review-bot/vcs/util.go

package vcs

import (
	"context"
	"fmt"
	"strconv"
	"strings"
)

const (
	// maxFilesInPath is the maximum number of files GetAllFilesInPath will fetch.
	// Prevents unbounded resource consumption on very large directory trees.
	maxFilesInPath = 10000

	// maxTotalBytesInPath is the maximum total bytes GetAllFilesInPath will accumulate.
	// Prevents memory exhaustion when fetching large repositories.
	maxTotalBytesInPath = 100 * 1024 * 1024 // 100 MB
)

// GetAllFilesInPath recursively fetches all file contents under a path using the
// provided FileReader. Returns a map of filepath -> content for all files found.
// If the path points to an empty directory, returns an empty map.
//
// This function uses fail-fast error handling: any error from ListContents or
// GetFileContent aborts the entire traversal and returns the error immediately.
// This differs from gitea.Client.GetAllFilesInPath, which logs errors and continues.
// The fail-fast contract ensures callers can trust that a nil error means all files
// were successfully fetched.
//
// Resource limits: the traversal is bounded by maxFilesInPath (file count) and
// maxTotalBytesInPath (total accumulated bytes). The context is checked before each
// recursive call and file fetch to respect cancellation.
func GetAllFilesInPath(ctx context.Context, client FileReader, owner, repo, path string) (map[string]string, error) {
	results := make(map[string]string)
	totalBytes := 0

	var walk func(string) error
	walk = func(dir string) error {
		if err := ctx.Err(); err != nil {
			return fmt.Errorf("context canceled during traversal: %w", err)
		}

		entries, err := client.ListContents(ctx, owner, repo, dir)
		if err != nil {
			return fmt.Errorf("list contents %q: %w", dir, err)
		}

		for _, entry := range entries {
			if err := ctx.Err(); err != nil {
				return fmt.Errorf("context canceled during traversal: %w", err)
			}

			switch entry.Type {
			case "file":
				if len(results) >= maxFilesInPath {
					return fmt.Errorf("exceeded max file count (%d) in path %q", maxFilesInPath, path)
				}

				content, err := client.GetFileContent(ctx, owner, repo, entry.Path, "")
				if err != nil {
					return fmt.Errorf("get file %q: %w", entry.Path, err)
				}

				totalBytes += len(content)
				if totalBytes > maxTotalBytesInPath {
					return fmt.Errorf("exceeded max total bytes (%d) in path %q", maxTotalBytesInPath, path)
				}

				results[entry.Path] = content
			case "dir":
				if err := walk(entry.Path); err != nil {
					return err
				}
			}
		}
		return nil
	}

	if err := walk(path); err != nil {
		return nil, err
	}
	return results, nil
}

// BuildLineToPositionMap parses a unified diff and returns a map of
// filename -> (new line number -> diff position). The diff position is a
// 1-indexed offset from the @@ hunk header line for each file.
// Only lines that appear in the new file (context lines and additions) are mapped.
// Deletion-only lines are not included.
func BuildLineToPositionMap(diff string) map[string]map[int]int {
	result := make(map[string]map[int]int)

	lines := strings.Split(diff, "\n")
	var currentFile string
	var position int
	var newLine int

	for _, line := range lines {
		// Detect new file in diff
		if strings.HasPrefix(line, "+++ b/") {
			currentFile = strings.TrimPrefix(line, "+++ b/")
			position = 0
			newLine = 0
			if result[currentFile] == nil {
				result[currentFile] = make(map[int]int)
			}
			continue
		}

		// Skip --- lines (old file header)
		if strings.HasPrefix(line, "--- ") {
			continue
		}

		// Skip diff --git lines
		if strings.HasPrefix(line, "diff --git") {
			continue
		}

		// Skip index lines
		if strings.HasPrefix(line, "index ") {
			continue
		}

		// Parse hunk headers
		if strings.HasPrefix(line, "@@") {
			position++
			// Extract new file start line from @@ -a,b +c,d @@
			newLine = parseHunkNewStart(line)
			continue
		}

		// We need a current file to map lines
		if currentFile == "" {
			continue
		}

		// Skip "\ No newline at end of file" markers — these are git diff
		// metadata and not part of the file content.
		if strings.HasPrefix(line, `\`) {
			continue
		}

		// Process diff content lines
		if strings.HasPrefix(line, "+") {
			position++
			result[currentFile][newLine] = position
			newLine++
		} else if strings.HasPrefix(line, "-") {
			position++
			// Deletion lines don't map to new line numbers
		} else if strings.HasPrefix(line, " ") {
			// Context line (space-prefixed).
			// Only map if position > 0, which means we've seen a hunk header.
			// Lines before the first hunk header (position == 0) are not part
			// of any diff hunk and should be skipped.
			if position > 0 {
				position++
				result[currentFile][newLine] = position
				newLine++
			}
		}
	}

	return result
}

// parseHunkNewStart extracts the new-file starting line number from a hunk header.
// Format: @@ -old_start[,old_count] +new_start[,new_count] @@
func parseHunkNewStart(hunkLine string) int {
	// Find the +N part
	plusIdx := strings.Index(hunkLine, "+")
	if plusIdx < 0 {
		return 1
	}
	rest := hunkLine[plusIdx+1:]

	// Find the end of the number (first non-digit after +)
	endIdx := 0
	for endIdx < len(rest) && rest[endIdx] >= '0' && rest[endIdx] <= '9' {
		endIdx++
	}

	if endIdx == 0 {
		return 1
	}

	n, err := strconv.Atoi(rest[:endIdx])
	if err != nil {
		return 1
	}
	return n
}