diff --git a/design/author.md b/design/author.md
index b1295956..6c8df1b0 100644
--- a/design/author.md
+++ b/design/author.md
@@ -40,7 +40,9 @@ to 3,000. Thirty things the code does not say by itself:
field the editor gives its rich-text widget (`widget: "html"`, or a ShowWhen that draws one) takes
markup with the tags the content check keeps for it (its own, `div`, `span`, `p`, `br`, and what
they imply), checked here rather than filtered later: any other tag, an event handler, a style
- the field does not allow, a script address or a comment is a problem by name. Plain text written
+ the field does not allow, a script address, a comment, or a quote a tag leaves open, read as a
+ browser reads its attributes, is a problem by name. The last is what a `>` inside an attribute makes: the content check ends the tag there,
+ while a browser reads on past it and loses the text in between (#222). Plain text written
into such a field is wrapped as that widget wraps typing: in `
` where the field's `enterMode`
is `p` or its tags offer headings, `pre`, `p` or `div`, in `
` otherwise. Every plain default
is escaped too, English or translated, and a `null` default (Course Presentation's colours) is
diff --git a/packages/author/lib/text.mjs b/packages/author/lib/text.mjs
index 5a8b9192..d09400e4 100644
--- a/packages/author/lib/text.mjs
+++ b/packages/author/lib/text.mjs
@@ -106,6 +106,29 @@ function stylePatterns(field) {
*/
const PIECE = /(<(?=[^a-zA-Z!/])||$)|<[^>]*(>|$)|>)/g
const ATTRIBUTE = /([a-zA-Z_:][-a-zA-Z0-9_:.]*)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s"'>]+)))?/g
+
+/**
+ * One attribute as a browser reads it in a tag: a name, which may begin with `=` and holds no
+ * space, `/` or `=`; then, after an `=`, its value, quoted whole, bare to the next space, or a quote
+ * that opens and does not close, captured.
+ */
+const ATTRIBUTE_READ = /[\s/]*=?[^\s/=]*\s*(?:=\s*(?:"[^"]*"|'[^']*'|(["'])|\S*))?/y
+
+/**
+ * Whether a tag's attributes, as the content check cut them at the tag's first `>`, leave a quoted
+ * value open, read in the order a browser reads them: a quote opens a value only where one begins,
+ * after an attribute's `=`, so in `title=x='y` it is part of a bare value.
+ * @param {string} attributes
+ */
+function quoteLeftOpen(attributes) {
+ ATTRIBUTE_READ.lastIndex = 0
+ while (ATTRIBUTE_READ.lastIndex < attributes.length) {
+ const read = ATTRIBUTE_READ.exec(attributes)
+ if (!read?.[0]) return false
+ if (read[1]) return true
+ }
+ return false
+}
const SAFE_URL = /^(?:https?:|mailto:|ftp:|[^:]*$)/i
/**
@@ -164,6 +187,13 @@ export function markupText(value, field, report, { saved = false } = {}) {
if (!match[1]) {
// The slash of `
` is not part of the last attribute, as the content check reads it.
const attributes = match[3].trimEnd().replace(/\/$/, '')
+ // A quote the tag leaves open: a `>` inside a quoted value ended the tag here, as the content
+ // check splits markup, or the quote was never closed. A browser reads the value whole and the
+ // tag on to a later `>`, so the text in between is lost and what follows is drawn inside the
+ // tag's element (#222).
+ if (quoteLeftOpen(attributes)) {
+ once(`has <${tag}> with a quote it does not close: a > inside an attribute ends the tag there, so write > for it`)
+ }
for (const [, name, double, single, bare] of attributes.matchAll(ATTRIBUTE)) {
const attribute = name.toLowerCase()
const content = double ?? single ?? bare ?? ''
diff --git a/packages/author/tests/text.test.ts b/packages/author/tests/text.test.ts
index f69630a3..44a0c947 100644
--- a/packages/author/tests/text.test.ts
+++ b/packages/author/tests/text.test.ts
@@ -78,6 +78,23 @@ describe('text as H5P stores it', () => {
expect(problems).toEqual(['has a comment, which H5P removes: take it out'])
})
+ it('refuses a quote a tag leaves open, as a > inside an attribute ends the tag there and a browser reads on past it (#222)', () => {
+ const problems = (value: string) => {
+ const said: string[] = []
+ markupText(value, { tags: ['a'] }, (message) => said.push(message))
+ return said
+ }
+ const open = 'has
with a quote it does not close: a > inside an attribute ends the tag there, so write > for it'
+ expect(problems('link after
')).toEqual([open])
+ expect(problems("link x
')).toEqual([])
+ // Read as a browser reads attributes: a quote opens a value only where one begins, after an
+ // attribute's `=`, so one inside a bare value opens nothing, and one after a closed value does.
+ expect(problems('z w
')).toEqual([])
+ expect(problems('