summercms/modules/tide/multipart.go

package tide

import (
	"bytes"
	"crypto/sha256"
	"encoding/hex"
	"fmt"
	"mime"
	"mime/multipart"
	"net/textproto"
	"os"
	"path"
	"path/filepath"
	"regexp"
	"strings"
)

// MultipartBoundary is the fixed boundary every multipart request body is
// encoded with, so a recording against one backend and a replay against
// another send byte-identical bodies.
const MultipartBoundary = "SummerTideMultipartBoundary7MA4YWxkTrZu0gW"

// Part is one field of a multipart/form-data request. A text field sets
// Value; a file field sets File, a path relative to the fixture directory
// (for example files/cover.png), whose bytes are never inlined into the
// YAML, with SHA256 pinning its content. Filename defaults to the base name
// of File and ContentType to application/octet-stream.
type Part struct {
	Name        string `yaml:"name"`
	Value       string `yaml:"value,omitempty"`
	File        string `yaml:"file,omitempty"`
	Filename    string `yaml:"filename,omitempty"`
	ContentType string `yaml:"content_type,omitempty"`
	SHA256      string `yaml:"sha256,omitempty"`
}

var sha256Hex = regexp.MustCompile(`^[0-9a-fA-F]{64}$`)

// validateParts checks the shape of a request's parts without reading files.
func validateParts(req Request) error {
	if len(req.Parts) == 0 {
		return nil
	}
	if req.Body != "" {
		return fmt.Errorf("request has both body and parts")
	}
	for i, p := range req.Parts {
		if strings.TrimSpace(p.Name) == "" {
			return fmt.Errorf("parts[%d] is missing name", i)
		}
		if p.File == "" {
			if p.Filename != "" || p.ContentType != "" || p.SHA256 != "" {
				return fmt.Errorf("part %q has file attributes but no file", p.Name)
			}
			continue
		}
		if p.Value != "" {
			return fmt.Errorf("part %q has both value and file", p.Name)
		}
		if err := validateSidecar(p.File); err != nil {
			return fmt.Errorf("part %q: %w", p.Name, err)
		}
		if !sha256Hex.MatchString(p.SHA256) {
			return fmt.Errorf("part %q needs the sha256 of its file", p.Name)
		}
	}
	return nil
}

// verifyPartFiles checks that every part file of flow exists under base and
// matches its sha256.
func verifyPartFiles(base string, flow Flow) error {
	for _, step := range flow.Steps {
		for _, p := range step.Request.Parts {
			if p.File == "" {
				continue
			}
			if _, err := readPartFile(base, p); err != nil {
				return fmt.Errorf("tide: step %s: %w", step.ID, err)
			}
		}
	}
	return nil
}

func readPartFile(base string, p Part) ([]byte, error) {
	if err := validateSidecar(p.File); err != nil {
		return nil, fmt.Errorf("part %q: %w", p.Name, err)
	}
	full := p.File
	if base != "" {
		full = filepath.Join(base, p.File)
	}
	resolved, err := resolvePath(full)
	if err != nil {
		return nil, fmt.Errorf("part %q file %s: %w", p.Name, p.File, err)
	}
	if base != "" {
		root, err := resolvePath(base)
		if err != nil {
			return nil, fmt.Errorf("fixture dir: %w", err)
		}
		if resolved != root && !strings.HasPrefix(resolved, root+string(os.PathSeparator)) {
			return nil, fmt.Errorf("part %q file %q escapes the fixture directory", p.Name, p.File)
		}
	}
	raw, err := os.ReadFile(resolved)
	if err != nil {
		return nil, fmt.Errorf("part %q file %s: %w", p.Name, p.File, err)
	}
	sum := sha256.Sum256(raw)
	if !strings.EqualFold(hex.EncodeToString(sum[:]), p.SHA256) {
		return nil, fmt.Errorf("part %q file %s sha256 mismatch", p.Name, p.File)
	}
	return raw, nil
}

var quoteEscaper = strings.NewReplacer(`\`, `\\`, `"`, `\"`)

// encodeParts writes parts in declaration order as a multipart/form-data
// body with MultipartBoundary and returns the body and its Content-Type.
func encodeParts(base string, parts []Part) ([]byte, string, error) {
	var buf bytes.Buffer
	w := multipart.NewWriter(&buf)
	if err := w.SetBoundary(MultipartBoundary); err != nil {
		return nil, "", err
	}
	marker := []byte(MultipartBoundary)
	for _, p := range parts {
		h := make(textproto.MIMEHeader)
		var content []byte
		if p.File == "" {
			h.Set("Content-Disposition", fmt.Sprintf(`form-data; name="%s"`, quoteEscaper.Replace(p.Name)))
			content = []byte(p.Value)
		} else {
			raw, err := readPartFile(base, p)
			if err != nil {
				return nil, "", err
			}
			filename := p.Filename
			if filename == "" {
				filename = path.Base(filepath.ToSlash(p.File))
			}
			ct := p.ContentType
			if ct == "" {
				ct = "application/octet-stream"
			}
			h.Set("Content-Disposition", fmt.Sprintf(`form-data; name="%s"; filename="%s"`, quoteEscaper.Replace(p.Name), quoteEscaper.Replace(filename)))
			h.Set("Content-Type", ct)
			content = raw
		}
		if bytes.Contains(content, marker) {
			return nil, "", fmt.Errorf("part %q contains the multipart boundary", p.Name)
		}
		pw, err := w.CreatePart(h)
		if err != nil {
			return nil, "", err
		}
		if _, err := pw.Write(content); err != nil {
			return nil, "", err
		}
	}
	if err := w.Close(); err != nil {
		return nil, "", err
	}
	return buf.Bytes(), w.FormDataContentType(), nil
}

// prepareRequest encodes a request's parts into its body. The multipart
// Content-Type replaces a recorded multipart Content-Type (whose boundary
// would be stale) and is added when none is recorded; a recorded non-
// multipart Content-Type is kept as it is.
func prepareRequest(req Request, base string) (Request, error) {
	if len(req.Parts) == 0 {
		return req, nil
	}
	if err := validateParts(req); err != nil {
		return Request{}, err
	}
	body, ct, err := encodeParts(base, req.Parts)
	if err != nil {
		return Request{}, err
	}
	out := req
	out.Body = Body(body)
	out.Headers = make(map[string]string, len(req.Headers)+1)
	set := false
	for k, v := range req.Headers {
		if strings.EqualFold(k, "Content-Type") {
			set = true
			if media, _, err := mime.ParseMediaType(v); err != nil || strings.HasPrefix(media, "multipart/") {
				v = ct
			}
		}
		out.Headers[k] = v
	}
	if !set {
		out.Headers["Content-Type"] = ct
	}
	return out, nil
}