databricks-cli/libs/patchwheel/patch.go

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

265 lines
6.5 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"
2025-03-04 17:27:45 +00:00
"github.com/databricks/cli/libs/log"
2025-03-03 15:05:18 +00:00
)
2025-03-05 13:44:10 +00:00
var versionKey []byte = []byte("Version:")
2025-03-05 16:03:30 +00:00
func findFile(r *zip.ReadCloser, filename string) *zip.File {
2025-03-03 15:05:18 +00:00
for _, f := range r.File {
2025-03-05 16:03:30 +00:00
if f.Name == filename {
return f
}
2025-03-03 15:05:18 +00:00
}
2025-03-05 16:03:30 +00:00
return nil
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() {
2025-03-05 13:44:10 +00:00
line := scanner.Bytes()
if versionValue, ok := bytes.CutPrefix(line, versionKey); ok {
foundVersion := string(bytes.TrimSpace(versionValue))
2025-03-04 11:08:35 +00:00
if foundVersion != oldVersion {
2025-03-05 13:44:10 +00:00
return nil, fmt.Errorf("Unexpected version in METADATA: %s (expected %s)", strings.TrimSpace(string(line)), oldVersion)
2025-03-04 11:08:35 +00:00
}
buf.WriteString(string(versionKey) + " " + newVersion + "\n")
2025-03-05 13:44:10 +00:00
} else {
buf.Write(line)
buf.WriteString("\n")
2025-03-03 15:05:18 +00:00
}
}
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) {
2025-03-05 13:44:10 +00:00
metadataPath := newDistInfoPrefix + "METADATA"
scanner := bufio.NewScanner(r)
2025-03-05 13:44:10 +00:00
var buf bytes.Buffer
2025-03-03 15:05:18 +00:00
for scanner.Scan() {
2025-03-05 13:44:10 +00:00
line := scanner.Bytes()
if len(bytes.TrimSpace(line)) == 0 {
2025-03-03 15:05:18 +00:00
continue
}
2025-03-05 13:44:10 +00:00
parts := strings.Split(string(line), ",")
2025-03-03 15:05:18 +00:00
if len(parts) < 3 {
// If the line doesn't have enough parts, preserve it as-is
2025-03-05 13:44:10 +00:00
buf.Write(line)
buf.WriteString("\n")
2025-03-03 15:05:18 +00:00
continue
}
2025-03-05 13:44:10 +00:00
2025-03-03 15:05:18 +00:00
origPath := parts[0]
2025-03-05 13:44:10 +00:00
pathSuffix, hasDistPrefix := strings.CutPrefix(origPath, oldDistInfoPrefix)
if hasDistPrefix {
parts[0] = newDistInfoPrefix + pathSuffix
2025-03-03 15:05:18 +00:00
}
2025-03-05 13:44:10 +00:00
if metadataPath == parts[0] {
2025-03-03 15:05:18 +00:00
parts[1] = "sha256=" + metadataHash
parts[2] = strconv.Itoa(metadataSize)
}
2025-03-05 13:44:10 +00:00
buf.WriteString(strings.Join(parts, ",") + "\n")
2025-03-03 15:05:18 +00:00
}
if err := scanner.Err(); err != nil {
return nil, err
}
2025-03-05 13:44:10 +00:00
return buf.Bytes(), nil
2025-03-03 15:05:18 +00:00
}
// PatchWheel patches a Python wheel file by updating its version in METADATA and RECORD.
// It returns the path to the new wheel.
2025-03-04 11:21:46 +00:00
// The version is updated according to the following rules:
// - if there is an existing part after + it is dropped
// - append +<mtime of the original wheel> to version
//
// The function is idempotent: repeated calls with the same input will produce the same output.
2025-03-04 11:21:46 +00:00
// If the target wheel already exists, it returns the path to the existing wheel without redoing the patching.
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
newVersion, newFilename := calculateNewVersion(wheelInfo, wheelMtime)
outpath := filepath.Join(outputDir, newFilename)
2025-03-03 19:53:43 +00:00
if _, err := os.Stat(outpath); err == nil {
2025-03-04 17:27:45 +00:00
log.Debugf(ctx, "Skipping patching of %s, already exists: %s", path, outpath)
return outpath, nil
}
2025-03-03 19:53:43 +00:00
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()
oldDistInfoPrefix := wheelInfo.Distribution + "-" + wheelInfo.Version + ".dist-info/"
2025-03-05 16:03:30 +00:00
metadataFile := findFile(r, oldDistInfoPrefix+"METADATA")
if metadataFile == nil {
return "", fmt.Errorf("wheel %s missing %sMETADATA", path, oldDistInfoPrefix)
}
2025-03-05 16:03:30 +00:00
recordFile := findFile(r, oldDistInfoPrefix+"RECORD")
if recordFile == nil {
return "", fmt.Errorf("wheel %s missing %sRECORD file", path, oldDistInfoPrefix)
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.
2025-03-04 14:37:58 +00:00
newDistInfoPrefix := strings.Replace(oldDistInfoPrefix, wheelInfo.Version, newVersion, 1)
if newDistInfoPrefix == oldDistInfoPrefix {
return "", fmt.Errorf("unexpected dist-info directory format: %s (version=%s)", oldDistInfoPrefix, wheelInfo.Version)
2025-03-03 15:05:18 +00:00
}
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
}