Allow Loading of Diffs that are too large (#17739)

* Allow Loading of Diffs that are too large

This PR allows the loading of diffs that are suppressed because the file
is too large. It does not handle diffs of files which have lines which
are too long.

Fix #17738

Signed-off-by: Andrew Thornton <art27@cantab.net>

2 files changed, 65 insertions, 42 deletions

diff --git a/services/gitdiff/gitdiff.go b/services/gitdiff/gitdiff.go
index c6d11ca89e..c303de0a01 100644
--- a/services/gitdiff/gitdiff.go
+++ b/services/gitdiff/gitdiff.go
@@ -695,6 +695,7 @@ const cmdDiffHead = "diff --git "
 
 // ParsePatch builds a Diff object from a io.Reader and some parameters.
 func ParsePatch(maxLines, maxLineCharacters, maxFiles int, reader io.Reader, skipToFile string) (*Diff, error) {
+	log.Debug("ParsePatch(%d, %d, %d, ..., %s)", maxLines, maxLineCharacters, maxFiles, skipToFile)
 	var curFile *DiffFile
 
 	skipping := skipToFile != ""
@@ -726,7 +727,7 @@ parsingLoop:
 			return diff, fmt.Errorf("invalid first file line: %s", line)
 		}
 
-		if len(diff.Files) >= maxFiles {
+		if maxFiles > -1 && len(diff.Files) >= maxFiles {
 			lastFile := createDiffFile(diff, line)
 			diff.End = lastFile.Name
 			diff.IsIncomplete = true
@@ -1038,7 +1039,7 @@ func parseHunks(curFile *DiffFile, maxLines, maxLineCharacters int, input *bufio
 
 		switch lineBytes[0] {
 		case '@':
-			if curFileLinesCount >= maxLines {
+			if maxLines > -1 && curFileLinesCount >= maxLines {
 				curFile.IsIncomplete = true
 				continue
 			}
@@ -1075,7 +1076,7 @@ func parseHunks(curFile *DiffFile, maxLines, maxLineCharacters int, input *bufio
 			rightLine = lineSectionInfo.RightIdx
 			continue
 		case '\\':
-			if curFileLinesCount >= maxLines {
+			if maxLines > -1 && curFileLinesCount >= maxLines {
 				curFile.IsIncomplete = true
 				continue
 			}
@@ -1090,7 +1091,7 @@ func parseHunks(curFile *DiffFile, maxLines, maxLineCharacters int, input *bufio
 		case '+':
 			curFileLinesCount++
 			curFile.Addition++
-			if curFileLinesCount >= maxLines {
+			if maxLines > -1 && curFileLinesCount >= maxLines {
 				curFile.IsIncomplete = true
 				continue
 			}
@@ -1114,7 +1115,7 @@ func parseHunks(curFile *DiffFile, maxLines, maxLineCharacters int, input *bufio
 		case '-':
 			curFileLinesCount++
 			curFile.Deletion++
-			if curFileLinesCount >= maxLines {
+			if maxLines > -1 && curFileLinesCount >= maxLines {
 				curFile.IsIncomplete = true
 				continue
 			}
@@ -1134,7 +1135,7 @@ func parseHunks(curFile *DiffFile, maxLines, maxLineCharacters int, input *bufio
 			curSection.Lines = append(curSection.Lines, diffLine)
 		case ' ':
 			curFileLinesCount++
-			if curFileLinesCount >= maxLines {
+			if maxLines > -1 && curFileLinesCount >= maxLines {
 				curFile.IsIncomplete = true
 				continue
 			}
@@ -1278,13 +1279,25 @@ func readFileName(rd *strings.Reader) (string, bool) {
 	return name[2:], ambiguity
 }
 
-// GetDiffRangeWithWhitespaceBehavior builds a Diff between two commits of a repository.
+// DiffOptions represents the options for a DiffRange
+type DiffOptions struct {
+	BeforeCommitID     string
+	AfterCommitID      string
+	SkipTo             string
+	MaxLines           int
+	MaxLineCharacters  int
+	MaxFiles           int
+	WhitespaceBehavior string
+	DirectComparison   bool
+}
+
+// GetDiff builds a Diff between two commits of a repository.
 // Passing the empty string as beforeCommitID returns a diff from the parent commit.
 // The whitespaceBehavior is either an empty string or a git flag
-func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID, afterCommitID, skipTo string, maxLines, maxLineCharacters, maxFiles int, whitespaceBehavior string, directComparison bool) (*Diff, error) {
+func GetDiff(gitRepo *git.Repository, opts *DiffOptions, files ...string) (*Diff, error) {
 	repoPath := gitRepo.Path
 
-	commit, err := gitRepo.GetCommit(afterCommitID)
+	commit, err := gitRepo.GetCommit(opts.AfterCommitID)
 	if err != nil {
 		return nil, err
 	}
@@ -1293,45 +1306,54 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 	defer cancel()
 
 	argsLength := 6
-	if len(whitespaceBehavior) > 0 {
+	if len(opts.WhitespaceBehavior) > 0 {
 		argsLength++
 	}
-	if len(skipTo) > 0 {
+	if len(opts.SkipTo) > 0 {
 		argsLength++
 	}
+	if len(files) > 0 {
+		argsLength += len(files) + 1
+	}
 
 	diffArgs := make([]string, 0, argsLength)
-	if (len(beforeCommitID) == 0 || beforeCommitID == git.EmptySHA) && commit.ParentCount() == 0 {
+	if (len(opts.BeforeCommitID) == 0 || opts.BeforeCommitID == git.EmptySHA) && commit.ParentCount() == 0 {
 		diffArgs = append(diffArgs, "diff", "--src-prefix=\\a/", "--dst-prefix=\\b/", "-M")
-		if len(whitespaceBehavior) != 0 {
-			diffArgs = append(diffArgs, whitespaceBehavior)
+		if len(opts.WhitespaceBehavior) != 0 {
+			diffArgs = append(diffArgs, opts.WhitespaceBehavior)
 		}
 		// append empty tree ref
 		diffArgs = append(diffArgs, "4b825dc642cb6eb9a060e54bf8d69288fbee4904")
-		diffArgs = append(diffArgs, afterCommitID)
+		diffArgs = append(diffArgs, opts.AfterCommitID)
 	} else {
-		actualBeforeCommitID := beforeCommitID
+		actualBeforeCommitID := opts.BeforeCommitID
 		if len(actualBeforeCommitID) == 0 {
 			parentCommit, _ := commit.Parent(0)
 			actualBeforeCommitID = parentCommit.ID.String()
 		}
 		diffArgs = append(diffArgs, "diff", "--src-prefix=\\a/", "--dst-prefix=\\b/", "-M")
-		if len(whitespaceBehavior) != 0 {
-			diffArgs = append(diffArgs, whitespaceBehavior)
+		if len(opts.WhitespaceBehavior) != 0 {
+			diffArgs = append(diffArgs, opts.WhitespaceBehavior)
 		}
 		diffArgs = append(diffArgs, actualBeforeCommitID)
-		diffArgs = append(diffArgs, afterCommitID)
-		beforeCommitID = actualBeforeCommitID
+		diffArgs = append(diffArgs, opts.AfterCommitID)
+		opts.BeforeCommitID = actualBeforeCommitID
 	}
 
 	// In git 2.31, git diff learned --skip-to which we can use to shortcut skip to file
 	// so if we are using at least this version of git we don't have to tell ParsePatch to do
 	// the skipping for us
-	parsePatchSkipToFile := skipTo
-	if skipTo != "" && git.CheckGitVersionAtLeast("2.31") == nil {
-		diffArgs = append(diffArgs, "--skip-to="+skipTo)
+	parsePatchSkipToFile := opts.SkipTo
+	if opts.SkipTo != "" && git.CheckGitVersionAtLeast("2.31") == nil {
+		diffArgs = append(diffArgs, "--skip-to="+opts.SkipTo)
 		parsePatchSkipToFile = ""
 	}
+
+	if len(files) > 0 {
+		diffArgs = append(diffArgs, "--")
+		diffArgs = append(diffArgs, files...)
+	}
+
 	cmd := exec.CommandContext(ctx, git.GitExecutable, diffArgs...)
 
 	cmd.Dir = repoPath
@@ -1349,16 +1371,16 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 	pid := process.GetManager().Add(fmt.Sprintf("GetDiffRange [repo_path: %s]", repoPath), cancel)
 	defer process.GetManager().Remove(pid)
 
-	diff, err := ParsePatch(maxLines, maxLineCharacters, maxFiles, stdout, parsePatchSkipToFile)
+	diff, err := ParsePatch(opts.MaxLines, opts.MaxLineCharacters, opts.MaxFiles, stdout, parsePatchSkipToFile)
 	if err != nil {
 		return nil, fmt.Errorf("unable to ParsePatch: %w", err)
 	}
-	diff.Start = skipTo
+	diff.Start = opts.SkipTo
 
 	var checker *git.CheckAttributeReader
 
 	if git.CheckGitVersionAtLeast("1.7.8") == nil {
-		indexFilename, worktree, deleteTemporaryFile, err := gitRepo.ReadTreeToTemporaryIndex(afterCommitID)
+		indexFilename, worktree, deleteTemporaryFile, err := gitRepo.ReadTreeToTemporaryIndex(opts.AfterCommitID)
 		if err == nil {
 			defer deleteTemporaryFile()
 
@@ -1370,12 +1392,12 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 			}
 			ctx, cancel := context.WithCancel(git.DefaultContext)
 			if err := checker.Init(ctx); err != nil {
-				log.Error("Unable to open checker for %s. Error: %v", afterCommitID, err)
+				log.Error("Unable to open checker for %s. Error: %v", opts.AfterCommitID, err)
 			} else {
 				go func() {
 					err := checker.Run()
 					if err != nil && err != ctx.Err() {
-						log.Error("Unable to open checker for %s. Error: %v", afterCommitID, err)
+						log.Error("Unable to open checker for %s. Error: %v", opts.AfterCommitID, err)
 					}
 					cancel()
 				}()
@@ -1426,7 +1448,7 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 			diffFile.IsGenerated = analyze.IsGenerated(diffFile.Name)
 		}
 
-		tailSection := diffFile.GetTailSection(gitRepo, beforeCommitID, afterCommitID)
+		tailSection := diffFile.GetTailSection(gitRepo, opts.BeforeCommitID, opts.AfterCommitID)
 		if tailSection != nil {
 			diffFile.Sections = append(diffFile.Sections, tailSection)
 		}
@@ -1437,19 +1459,19 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 	}
 
 	separator := "..."
-	if directComparison {
+	if opts.DirectComparison {
 		separator = ".."
 	}
 
-	shortstatArgs := []string{beforeCommitID + separator + afterCommitID}
-	if len(beforeCommitID) == 0 || beforeCommitID == git.EmptySHA {
-		shortstatArgs = []string{git.EmptyTreeSHA, afterCommitID}
+	shortstatArgs := []string{opts.BeforeCommitID + separator + opts.AfterCommitID}
+	if len(opts.BeforeCommitID) == 0 || opts.BeforeCommitID == git.EmptySHA {
+		shortstatArgs = []string{git.EmptyTreeSHA, opts.AfterCommitID}
 	}
 	diff.NumFiles, diff.TotalAddition, diff.TotalDeletion, err = git.GetDiffShortStat(repoPath, shortstatArgs...)
 	if err != nil && strings.Contains(err.Error(), "no merge base") {
 		// git >= 2.28 now returns an error if base and head have become unrelated.
 		// previously it would return the results of git diff --shortstat base head so let's try that...
-		shortstatArgs = []string{beforeCommitID, afterCommitID}
+		shortstatArgs = []string{opts.BeforeCommitID, opts.AfterCommitID}
 		diff.NumFiles, diff.TotalAddition, diff.TotalDeletion, err = git.GetDiffShortStat(repoPath, shortstatArgs...)
 	}
 	if err != nil {
@@ -1459,12 +1481,6 @@ func GetDiffRangeWithWhitespaceBehavior(gitRepo *git.Repository, beforeCommitID,
 	return diff, nil
 }
 
-// GetDiffCommitWithWhitespaceBehavior builds a Diff representing the given commitID.
-// The whitespaceBehavior is either an empty string or a git flag
-func GetDiffCommitWithWhitespaceBehavior(gitRepo *git.Repository, commitID, skipTo string, maxLines, maxLineCharacters, maxFiles int, whitespaceBehavior string, directComparison bool) (*Diff, error) {
-	return GetDiffRangeWithWhitespaceBehavior(gitRepo, "", commitID, skipTo, maxLines, maxLineCharacters, maxFiles, whitespaceBehavior, directComparison)
-}
-
 // CommentAsDiff returns c.Patch as *Diff
 func CommentAsDiff(c *models.Comment) (*Diff, error) {
 	diff, err := ParsePatch(setting.Git.MaxGitDiffLines,
diff --git a/services/gitdiff/gitdiff_test.go b/services/gitdiff/gitdiff_test.go
index 76a8b4e7ca..7d63beffeb 100644
--- a/services/gitdiff/gitdiff_test.go
+++ b/services/gitdiff/gitdiff_test.go
@@ -693,8 +693,15 @@ func TestGetDiffRangeWithWhitespaceBehavior(t *testing.T) {
 	}
 	defer gitRepo.Close()
 	for _, behavior := range []string{"-w", "--ignore-space-at-eol", "-b", ""} {
-		diffs, err := GetDiffRangeWithWhitespaceBehavior(gitRepo, "559c156f8e0178b71cb44355428f24001b08fc68", "bd7063cc7c04689c4d082183d32a604ed27a24f9", "",
-			setting.Git.MaxGitDiffLines, setting.Git.MaxGitDiffLines, setting.Git.MaxGitDiffFiles, behavior, false)
+		diffs, err := GetDiff(gitRepo,
+			&DiffOptions{
+				AfterCommitID:      "bd7063cc7c04689c4d082183d32a604ed27a24f9",
+				BeforeCommitID:     "559c156f8e0178b71cb44355428f24001b08fc68",
+				MaxLines:           setting.Git.MaxGitDiffLines,
+				MaxLineCharacters:  setting.Git.MaxGitDiffLineCharacters,
+				MaxFiles:           setting.Git.MaxGitDiffFiles,
+				WhitespaceBehavior: behavior,
+			})
 		assert.NoError(t, err, fmt.Sprintf("Error when diff with %s", behavior))
 		for _, f := range diffs.Files {
 			assert.True(t, len(f.Sections) > 0, fmt.Sprintf("%s should have sections", f.Name))