mesh-catalog/modules/gitea/cmd/forge/applypatch.go

package main

import (
	"encoding/base64"
	"fmt"
	"regexp"
	"sort"
	"strconv"
	"strings"
	"unicode/utf8"
)

// A unified diff applied by the tool itself (novox/hq issue 1604). The forge's diffpatch answers 500 with an empty
// message when an unchanged context line holds a non-ASCII character, so a patch the forge did not take is applied
// here: each touched file is read at the commit the commit goes on, its hunks are applied, and the resulting
// files are committed through the same call a files commit uses. The forge's files call carries no file mode, so
// a patch that creates an executable file, changes a mode or touches a symlink is not applied here: the tool says so.

// fileSource reads a file at the commit the patch is applied to; ok is false when there is no such file.
type fileSource func(path string) (content []byte, mode string, ok bool, err error)

// patchFile is one file section of a patch.
type patchFile struct {
	oldPath, newPath string // "" when /dev/null
	create, remove   bool
	renamed          bool
	oldSeen          bool
	hunks            []hunk
	modeBlock        string // why the files call cannot carry this section, "" when it can
}

type hunk struct {
	oldStart         int
	old, new         []string
	oldNoNL, newNoNL bool
	leading          int // unchanged lines before the first change
	trailing         int // unchanged lines after the last change
}

var hunkHeader = regexp.MustCompile(`^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@`)

// gitPath reads a path of a diff header: git quotes one holding unusual bytes in C style.
func gitPath(s string) string {
	s = strings.TrimRight(s, "\t ")
	if strings.HasPrefix(s, `"`) {
		if u, err := strconv.Unquote(s); err == nil {
			return u
		}
	} else if i := strings.IndexByte(s, '\t'); i >= 0 {
		s = s[:i]
	}
	return s
}

func stripPrefix(p string) string {
	if p == "/dev/null" {
		return ""
	}
	if i := strings.IndexByte(p, '/'); i >= 0 && (p[:i] == "a" || p[:i] == "b") {
		return p[i+1:]
	}
	return p
}

func nonRegular(mode string) bool { return mode != "" && mode != "100644" }

// parsePatch reads a unified diff into its file sections.
func parsePatch(text string) ([]*patchFile, error) {
	lines := strings.Split(text, "\n")
	if n := len(lines); n > 0 && lines[n-1] == "" {
		lines = lines[:n-1]
	}
	var files []*patchFile
	var cur *patchFile
	block := func(f *patchFile, what string) {
		if f.modeBlock == "" {
			f.modeBlock = what
		}
	}
	for i := 0; i < len(lines); i++ {
		l := strings.TrimSuffix(lines[i], "\r")
		switch {
		case strings.HasPrefix(l, "diff --git "):
			cur = &patchFile{}
			files = append(files, cur)
			rest := strings.TrimPrefix(l, "diff --git ")
			if k := strings.Index(rest, " b/"); k > 0 && !strings.HasPrefix(rest, `"`) {
				cur.oldPath, cur.newPath = stripPrefix(rest[:k]), stripPrefix(rest[k+1:])
			}
		case cur != nil && strings.HasPrefix(l, "new file mode "):
			cur.create = true
			if m := strings.TrimPrefix(l, "new file mode "); nonRegular(m) {
				block(cur, "creates a file of mode "+m)
			}
		case cur != nil && strings.HasPrefix(l, "deleted file mode "):
			cur.remove = true
		case cur != nil && strings.HasPrefix(l, "old mode "):
			block(cur, "changes the file mode ("+l+")")
		case cur != nil && strings.HasPrefix(l, "new mode "):
			block(cur, "changes the file mode ("+l+")")
		case cur != nil && strings.HasPrefix(l, "index "):
			if f := strings.Fields(l); len(f) == 3 && nonRegular(f[2]) {
				block(cur, "touches a file of mode "+f[2])
			}
		case cur != nil && strings.HasPrefix(l, "rename from "):
			cur.oldPath, cur.renamed = gitPath(strings.TrimPrefix(l, "rename from ")), true
		case cur != nil && strings.HasPrefix(l, "rename to "):
			cur.newPath, cur.renamed = gitPath(strings.TrimPrefix(l, "rename to ")), true
		case strings.HasPrefix(l, "copy from "), strings.HasPrefix(l, "copy to "):
			return nil, fmt.Errorf("a copy in a patch (%s) is not applied by the tool", l)
		case strings.HasPrefix(l, "GIT binary patch"), strings.HasPrefix(l, "Binary files "):
			return nil, fmt.Errorf("a binary patch is not applied by the tool")
		case strings.HasPrefix(l, "--- "):
			if cur == nil || cur.oldSeen {
				cur = &patchFile{}
				files = append(files, cur)
			}
			p := stripPrefix(gitPath(strings.TrimPrefix(l, "--- ")))
			cur.oldPath = p
			if p == "" {
				cur.create = true
			}
			cur.oldSeen = true
			if i+1 >= len(lines) || !strings.HasPrefix(lines[i+1], "+++ ") {
				return nil, fmt.Errorf("the line %d, %q, is not followed by a +++ line", i+1, l)
			}
			i++
			q := stripPrefix(gitPath(strings.TrimPrefix(strings.TrimSuffix(lines[i], "\r"), "+++ ")))
			cur.newPath = q
			if q == "" {
				cur.remove = true
			}
		case cur != nil && hunkHeader.MatchString(l):
			m := hunkHeader.FindStringSubmatch(l)
			atoi := func(s string, def int) int {
				if s == "" {
					return def
				}
				n, _ := strconv.Atoi(s)
				return n
			}
			h := hunk{oldStart: atoi(m[1], 0)}
			oldN, newN := atoi(m[2], 1), atoi(m[4], 1)
			startLine := i + 1
			noNL := func() { // "\ No newline at end of file" is about the line before it
				prev := lines[i-1]
				switch {
				case prev != "" && prev[0] == '-':
					h.oldNoNL = true
				case prev != "" && prev[0] == '+':
					h.newNoNL = true
				default:
					h.oldNoNL, h.newNoNL = true, true
				}
			}
			changed := false
			for oldN > 0 || newN > 0 {
				i++
				if i >= len(lines) {
					return nil, fmt.Errorf("the hunk of %s at patch line %d ends before its %d old and %d new lines are all there", cur.name(), startLine, oldN, newN)
				}
				b := strings.TrimSuffix(lines[i], "\r")
				if b == "" {
					b = " " // a blank context line some tools strip to nothing, or a bare CR
				}
				switch b[0] {
				case '\\':
					noNL()
				case ' ':
					text := lines[i]
					if strings.HasPrefix(text, " ") {
						text = text[1:] // a bare "" or "\r" is the whole blank line, its leading space stripped
					}
					h.old, h.new = append(h.old, text), append(h.new, text)
					oldN--
					newN--
					if changed {
						h.trailing++
					} else {
						h.leading++
					}
				case '-':
					h.old = append(h.old, lines[i][1:])
					oldN--
					changed, h.trailing = true, 0
				case '+':
					h.new = append(h.new, lines[i][1:])
					newN--
					changed, h.trailing = true, 0
				default:
					return nil, fmt.Errorf("the hunk of %s at patch line %d has a line %d that is neither context, - nor +: %q", cur.name(), startLine, i+1, lines[i])
				}
			}
			for i+1 < len(lines) && strings.HasPrefix(lines[i+1], `\`) {
				i++
				noNL()
			}
			cur.hunks = append(cur.hunks, h)
		}
	}
	for _, f := range files {
		if f.newPath == "" && f.oldPath == "" {
			return nil, fmt.Errorf("a file section of the patch names no file")
		}
	}
	if len(files) == 0 {
		return nil, fmt.Errorf("the patch holds no file section")
	}
	return files, nil
}

func (f *patchFile) name() string {
	if f.newPath != "" {
		return f.newPath
	}
	return f.oldPath
}

// textFile is a file's content as lines, and whether its last line ends with a newline.
type textFile struct {
	lines []string
	endNL bool
}

func splitText(b []byte) textFile {
	if len(b) == 0 {
		return textFile{endNL: true}
	}
	s := string(b)
	endNL := strings.HasSuffix(s, "\n")
	s = strings.TrimSuffix(s, "\n")
	return textFile{lines: strings.Split(s, "\n"), endNL: endNL}
}

func (t textFile) bytes() []byte {
	if len(t.lines) == 0 {
		return nil
	}
	s := strings.Join(t.lines, "\n")
	if t.endNL {
		s += "\n"
	}
	return []byte(s)
}

// apply puts one hunk on the file, at where the hunk says or, as git apply does, at the nearest place its old
// lines are found exactly; delta is how far earlier hunks of this file moved lines.
func (t textFile) apply(h hunk, delta int) (textFile, int, error) {
	want := h.oldStart - 1 + delta
	if len(h.old) == 0 && h.oldStart > 0 {
		want = h.oldStart + delta // a pure insertion is "after" the line it names
	}
	if want < 0 {
		want = 0
	}
	matches := func(at int) bool {
		if at < 0 || at+len(h.old) > len(t.lines) {
			return false
		}
		for k, l := range h.old {
			if t.lines[at+k] != l {
				return false
			}
		}
		// the old side's last line lacks a newline only when the file's does
		if at+len(h.old) == len(t.lines) && len(h.old) > 0 && h.oldNoNL == t.endNL {
			return false
		}
		return true
	}
	// As git apply does: a hunk that starts at line 1 (or 0) begins the file, and a hunk with context but none
	// after its last change ends it; a hunk whose old side lacks the final newline ends it too.
	atStart := h.oldStart <= 1
	atEnd := (h.trailing == 0 && h.leading > 0) || h.oldNoNL
	at := -1
	switch {
	case atStart && atEnd:
		if len(t.lines) == len(h.old) && matches(0) {
			at = 0
		}
	case atStart:
		if matches(0) {
			at = 0
		}
	case atEnd:
		if e := len(t.lines) - len(h.old); matches(e) {
			at = e
		}
	}
	for d := 0; !atStart && !atEnd && d <= len(t.lines)+1 && at < 0; d++ {
		switch {
		case matches(want - d):
			at = want - d
		case matches(want + d):
			at = want + d
		}
	}
	if at < 0 {
		return t, delta, fmt.Errorf("the hunk at line %d does not apply: its context and removed lines are not in the file (first of them: %q)", h.oldStart, firstOf(h.old))
	}
	out := append(append(append([]string{}, t.lines[:at]...), h.new...), t.lines[at+len(h.old):]...)
	endNL := t.endNL
	if at+len(h.old) == len(t.lines) && (len(h.new) > 0 || len(h.old) > 0) {
		endNL = !h.newNoNL
	}
	return textFile{lines: out, endNL: endNL}, delta + len(h.new) - len(h.old), nil
}

func firstOf(l []string) string {
	if len(l) == 0 {
		return ""
	}
	return l[0]
}

// patchAsFiles applies the patch to the files read through read, and says the files to commit.
func patchAsFiles(text string, read fileSource) ([]FileChange, error) {
	sections, err := parsePatch(text)
	if err != nil {
		return nil, err
	}
	for _, s := range sections {
		if s.modeBlock != "" {
			return nil, fmt.Errorf("%s %s, and the forge's files call carries no file mode", s.name(), s.modeBlock)
		}
	}
	work := map[string][]byte{} // the files as they are being changed; nil means absent
	orig := map[string]bool{}   // which paths existed
	load := func(p string) error {
		if _, seen := work[p]; seen {
			return nil
		}
		b, mode, ok, err := read(p)
		if err != nil {
			return err
		}
		if ok && mode != "" && mode != "100644" {
			return fmt.Errorf("%s is a file of mode %s, and the forge's files call carries no file mode, so committing it that way would make it 100644", p, mode)
		}
		orig[p] = ok
		if ok {
			if b == nil {
				b = []byte{}
			}
			work[p] = b
		} else {
			work[p] = nil
		}
		return nil
	}
	for _, s := range sections {
		from, to := s.oldPath, s.newPath
		if s.create {
			from = ""
		}
		var t textFile
		if from != "" {
			if err := load(from); err != nil {
				return nil, err
			}
			if work[from] == nil {
				return nil, fmt.Errorf("%s: the file is not there at the commit the patch goes on, so the patch does not apply", from)
			}
			if !utf8.Valid(work[from]) && len(s.hunks) > 0 {
				return nil, fmt.Errorf("%s is not UTF-8 text: the tool applies text patches only", from)
			}
			t = splitText(work[from])
		} else {
			if err := load(to); err != nil {
				return nil, err
			}
			if work[to] != nil {
				return nil, fmt.Errorf("%s: the patch creates it, but it is already there at the commit the patch goes on", to)
			}
			t = textFile{endNL: true}
		}
		delta := 0
		for _, h := range s.hunks {
			if t, delta, err = t.apply(h, delta); err != nil {
				return nil, fmt.Errorf("%s: %w", s.name(), err)
			}
		}
		switch {
		case s.remove:
			if len(t.lines) != 0 {
				return nil, fmt.Errorf("%s: the patch deletes the file, but lines are left after its hunks", from)
			}
			work[from] = nil
		default:
			if s.renamed && from != to {
				work[from] = nil
				if err := load(to); err != nil {
					return nil, err
				}
			}
			body := t.bytes()
			if body == nil {
				body = []byte{}
			}
			work[to] = body
		}
	}
	var paths []string
	for p := range work {
		paths = append(paths, p)
	}
	sort.Strings(paths)
	var out []FileChange
	for _, p := range paths {
		switch {
		case work[p] == nil && !orig[p]:
		case work[p] == nil:
			out = append(out, FileChange{Path: p, Delete: true})
		default:
			fc := FileChange{Path: p, size: len(work[p])}
			if utf8.Valid(work[p]) {
				s := string(work[p])
				fc.Content = &s
			} else {
				s := base64Std(work[p])
				fc.ContentBase64 = &s
			}
			out = append(out, fc)
		}
	}
	if len(out) == 0 {
		return nil, fmt.Errorf("the patch changes no file")
	}
	return out, nil
}

func base64Std(b []byte) string { return base64.StdEncoding.EncodeToString(b) }