// Package pdfinfo extracts basic facts from PDF bytes with a tiny parser —
// no third-party PDF library, just enough to count pages in golden tests
// and smoke checks.
package pdfinfo

import (
	"bytes"
	"fmt"
	"regexp"
	"strconv"
)

// Pages returns the page count of a PDF by scanning for the /Type /Pages
// objects' /Count entries (the catalog root carries the total; we take the
// maximum to be safe with nested page trees).
func Pages(pdf []byte) (int, error) {
	if len(pdf) == 0 || !bytes.HasPrefix(pdf, []byte("%PDF-")) {
		return 0, fmt.Errorf("pdfinfo: not a PDF (missing %%%%PDF- header)")
	}
	re := regexp.MustCompile(`/Type\s*/Pages[^>]*?/Count\s+(\d+)`)
	best := 0
	for _, m := range re.FindAllStringSubmatch(string(pdf), -1) {
		if n, err := strconv.Atoi(m[1]); err == nil && n > best {
			best = n
		}
	}
	if best == 0 {
		// Compressed object streams can hide the count; require at least
		// the header then fall back to /Type /Page occurrences (\b keeps
		// /Pages from matching).
		pageRe := regexp.MustCompile(`/Type\s*/Page\b`)
		best = len(pageRe.FindAll(pdf, -1))
		if best == 0 {
			return 0, fmt.Errorf("pdfinfo: no page count found")
		}
	}
	return best, nil
}