` elements in the document and replaces them in-place
+ * with `` blocks containing a pipe-based text representation.
+ */
+fun Document.convertTablesToText() {
+ // Snapshot the list to avoid concurrent modification
+ val tables = getElementsByTag("table").toList()
+ for (table in tables) {
+ val rows = extractRows(table)
+ if (rows.isEmpty()) {
+ table.remove()
+ continue
+ }
+
+ val maxCols = rows.maxOf { it.size }
+ // Pad rows with fewer cells
+ val normalizedRows = rows.map { row ->
+ row + List(maxCols - row.size) { "" }
+ }
+
+ val headerRowCount = detectHeaderRowCount(table)
+
+ val colWidths = IntArray(maxCols) { col ->
+ normalizedRows.maxOf { it[col].length }.coerceAtLeast(1)
+ }
+
+ val lines = buildList {
+ for ((i, row) in normalizedRows.withIndex()) {
+ add(formatRow(row, colWidths))
+ if (i == headerRowCount - 1 && headerRowCount > 0) {
+ add(formatSeparator(colWidths))
+ }
+ }
+ }
+
+ val text = lines.joinToString("\n")
+ val pre = Element("pre")
+ val code = Element("code")
+ code.appendText(text)
+ pre.appendChild(code)
+ table.replaceWith(pre)
+ }
+}
+
+private fun extractRows(table: Element): List> {
+ val rows = mutableListOf>()
+
+ val thead = table.getElementsByTag("thead").first()
+ val tbody = table.getElementsByTag("tbody").first()
+
+ if (thead != null) {
+ for (tr in thead.getElementsByTag("tr")) {
+ rows.add(extractCells(tr))
+ }
+ }
+ if (tbody != null) {
+ for (tr in tbody.getElementsByTag("tr")) {
+ rows.add(extractCells(tr))
+ }
+ }
+
+ // If no thead/tbody, get tr elements directly from the table
+ if (thead == null && tbody == null) {
+ for (tr in table.getElementsByTag("tr")) {
+ rows.add(extractCells(tr))
+ }
+ }
+
+ return rows
+}
+
+private fun extractCells(tr: Element): List {
+ return tr.children()
+ .filter { it.tagName() == "th" || it.tagName() == "td" }
+ .map { it.text().trim() }
+}
+
+/**
+ * Detects the number of header rows.
+ * If `` exists, its row count is used.
+ * Otherwise, if the first row contains only `| ` elements, it's treated as a header.
+ */
+private fun detectHeaderRowCount(table: Element): Int {
+ val thead = table.getElementsByTag("thead").first()
+ if (thead != null) {
+ return thead.getElementsByTag("tr").size
+ }
+
+ // Check if the first | contains only elements
+ val firstTr = table.getElementsByTag("tr").firstOrNull() ?: return 0
+ val cells = firstTr.children().filter { it.tagName() == "th" || it.tagName() == "td" }
+ return if (cells.isNotEmpty() && cells.all { it.tagName() == "th" }) 1 else 0
+}
+
+private fun formatRow(cells: List, colWidths: IntArray): String {
+ return cells.mapIndexed { i, cell ->
+ cell.padEnd(colWidths[i])
+ }.joinToString(" | ")
+}
+
+private fun formatSeparator(colWidths: IntArray): String {
+ return colWidths.joinToString("-+-") { "-".repeat(it) }
+}
diff --git a/libraries/matrixui/src/main/kotlin/io/element/android/libraries/matrix/ui/messages/ToHtmlDocument.kt b/libraries/matrixui/src/main/kotlin/io/element/android/libraries/matrix/ui/messages/ToHtmlDocument.kt
index ee9de516817..c3f10e2824c 100644
--- a/libraries/matrixui/src/main/kotlin/io/element/android/libraries/matrix/ui/messages/ToHtmlDocument.kt
+++ b/libraries/matrixui/src/main/kotlin/io/element/android/libraries/matrix/ui/messages/ToHtmlDocument.kt
@@ -13,6 +13,7 @@ import io.element.android.libraries.matrix.api.permalink.PermalinkParser
import io.element.android.libraries.matrix.api.timeline.item.event.FormattedBody
import io.element.android.libraries.matrix.api.timeline.item.event.MessageFormat
import io.element.android.wysiwyg.utils.HtmlToDomParser
+import org.jsoup.Jsoup
import org.jsoup.nodes.Document
/**
@@ -33,11 +34,12 @@ fun FormattedBody.toHtmlDocument(
// We don't trim the start in case it's used as indentation.
?.trimEnd()
?.let { formattedBody ->
- val dom = if (prefix != null) {
- HtmlToDomParser.document("$prefix $formattedBody")
- } else {
- HtmlToDomParser.document(formattedBody)
- }
+ val htmlToProcess = if (prefix != null) "$prefix $formattedBody" else formattedBody
+
+ // Convert tables to pre/code blocks before the wysiwyg safelist strips them
+ val processedHtml = convertTablesInHtml(htmlToProcess)
+
+ val dom = HtmlToDomParser.document(processedHtml)
// Prepend `@` to mentions
fixMentions(dom, permalinkParser)
@@ -46,6 +48,13 @@ fun FormattedBody.toHtmlDocument(
}
}
+private fun convertTablesInHtml(html: String): String {
+ if ("| A | B | | C | D | |
"
+ )
+ doc.convertTablesToText()
+
+ val code = doc.selectFirst("pre > code")
+ assertThat(code).isNotNull()
+ assertThat(code!!.wholeText()).isEqualTo(
+ "A | B\n" +
+ "C | D"
+ )
+ }
+
+ @Test
+ fun `table with thead header`() {
+ val doc = Jsoup.parse(
+ "