edge/pkg/bundle/zim_bundle.go

package bundle

import (
	"bytes"
	"context"
	"io"
	"io/fs"
	"os"
	"path/filepath"
	"strings"
	"sync"
	"time"

	"golang.org/x/net/html"

	"forge.cadoles.com/arcad/edge/pkg/bundle/zim"
	lru "github.com/hashicorp/golang-lru/v2"
	"github.com/pkg/errors"
	"gitlab.com/wpetit/goweb/logger"
	"gopkg.in/yaml.v2"
)

type ZimBundle struct {
	archivePath string

	initOnce sync.Once
	initErr  error

	reader            *zim.Reader
	urlNamespaceCache *lru.Cache[string, zim.Namespace]
}

func (b *ZimBundle) File(filename string) (io.ReadCloser, os.FileInfo, error) {
	ctx := logger.With(
		context.Background(),
		logger.F("filename", filename),
	)

	logger.Debug(ctx, "opening file")

	switch filename {
	case "manifest.yml":
		return b.renderFakeManifest(ctx)
	case "server/main.js":
		return b.renderFakeServerMain(ctx)
	case "public":
		return b.renderDirectory(ctx, filename)
	case "public/index.html":
		return b.renderMainPage(ctx, filename)

	default:
		return b.renderURL(ctx, filename)
	}
}

func (b *ZimBundle) Dir(dirname string) ([]os.FileInfo, error) {
	files := make([]os.FileInfo, 0)
	return files, nil
}

func (b *ZimBundle) renderFakeManifest(ctx context.Context) (io.ReadCloser, os.FileInfo, error) {
	if err := b.init(); err != nil {
		return nil, nil, errors.WithStack(err)
	}

	metadata, err := b.reader.Metadata()
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	manifest := map[string]any{}

	manifest["version"] = "0.0.0"

	if name, exists := metadata[zim.MetadataName]; exists {
		replacer := strings.NewReplacer(
			"_", "",
			" ", "",
		)

		manifest["id"] = strings.ToLower(replacer.Replace(name)) + ".zim.edge.app"
	} else {
		manifest["id"] = b.reader.UUID() + ".zim.edge.app"
	}

	if title, exists := metadata[zim.MetadataTitle]; exists {
		manifest["title"] = title
	} else {
		manifest["title"] = "Unknown"
	}

	if description, exists := metadata[zim.MetadataDescription]; exists {
		manifest["description"] = description
	}

	favicon, err := b.reader.Favicon()
	if err != nil && !errors.Is(err, zim.ErrNotFound) {
		return nil, nil, errors.WithStack(err)
	}

	if favicon != nil {
		manifestMeta, exists := manifest["metadata"].(map[string]any)
		if !exists {
			manifestMeta = make(map[string]any)
			manifest["metadata"] = manifestMeta
		}

		paths, exists := manifestMeta["paths"].(map[string]any)
		if !exists {
			paths = make(map[string]any)
			manifestMeta["paths"] = paths
		}

		paths["icon"] = "/" + favicon.FullURL()
	}

	data, err := yaml.Marshal(manifest)
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	stat := &zimFileInfo{
		isDir:   false,
		modTime: time.Time{},
		mode:    0,
		name:    "manifest.yml",
		size:    int64(len(data)),
	}

	buf := bytes.NewBuffer(data)
	file := io.NopCloser(buf)

	return file, stat, nil
}

func (b *ZimBundle) renderFakeServerMain(ctx context.Context) (io.ReadCloser, os.FileInfo, error) {
	stat := &zimFileInfo{
		isDir:   false,
		modTime: time.Time{},
		mode:    0,
		name:    "server/main.js",
		size:    0,
	}

	buf := bytes.NewBuffer(nil)
	file := io.NopCloser(buf)

	return file, stat, nil
}

func (b *ZimBundle) renderURL(ctx context.Context, url string) (io.ReadCloser, os.FileInfo, error) {
	if err := b.init(); err != nil {
		return nil, nil, errors.WithStack(err)
	}

	filename := filepath.Base(url)
	url = strings.TrimPrefix(url, "public/")

	entry, err := b.searchEntryFromURL(ctx, url)
	if err != nil {
		if errors.Is(err, zim.ErrNotFound) {
			return nil, nil, os.ErrNotExist
		}

		return nil, nil, errors.WithStack(err)
	}

	logger.Debug(
		ctx, "found zim entry",
		logger.F("webURL", url),
		logger.F("zimFullURL", entry.FullURL()),
	)

	content, err := entry.Redirect()
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	contentReader, err := content.Reader()
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	size, err := contentReader.Size()
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	mimeType := content.MimeType()
	if mimeType != "text/html" {
		zimFile := &zimFile{
			fileInfo: &zimFileInfo{
				isDir:   false,
				modTime: time.Time{},
				mode:    0,
				name:    filename,
				size:    size,
			},
			reader: contentReader,
		}

		return zimFile, zimFile.fileInfo, nil
	}

	// Read HTML file and inject Edge scripts

	data, err := io.ReadAll(contentReader)
	if err != nil {
		return nil, nil, err
	}

	injected, err := b.injectEdgeScriptTag(data)
	if err != nil {
		logger.Error(ctx, "could not inject edge script", logger.E(errors.WithStack(err)))
	} else {
		data = injected
	}

	zimFile := &zimFile{
		fileInfo: &zimFileInfo{
			isDir:   false,
			modTime: time.Time{},
			mode:    0,
			name:    filename,
			size:    size,
		},
		reader: io.NopCloser(bytes.NewBuffer(data)),
	}

	return zimFile, zimFile.fileInfo, nil
}

func (b *ZimBundle) searchEntryFromURL(ctx context.Context, url string) (zim.Entry, error) {
	ctx = logger.With(ctx, logger.F("webURL", url))

	logger.Debug(ctx, "searching entry namespace in local cache")

	// Search URL namespace from cache
	if namespace, found := b.urlNamespaceCache.Get(url); found {
		logger.Debug(ctx, "found entry namespace in cache")

		entry, err := b.reader.EntryWithURL(namespace, url)
		if err != nil {
			return nil, errors.WithStack(err)
		}

		return entry, nil
	}

	// Try to access entry directly if the URL match the pattern <NS>/<URL>
	urlParts := strings.SplitN(url, "/", 2)
	if len(urlParts) == 2 && len(urlParts[0]) == 1 {
		namespace := zim.Namespace(urlParts[0])
		url = urlParts[1]

		logger.Debug(
			ctx, "trying to access entry directly",
			logger.F("zimNamespace", namespace),
			logger.F("zimURL", url),
		)

		entry, err := b.reader.EntryWithURL(namespace, url)
		if err != nil && !errors.Is(err, zim.ErrNotFound) {
			return nil, errors.WithStack(err)
		}

		if entry != nil {
			b.urlNamespaceCache.Add(url, entry.Namespace())
			return entry, nil
		}
	}

	contentNamespaces := []zim.Namespace{
		zim.V6NamespaceContent,
		zim.V6NamespaceMetadata,
		zim.V5NamespaceLayout,
		zim.V5NamespaceArticle,
		zim.V5NamespaceImageFile,
		zim.V5NamespaceMetadata,
	}

	logger.Debug(
		ctx, "make educated guesses about potential url namespace",
		logger.F("zimNamespaces", contentNamespaces),
	)

	for _, ns := range contentNamespaces {
		logger.Debug(
			ctx, "trying to access entry directly",
			logger.F("zimNamespace", ns),
			logger.F("zimURL", url),
		)

		entry, err := b.reader.EntryWithURL(ns, url)
		if err != nil && !errors.Is(err, zim.ErrNotFound) {
			return nil, errors.WithStack(err)
		}

		if entry != nil {
			b.urlNamespaceCache.Add(url, entry.Namespace())
			return entry, nil
		}
	}

	logger.Debug(ctx, "doing full entries scan")

	var entry zim.Entry

	iterator := b.reader.Entries()
	for iterator.Next() {
		current := iterator.Entry()

		if current.FullURL() != url && current.URL() != url {
			continue
		}

		entry = current
		b.urlNamespaceCache.Add(url, entry.Namespace())
		break
	}
	if err := iterator.Err(); err != nil {
		return nil, errors.WithStack(err)
	}

	if entry == nil {
		return nil, errors.WithStack(zim.ErrNotFound)
	}

	return entry, nil
}

func (b *ZimBundle) renderDirectory(ctx context.Context, filename string) (io.ReadCloser, os.FileInfo, error) {
	zimFile := &zimFile{
		fileInfo: &zimFileInfo{
			isDir:   true,
			modTime: time.Time{},
			mode:    0,
			name:    filename,
			size:    0,
		},
		reader: io.NopCloser(bytes.NewBuffer(nil)),
	}

	return zimFile, zimFile.fileInfo, nil
}

func (b *ZimBundle) renderMainPage(ctx context.Context, filename string) (io.ReadCloser, os.FileInfo, error) {
	if err := b.init(); err != nil {
		return nil, nil, errors.WithStack(err)
	}

	main, err := b.reader.MainPage()
	if err != nil {
		return nil, nil, errors.WithStack(err)
	}

	return b.renderURL(ctx, main.FullURL())
}

func (b *ZimBundle) injectEdgeScriptTag(data []byte) ([]byte, error) {
	buff := bytes.NewBuffer(data)
	doc, err := html.Parse(buff)
	if err != nil {
		return nil, errors.WithStack(err)
	}

	var f func(*html.Node) bool
	f = func(n *html.Node) bool {
		if n.Type == html.ElementNode && n.Data == "head" {
			script := &html.Node{
				Type: html.ElementNode,
				Data: "script",
				Attr: []html.Attribute{
					{
						Key: "src",
						Val: "/edge/sdk/client.js",
					},
				},
			}

			n.AppendChild(script)

			return false
		}

		for c := n.FirstChild; c != nil; c = c.NextSibling {
			if keepWalking := f(c); !keepWalking {
				return false
			}
		}

		return true
	}

	f(doc)

	buff.Reset()

	if err := html.Render(buff, doc); err != nil {
		return nil, errors.WithStack(err)
	}

	return buff.Bytes(), nil
}

func (b *ZimBundle) init() error {
	b.initOnce.Do(func() {
		reader, err := zim.Open(b.archivePath)
		if err != nil {
			b.initErr = errors.Wrapf(err, "could not open '%v'", b.archivePath)
			return
		}

		b.reader = reader

		cache, err := lru.New[string, zim.Namespace](128)
		if err != nil {
			b.initErr = errors.Wrap(err, "could not initialize cache")
			return
		}

		b.urlNamespaceCache = cache
	})
	if b.initErr != nil {
		return errors.WithStack(b.initErr)
	}

	return nil
}

func NewZimBundle(archivePath string) *ZimBundle {
	return &ZimBundle{
		archivePath: archivePath,
	}
}

type zimFile struct {
	fileInfo *zimFileInfo
	reader   io.ReadCloser
}

// Close implements fs.File.
func (f *zimFile) Close() error {
	if err := f.reader.Close(); err != nil {
		return errors.WithStack(err)
	}

	return nil
}

// Read implements fs.File.
func (f *zimFile) Read(d []byte) (int, error) {
	n, err := f.reader.Read(d)
	if err != nil {
		if errors.Is(err, io.EOF) {
			return n, err
		}

		return n, errors.WithStack(err)
	}

	return n, nil
}

// Stat implements fs.File.
func (f *zimFile) Stat() (fs.FileInfo, error) {
	return f.fileInfo, nil
}

var _ fs.File = &zimFile{}

type zimFileInfo struct {
	isDir   bool
	modTime time.Time
	mode    fs.FileMode
	name    string
	size    int64
}

// IsDir implements fs.FileInfo.
func (i *zimFileInfo) IsDir() bool {
	return i.isDir
}

// ModTime implements fs.FileInfo.
func (i *zimFileInfo) ModTime() time.Time {
	return i.modTime
}

// Mode implements fs.FileInfo.
func (i *zimFileInfo) Mode() fs.FileMode {
	return i.mode
}

// Name implements fs.FileInfo.
func (i *zimFileInfo) Name() string {
	return i.name
}

// Size implements fs.FileInfo.
func (i *zimFileInfo) Size() int64 {
	return i.size
}

// Sys implements fs.FileInfo.
func (*zimFileInfo) Sys() any {
	return nil
}

var _ fs.FileInfo = &zimFileInfo{}