@sdxc/mail 2026.10.9 → 2026.10.10

package/dist/lib/html-to-text.js

@@ -7,26 +7,19 @@
  * @author [Sergio Xalambrí](https://sergiodxa.com)
  * @copyright Sergio Xalambrí 2026
  */
-/** HTML comments, which never carry reader-visible content. */
-const COMMENT = /<!--[\s\S]*?-->/g;
-/** A document type declaration, which describes the document format. */
-const DOCTYPE = /<!doctype[^>]*>/gi;
-/**
- * Blocks hidden from sighted readers with `display:none`, such as a preheader.
- * Non-greedy matching closes at the first same-tag closing tag, which is correct
- * for the hidden single-element blocks email layouts use.
- */
-const HIDDEN_BLOCK = /<(div|span|p)\b[^>]*display\s*:\s*none[^>]*>[\s\S]*?<\/\1>/gi;
+/** Elements whose content belongs to the document structure. */
+const DROPPED_ELEMENTS = new Set(["head", "script", "style", "title"]);
+/** Elements an email layout hides with `display:none`, such as a preheader. */
+const HIDEABLE_ELEMENTS = new Set(["div", "span", "p"]);
+/** The name an opening tag starts with, lowercased by the caller. */
+const OPENING_NAME = /^<(\w+)/;
+/** An inline style that hides an element from sighted readers. */
+const DISPLAY_NONE = /display\s*:\s*none/i;
 /**
- * Blocks marked `data-skip-in-text`, the explicit signal an author gives that an
- * element belongs to the HTML part alone, covering visible content such as a
- * decorative rule, a spacer, or a logo's alt text.
+ * The explicit signal an author gives that an element belongs to the HTML part
+ * alone, covering visible content such as a decorative rule, a spacer, or a logo's alt text.
  */
-const SKIPPED_BLOCK = /<(\w+)\b[^>]*\bdata-skip-in-text\b[^>]*>[\s\S]*?<\/\1>/gi;
-/** The same marker on an element that closes itself, which has nothing to drop between tags. */
-const SKIPPED_VOID = /<\w+\b[^>]*\bdata-skip-in-text\b[^>]*\/?>/gi;
-/** Elements whose content belongs to the document structure. */
-const DROPPED_BLOCK = /<(head|script|style|title)\b[^>]*>[\s\S]*?<\/\1>/gi;
+const SKIP_IN_TEXT = /\bdata-skip-in-text\b/i;
 /** Explicit line breaks, the one inline element that carries layout meaning. */
 const LINE_BREAK = /<br\s*\/?>/gi;
 /** A link with its target, captured as double-quoted, single-quoted, or bare. */
@@ -52,8 +45,6 @@ const PARAGRAPH_BOUNDARY = /<\/?(p|div|h[1-6]|table|ul|ol|dl|blockquote|section|
 const CELL_END = /<\/(td|th)\s*>/gi;
 /** End of a table row or definition entry, which ends the line so the next row starts immediately below it. */
 const ROW_END = /<\/(tr|dt|dd|caption)\s*>/gi;
-/** Any remaining tag, dropped once its structural meaning has been applied. */
-const TAG = /<[^>]+>/g;
 /** A character reference in named, decimal, or hexadecimal form. */
 const ENTITY = /&(#\d+|#x[0-9a-f]+|[a-z]+);/gi;
 /** Named references worth decoding: the ones a renderer emits or copy commonly uses. */
@@ -72,9 +63,88 @@ const NAMED_ENTITIES = {
 const HORIZONTAL_WHITESPACE = /[ \t\r\f\v\u00a0]+/g;
 /** Three or more newlines, collapsed so structure never turns into empty screens. */
 const EXTRA_NEWLINES = /\n{3,}/g;
-/** Removes markup from a fragment that has already had its structure applied. */
+/**
+ * Removes markup from a fragment that has already had its structure applied. Each
+ * tag runs from a `<` to the next `>`, and a `<` with no `>` after it stays as text,
+ * so one forward pass leaves nothing that reads as a tag.
+ */
 function stripTags(html) {
-    return html.replace(TAG, "");
+    let out = "";
+    let index = 0;
+    while (index < html.length) {
+        let open = html.indexOf("<", index);
+        if (open === -1)
+            break;
+        let close = html.indexOf(">", open + 1);
+        if (close === -1)
+            break;
+        out += html.slice(index, open);
+        index = close + 1;
+    }
+    return out + html.slice(index);
+}
+/**
+ * The element an opening tag starts that the text part drops together with its
+ * content, or `null` when the tag stays. A document type declaration counts as an
+ * element with no content.
+ */
+function droppedElement(tag) {
+    let lower = tag.toLowerCase();
+    if (lower.startsWith("<!doctype"))
+        return "";
+    let name = OPENING_NAME.exec(lower)?.[1];
+    if (name === undefined)
+        return null;
+    if (DROPPED_ELEMENTS.has(name) || SKIP_IN_TEXT.test(tag))
+        return name;
+    if (HIDEABLE_ELEMENTS.has(name) && DISPLAY_NONE.test(tag))
+        return name;
+    return null;
+}
+/**
+ * Drops comments, the doctype, and every element a reader cannot see in one forward
+ * pass, so time stays linear and removing a region never joins its neighbours into a
+ * new tag. An element ends at its first same-name closing tag, or loses only its opening tag.
+ */
+function dropInvisible(html) {
+    let lower = html.toLowerCase();
+    let closings = new Map();
+    let out = "";
+    let index = 0;
+    while (index < html.length) {
+        let open = html.indexOf("<", index);
+        if (open === -1)
+            break;
+        out += html.slice(index, open);
+        if (html.startsWith("<!--", open)) {
+            let end = html.indexOf("-->", open + 4);
+            index = end === -1 ? html.length : end + 3;
+            continue;
+        }
+        let close = html.indexOf(">", open + 1);
+        if (close === -1) {
+            index = open;
+            break;
+        }
+        let tag = html.slice(open, close + 1);
+        let name = droppedElement(tag);
+        index = close + 1;
+        if (name === null) {
+            out += tag;
+            continue;
+        }
+        if (name === "" || tag.endsWith("/>"))
+            continue;
+        let closing = `</${name}>`;
+        let found = closings.get(name);
+        if (found === undefined || (found !== -1 && found < index)) {
+            found = lower.indexOf(closing, index);
+            closings.set(name, found);
+        }
+        if (found !== -1)
+            index = found + closing.length;
+    }
+    return out + html.slice(index);
 }
 /** Resolves character references to the characters they stand for, leaving unknown ones intact. */
 function decodeEntities(text) {
@@ -118,13 +188,7 @@ function formatLink(href, label) {
  * @example htmlToText('<p>Hi <a href="https://x.dev">here</a></p>'); // "Hi here (https://x.dev)"
  */
 export function htmlToText(html) {
-    let text = html
-        .replace(COMMENT, "")
-        .replace(DOCTYPE, "")
-        .replace(HIDDEN_BLOCK, "")
-        .replace(SKIPPED_BLOCK, "")
-        .replace(SKIPPED_VOID, "")
-        .replace(DROPPED_BLOCK, "")
+    let text = dropInvisible(html)
         .replace(LINE_BREAK, "\n")
         .replace(ANCHOR, (_match, _target, double, single, bare, label) => formatLink(double ?? single ?? bare ?? "", stripTags(label)))
         .replace(IMAGE, (_match, _quoted, double, single) => {

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@sdxc/mail",
-	"version": "2026.10.9",
+	"version": "2026.10.10",
 	"description": "Transport-agnostic transactional email with pluggable transports and remix/component rendering",
 	"license": "MIT",
 	"type": "module",
@@ -13,14 +13,14 @@
 		"./unsubscribe": "./dist/unsubscribe.js"
 	},
 	"dependencies": {
-		"@sdxc/crypto": "2026.10.9",
-		"@sdxc/highlight": "2026.10.9",
-		"@sdxc/logger": "2026.10.9",
-		"@sdxc/markdown": "2026.10.9",
+		"@sdxc/crypto": "2026.10.10",
+		"@sdxc/highlight": "2026.10.10",
+		"@sdxc/logger": "2026.10.10",
+		"@sdxc/markdown": "2026.10.10",
 		"@sdxc/result": "2026.10.6",
 		"remix": "3.0.0"
 	},
-	"gitHead": "64fa732ed0a8ca8693a35aeef6345ae5be2534ef",
+	"gitHead": "b61f19c59866708a2d89a5a8f4655ebbc16941b0",
 	"publishConfig": {
 		"access": "public"
 	},