mopac-pdf v0: markdown to beautiful PDFs via digest-pinned typst

Go CLI that turns the markdown currency of this stack (Redmine notes,
Discourse posts, briefing output) into typeset PDFs: two embedded typst
templates (report with title page/TOC/headers, dense brief), front-matter
(title/subtitle/author/date/classification/template), GFM tables, and
bar charts rendered in pure Go (go-chart) from fenced chart data blocks.
Engine is a prebuilt typst 0.15.1 container pinned by digest and run
--network none; PDF bytes to stdout or -o. Exit codes 0/1/2.

All dev in docker (dev.sh/Makefile); unit tests + golden typst fixtures
plus a host-side smoke against the real engine container.

Generated with Crush

Assisted-by: Crush:glm-5.2

internal/pdfinfo/pdfinfo.go

@@ -0,0 +1,38 @@
+// 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
+}

internal/pdfinfo/pdfinfo_test.go

@@ -0,0 +1,58 @@
+package pdfinfo
+
+import (
+	"fmt"
+	"strings"
+	"testing"
+)
+
+// fakePDF builds a minimal but structurally scannable PDF with a page-tree
+// /Count and N /Type /Page objects.
+func fakePDF(pages, declared int) []byte {
+	var objs []string
+	for i := 1; i <= pages; i++ {
+		objs = append(objs, fmt.Sprintf("%d 0 obj<</Type/Page/Parent 99 0 R>>endobj", i))
+	}
+	objs = append(objs, fmt.Sprintf("99 0 obj<</Type/Pages/Count %d/Kids[1 0 R]>>endobj", declared))
+	body := strings.Join(objs, "\n")
+	return []byte("%PDF-1.7\n" + body + "\n%%EOF")
+}
+
+func TestPagesCount(t *testing.T) {
+	for _, n := range []int{1, 2, 7, 23} {
+		got, err := Pages(fakePDF(n, n))
+		if err != nil {
+			t.Fatalf("n=%d: %v", n, err)
+		}
+		if got != n {
+			t.Fatalf("n=%d got %d", n, got)
+		}
+	}
+}
+
+func TestPagesFallsBackToPageObjects(t *testing.T) {
+	// No /Pages /Count reachable: count /Type /Page objects instead.
+	pdf := []byte("%PDF-1.7\n1 0 obj<</Type/Page>>endobj\n2 0 obj<</Type/Page>>endobj\n%%EOF")
+	got, err := Pages(pdf)
+	if err != nil || got != 2 {
+		t.Fatalf("got=%d err=%v", got, err)
+	}
+}
+
+func TestPagesRejectsNonPDF(t *testing.T) {
+	if _, err := Pages([]byte("hello")); err == nil {
+		t.Fatal("want error for non-PDF")
+	}
+	if _, err := Pages(nil); err == nil {
+		t.Fatal("want error for empty input")
+	}
+}
+
+func TestPagesDoesNotCountPagesType(t *testing.T) {
+	// "/Type /Pages" must not match the /Type /Page fallback.
+	pdf := []byte("%PDF-1.7\n1 0 obj<</Type/Page>>endobj\n2 0 obj<</Type/Pages/Count 5>>endobj\n%%EOF")
+	got, err := Pages(pdf)
+	if err != nil || got != 5 {
+		t.Fatalf("got=%d err=%v", got, err)
+	}
+}