Refactor keyword parsing/rendering & add support for ATTR_HTML

org/html.go

@@ -4,6 +4,9 @@ import (
 	"fmt"
 	"html"
 	"strings"
+
+	h "golang.org/x/net/html"
+	"golang.org/x/net/html/atom"
 )
 
 type HTMLWriter struct {
@@ -45,10 +48,14 @@ func (w *HTMLWriter) emptyClone() *HTMLWriter {
 	return &wcopy
 }
 
-func (w *HTMLWriter) before(d *Document) {
-	w.document = d
+func (w *HTMLWriter) nodesAsString(nodes ...Node) string {
+	tmp := w.emptyClone()
+	tmp.writeNodes(nodes...)
+	return tmp.String()
 }
 
+func (w *HTMLWriter) before(d *Document) {}
+
 func (w *HTMLWriter) after(d *Document) {
 	w.writeFootnotes(d)
 }
@@ -199,9 +206,7 @@ func (w *HTMLWriter) writeRegularLink(l RegularLink) {
 	}
 	description := url
 	if l.Description != nil {
-		descriptionWriter := w.emptyClone()
-		descriptionWriter.writeNodes(l.Description...)
-		description = descriptionWriter.String()
+		description = w.nodesAsString(l.Description...)
 	}
 	switch l.Kind() {
 	case "image":
@@ -245,17 +250,27 @@ func (w *HTMLWriter) writeHorizontalRule(h HorizontalRule) {
 	w.WriteString("<hr>\n")
 }
 
-func (w *HTMLWriter) writeNodeWithMeta(m NodeWithMeta) {
-	nodeW := w.emptyClone()
-	nodeW.writeNodes(m.Node)
-	nodeString := nodeW.String()
-	if rawCaption, ok := m.Meta["CAPTION"]; ok {
-		nodes, captionW := w.document.parseInline(rawCaption), w.emptyClone()
-		captionW.writeNodes(nodes...)
-		caption := `<p class="caption">` + "\n" + captionW.String() + "\n</p>\n"
-		nodeString = `<div class="captioned">` + "\n" + nodeString + caption + `</div>` + "\n"
+func (w *HTMLWriter) writeNodeWithMeta(n NodeWithMeta) {
+	out := w.nodesAsString(n.Node)
+	if p, ok := n.Node.(Paragraph); ok {
+		if len(p.Children) == 1 && isImageOrVideoLink(p.Children[0]) {
+			out = w.nodesAsString(p.Children[0])
+		}
 	}
-	w.WriteString(nodeString)
+	for _, attributes := range n.Meta.HTMLAttributes {
+		out = withHTMLAttributes(out, attributes...) + "\n"
+	}
+	if len(n.Meta.Caption) != 0 {
+		caption := ""
+		for i, ns := range n.Meta.Caption {
+			if i != 0 {
+				caption += " "
+			}
+			caption += w.nodesAsString(ns...)
+		}
+		out = fmt.Sprintf("<figure>\n%s<figcaption>\n%s\n</figcaption>\n</figure>\n", out, caption)
+	}
+	w.WriteString(out)
 }
 
 func (w *HTMLWriter) writeTable(t Table) {
@@ -289,3 +304,24 @@ func (w *HTMLWriter) writeTableHeader(t TableHeader) {
 func (w *HTMLWriter) writeTableSeparator(t TableSeparator) {
 	w.WriteString("<tr></tr>\n")
 }
+
+func withHTMLAttributes(input string, kvs ...string) string {
+	if len(kvs)%2 != 0 {
+		panic(fmt.Sprintf("len of kvs must be even: %#v", kvs))
+	}
+	context := &h.Node{Type: h.ElementNode, Data: "body", DataAtom: atom.Body}
+	nodes, err := h.ParseFragment(strings.NewReader(strings.TrimSpace(input)), context)
+	if err != nil || len(nodes) != 1 {
+		panic(fmt.Sprintf("could not extend html attributes of %s: %v (%s)", input, len(nodes), err))
+	}
+	out, node := strings.Builder{}, nodes[0]
+	for i := 0; i < len(kvs)-1; i += 2 {
+		k, v := strings.TrimPrefix(kvs[i], ":"), kvs[i+1]
+		node.Attr = append(node.Attr, h.Attribute{Namespace: "", Key: k, Val: v})
+	}
+	err = h.Render(&out, nodes[0])
+	if err != nil {
+		panic(fmt.Sprintf("could not extend html attributes of %s: %#v (%s)", input, nodes, err))
+	}
+	return out.String()
+}

org/keyword.go

@@ -1,6 +1,7 @@
 package org
 
 import (
+	"encoding/csv"
 	"regexp"
 	"strings"
 )
@@ -12,14 +13,18 @@ type Keyword struct {
 
 type NodeWithMeta struct {
 	Node Node
-	Meta map[string]string
+	Meta Metadata
+}
+
+type Metadata struct {
+	Caption        [][]Node
+	HTMLAttributes [][]string
 }
 
 type Comment struct{ Content string }
 
 var keywordRegexp = regexp.MustCompile(`^(\s*)#\+([^:]+):(\s+(.*)|(\s*)$)`)
 var commentRegexp = regexp.MustCompile(`^(\s*)#(.*)`)
-var affiliatedKeywordRegexp = regexp.MustCompile(`^(CAPTION)$`)
 
 func lexKeywordOrComment(line string) (token, bool) {
 	if m := keywordRegexp.FindStringSubmatch(line); m != nil {
@@ -30,47 +35,44 @@ func lexKeywordOrComment(line string) (token, bool) {
 	return nilToken, false
 }
 
-func (d *Document) parseKeyword(i int, stop stopFn) (int, Node) {
-	k := parseKeyword(d.tokens[i])
-	if affiliatedKeywordRegexp.MatchString(k.Key) {
-		consumed, node := d.parseAffiliated(i, stop)
-		if consumed != 0 {
-			return consumed, node
-		}
-	} else {
-		d.BufferSettings[k.Key] = strings.Join([]string{d.BufferSettings[k.Key], k.Value}, "\n")
-	}
-	return 1, k
-}
-
 func (d *Document) parseComment(i int, stop stopFn) (int, Node) {
 	return 1, Comment{d.tokens[i].content}
 }
 
-func (d *Document) parseAffiliated(i int, stop stopFn) (int, Node) {
-	start, meta := i, map[string]string{}
-	for ; !stop(d, i) && d.tokens[i].kind == "keyword"; i++ {
-		k := parseKeyword(d.tokens[i])
-		if !affiliatedKeywordRegexp.MatchString(k.Key) {
-			return 0, nil
+func (d *Document) parseKeyword(i int, stop stopFn) (int, Node) {
+	k := parseKeyword(d.tokens[i])
+	if k.Key == "CAPTION" || k.Key == "ATTR_HTML" {
+		consumed, node := d.parseAffiliated(i, stop)
+		if consumed != 0 {
+			return consumed, node
 		}
-		if value, ok := meta[k.Key]; ok {
-			meta[k.Key] = value + " " + k.Value
-		} else {
-			meta[k.Key] = k.Value
+	}
+	d.BufferSettings[k.Key] = strings.Join([]string{d.BufferSettings[k.Key], k.Value}, "\n")
+	return 1, k
+}
+
+func (d *Document) parseAffiliated(i int, stop stopFn) (int, Node) {
+	start, meta := i, Metadata{}
+	for ; !stop(d, i) && d.tokens[i].kind == "keyword"; i++ {
+		switch k := parseKeyword(d.tokens[i]); k.Key {
+		case "CAPTION":
+			meta.Caption = append(meta.Caption, d.parseInline(k.Value))
+		case "ATTR_HTML":
+			r := csv.NewReader(strings.NewReader(k.Value))
+			r.Comma = ' '
+			attributes, err := r.Read()
+			if err != nil {
+				return 0, nil
+			}
+			meta.HTMLAttributes = append(meta.HTMLAttributes, attributes)
+		default:
+			return 0, nil
 		}
 	}
 	if stop(d, i) {
 		return 0, nil
 	}
-	consumed, node := 0, (Node)(nil)
-	if t := d.tokens[i]; t.kind == "text" {
-		if nodes := d.parseInline(t.content); len(nodes) == 1 && isImageOrVideoLink(nodes[0]) {
-			consumed, node = 1, Paragraph{nodes[:1]}
-		}
-	} else {
-		consumed, node = d.parseOne(i, stop)
-	}
+	consumed, node := d.parseOne(i, stop)
 	if consumed == 0 || node == nil {
 		return 0, nil
 	}

org/org.go

@@ -173,11 +173,25 @@ func (w *OrgWriter) writeKeyword(k Keyword) {
 	w.WriteString(w.indent + fmt.Sprintf("#+%s: %s\n", k.Key, k.Value))
 }
 
-func (w *OrgWriter) writeNodeWithMeta(m NodeWithMeta) {
-	for k, v := range m.Meta {
-		w.writeNodes(Keyword{k, v})
+func (w *OrgWriter) writeNodeWithMeta(n NodeWithMeta) {
+	for _, ns := range n.Meta.Caption {
+		w.WriteString("#+CAPTION: ")
+		w.writeNodes(ns...)
+		w.WriteString("\n")
 	}
-	w.writeNodes(m.Node)
+	for _, attributes := range n.Meta.HTMLAttributes {
+		w.WriteString("#+ATTR_HTML: ")
+		for i := 0; i < len(attributes)-1; i += 2 {
+			w.WriteString(attributes[i] + " ")
+			if strings.ContainsAny(attributes[i+1], "\t ") {
+				w.WriteString(`"` + attributes[i+1] + `"`)
+			} else {
+				w.WriteString(attributes[i+1])
+			}
+		}
+		w.WriteString("\n")
+	}
+	w.writeNodes(n.Node)
 }
 
 func (w *OrgWriter) writeComment(c Comment) {

org/testdata/blocks.html

@@ -1,4 +1,4 @@
-<div class="captioned">
+<figure>
 <div class="highlight">
 <pre>
 echo &#34;a bash source block&#34;
@@ -10,10 +10,10 @@ function hello {
 hello
 </pre>
 </div>
-<p class="caption">
+<figcaption>
 block caption
-</p>
-</div>
+</figcaption>
+</figure>
 <div class="highlight">
 <pre>
 a source block without a language

org/testdata/captions.html

@@ -1,27 +1,30 @@
 <p>
 Anything can be captioned. Also captions are not real, correct captions but just a paragraph below the element (bothe wrapped into a div)
 </p>
-<div class="captioned">
+<figure>
 <div class="highlight">
 <pre>
 echo &#34;i have a caption!&#34;
 </pre>
 </div>
-<p class="caption">
+<figcaption>
 captioned soure block
-</p>
-</div>
-<div class="captioned">
+</figcaption>
+</figure>
+<figure>
+<img src="http://placekitten.com/200/200#.png" alt="http://placekitten.com/200/200#.png" title="http://placekitten.com/200/200#.png" /><figcaption>
+captioned link (image in this case)
+</figcaption>
+</figure>
 <p>
-<video src="my-video.mp4" title="my-video.mp4">my-video.mp4</video>
+note that the whole paragraph is captioned, so a linebreak is needed for images to caption correctly
 </p>
-<p class="caption">
-captioned link (video in this case)
-</p>
-</div>
+<figure>
 <p>
-note that only that one line is captioned, not the whole paragraph
-</p>
-<p>
-also, normal text lines can&#39;t be captioned
+<img src="http://placekitten.com/200/200#.png" alt="http://placekitten.com/200/200#.png" title="http://placekitten.com/200/200#.png" />
+see?
 </p>
+<figcaption>
+captioned link (image in this case)
+</figcaption>
+</figure>

org/testdata/captions.org

@@ -5,9 +5,12 @@ Anything can be captioned. Also captions are not real, correct captions but just
 echo "i have a caption!"
 #+END_SRC
 
-#+CAPTION: captioned link (video in this case)
-[[my-video.mp4]]
-note that only that one line is captioned, not the whole paragraph
+#+CAPTION: captioned link (image in this case)
+[[http://placekitten.com/200/200#.png]]
+
+note that the whole paragraph is captioned, so a linebreak is needed for images to caption correctly
+
+#+CAPTION: captioned link (image in this case)
+[[http://placekitten.com/200/200#.png]]
+see?
 
-#+CAPTION: not happening!
-also, normal text lines can't be captioned

org/testdata/keywords.html

@@ -0,0 +1,18 @@
+<figure>
+<div class="highlight" class="a b c" id="it">
+<pre>echo &#34;a bash source block with custom html attributes&#34;
+</pre>
+</div>
+<figcaption>
+and <span style="text-decoration: underline;">multiple</span> lines of <strong>captions</strong>!
+</figcaption>
+</figure>
+<p>
+and an image with custom html attributes and a caption
+</p>
+<figure>
+<img src="http://placekitten.com/200/200#.png" alt="http://placekitten.com/200/200#.png" title="http://placekitten.com/200/200#.png" style="border: 10px solid black"/>
+<figcaption>
+kittens!
+</figcaption>
+</figure>

org/testdata/keywords.org

@@ -0,0 +1,13 @@
+
+#+CAPTION: and _multiple_
+#+CAPTION: lines of *captions*!
+#+ATTR_HTML: :class "a b c"
+#+ATTR_HTML: :id it
+#+BEGIN_SRC sh
+echo "a bash source block with custom html attributes"
+#+END_SRC
+
+and an image with custom html attributes and a caption
+#+CAPTION: kittens!
+#+ATTR_HTML: :style "border: 10px solid black"
+[[http://placekitten.com/200/200#.png]]