databricks-cli/libs/patchwheel/patch.go

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

293 lines
7.2 KiB
Go
Raw Normal View History

2025-03-03 15:05:18 +00:00
package patchwheel
import (
"archive/zip"
"bufio"
"bytes"
"context"
2025-03-03 15:05:18 +00:00
"crypto/sha256"
"encoding/base64"
2025-03-04 11:19:11 +00:00
"errors"
2025-03-03 15:05:18 +00:00
"fmt"
"io"
"os"
"path/filepath"
"strconv"
"strings"
)
const (
versionKey = "Version:"
nameKey = "Name:"
)
2025-03-03 15:05:18 +00:00
// readMetadataAndRecord scans the zip file for files matching the patterns
// "*.dist-info/METADATA" and "*.dist-info/RECORD". If multiple are found, it picks
// the first one encountered.
func readMetadataAndRecord(r *zip.ReadCloser) (metadataFile, recordFile *zip.File, oldDistInfoPrefix string) {
2025-03-03 15:05:18 +00:00
for _, f := range r.File {
if metadataFile == nil {
matched, _ := filepath.Match("*.dist-info/METADATA", f.Name)
if matched {
2025-03-03 15:05:18 +00:00
metadataFile = f
// Determine the old dist-info directory prefix.
if i := strings.LastIndex(f.Name, "/"); i != -1 {
oldDistInfoPrefix = f.Name[:i+1]
}
if recordFile != nil {
break
}
2025-03-04 11:01:17 +00:00
continue
2025-03-03 15:05:18 +00:00
}
}
if recordFile == nil {
2025-03-04 11:01:17 +00:00
matched, _ := filepath.Match("*.dist-info/RECORD", f.Name)
if matched {
recordFile = f
if metadataFile != nil {
break
}
}
}
2025-03-03 15:05:18 +00:00
}
return metadataFile, recordFile, oldDistInfoPrefix
2025-03-03 15:05:18 +00:00
}
2025-03-04 11:08:35 +00:00
// patchMetadata returns new METADATA content with an updated "Version:" field and validates that previous version matches oldVersion
func patchMetadata(r io.Reader, oldVersion, newVersion string) ([]byte, error) {
scanner := bufio.NewScanner(r)
2025-03-03 15:05:18 +00:00
var buf bytes.Buffer
for scanner.Scan() {
line := scanner.Text()
2025-03-04 11:08:35 +00:00
if strings.HasPrefix(line, versionKey) {
foundVersion := strings.TrimSpace(line[len(versionKey):])
if foundVersion != oldVersion {
return nil, fmt.Errorf("Unexpected version in METADATA: %s (expected %s)", strings.TrimSpace(line), oldVersion)
}
line = versionKey + newVersion
2025-03-03 15:05:18 +00:00
}
buf.WriteString(line + "\n")
}
if err := scanner.Err(); err != nil {
return nil, err
}
return buf.Bytes(), nil
}
// patchRecord updates RECORD content: it replaces the old dist-info prefix with the new one
// in all file paths and, for the METADATA entry, updates the hash and size.
func patchRecord(r io.Reader, oldDistInfoPrefix, newDistInfoPrefix, metadataHash string, metadataSize int) ([]byte, error) {
scanner := bufio.NewScanner(r)
2025-03-03 15:05:18 +00:00
var newLines []string
for scanner.Scan() {
line := scanner.Text()
if strings.TrimSpace(line) == "" {
continue
}
parts := strings.Split(line, ",")
if len(parts) < 3 {
// If the line doesn't have enough parts, preserve it as-is
2025-03-03 15:05:18 +00:00
newLines = append(newLines, line)
continue
}
origPath := parts[0]
if strings.HasPrefix(origPath, oldDistInfoPrefix) {
parts[0] = newDistInfoPrefix + origPath[len(oldDistInfoPrefix):]
}
// For the METADATA file entry, update hash and size.
if strings.HasSuffix(parts[0], "METADATA") {
parts[1] = "sha256=" + metadataHash
parts[2] = strconv.Itoa(metadataSize)
}
newLines = append(newLines, strings.Join(parts, ","))
}
if err := scanner.Err(); err != nil {
return nil, err
}
return []byte(strings.Join(newLines, "\n") + "\n"), nil
}
// PatchWheel patches a Python wheel file by updating its version in METADATA and RECORD.
// It returns the path to the new wheel.
// The function is idempotent: repeated calls with the same input will produce the same output.
// If the target wheel already exists, it returns the path to the existing wheel without processing.
func PatchWheel(ctx context.Context, path, outputDir string) (string, error) {
fileInfo, err := os.Stat(path)
if err != nil {
return "", err
}
wheelMtime := fileInfo.ModTime().UTC()
2025-03-03 19:53:43 +00:00
filename := filepath.Base(path)
wheelInfo, err := ParseWheelFilename(filename)
if err != nil {
return "", err
}
2025-03-03 19:53:43 +00:00
baseVersion := strings.SplitN(wheelInfo.Version, "+", 2)[0]
2025-03-03 19:53:43 +00:00
dt := strings.Replace(wheelMtime.Format("20060102150405.00"), ".", "", 1)
dt = strings.Replace(dt, ".", "", 1)
newVersion := baseVersion + "+" + dt
2025-03-03 19:53:43 +00:00
newFilename := fmt.Sprintf("%s-%s-%s.whl",
wheelInfo.Distribution,
newVersion,
strings.Join(wheelInfo.Tags, "-"))
outpath := filepath.Join(outputDir, newFilename)
2025-03-03 19:53:43 +00:00
if _, err := os.Stat(outpath); err == nil {
// Target wheel already exists, return its path
return outpath, nil
}
2025-03-03 19:53:43 +00:00
// Target wheel doesn't exist, proceed with patching
// Create a temporary file in the same directory with a unique name
2025-03-04 11:19:11 +00:00
tmpFilename := outpath + fmt.Sprintf(".tmp%d", os.Getpid())
2025-03-03 20:31:58 +00:00
2025-03-04 08:59:55 +00:00
needRemoval := true
defer func() {
2025-03-04 08:59:55 +00:00
if needRemoval {
2025-03-04 11:19:11 +00:00
_ = os.Remove(tmpFilename)
}
}()
2025-03-03 20:31:58 +00:00
2025-03-03 15:05:18 +00:00
r, err := zip.OpenReader(path)
if err != nil {
return "", err
}
defer r.Close()
metadataFile, recordFile, oldDistInfoPrefix := readMetadataAndRecord(r)
if metadataFile == nil {
return "", fmt.Errorf("wheel %s missing METADATA file", path)
}
if recordFile == nil {
return "", fmt.Errorf("wheel %s missing RECORD file", path)
2025-03-03 15:05:18 +00:00
}
metadataReader, err := metadataFile.Open()
2025-03-03 15:05:18 +00:00
if err != nil {
return "", err
}
defer metadataReader.Close()
2025-03-03 15:05:18 +00:00
2025-03-04 11:08:35 +00:00
newMetadata, err := patchMetadata(metadataReader, wheelInfo.Version, newVersion)
2025-03-03 15:05:18 +00:00
if err != nil {
return "", err
}
h := sha256.New()
h.Write(newMetadata)
metadataHash := base64.URLEncoding.WithPadding(base64.NoPadding).EncodeToString(h.Sum(nil))
metadataSize := len(newMetadata)
// Compute the new dist-info directory prefix.
newDistInfoPrefix := ""
if idx := strings.LastIndex(oldDistInfoPrefix, "-"); idx != -1 {
base := oldDistInfoPrefix[:idx]
newDistInfoPrefix = base + "-" + newVersion + ".dist-info/"
} else {
return "", fmt.Errorf("unexpected dist-info directory format: %s", oldDistInfoPrefix)
}
2025-03-04 11:19:11 +00:00
recordReader, err := recordFile.Open()
if err != nil {
return "", err
}
defer recordReader.Close()
newRecord, err := patchRecord(recordReader, oldDistInfoPrefix, newDistInfoPrefix, metadataHash, metadataSize)
2025-03-03 15:05:18 +00:00
if err != nil {
return "", err
}
2025-03-04 11:19:11 +00:00
outFile, err := os.Create(tmpFilename)
2025-03-03 15:05:18 +00:00
if err != nil {
return "", err
}
defer outFile.Close()
2025-03-04 11:19:11 +00:00
metadataUpdated := 0
recordUpdated := 0
2025-03-03 15:05:18 +00:00
zipw := zip.NewWriter(outFile)
for _, f := range r.File {
// If the file is inside the old dist-info directory, update its name.
newName := f.Name
if strings.HasPrefix(f.Name, oldDistInfoPrefix) {
newName = newDistInfoPrefix + f.Name[len(oldDistInfoPrefix):]
}
header := &zip.FileHeader{
Name: newName,
Method: f.Method,
}
header.Modified = f.ModTime()
2025-03-03 15:05:18 +00:00
header.Comment = f.Comment
if f.FileInfo().IsDir() && !strings.HasSuffix(header.Name, "/") {
header.Name += "/"
}
2025-03-03 15:05:18 +00:00
writer, err := zipw.CreateHeader(header)
if err != nil {
return "", err
}
2025-03-04 11:19:11 +00:00
if f.Name == metadataFile.Name {
2025-03-03 15:05:18 +00:00
_, err = writer.Write(newMetadata)
if err != nil {
return "", err
}
2025-03-04 11:19:11 +00:00
metadataUpdated += 1
} else if f.Name == recordFile.Name {
2025-03-03 15:05:18 +00:00
_, err = writer.Write(newRecord)
if err != nil {
return "", err
}
2025-03-04 11:19:11 +00:00
recordUpdated += 1
2025-03-03 15:05:18 +00:00
} else {
rc, err := f.Open()
if err != nil {
return "", err
}
_, err = io.Copy(writer, rc)
if err != nil {
rc.Close()
return "", err
}
if err := rc.Close(); err != nil {
2025-03-03 15:05:18 +00:00
return "", err
}
}
}
2025-03-03 15:05:18 +00:00
if err := zipw.Close(); err != nil {
return "", err
}
2025-03-03 20:31:58 +00:00
outFile.Close()
2025-03-03 20:31:58 +00:00
2025-03-04 11:19:11 +00:00
if metadataUpdated != 1 {
return "", errors.New("Could not update METADATA")
}
2025-03-04 11:19:11 +00:00
if recordUpdated != 1 {
return "", errors.New("Could not update RECORD")
}
if err := os.Rename(tmpFilename, outpath); err != nil {
return "", err
}
2025-03-04 08:59:55 +00:00
needRemoval = false
2025-03-04 11:19:11 +00:00
2025-03-03 15:05:18 +00:00
return outpath, nil
}