paliad/pkg/docforge/docx/markdown.go

package docx

// Markdown → OOXML rendering for Composer section content (t-paliad-313
// Slice B/D; restructured in t-paliad-349 slice 8).
//
// Parsing now lives in pkg/docforge/markdown, which produces the neutral
// docforge.Document. This file renders that Document into OOXML paragraph
// elements (<w:p>…</w:p>) ready to splice into a .docx body. There is one
// Markdown parser for docforge; this is the .docx exporter for its model.
//
// Output uses the base's stylemap entry for each block kind on the
// <w:pStyle>, so styling matches the base's typography (HLpat-Body-B0 on
// the HLC base, Normal on the neutral base, etc.). Placeholders ({{key}})
// ride through as literal run text and are substituted by the placeholder
// pass after assembly.

import (
	"strconv"
	"strings"

	"mgit.msbls.de/m/paliad/pkg/docforge"
	"mgit.msbls.de/m/paliad/pkg/docforge/markdown"
)

// HyperlinkAllocator hands the renderer a `rId` for each external URL it
// encounters in `[label](url)` inline links. The composer's post-pass uses
// these allocations to mutate `word/_rels/document.xml.rels` so the emitted
// `<w:hyperlink r:id="…">` elements resolve. Pass nil to drop links to
// plain text (the label survives, the URL doesn't render). t-paliad-316.
type HyperlinkAllocator func(url string) string

// RenderMarkdownToOOXML renders Markdown into OOXML paragraphs with a
// single paragraph style. Slice B back-compat wrapper.
func RenderMarkdownToOOXML(md, paragraphStyle string) string {
	return RenderMarkdownToOOXMLWithStyles(md, map[string]string{"paragraph": paragraphStyle}, nil)
}

// RenderMarkdownToOOXMLWithStyles parses Markdown into a docforge.Document
// and renders it to OOXML. stylemap maps each block kind (paragraph,
// heading_1/2/3, list_bullet, list_numbered, blockquote) to a Word
// paragraph style; missing entries fall back to the "paragraph" style.
func RenderMarkdownToOOXMLWithStyles(md string, stylemap map[string]string, links HyperlinkAllocator) string {
	return RenderDocumentToOOXML(markdown.Import(md), stylemap, links)
}

// RenderDocumentToOOXML renders a neutral Document to OOXML paragraphs —
// the .docx side of the docforge importer→model→exporter pipeline. Any
// Document (Markdown today, a foreign-doc importer later) renders the same
// way.
func RenderDocumentToOOXML(doc docforge.Document, stylemap map[string]string, links HyperlinkAllocator) string {
	defaultStyle := stylemap["paragraph"]
	// Numbered-list counter resets on every non-numbered block so
	// "1. A\n2. B\n\n1. C" renders 1./2./1. — the input determined the
	// ordinal, the renderer just emits it.
	numbered := 0
	var b strings.Builder
	for _, blk := range doc.Blocks {
		style := stylemap[string(blk.Kind)]
		if style == "" {
			style = defaultStyle
		}
		if blk.Kind == docforge.KindListNumbered {
			numbered++
		} else {
			numbered = 0
		}
		b.WriteString(renderBlock(blk, style, links, numbered))
	}
	return b.String()
}

// renderBlock emits one <w:p> for a block. List blocks get a visible
// "• " / "N. " prefix run (the base stylemap handles indentation if it
// defines a list style; the prefix at least surfaces the structure).
func renderBlock(blk docforge.Block, paragraphStyle string, links HyperlinkAllocator, numberedOrdinal int) string {
	var b strings.Builder
	b.WriteString(`<w:p>`)
	if paragraphStyle != "" {
		b.WriteString(`<w:pPr><w:pStyle w:val="`)
		b.WriteString(xmlAttrEscape(paragraphStyle))
		b.WriteString(`"/></w:pPr>`)
	}
	// An empty block is an intentional empty paragraph: one empty run.
	if len(blk.Spans) == 0 {
		b.WriteString(`<w:r><w:t xml:space="preserve"></w:t></w:r></w:p>`)
		return b.String()
	}
	switch blk.Kind {
	case docforge.KindListBullet:
		b.WriteString(`<w:r><w:t xml:space="preserve">• </w:t></w:r>`)
	case docforge.KindListNumbered:
		ordinal := numberedOrdinal
		if ordinal <= 0 {
			ordinal = 1
		}
		b.WriteString(`<w:r><w:t xml:space="preserve">`)
		b.WriteString(strconv.Itoa(ordinal))
		b.WriteString(`. </w:t></w:r>`)
	}
	for _, span := range blk.Spans {
		b.WriteString(renderInlineSpan(span, links))
	}
	b.WriteString(`</w:p>`)
	return b.String()
}

// renderInlineSpan emits one span. A hyperlink span (Link != "") becomes a
// <w:hyperlink r:id="…"> wrapping its children when an allocator yields a
// rId; otherwise the label children render as plain runs (URL dropped).
func renderInlineSpan(span docforge.InlineSpan, links HyperlinkAllocator) string {
	if span.Link != "" {
		if links != nil {
			if rid := links(span.Link); rid != "" {
				var hb strings.Builder
				hb.WriteString(`<w:hyperlink r:id="`)
				hb.WriteString(xmlAttrEscape(rid))
				hb.WriteString(`">`)
				for _, child := range span.Children {
					hb.WriteString(renderRunWithLinkStyle(child))
				}
				hb.WriteString(`</w:hyperlink>`)
				return hb.String()
			}
		}
		// No allocator / no rId — render the label as plain runs.
		var fb strings.Builder
		for _, child := range span.Children {
			fb.WriteString(renderRun(child))
		}
		return fb.String()
	}
	return renderRun(span)
}

// renderRunWithLinkStyle emits a hyperlink child run with Word's built-in
// "Hyperlink" character style (colour + underline), plus B/I.
func renderRunWithLinkStyle(span docforge.InlineSpan) string {
	var b strings.Builder
	b.WriteString(`<w:r><w:rPr><w:rStyle w:val="Hyperlink"/>`)
	if span.Bold {
		b.WriteString(`<w:b/>`)
	}
	if span.Italic {
		b.WriteString(`<w:i/>`)
	}
	b.WriteString(`</w:rPr><w:t xml:space="preserve">`)
	b.WriteString(xmlTextEscape(span.Text))
	b.WriteString(`</w:t></w:r>`)
	return b.String()
}

// renderRun emits one <w:r> for a plain (text/bold/italic) span.
func renderRun(span docforge.InlineSpan) string {
	var b strings.Builder
	b.WriteString(`<w:r>`)
	if span.Bold || span.Italic {
		b.WriteString(`<w:rPr>`)
		if span.Bold {
			b.WriteString(`<w:b/>`)
		}
		if span.Italic {
			b.WriteString(`<w:i/>`)
		}
		b.WriteString(`</w:rPr>`)
	}
	b.WriteString(`<w:t xml:space="preserve">`)
	b.WriteString(xmlTextEscape(span.Text))
	b.WriteString(`</w:t></w:r>`)
	return b.String()
}

// xmlTextEscape escapes the XML-significant characters for <w:t> content.
// Quotes/apostrophes are legal in element text — not escaped.
func xmlTextEscape(s string) string {
	s = strings.ReplaceAll(s, "&", "&amp;")
	s = strings.ReplaceAll(s, "<", "&lt;")
	s = strings.ReplaceAll(s, ">", "&gt;")
	return s
}

// xmlAttrEscape escapes for an attribute value (e.g. <w:pStyle w:val="…"/>).
func xmlAttrEscape(s string) string {
	s = strings.ReplaceAll(s, "&", "&amp;")
	s = strings.ReplaceAll(s, "<", "&lt;")
	s = strings.ReplaceAll(s, ">", "&gt;")
	s = strings.ReplaceAll(s, `"`, "&quot;")
	return s
}