		case *fs.PathError:
			slog.Info("pathError")
			slog.Error(fmt.Sprintf("%s", err))
		case toml.ParseError:
			slog.Info("parseError")
			slog.Error(fmt.Sprintf("%s", err))
		default:
			slog.Info("other")
			slog.Error(fmt.Sprintf("cause:%+v", errors.Cause(cause)))
			slog.Error(fmt.Sprintf("%+v", cause))
		}
		os.Exit(1)
	}

	slog.Debug("done")
}