review-bot/gitea/position.go

package gitea

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

// PositionMap holds a per-file mapping of GitHub diff-position to new-file line number.
// Position is a 1-indexed offset from the @@ hunk header line in the unified diff.
type PositionMap struct {
	// files maps filename → (position → new-file line number).
	// Deletion lines are mapped to -1 (no new-file line).
	// Hunk-header lines are mapped to 0 (no new-file line).
	files map[string]map[int]int
	// maxPositions caches the highest position number per file,
	// tracked during construction to avoid O(n) scans at translate time.
	maxPositions map[string]int
}

// Translate converts a GitHub diff-position to a new-file line number for a given file.
// Returns an error if the file is not in the diff or the position is out of range.
// If the position targets a deletion or hunk-header line, it maps to the nearest
// context/addition line below; if no such line exists, returns an error.
func (pm *PositionMap) Translate(file string, position int) (int, error) {
	if pm == nil || pm.files == nil {
		return 0, fmt.Errorf("empty position map")
	}

	fileMap, ok := pm.files[file]
	if !ok {
		return 0, fmt.Errorf("file %q not found in diff", file)
	}

	if position < 1 {
		return 0, fmt.Errorf("position %d out of range (must be >= 1)", position)
	}

	lineNum, ok := fileMap[position]
	if !ok {
		return 0, fmt.Errorf("position %d out of range for file %q", position, file)
	}

	// lineNum == -1 means this position is a deletion line.
	// lineNum == 0 means this position is a hunk-header line.
	// Both map to the nearest context/addition line below.
	if lineNum <= 0 {
		maxPos := pm.maxPosition(file)
		for p := position + 1; p <= maxPos; p++ {
			if ln, exists := fileMap[p]; exists && ln > 0 {
				return ln, nil
			}
		}
		if lineNum == 0 {
			return 0, fmt.Errorf("position %d targets a hunk-header line with no subsequent new-file line in %q", position, file)
		}
		return 0, fmt.Errorf("position %d targets a deletion line with no subsequent new-file line in %q", position, file)
	}

	return lineNum, nil
}

// maxPosition returns the highest position number for a file.
// O(1) — the maximum is tracked during map construction.
func (pm *PositionMap) maxPosition(file string) int {
	return pm.maxPositions[file]
}

// BuildPositionToLineMap parses a unified diff and builds a PositionMap
// mapping diff-position → new-file line number per file.
//
// Diff-position counting rules (GitHub spec):
//   - The @@ hunk header line is position 1 for the file's first hunk
//   - Every subsequent line increments position by 1 — context, additions, AND deletions
//   - A new @@ hunk within the same file continues incrementing (does not reset)
//   - Position maps to the new file line number for additions and context lines
//   - Deletion lines have a position but no new-file line number (stored as -1)
//   - Hunk-header lines have a position but no new-file line number (stored as 0)
func BuildPositionToLineMap(diff string) *PositionMap {
	pm := &PositionMap{
		files:        make(map[string]map[int]int),
		maxPositions: make(map[string]int),
	}

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

	for _, line := range lines {
		// Detect new file in diff.
		// "+++ b/" is checked before "+++ /dev/null" — the two prefixes are
		// non-overlapping ("+++ /dev/null" does not start with "+++ b/"), so
		// ordering is independent. Checking the common case first for clarity.
		if strings.HasPrefix(line, "+++ b/") {
			currentFile = strings.TrimPrefix(line, "+++ b/")
			position = 0
			newLine = 0
			if pm.files[currentFile] == nil {
				pm.files[currentFile] = make(map[int]int)
			}
			continue
		}

		// Deleted file: +++ /dev/null means the file is being deleted
		if strings.HasPrefix(line, "+++ /dev/null") {
			currentFile = ""
			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
		}

		// Binary file detection
		if strings.HasPrefix(line, "Binary files") {
			currentFile = ""
			continue
		}

		// Parse hunk headers
		if strings.HasPrefix(line, "@@") && currentFile != "" {
			position++
			pm.files[currentFile][position] = 0 // sentinel: hunk-header has no new-file line
			pm.maxPositions[currentFile] = position
			newLine = parseHunkStart(line)
			continue
		}

		if currentFile == "" {
			continue
		}

		// Skip "\ No newline at end of file" markers
		if strings.HasPrefix(line, `\`) {
			continue
		}

		// Process diff content lines
		if strings.HasPrefix(line, "+") {
			// Addition: has a new-file line number
			position++
			pm.files[currentFile][position] = newLine
			pm.maxPositions[currentFile] = position
			newLine++
		} else if strings.HasPrefix(line, "-") {
			// Deletion: has a position but no new-file line number
			position++
			pm.files[currentFile][position] = -1
			pm.maxPositions[currentFile] = position
		} else if strings.HasPrefix(line, " ") {
			// Context line
			position++
			pm.files[currentFile][position] = newLine
			pm.maxPositions[currentFile] = position
			newLine++
		}
	}

	return pm
}

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

	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
}