Warn when a Git repository is uploaded with Git LFS-tracked files.

src/fetch.go

@@ -9,6 +9,7 @@ import (
 	"net/url"
 	"os"
 	"slices"
+	"strings"
 
 	"github.com/c2h5oh/datasize"
 	"github.com/go-git/go-billy/v6/osfs"
@@ -209,6 +210,8 @@ func FetchRepository(
 		datasize.ByteSize(dataBytesTransferred).HR(),
 	)
 
+	warnAboutGitLFS(ctx, manifest)
+
 	return manifest, nil
 }
 
@@ -254,3 +257,18 @@ func readGitBlob(
 
 	return nil
 }
+
+func warnAboutGitLFS(ctx context.Context, manifest *Manifest) {
+	gitattributes := ReadGitAttributes(ctx, manifest)
+	for _, name := range slices.Sorted(maps.Keys(manifest.GetContents())) {
+		entry := manifest.GetContents()[name]
+		if !IsEntryRegularFile(entry) {
+			continue
+		}
+		parts := strings.Split(name, "/")
+		attrs, _ := gitattributes.Match(parts, nil)
+		if attr, ok := attrs["filter"]; ok && attr.Value() == "lfs" {
+			AddProblem(manifest, name, "git-pages does not support Git LFS; move this file into Git or use incremental uploads")
+		}
+	}
+}

src/gitattributes.go

@@ -0,0 +1,61 @@
+package git_pages
+
+import (
+	"bytes"
+	"cmp"
+	"context"
+	"slices"
+	"strings"
+
+	"github.com/go-git/go-git/v6/plumbing/format/gitattributes"
+)
+
+func ReadGitAttributes(ctx context.Context, manifest *Manifest) gitattributes.Matcher {
+	type entryPair struct {
+		parts []string
+		entry *Entry
+	}
+
+	// Collect all .gitattributes files.
+	var files []entryPair
+	for name, entry := range manifest.GetContents() {
+		switch entry.GetType() {
+		case Type_InlineFile, Type_ExternalFile:
+			parts := strings.Split(name, "/")
+			if parts[len(parts)-1] == ".gitattributes" {
+				files = append(files, entryPair{parts, entry})
+			}
+		}
+	}
+
+	// Sort the file list by depth, then by name.
+	slices.SortFunc(files, func(a entryPair, b entryPair) int {
+		return cmp.Or(
+			cmp.Compare(len(a.parts), len(b.parts)),
+			slices.Compare(a.parts, b.parts),
+		)
+	})
+
+	// Gather all .gitattributes rules, sorted by depth.
+	var rules []gitattributes.MatchAttribute
+	for _, pair := range files {
+		parts, entry := pair.parts, pair.entry
+		data, err := GetEntryContents(ctx, entry)
+		if err != nil {
+			continue
+		}
+		dirs := parts[:len(parts)-1]
+		isRoot := len(parts) == 1
+		newRules, err := gitattributes.ReadAttributes(bytes.NewReader(data), dirs, isRoot)
+		if err != nil {
+			AddProblem(manifest, strings.Join(parts, "/"), "parsing .gitattributes: %v", err)
+			continue
+		}
+		rules = append(rules, newRules...)
+	}
+
+	// gitattributes.Matcher applies rules in reverse.
+	slices.Reverse(rules)
+	matcher := gitattributes.NewMatcher(rules)
+	return matcher
+}

src/headers.go

@@ -1,6 +1,7 @@
 package git_pages
 
 import (
+	"context"
 	"errors"
 	"fmt"
 	"net/http"
@@ -85,17 +86,22 @@ func validateHeaderRule(rule headers.Rule) error {
 }
 
 // Parses redirects file and injects rules into the manifest.
-func ProcessHeadersFile(manifest *Manifest) error {
+func ProcessHeadersFile(ctx context.Context, manifest *Manifest) error {
 	headersEntry := manifest.Contents[HeadersFileName]
 	delete(manifest.Contents, HeadersFileName)
 	if headersEntry == nil {
 		return nil
-	} else if headersEntry.GetType() != Type_InlineFile {
-		return AddProblem(manifest, HeadersFileName,
-			"not a regular file")
 	}
 
-	rules, err := headers.ParseString(string(headersEntry.GetData()))
+	data, err := GetEntryContents(ctx, headersEntry)
+	if errors.Is(err, ErrNotRegularFile) {
+		return AddProblem(manifest, HeadersFileName,
+			"not a regular file")
+	} else if err != nil {
+		return err
+	}
+
+	rules, err := headers.ParseString(string(data))
 	if err != nil {
 		return AddProblem(manifest, HeadersFileName,
 			"syntax error: %s", err)

src/manifest.go

@@ -8,6 +8,7 @@ import (
 	"crypto/sha256"
 	"errors"
 	"fmt"
+	"io"
 	"mime"
 	"net/http"
 	"path"
@@ -144,6 +145,44 @@ func AddProblem(manifest *Manifest, pathName, format string, args ...any) error
 	return fmt.Errorf("%s: %s", pathName, cause)
 }
 
+func IsEntryRegularFile(entry *Entry) bool {
+	return entry.GetType() == Type_InlineFile ||
+		entry.GetType() == Type_ExternalFile
+}
+
+var ErrNotRegularFile = errors.New("not a regular file")
+
+func GetEntryContents(ctx context.Context, entry *Entry) (data []byte, err error) {
+	switch entry.GetType() {
+	case Type_InlineFile:
+		data = entry.GetData()
+	case Type_ExternalFile:
+		reader, _, err := backend.GetBlob(ctx, string(entry.GetData()))
+		if err != nil {
+			return nil, err
+		}
+		data, err = io.ReadAll(reader)
+		if err != nil {
+			return nil, err
+		}
+	default:
+		return nil, ErrNotRegularFile
+	}
+
+	switch entry.GetTransform() {
+	case Transform_Identity:
+	case Transform_Zstd:
+		data, err = zstdDecoder.DecodeAll(data, []byte{})
+		if err != nil {
+			return nil, err
+		}
+	default:
+		return nil, fmt.Errorf("unexpected transform")
+	}
+
+	return
+}
+
 // EnsureLeadingDirectories adds directory entries for any parent directories
 // that are implicitly referenced by files in the manifest but don't have
 // explicit directory entries. (This can be the case if an archive is created
@@ -275,7 +314,7 @@ func CompressFiles(ctx context.Context, manifest *Manifest) {
 // (Perhaps in the future they could be exposed at `.git-pages/status.txt`?)
 func PrepareManifest(ctx context.Context, manifest *Manifest) error {
 	// Parse Netlify-style `_redirects`.
-	if err := ProcessRedirectsFile(manifest); err != nil {
+	if err := ProcessRedirectsFile(ctx, manifest); err != nil {
 		logc.Printf(ctx, "redirects err: %s\n", err)
 	} else if len(manifest.Redirects) > 0 {
 		logc.Printf(ctx, "redirects ok: %d rules\n", len(manifest.Redirects))
@@ -285,7 +324,7 @@ func PrepareManifest(ctx context.Context, manifest *Manifest) error {
 	LintRedirects(manifest)
 
 	// Parse Netlify-style `_headers`.
-	if err := ProcessHeadersFile(manifest); err != nil {
+	if err := ProcessHeadersFile(ctx, manifest); err != nil {
 		logc.Printf(ctx, "headers err: %s\n", err)
 	} else if len(manifest.Headers) > 0 {
 		logc.Printf(ctx, "headers ok: %d rules\n", len(manifest.Headers))

src/redirects.go

@@ -1,6 +1,8 @@
 package git_pages
 
 import (
+	"context"
+	"errors"
 	"fmt"
 	"net/http"
 	"net/url"
@@ -96,17 +98,22 @@ func validateRedirectRule(rule *redirects.Rule) error {
 }
 
 // Parses redirects file and injects rules into the manifest.
-func ProcessRedirectsFile(manifest *Manifest) error {
+func ProcessRedirectsFile(ctx context.Context, manifest *Manifest) error {
 	redirectsEntry := manifest.Contents[RedirectsFileName]
 	delete(manifest.Contents, RedirectsFileName)
 	if redirectsEntry == nil {
 		return nil
-	} else if redirectsEntry.GetType() != Type_InlineFile {
-		return AddProblem(manifest, RedirectsFileName,
-			"not a regular file")
 	}
 
-	rules, err := redirects.ParseString(string(redirectsEntry.GetData()))
+	data, err := GetEntryContents(ctx, redirectsEntry)
+	if errors.Is(err, ErrNotRegularFile) {
+		return AddProblem(manifest, RedirectsFileName,
+			"not a regular file")
+	} else if err != nil {
+		return err
+	}
+
+	rules, err := redirects.ParseString(string(data))
 	if err != nil {
 		return AddProblem(manifest, RedirectsFileName,
 			"syntax error: %s", err)