Merge: t-paliad-349 docforge slice 6a — authoring core + TemplateStore wiring (m/paliad#157)

cmd/server/main.go

@@ -174,6 +174,9 @@ func main() {
 		submissionComposerSvc := services.NewSubmissionComposer(submissionRenderer)
 		// t-paliad-315 Slice C — building-block library.
 		submissionBuildingBlockSvc := services.NewBuildingBlockService(pool, branding.Name)
+		// t-paliad-349 docforge slice 4/6 — uploaded-template store
+		// (Postgres bytea) backing the authoring surface.
+		templateStoreSvc := services.NewPgTemplateStore(pool)
 		// t-paliad-225 Slice A — user-authored checklist templates.
 		// Slice B adds checklist_shares grants + admin promotion.
 		checklistCatalogSvc := services.NewChecklistCatalogService(pool)
@@ -190,6 +193,7 @@ func main() {
 			SubmissionSection:       submissionSectionSvc,
 			SubmissionComposer:      submissionComposerSvc,
 			SubmissionBuildingBlock: submissionBuildingBlockSvc,
+			TemplateStore:           templateStoreSvc,
 			Deadline:       deadlineSvc,
 			Appointment:    appointmentSvc,
 			CalDAV:         caldavSvc,

internal/handlers/handlers.go

@@ -128,6 +128,10 @@ type Services struct {
 	// editor. Per Q2: paste sources only, no lineage on sections.
 	SubmissionBuildingBlock *services.BuildingBlockService
 
+	// t-paliad-349 docforge slice 4/6 — uploaded-template store backing
+	// the authoring surface.
+	TemplateStore *services.PgTemplateStore
+
 	// t-paliad-265 / m/paliad#96 — per-event-card optional choices on
 	// the Verfahrensablauf timeline.
 	EventChoice *services.EventChoiceService
@@ -215,6 +219,7 @@ func Register(mux *http.ServeMux, client *auth.Client, giteaAPIToken string, svc
 			submissionSection:       svc.SubmissionSection,
 			submissionComposer:      svc.SubmissionComposer,
 			submissionBuildingBlock: svc.SubmissionBuildingBlock,
+			templateStore:           svc.TemplateStore,
 			eventChoice:       svc.EventChoice,
 			scenario:          svc.Scenario,
 			scenarioFlags:     svc.ScenarioFlags,

internal/handlers/projects.go

@@ -77,6 +77,9 @@ type dbServices struct {
 	submissionComposer      *services.SubmissionComposer
 	submissionBuildingBlock *services.BuildingBlockService
 
+	// t-paliad-349 docforge slice 4/6 — uploaded-template store.
+	templateStore *services.PgTemplateStore
+
 	// t-paliad-265 — per-event-card optional choices.
 	eventChoice *services.EventChoiceService
 

pkg/docforge/docx/authoring.go

@@ -0,0 +1,172 @@
+package docx
+
+// Authoring support — the .docx side of the docforge authoring surface
+// (t-paliad-349 slice 6). Two operations back the "upload a base .docx →
+// place variable slots" flow:
+//
+//   ImportForAuthoring — parse an uploaded .docx into a run-addressable
+//     preview (one <span data-run="N"> per <w:t>, in document order) plus
+//     the slots already present in the carrier.
+//   InjectSlot — replace a selected piece of text inside run N with a
+//     {{slot_key}} placeholder, returning the new carrier bytes. The
+//     placeholder is the sentinel that locates the slot (PRD §5 lean) and
+//     the same token the generation-time renderer substitutes.
+//
+// Both walk runs in the same order (paragraphs, then <w:t> within), so the
+// data-run indices the preview emits address exactly the runs InjectSlot
+// targets. Injection keys on the selected text
+// (not a byte/UTF-16 offset) so umlauts in German prose can't desync the
+// client's selection from the server's slice.
+//
+// v1 scope (PRD §2.1): text-level slots inside body paragraphs. A run is a
+// <w:t> within a <w:p>; selections spanning runs or sitting in
+// headers/footers/tables are out of scope and surface as an error the UI
+// turns into "select within a single text span".
+
+import (
+	"bytes"
+	"fmt"
+	"strconv"
+	"strings"
+
+	"mgit.msbls.de/m/paliad/pkg/docforge"
+)
+
+// AuthoringView is the parsed, run-addressable form of an uploaded
+// template, ready for the authoring editor.
+type AuthoringView struct {
+	// PreviewHTML is the body rendered as paragraphs of run spans:
+	// <p>…<span class="docforge-run" data-run="N">text</span>…</p>.
+	// The client attaches selection handling to the run spans; data-run
+	// is the index InjectSlot expects.
+	PreviewHTML string
+	// Slots are the {{placeholder}} tokens already present in the
+	// carrier (so re-opening a saved template shows its slots).
+	Slots []docforge.TemplateSlot
+}
+
+// ImportForAuthoring parses carrierBytes (any .docx/.dotm/...) into an
+// AuthoringView. Runs the .dotm→.docx pre-pass so macro templates import
+// cleanly.
+func ImportForAuthoring(carrierBytes []byte) (*AuthoringView, error) {
+	clean, err := ConvertDotmToDocx(carrierBytes)
+	if err != nil {
+		return nil, fmt.Errorf("authoring import: convert: %w", err)
+	}
+	documentXML, _, err := splitBaseZip(clean)
+	if err != nil {
+		return nil, fmt.Errorf("authoring import: %w", err)
+	}
+	return &AuthoringView{
+		PreviewHTML: authoringPreviewHTML(documentXML),
+		Slots:       detectSlots(documentXML),
+	}, nil
+}
+
+// authoringPreviewHTML renders the body as run-addressable HTML. One <p>
+// per <w:p>; one <span class="docforge-run" data-run="N"> per <w:t>, with
+// the decoded run text HTML-escaped. N is the global run index in
+// document-then-paragraph order — the same order InjectSlot walks.
+func authoringPreviewHTML(documentXML []byte) string {
+	var out bytes.Buffer
+	runIdx := 0
+	paras := wParagraphRegex.FindAll(documentXML, -1)
+	for _, para := range paras {
+		out.WriteString("<p>")
+		for _, m := range wTextNodeRegex.FindAllSubmatch(para, -1) {
+			text := xmlDecode(string(m[2]))
+			out.WriteString(`<span class="docforge-run" data-run="`)
+			out.WriteString(strconv.Itoa(runIdx))
+			out.WriteString(`">`)
+			out.WriteString(htmlEscape(text))
+			out.WriteString(`</span>`)
+			runIdx++
+		}
+		out.WriteString("</p>\n")
+	}
+	if out.Len() == 0 {
+		return "<p></p>"
+	}
+	return out.String()
+}
+
+// detectSlots returns the distinct {{placeholder}} tokens present in the
+// document body, in first-appearance order.
+func detectSlots(documentXML []byte) []docforge.TemplateSlot {
+	seen := map[string]bool{}
+	var slots []docforge.TemplateSlot
+	// Match against decoded text so a placeholder split by an entity is
+	// still found the same way the renderer would substitute it.
+	for _, m := range wTextNodeRegex.FindAllSubmatch(documentXML, -1) {
+		text := xmlDecode(string(m[2]))
+		for _, pm := range placeholderRegex.FindAllStringSubmatch(text, -1) {
+			key := pm[1]
+			if seen[key] {
+				continue
+			}
+			seen[key] = true
+			slots = append(slots, docforge.TemplateSlot{
+				Key:        key,
+				Anchor:     "{{" + key + "}}",
+				OrderIndex: len(slots),
+			})
+		}
+	}
+	return slots
+}
+
+// InjectSlot replaces the first occurrence of selectedText inside run
+// runIndex with a {{slotKey}} placeholder and returns the new carrier
+// bytes. Errors when the run is out of range or selectedText isn't found
+// in that run (a render/selection desync, or a cross-run selection).
+func InjectSlot(carrierBytes []byte, runIndex int, selectedText, slotKey string) ([]byte, error) {
+	if selectedText == "" {
+		return nil, fmt.Errorf("authoring inject: empty selection")
+	}
+	if !placeholderRegex.MatchString("{{" + slotKey + "}}") {
+		return nil, fmt.Errorf("authoring inject: invalid slot key %q", slotKey)
+	}
+	clean, err := ConvertDotmToDocx(carrierBytes)
+	if err != nil {
+		return nil, fmt.Errorf("authoring inject: convert: %w", err)
+	}
+	documentXML, parts, err := splitBaseZip(clean)
+	if err != nil {
+		return nil, fmt.Errorf("authoring inject: %w", err)
+	}
+
+	runIdx := 0
+	injected := false
+	newDoc := wParagraphRegex.ReplaceAllFunc(documentXML, func(para []byte) []byte {
+		return wTextNodeRegex.ReplaceAllFunc(para, func(tnode []byte) []byte {
+			idx := runIdx
+			runIdx++
+			if injected || idx != runIndex {
+				return tnode
+			}
+			sub := wTextNodeRegex.FindSubmatch(tnode)
+			attrs := string(sub[1])
+			content := xmlDecode(string(sub[2]))
+			before, after, found := strings.Cut(content, selectedText)
+			if !found {
+				return tnode // not found here — reported after the walk
+			}
+			newContent := before + "{{" + slotKey + "}}" + after
+			if !strings.Contains(attrs, "xml:space") &&
+				(strings.HasPrefix(newContent, " ") || strings.HasSuffix(newContent, " ")) {
+				attrs += ` xml:space="preserve"`
+			}
+			injected = true
+			return []byte(`<w:t` + attrs + `>` + xmlEncode(newContent) + `</w:t>`)
+		})
+	})
+	if !injected {
+		return nil, fmt.Errorf("authoring inject: selection %q not found in run %d", selectedText, runIndex)
+	}
+
+	repacked, err := repackBaseZip(parts, newDoc)
+	if err != nil {
+		return nil, fmt.Errorf("authoring inject: %w", err)
+	}
+	return repacked, nil
+}

pkg/docforge/docx/authoring_test.go

@@ -0,0 +1,111 @@
+package docx
+
+import (
+	"strings"
+	"testing"
+)
+
+// docBody wraps a <w:body> inner string into a full document.xml.
+func docBody(inner string) string {
+	return `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>` +
+		`<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">` +
+		`<w:body>` + inner + `</w:body></w:document>`
+}
+
+func TestImportForAuthoring_PreviewIsRunAddressable(t *testing.T) {
+	body := docBody(
+		`<w:p><w:r><w:t>Az. 4c O 12/23</w:t></w:r></w:p>` +
+			`<w:p><w:r><w:t>Klägerin</w:t></w:r><w:r><w:t xml:space="preserve"> GmbH</w:t></w:r></w:p>`)
+	view, err := ImportForAuthoring(minimalMergeDOCX(t, body))
+	if err != nil {
+		t.Fatalf("ImportForAuthoring: %v", err)
+	}
+	// Three <w:t> → three run spans, indexed 0,1,2 in document order.
+	for i, want := range []string{`data-run="0"`, `data-run="1"`, `data-run="2"`} {
+		if !strings.Contains(view.PreviewHTML, want) {
+			t.Errorf("preview missing %s (run %d); html=%s", want, i, view.PreviewHTML)
+		}
+	}
+	if !strings.Contains(view.PreviewHTML, "Az. 4c O 12/23") {
+		t.Errorf("preview missing run text; html=%s", view.PreviewHTML)
+	}
+	// Two paragraphs.
+	if n := strings.Count(view.PreviewHTML, "<p>"); n != 2 {
+		t.Errorf("paragraph count = %d; want 2", n)
+	}
+	if len(view.Slots) != 0 {
+		t.Errorf("fresh doc should have no slots; got %v", view.Slots)
+	}
+}
+
+func TestImportForAuthoring_DetectsExistingSlots(t *testing.T) {
+	body := docBody(`<w:p><w:r><w:t>Az. {{project.case_number}} vor {{project.court}}</w:t></w:r></w:p>`)
+	view, err := ImportForAuthoring(minimalMergeDOCX(t, body))
+	if err != nil {
+		t.Fatalf("ImportForAuthoring: %v", err)
+	}
+	if len(view.Slots) != 2 {
+		t.Fatalf("slots = %d; want 2 (%v)", len(view.Slots), view.Slots)
+	}
+	if view.Slots[0].Key != "project.case_number" || view.Slots[0].Anchor != "{{project.case_number}}" {
+		t.Errorf("slot[0] = %+v; want project.case_number", view.Slots[0])
+	}
+	if view.Slots[1].Key != "project.court" {
+		t.Errorf("slot[1].Key = %q; want project.court", view.Slots[1].Key)
+	}
+}
+
+func TestInjectSlot_ReplacesSelectionWithPlaceholder(t *testing.T) {
+	body := docBody(`<w:p><w:r><w:t>Az. 4c O 12/23 vor dem LG</w:t></w:r></w:p>`)
+	out, err := InjectSlot(minimalMergeDOCX(t, body), 0, "4c O 12/23", "project.case_number")
+	if err != nil {
+		t.Fatalf("InjectSlot: %v", err)
+	}
+	doc := readMergeDocumentXML(t, out)
+	if !strings.Contains(doc, "Az. {{project.case_number}} vor dem LG") {
+		t.Errorf("injected doc wrong; got %s", doc)
+	}
+	// Round-trips: re-importing finds the new slot.
+	view, err := ImportForAuthoring(out)
+	if err != nil {
+		t.Fatalf("re-import: %v", err)
+	}
+	if len(view.Slots) != 1 || view.Slots[0].Key != "project.case_number" {
+		t.Errorf("re-imported slots = %v; want [project.case_number]", view.Slots)
+	}
+}
+
+func TestInjectSlot_TargetsTheNamedRun(t *testing.T) {
+	// "GmbH" appears in run 1 only; "Müller" (with umlaut) in run 0.
+	body := docBody(
+		`<w:p><w:r><w:t>Müller</w:t></w:r><w:r><w:t xml:space="preserve"> GmbH</w:t></w:r></w:p>`)
+	out, err := InjectSlot(minimalMergeDOCX(t, body), 0, "Müller", "parties.claimant.name")
+	if err != nil {
+		t.Fatalf("InjectSlot: %v", err)
+	}
+	doc := readMergeDocumentXML(t, out)
+	if !strings.Contains(doc, "{{parties.claimant.name}}") {
+		t.Errorf("umlaut selection not replaced; got %s", doc)
+	}
+	if !strings.Contains(doc, " GmbH") {
+		t.Errorf("run 1 should be untouched; got %s", doc)
+	}
+}
+
+func TestInjectSlot_ErrorsWhenSelectionNotInRun(t *testing.T) {
+	body := docBody(`<w:p><w:r><w:t>Hello</w:t></w:r></w:p>`)
+	if _, err := InjectSlot(minimalMergeDOCX(t, body), 0, "Goodbye", "firm.name"); err == nil {
+		t.Error("expected error when selection absent from run; got nil")
+	}
+	// Out-of-range run index.
+	if _, err := InjectSlot(minimalMergeDOCX(t, body), 9, "Hello", "firm.name"); err == nil {
+		t.Error("expected error for out-of-range run index; got nil")
+	}
+}
+
+func TestInjectSlot_RejectsInvalidSlotKey(t *testing.T) {
+	body := docBody(`<w:p><w:r><w:t>Hello</w:t></w:r></w:p>`)
+	if _, err := InjectSlot(minimalMergeDOCX(t, body), 0, "Hello", "9bad-key!"); err == nil {
+		t.Error("expected error for invalid slot key; got nil")
+	}
+}