kiln/dir.go

package main

import (
	"bytes"
	"fmt"
	"io"
	"io/ioutil"
	"log"
	"os"
	"os/exec"
	pathpkg "path"
	"sort"
	"strings"
	"time"

	"gopkg.in/yaml.v3"
)

// Dir represents a directory.
type Dir struct {
	Path  string
	Pages []*Page
	Dirs  []*Dir
	index *Page  // The index page.
	feed  []byte // Atom feed.
}

// Page represents a page.
type Page struct {
	Title    string
	Date     time.Time
	Weight   int
	Path     string `yaml:"-"`
	FilePath string `yaml:"-"`
	Content  string `yaml:"-"`
	Params   map[string]string
	Prev     *Page `yaml:"-"`
	Next     *Page `yaml:"-"`
}

// read reads from a directory and indexes the files and directories within it.
func (d *Dir) read(srcDir string, task *Task, cfg *Site) error {
	return d._read(srcDir, "", task, cfg)
}

func (d *Dir) _read(srcDir, path string, task *Task, cfg *Site) error {
	entries, err := ioutil.ReadDir(pathpkg.Join(srcDir, path))
	if err != nil {
		return err
	}
	for _, entry := range entries {
		name := entry.Name()
		path := pathpkg.Join(path, name)
		if entry.IsDir() {
			// Ignore directories beginning with "_"
			if strings.HasPrefix(name, "_") {
				continue
			}
			// Gather directory data
			dir := &Dir{Path: "/" + path + "/"}
			if err := dir._read(srcDir, path, task, cfg); err != nil {
				return err
			}
			d.Dirs = append(d.Dirs, dir)
		} else if ext := pathpkg.Ext(name); task.Match(ext) {
			// Ignore pages beginning with "_" with the exception of _index pages
			namePrefix := strings.TrimSuffix(name, ext)
			if strings.HasPrefix(name, "_") && namePrefix != "_index" {
				continue
			}

			srcPath := pathpkg.Join(srcDir, path)
			content, err := ioutil.ReadFile(srcPath)
			if err != nil {
				return err
			}

			page := &Page{}

			// Try to parse the date from the page filename
			const layout = "2006-01-02"
			base := pathpkg.Base(path)
			if len(base) >= len(layout) {
				dateStr := base[:len(layout)]
				if time, err := time.Parse(layout, dateStr); err == nil {
					page.Date = time
					// Remove the date from the path
					base = base[len(layout):]
					if len(base) > 0 {
						// Remove a leading dash
						if base[0] == '-' {
							base = base[1:]
						}
						if len(base) > 0 {
							dir := pathpkg.Dir(path)
							if dir == "." {
								dir = ""
							}
							path = pathpkg.Join(dir, base)
						}
					}
				}
			}

			// Extract frontmatter from content
			frontmatter, content := extractFrontmatter(content)
			if len(frontmatter) != 0 {
				if err := yaml.Unmarshal(frontmatter, page); err != nil {
					log.Printf("failed to parse frontmatter for %q: %v", path, err)
				}

				// Trim leading newlines from content
				content = bytes.TrimLeft(content, "\r\n")
			}

			if cmd, ok := task.Preprocess[strings.TrimPrefix(ext, ".")]; ok {
				content = process(cmd, bytes.NewReader(content))
			}
			page.Content = string(content)

			page.FilePath = path

			if namePrefix == "_index" {
				page.Path = d.Path
				d.index = page
			} else {
				if namePrefix == "index" {
					path = "/" + strings.TrimSuffix(path, name)
				} else {
					path = "/" + strings.TrimSuffix(path, ext)
					if task.UglyURLs {
						path += task.OutputExt
					} else {
						path += "/"
					}
				}
				page.Path = path
				if permalink, ok := cfg.permalinks[d.Path]; ok {
					var b strings.Builder
					permalink.Execute(&b, page)
					page.Path = b.String()
				}
				d.Pages = append(d.Pages, page)
			}
		}
	}
	return nil
}

// process processes the directory's contents.
func (d *Dir) process(cfg *Site, task *Task) error {
	if task.TemplateExt != "" {
		// Create index
		if d.index != nil {
			tmpl, ok := cfg.templates.FindTemplate(d.Path, "index"+task.TemplateExt)
			if ok {
				var b strings.Builder
				if err := tmpl.Execute(&b, d); err != nil {
					return err
				}
				d.index.Content = b.String()
			}
		}

		// Process pages
		for i := range d.Pages {
			var b strings.Builder
			tmpl, ok := cfg.templates.FindTemplate(d.Path, "page"+task.TemplateExt)
			if ok {
				if err := tmpl.Execute(&b, d.Pages[i]); err != nil {
					return err
				}
				d.Pages[i].Content = b.String()
			}
		}
	}

	// Feed represents a feed.
	type Feed struct {
		Title   string    // Feed title.
		Path    string    // Feed path.
		Updated time.Time // Last updated time.
		Entries []*Page   // Feed entries.
	}

	// Create feeds
	if title, ok := cfg.Feeds[d.Path]; ok {
		var b bytes.Buffer
		feed := &Feed{
			Title:   title,
			Path:    d.Path,
			Updated: time.Now(),
			Entries: d.Pages,
		}
		tmpl, ok := cfg.templates.FindTemplate(d.Path, "atom.xml")
		if ok {
			if err := tmpl.Execute(&b, feed); err != nil {
				return err
			}
			d.feed = b.Bytes()
		} else {
			fmt.Printf("Warning: failed to generate feed %q: missing template \"atom.xml\"\n", title)
		}
	}

	// Process subdirectories
	for _, d := range d.Dirs {
		if err := d.process(cfg, task); err != nil {
			return err
		}
	}
	return nil
}

// write writes the directory's contents to the provided destination path.
func (d *Dir) write(dstDir string, task *Task) error {
	dirPath := pathpkg.Join(dstDir, d.Path)

	// Write pages
	pages := d.Pages
	if d.index != nil {
		pages = append(pages, d.index)
	}
	for _, page := range pages {
		path := page.Path
		if !task.UglyURLs || page == d.index {
			path = pathpkg.Join(path, "index"+task.OutputExt)
		}
		var content []byte
		if cmd := task.Postprocess; cmd != "" {
			content = process(cmd, strings.NewReader(page.Content))
		} else {
			content = []byte(page.Content)
		}

		dstPath := pathpkg.Join(dstDir, path)
		dir := pathpkg.Dir(dstPath)
		os.MkdirAll(dir, 0755)
		if err := os.WriteFile(dstPath, content, 0644); err != nil {
			return err
		}
	}

	// Write the atom feed
	if d.feed != nil {
		const path = "atom.xml"
		dstPath := pathpkg.Join(dirPath, path)
		os.MkdirAll(dirPath, 0755)
		if err := os.WriteFile(dstPath, d.feed, 0644); err != nil {
			return err
		}
	}

	// Write subdirectories
	for _, dir := range d.Dirs {
		dir.write(dstDir, task)
	}
	return nil
}

// sort sorts the directory's pages by weight, then date, then filepath.
func (d *Dir) sort() {
	sort.Slice(d.Pages, func(i, j int) bool {
		pi, pj := d.Pages[i], d.Pages[j]
		return pi.Weight < pj.Weight || pi.Date.Before(pj.Date) ||
			pi.FilePath < pj.FilePath
	})

	for i := range d.Pages {
		if i-1 >= 0 {
			d.Pages[i].Prev = d.Pages[i-1]
		}
		if i+1 < len(d.Pages) {
			d.Pages[i].Next = d.Pages[i+1]
		}
	}

	// Sort subdirectories
	for _, d := range d.Dirs {
		d.sort()
	}
}

// process runs a process command.
func process(command string, input io.Reader) []byte {
	split := strings.Split(command, " ")
	cmd := exec.Command(split[0], split[1:]...)
	cmd.Stdin = input
	cmd.Stderr = os.Stderr
	output, err := cmd.Output()
	if err != nil {
		log.Fatal(err)
	}
	return output
}

func (d *Dir) Title() string {
	return d.index.Title
}

func (d *Dir) Date() time.Time {
	return d.index.Date
}

func (d *Dir) Content() string {
	return d.index.Content
}

func (d *Dir) getDir(path string) *Dir {
	// XXX: This is inefficient
	if d.Path == path {
		return d
	}
	for _, dir := range d.Dirs {
		if dir.Path == path {
			return dir
		}
	}
	for i := range d.Dirs {
		if dir := d.Dirs[i].getDir(path); dir != nil {
			return dir
		}
	}
	return nil
}

func (d *Dir) getPage(path string) *Page {
	// XXX: This is inefficient
	for _, page := range d.Pages {
		if page.FilePath == path {
			return page
		}
	}
	for _, dir := range d.Dirs {
		if page := dir.getPage(path); page != nil {
			return page
		}
	}
	return nil
}