From 2745531bc541b9e526f0b25ffd396dad4749546b Mon Sep 17 00:00:00 2001
From: nsfisis <nsfisis@gmail.com>
Date: Wed, 15 Mar 2023 01:37:33 +0900
Subject: feat(nuldoc): support CDATA section

---
 nuldoc-src/xml.ts | 37 +++++++++++++++++++++++++++++++++++--
 1 file changed, 35 insertions(+), 2 deletions(-)

(limited to 'nuldoc-src/xml.ts')

diff --git a/nuldoc-src/xml.ts b/nuldoc-src/xml.ts
index 0bfbd8d..bb7d499 100644
--- a/nuldoc-src/xml.ts
+++ b/nuldoc-src/xml.ts
@@ -3,7 +3,6 @@ import { XmlParseError } from "./errors.ts";
 
 // TODO
 // Support comment? <!-- -->
-// Support CDATA
 
 export async function parseXmlFile(filePath: string): Promise<Element> {
   const source = await Deno.readTextFile(filePath);
@@ -64,8 +63,11 @@ function parseChildNodes(p: Parser): Node[] {
     if (c === "<") {
       if (c2 === "/") {
         break;
+      } else if (c2 === "!") {
+        nodes.push(parseCdata(p));
+      } else {
+        nodes.push(parseXmlElement(p));
       }
-      nodes.push(parseXmlElement(p));
     } else {
       nodes.push(parseTextNode(p));
     }
@@ -81,6 +83,37 @@ function parseTextNode(p: Parser): Text {
   };
 }
 
+function parseCdata(p: Parser): Text {
+  expect(p, "<![CDATA[");
+  const content = skipTo(p, "]]>");
+  next(p, "]]>".length);
+  return {
+    kind: "text",
+    content: formatCdata(content),
+  };
+}
+
+function formatCdata(s: string): string {
+  // <![CDATA[
+  //   foo
+  //     bar
+  //   baz
+  // ]]>
+  // => "foo\n  bar\nbaz"
+  s = s.replace(/^\n(.*)\n *$/s, "$1");
+  const ls = s.split("\n");
+  const n = Math.min(
+    ...ls.filter((l) => l !== "").map((l) =>
+      l.match(/^( *)/)?.[0]?.length ?? 0
+    ),
+  );
+  let z = "";
+  for (const p of s.split("\n")) {
+    z += p.slice(n) + "\n";
+  }
+  return z.slice(0, -1);
+}
+
 function parseStartTag(
   p: Parser,
 ): { name: string; attributes: Map<string, string>; closed: boolean } {
-- 
cgit v1.2.3-70-g09d2