descrybe/apps/api/internal/feeds/source.go

package feeds

import (
	"context"
	"errors"
	"fmt"
	"io"
	"os"
	"path/filepath"
	"strings"

	"github.com/google/uuid"
)

var (
	errSourceRequired = errors.New("feed url or uploaded CSV source required")
	errLocalSource    = errors.New("local feed source unavailable")
)

// feedBlob is feed content on disk. Close removes owned temp files (HTTP downloads).
// Local uploads reference the existing path and Close is a no-op.
type feedBlob struct {
	path        string
	contentType string
	size        int64
	owned       bool
}

// Close removes the temp file when this blob owns it.
func (b *feedBlob) Close() error {
	if b == nil || !b.owned || b.path == "" {
		return nil
	}
	err := os.Remove(b.path)
	b.path = ""
	b.owned = false
	return err
}

// Open returns a new read handle at the start of the blob.
func (b *feedBlob) Open() (*os.File, error) {
	if b == nil || b.path == "" {
		return nil, errors.New("feed blob closed or empty")
	}
	return os.Open(b.path)
}

// Sniff reads up to n bytes from the start of the blob (for format detection).
func (b *feedBlob) Sniff(n int) ([]byte, error) {
	if n <= 0 {
		return nil, nil
	}
	f, err := b.Open()
	if err != nil {
		return nil, err
	}
	defer f.Close()
	buf := make([]byte, n)
	nr, err := io.ReadFull(f, buf)
	if err == io.EOF || err == io.ErrUnexpectedEOF {
		err = nil
	}
	if err != nil {
		return nil, err
	}
	return buf[:nr], nil
}

// loadFeedSource returns on-disk feed content from a local upload or HTTP(S) URL.
// Callers must Close the blob when finished.
func (s *Service) loadFeedSource(ctx context.Context, companyID uuid.UUID, feed map[string]any) (*feedBlob, error) {
	if path := sourcePathFromOptions(feed["options"]); path != "" {
		return s.readLocalFeed(companyID, path)
	}
	urlStr, _ := feed["url"].(string)
	urlStr = strings.TrimSpace(urlStr)
	if urlStr == "" {
		return nil, errSourceRequired
	}
	return downloadFeed(ctx, urlStr)
}

func sourcePathFromOptions(raw any) string {
	opts, ok := raw.(map[string]any)
	if !ok || opts == nil {
		return ""
	}
	for _, key := range []string{"source_path", "local_path", "file_path"} {
		if v, ok := opts[key].(string); ok {
			if p := strings.TrimSpace(v); p != "" {
				return p
			}
		}
	}
	return ""
}

func (s *Service) readLocalFeed(companyID uuid.UUID, rel string) (*feedBlob, error) {
	uploadDir := strings.TrimSpace(s.UploadDir)
	if uploadDir == "" {
		return nil, ClientMsg("upload directory not configured")
	}
	abs, err := resolveCompanyPath(uploadDir, companyID, rel)
	if err != nil {
		return nil, err
	}
	info, err := os.Stat(abs)
	if err != nil {
		if os.IsNotExist(err) {
			return nil, fmt.Errorf("%w: file missing", errLocalSource)
		}
		return nil, err
	}
	if !info.Mode().IsRegular() {
		return nil, ClientMsg("invalid source path")
	}
	if info.Size() > maxDownloadBytes {
		return nil, downloadTooLarge()
	}
	ct := "text/csv"
	lower := strings.ToLower(abs)
	if strings.HasSuffix(lower, ".xml") {
		ct = "application/xml"
	}
	return &feedBlob{
		path:        abs,
		contentType: ct,
		size:        info.Size(),
		owned:       false,
	}, nil
}

func resolveCompanyPath(uploadDir string, companyID uuid.UUID, rel string) (string, error) {
	rel = filepath.ToSlash(strings.TrimSpace(rel))
	if rel == "" || strings.Contains(rel, "..") {
		return "", ClientMsg("invalid source path")
	}
	prefix := companyID.String() + "/"
	if !strings.HasPrefix(rel, prefix) {
		return "", ClientMsg("forbidden source path")
	}
	base, err := filepath.Abs(uploadDir)
	if err != nil {
		return "", err
	}
	abs, err := filepath.Abs(filepath.Join(uploadDir, filepath.FromSlash(rel)))
	if err != nil {
		return "", err
	}
	sep := string(os.PathSeparator)
	if abs != base && !strings.HasPrefix(abs, base+sep) {
		return "", ClientMsg("forbidden source path")
	}
	return abs, nil
}