autogits/bots-common/git_utils.go

package common

/*
 * This file is part of Autogits.
 *
 * Copyright © 2024 SUSE LLC
 *
 * Autogits is free software: you can redistribute it and/or modify it under
 * the terms of the GNU General Public License as published by the Free Software
 * Foundation, either version 2 of the License, or (at your option) any later
 * version.
 *
 * Autogits is distributed in the hope that it will be useful, but WITHOUT ANY
 * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 * PARTICULAR PURPOSE. See the GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * Foobar. If not, see <https://www.gnu.org/licenses/>.
 */

import (
	"fmt"
	"io"
	"log"
	"os"
	"os/exec"
	"path"
	"path/filepath"
	"strings"
	"sync"
)

type GitHandler struct {
	DebugLogger bool

	GitPath     string
	GitCommiter string
	GitEmail    string
}

func CreateGitHandler(git_author, email, name string) (*GitHandler, error) {
	var err error

	git := new(GitHandler)
	git.GitCommiter = git_author
	git.GitPath, err = os.MkdirTemp("", name)
	if err != nil {
		return nil, fmt.Errorf("Cannot create temp dir: %w", err)
	}

	if err = os.Chmod(git.GitPath, 0700); err != nil {
		return nil, fmt.Errorf("Cannot fix permissions of temp dir: %w", err)
	}

	return git, nil
}

//func (h *GitHandler) ProcessBranchList() []string {
//	if h.HasError() {
//		return make([]string, 0)
//	}
//
//	trackedBranches, err := os.ReadFile(path.Join(h.GitPath, DefaultGitPrj, TrackedBranchesFile))
//	if err != nil {
//		if errors.Is(err, os.ErrNotExist) {
//			trackedBranches = []byte("factory")
//		} else {
//			h.LogError("file error reading '%s' file in repo", TrackedBranchesFile)
//			h.Error = err
//			return make([]string, 0)
//		}
//	}
//
//	return strings.Split(string(trackedBranches), "\n")
//}

type GitReference struct {
	Branch string
	Id     string
}

type GitReferences struct {
	refs []GitReference
}

func (refs *GitReferences) addReference(id, branch string) {
	for _, ref := range refs.refs {
		if ref.Id == id && ref.Branch == branch {
			return
		}
	}

	refs.refs = append(refs.refs, GitReference{Branch: branch, Id: id})
}

func processRefs(gitDir string) ([]GitReference, error) {
	packedRefsPath := path.Join(gitDir, "packed-refs")
	stat, err := os.Stat(packedRefsPath)
	if err != nil {
		return nil, err
	}

	if stat.Size() > 10000 || stat.IsDir() {
		return nil, fmt.Errorf("Funny business with 'packed-refs' in '%s'", gitDir)
	}

	data, err := os.ReadFile(packedRefsPath)
	if err != nil {
		return nil, err
	}

	var references GitReferences
	for _, line := range strings.Split(string(data), "\n") {
		if len(line) < 1 || line[0] == '#' {
			continue
		}

		splitLine := strings.Split(line, " ")
		if len(splitLine) != 2 {
			return nil, fmt.Errorf("Unexpected packaged-refs entry '%#v' in '%s'", splitLine, packedRefsPath)
		}
		id, ref := splitLine[0], splitLine[1]
		const remoteRefPrefix = "refs/remotes/origin/"
		if ref[0:len(remoteRefPrefix)] != remoteRefPrefix {
			continue
		}

		references.addReference(id, ref[len(remoteRefPrefix):])
	}

	return references.refs, nil
}

func findGitDir(p string) (string, error) {
	gitFile := path.Join(p, ".git")
	stat, err := os.Stat(gitFile)
	if err != nil {
		return "", err
	}

	if stat.IsDir() {
		return path.Join(p, ".git"), nil
	}

	data, err := os.ReadFile(gitFile)
	if err != nil {
		return "", err
	}

	for _, line := range strings.Split(string(data), "\n") {
		refs := strings.Split(line, ":")
		if len(refs) != 2 {
			return "", fmt.Errorf("Unknown format of .git file: '%s'\n", line)
		}

		if refs[0] != "gitdir" {
			return "", fmt.Errorf("Unknown header of .git file: '%s'\n", refs[0])
		}

		return path.Join(p, strings.TrimSpace(refs[1])), nil
	}

	return "", fmt.Errorf("Can't find git subdirectory in '%s'", p)
}

func (e *GitHandler) GitBranchHead(gitDir, branchName string) (string, error) {
	path, err := findGitDir(path.Join(e.GitPath, gitDir))
	if err != nil {
		return "", fmt.Errorf("Error identifying gitdir in `%s`: %w", gitDir, err)
	}

	refs, err := processRefs(path)
	if err != nil {
		return "", fmt.Errorf("Error finding branches (%s): %w\n", branchName, err)
	}

	for _, ref := range refs {
		if ref.Branch == branchName {
			return ref.Id, nil
		}
	}

	return "", fmt.Errorf("Can't find default remote branch: %s", branchName)
}

func (e *GitHandler) Close() error {
	if err := os.RemoveAll(e.GitPath); err != nil {
		return err
	}
	e.GitPath = ""
	return nil
}

type writeFunc func(data []byte) (int, error)

func (f writeFunc) Write(data []byte) (int, error) {
	return f(data)
}

func (h writeFunc) UnmarshalText(text []byte) error {
	_, err := h.Write(text)
	return err
}

func (h writeFunc) Close() error {
	_, err := h.Write(nil)
	return err
}

func (e *GitHandler) GitExec(cwd string, params ...string) error {
	cmd := exec.Command("/usr/bin/git", params...)
	cmd.Env = []string{
		"GIT_CEILING_DIRECTORIES=" + e.GitPath,
		"GIT_CONFIG_GLOBAL=/dev/null",
		"GIT_AUTHOR_NAME=" + e.GitCommiter,
		"GIT_COMMITTER_NAME=" + e.GitCommiter,
		"EMAIL=not@exist@src.opensuse.org",
		"GIT_LFS_SKIP_SMUDGE=1",
		"GIT_SSH_COMMAND=/usr/bin/ssh -o StrictHostKeyChecking=yes",
	}
	cmd.Dir = filepath.Join(e.GitPath, cwd)
	cmd.Stdin = nil

	if e.DebugLogger {
		log.Printf("git execute: %#v\n", cmd.Args)
	}
	out, err := cmd.CombinedOutput()
	if e.DebugLogger {
		log.Println(string(out))
	}
	if err != nil {
		if e.DebugLogger {
			log.Printf(" *** error: %v\n", err)
		}
		return fmt.Errorf("error executing: git %#v \n%s\n err: %w", cmd.Args, out, err)
	}

	return nil
}

type ChanIO struct {
	ch chan byte
}

func (c *ChanIO) Write(p []byte) (int, error) {
	for _, b := range p {
		c.ch <- b
	}
	return len(p), nil
}

// read at least 1 byte, but don't block if nothing more in channel
func (c *ChanIO) Read(data []byte) (idx int, err error) {
	var ok bool

	data[idx], ok = <-c.ch
	if !ok {
		err = io.EOF
		return
	}
	idx++

	for len(c.ch) > 0 && idx < len(data) {
		data[idx], ok = <-c.ch
		if !ok {
			err = io.EOF
			return
		}

		idx++
	}

	return
}

type gitMsg struct {
	hash     string
	itemType string
	size     int
}

type commit struct {
	Tree string
	Msg  string
}

type tree_entry struct {
	name string
	mode int
	hash string

	size int
}

type tree struct {
	items []tree_entry
}

func (t *tree_entry) isSubmodule() bool {
	return (t.mode & 0170000) == 0160000
}

func (t *tree_entry) isTree() bool {
	return (t.mode & 0170000) == 0040000
}

func (t *tree_entry) isBlob() bool {
	return !t.isTree() && !t.isSubmodule()
}

func parseGitMsg(data <-chan byte) (gitMsg, error) {
	var id []byte = make([]byte, 64)
	var msgType []byte = make([]byte, 16)
	var size int

	pos := 0
	for c := <-data; c != ' '; c = <-data {
		if (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f') {
			id[pos] = c
			pos++
		} else {
			return gitMsg{}, fmt.Errorf("Invalid character during object hash parse '%c' at %d", c, pos)
		}
	}
	id = id[:pos]

	pos = 0
	var c byte
	for c = <-data; c != ' ' && c != '\x00'; c = <-data {
		if c >= 'a' && c <= 'z' {
			msgType[pos] = c
			pos++
		} else {
			return gitMsg{}, fmt.Errorf("Invalid character during object type parse '%c' at %d", c, pos)
		}
	}
	msgType = msgType[:pos]

	switch string(msgType) {
	case "commit", "tree", "blob":
		break
	case "missing":
		if c != '\x00' {
			return gitMsg{}, fmt.Errorf("Missing format weird")
		}
		return gitMsg{
			hash:     string(id[:]),
			itemType: "missing",
			size:     0,
		}, fmt.Errorf("Object not found: '%s'", string(id))
	default:
		return gitMsg{}, fmt.Errorf("Invalid object type: '%s'", string(msgType))
	}

	for c = <-data; c != '\000'; c = <-data {
		if c >= '0' && c <= '9' {
			size = size*10 + (int(c) - '0')
		} else {
			return gitMsg{}, fmt.Errorf("Invalid character during object size parse: '%c'", c)
		}
	}

	return gitMsg{
		hash:     string(id[:]),
		itemType: string(msgType),
		size:     size,
	}, nil
}

func parseGitCommitHdr(data <-chan byte) ([2]string, error) {
	hdr := make([]byte, 0, 60)
	val := make([]byte, 0, 1000)

	c := <-data
	if c != '\n' { // end of header marker
		for ; c != ' '; c = <-data {
			hdr = append(hdr, c)
		}
		for c := <-data; c != '\n'; c = <-data {
			val = append(val, c)
		}
	}

	return [2]string{string(hdr), string(val)}, nil
}

func parseGitCommitMsg(data <-chan byte, l int) (string, error) {
	msg := make([]byte, 0, l)

	for c := <-data; c != '\x00'; c = <-data {
		msg = append(msg, c)
		l--
	}
	//	l--

	if l != 0 {
		return "", fmt.Errorf("Unexpected data in the git commit msg: l=%d", l)
	}

	return string(msg), nil
}

func parseGitCommit(data <-chan byte) (commit, error) {
	hdr, err := parseGitMsg(data)
	if err != nil {
		return commit{}, err
	} else if hdr.itemType != "commit" {
		return commit{}, fmt.Errorf("expected commit but parsed %s", hdr.itemType)
	}

	var c commit
	l := hdr.size
	for {
		hdr, err := parseGitCommitHdr(data)
		if err != nil {
			return commit{}, nil
		}

		if len(hdr[0])+len(hdr[1]) == 0 { // hdr end marker
			break
		}

		switch hdr[0] {
		case "tree":
			c.Tree = hdr[1]
		}

		l -= len(hdr[0]) + len(hdr[1]) + 2
	}
	l--

	c.Msg, err = parseGitCommitMsg(data, l)
	return c, err
}

func parseTreeEntry(data <-chan byte, hashLen int) (tree_entry, error) {
	var e tree_entry

	for c := <-data; c != ' '; c = <-data {
		e.mode = e.mode*8 + int(c-'0')
		e.size++
	}
	e.size++

	name := make([]byte, 0, 128)
	for c := <-data; c != '\x00'; c = <-data {
		name = append(name, c)
		e.size++
	}
	e.size++
	e.name = string(name)

	const hexBinToAscii = "0123456789abcdef"

	hash := make([]byte, 0, hashLen*2)
	for range hashLen {
		c := <-data
		hash = append(hash, hexBinToAscii[((c&0xF0)>>4)], hexBinToAscii[c&0xF])
	}
	e.hash = string(hash)
	e.size += hashLen

	return e, nil
}

func parseGitTree(data <-chan byte) (tree, error) {

	hdr, err := parseGitMsg(data)
	if err != nil {
		return tree{}, err
	}

	// max capacity to length of hash
	t := tree{items: make([]tree_entry, 0, hdr.size/len(hdr.hash))}
	parsedLen := 0
	for parsedLen < hdr.size {
		entry, err := parseTreeEntry(data, len(hdr.hash)/2)
		if err != nil {
			return tree{}, nil
		}

		t.items = append(t.items, entry)
		parsedLen += entry.size
	}
	c := <-data // \0 read

	if c != '\x00' {
		return t, fmt.Errorf("Unexpected character during git tree data read")
	}

	if parsedLen != hdr.size {
		return t, fmt.Errorf("Invalid size of git tree data")
	}

	return t, nil
}

func parseGitBlob(data <-chan byte) ([]byte, error) {
	hdr, err := parseGitMsg(data)
	if err != nil {
		return []byte{}, err
	}

	d := make([]byte, hdr.size)
	for l := 0; l < hdr.size; l++ {
		d[l] = <-data
	}
	eob := <-data
	if eob != '\x00' {
		return d, fmt.Errorf("invalid byte read in parseGitBlob")
	}

	return d, nil
}

// TODO: support sub-trees
func (e *GitHandler) GitCatFile(cwd, commitId, filename string) (data []byte, err error) {
	var done sync.Mutex

	done.Lock()
	data_in, data_out := ChanIO{make(chan byte, 256)}, ChanIO{make(chan byte, 70)}

	go func() {
		defer done.Unlock()
		defer close(data_out.ch)

		data_out.Write([]byte(commitId))
		data_out.ch <- '\x00'
		c, err := parseGitCommit(data_in.ch)
		if err != nil {
			log.Printf("Error parsing git commit: %v\n", err)
			return
		}
		data_out.Write([]byte(c.Tree))
		data_out.ch <- '\x00'
		tree, err := parseGitTree(data_in.ch)

		if err != nil {
			if e.DebugLogger {
				log.Printf("Error parsing git tree: %v\n", err)
			}
			return
		}

		for _, te := range tree.items {
			if te.isBlob() && te.name == filename {
				data_out.Write([]byte(te.hash))
				data_out.ch <- '\x00'
				data, err = parseGitBlob(data_in.ch)
				return
			}
		}

		err = fmt.Errorf("file not found: '%s'", filename)
	}()

	cmd := exec.Command("/usr/bin/git", "cat-file", "--batch", "-Z")
	cmd.Env = []string{
		"GIT_CEILING_DIRECTORIES=" + e.GitPath,
		"GIT_CONFIG_GLOBAL=/dev/null",
	}
	cmd.Dir = filepath.Join(e.GitPath, cwd)
	cmd.Stdout = &data_in
	cmd.Stdin = &data_out
	cmd.Stderr = writeFunc(func(data []byte) (int, error) {
		if e.DebugLogger {
			log.Printf(string(data))
		}
		return len(data), nil
	})
	if e.DebugLogger {
		log.Printf("command run: %v\n", cmd.Args)
	}
	err = cmd.Run()

	done.Lock()
	return
}

// return (filename) -> (hash) map for all submodules
// TODO: recursive? map different orgs, not just assume '.' for path
func (e *GitHandler) GitSubmoduleList(cwd, commitId string) (submoduleList map[string]string, err error) {
	var done sync.Mutex
	submoduleList = make(map[string]string)

	done.Lock()
	data_in, data_out := ChanIO{make(chan byte, 256)}, ChanIO{make(chan byte, 70)}

	go func() {
		defer done.Unlock()
		defer close(data_out.ch)

		data_out.Write([]byte(commitId))
		data_out.ch <- '\x00'
		var c commit
		c, err = parseGitCommit(data_in.ch)
		if err != nil {
			err = fmt.Errorf("Error parsing git commit. Err: %w", err)
			return
		}
		data_out.Write([]byte(c.Tree))
		data_out.ch <- '\x00'
		var tree tree
		tree, err = parseGitTree(data_in.ch)

		if err != nil {
			err = fmt.Errorf("Error parsing git tree: %w", err)
			return
		}

		for _, te := range tree.items {
			if te.isSubmodule() {
				submoduleList[te.name] = te.hash
			}
		}
	}()

	cmd := exec.Command("/usr/bin/git", "cat-file", "--batch", "-Z")
	cmd.Env = []string{
		"GIT_CEILING_DIRECTORIES=" + e.GitPath,
		"GIT_CONFIG_GLOBAL=/dev/null",
	}
	cmd.Dir = filepath.Join(e.GitPath, cwd)
	cmd.Stdout = &data_in
	cmd.Stdin = &data_out
	cmd.Stderr = writeFunc(func(data []byte) (int, error) {
		if e.DebugLogger {
			log.Println(string(data))
		}
		return len(data), nil
	})
	if e.DebugLogger {
		log.Printf("command run: %v\n", cmd.Args)
	}
	err = cmd.Run()

	done.Lock()
	return submoduleList, err
}

func (e *GitHandler) GitSubmoduleCommitId(cwd, packageName, commitId string) (subCommitId string, valid bool) {
	defer func() {
		if recover() != nil {
			commitId = ""
			valid = false
		}
	}()

	data_in, data_out := ChanIO{make(chan byte, 256)}, ChanIO{make(chan byte, 70)}
	var wg sync.WaitGroup

	wg.Add(1)

	if e.DebugLogger {
		log.Printf("getting commit id '%s' from git at '%s' with packageName: %s\n", commitId, cwd, packageName)
	}

	go func() {
		defer wg.Done()
		defer close(data_out.ch)

		data_out.Write([]byte(commitId))
		data_out.ch <- '\x00'
		c, err := parseGitCommit(data_in.ch)
		if err != nil {
			log.Panicf("Error parsing git commit: %v\n", err)
		}
		data_out.Write([]byte(c.Tree))
		data_out.ch <- '\x00'
		tree, err := parseGitTree(data_in.ch)

		if err != nil {
			log.Panicf("Error parsing git tree: %v\n", err)
		}

		for _, te := range tree.items {
			if te.name == packageName && te.isSubmodule() {
				subCommitId = te.hash
				return
			}
		}
	}()

	cmd := exec.Command("/usr/bin/git", "cat-file", "--batch", "-Z")
	cmd.Env = []string{
		"GIT_CEILING_DIRECTORIES=" + e.GitPath,
		"GIT_CONFIG_GLOBAL=/dev/null",
	}
	cmd.Dir = filepath.Join(e.GitPath, cwd)
	cmd.Stdout = &data_in
	cmd.Stdin = &data_out
	cmd.Stderr = writeFunc(func(data []byte) (int, error) {
		log.Println(string(data))
		return len(data), nil
	})
	if e.DebugLogger {
		log.Printf("command run: %v\n", cmd.Args)
	}
	if err := cmd.Run(); err != nil {
		log.Printf("Error running command %v, err: %v", cmd.Args, err)
	}

	wg.Wait()
	return subCommitId, len(subCommitId) == len(commitId)
}