PrintToVeraeTimeStamp/internal/job/job.go

package job

import (
	"encoding/json"
	"fmt"
	"os"
	"path/filepath"
	"regexp"
	"strings"
	"time"

	"github.com/VeraeLLC/PrintToVeraeTimeStamp/internal/auth"
	"github.com/VeraeLLC/PrintToVeraeTimeStamp/internal/hash"
	"github.com/VeraeLLC/PrintToVeraeTimeStamp/internal/pdfcert"
	"github.com/VeraeLLC/PrintToVeraeTimeStamp/internal/veraeapi"
)

const HelperVersion = "0.1.0"

// Result paths written for one print job.
type Result struct {
	OriginalPDF     string
	CertificatePDF  string
	ReceiptJSON     string
	ReceiptID       string
	SHA256          string
	Channel         auth.Channel
}

// Options for processing a captured print PDF.
type Options struct {
	SourcePDF    string // temp PDF from macOS PDF Services
	OutDir       string
	DocumentName string // optional override for stem; default from source basename
	Channel      auth.Channel
	Client       *veraeapi.Client
}

var unsafeName = regexp.MustCompile(`[^A-Za-z0-9._-]+`)

// SanitizeReceiptID makes a receipt id safe for filenames.
func SanitizeReceiptID(id string) string {
	id = strings.TrimSpace(id)
	id = unsafeName.ReplaceAllString(id, "-")
	id = strings.Trim(id, "-.")
	if id == "" {
		id = "unknown"
	}
	// Keep under common NAME_MAX budget with prefix
	const max = 180
	if len(id) > max {
		id = id[:max]
	}
	return id
}

// StemFromPath returns the document stem without .pdf.
func StemFromPath(path, override string) string {
	if override != "" {
		s := strings.TrimSuffix(override, filepath.Ext(override))
		return unsafeName.ReplaceAllString(s, "-")
	}
	base := filepath.Base(path)
	stem := strings.TrimSuffix(base, filepath.Ext(base))
	stem = unsafeName.ReplaceAllString(stem, "-")
	stem = strings.Trim(stem, "-.")
	if stem == "" {
		stem = "document"
	}
	return stem
}

// DefaultOutDir is ~/Documents/Verae Timestamps.
func DefaultOutDir() (string, error) {
	home, err := os.UserHomeDir()
	if err != nil {
		return "", err
	}
	return filepath.Join(home, "Documents", "Verae Timestamps"), nil
}

// Run copies the source PDF, stamps digests-only, writes certificate + JSON.
func Run(opt Options) (Result, error) {
	if opt.SourcePDF == "" {
		return Result{}, fmt.Errorf("source PDF required")
	}
	if opt.Client == nil {
		return Result{}, fmt.Errorf("client required")
	}
	outDir := opt.OutDir
	if outDir == "" {
		d, err := DefaultOutDir()
		if err != nil {
			return Result{}, err
		}
		outDir = d
	}
	if err := os.MkdirAll(outDir, 0o700); err != nil {
		return Result{}, err
	}

	stem := StemFromPath(opt.SourcePDF, opt.DocumentName)
	origPath := filepath.Join(outDir, stem+".pdf")
	srcBytes, err := os.ReadFile(opt.SourcePDF)
	if err != nil {
		return Result{}, err
	}
	if err := os.WriteFile(origPath, srcBytes, 0o600); err != nil {
		return Result{}, err
	}

	digest, err := hash.FileSHA256(origPath)
	if err != nil {
		return Result{}, err
	}

	client := *opt.Client
	client.Channel = opt.Channel
	rec, err := client.Stamp(digest, stem+".pdf")
	if err != nil {
		return Result{}, err
	}
	if rec.ReceiptID == "" {
		return Result{}, fmt.Errorf("stamp returned empty receipt id")
	}

	safeID := SanitizeReceiptID(rec.ReceiptID)
	certPath := filepath.Join(outDir, fmt.Sprintf("%s-verae-timestamp-%s.pdf", stem, safeID))

	// Always prefer the branded Registration Document (digests-only wording,
	// no mcp.verae.ai). Edge attachPdf is optional and may still say MCP.
	pdfBytes, err := pdfcert.Build(pdfcert.Input{
		DocumentName: stem + ".pdf",
		SHA256:       digest,
		ReceiptID:    rec.ReceiptID,
		CreatedAt:    rec.CreatedAt,
		Merkle:       rec.Merkle,
		Scope:        "global",
		Channel:      string(opt.Channel),
		ByteLength:   int64(len(srcBytes)),
	})
	if err != nil {
		return Result{}, err
	}
	_ = rec.PDF // upstream PDF ignored in favor of branded local certificate
	if err := os.WriteFile(certPath, pdfBytes, 0o600); err != nil {
		return Result{}, err
	}

	sidecar := map[string]any{
		"helper_version": HelperVersion,
		"document_name":  stem + ".pdf",
		"sha256":         digest,
		"receipt_id":     rec.ReceiptID,
		"created_at":     rec.CreatedAt,
		"merkle":         rec.Merkle,
		"channel":        string(opt.Channel),
		"base_url":       rec.BaseURL,
		"original_pdf":   filepath.Base(origPath),
		"certificate_pdf": filepath.Base(certPath),
		"byte_length":    len(srcBytes),
		"saved_at":       time.Now().UTC().Format(time.RFC3339),
	}
	jsonPath := filepath.Join(outDir, stem+".verae.json")
	jb, err := json.MarshalIndent(sidecar, "", "  ")
	if err != nil {
		return Result{}, err
	}
	if err := os.WriteFile(jsonPath, jb, 0o600); err != nil {
		return Result{}, err
	}

	return Result{
		OriginalPDF:    origPath,
		CertificatePDF: certPath,
		ReceiptJSON:    jsonPath,
		ReceiptID:      rec.ReceiptID,
		SHA256:         digest,
		Channel:        opt.Channel,
	}, nil
}