From d0bd26b9593c5d2f80c6ad5aea3688a1f4a91826 Mon Sep 17 00:00:00 2001
From: Nulo <nulo@sutty.nl>
Date: Fri, 25 Nov 2022 18:47:57 -0300
Subject: [PATCH] Recurse directory instead of pulling stdin

---
 index.js | 89 +++++++++++++++++++++++++++++++-------------------------
 1 file changed, 50 insertions(+), 39 deletions(-)

diff --git a/index.js b/index.js
index c82c48e..461bb5c 100644
--- a/index.js
+++ b/index.js
@@ -1,41 +1,39 @@
 import { Parser } from "htmlparser2";
 import { DomHandler } from "domhandler";
+import { readdir, readFile } from "fs/promises";
+import { join } from "path";
 
-const noop = () => {};
-// const log = noop;
-const { log } = console;
+const { argv } = process;
+const dirPath = argv[2] || ".";
 
-const stdin = await readStdin();
-const rawHtml = stdin;
-
-function recursive(el) {
+function recurseElement(report, el) {
   if (el.name === "a") {
     if (el.attribs.href) {
-      checkUrl("link", el, el.attribs.href);
+      checkUrl(report, "link", el, el.attribs.href);
     } else {
-      log("Link with no href:", getHtml(el));
+      report.warnings.push(`Link with no href: ${getHtml(el)}`);
     }
   }
   if (["audio", "video", "img"].includes(el.name)) {
     if (el.attribs.src) {
-      checkUrl(el.name, el, el.attribs.src);
+      checkUrl(report, el.name, el, el.attribs.src);
     } else {
-      log(`${el.name} with no src:`, getHtml(el));
+      report.warnings.push(`${el.name} with no src: ${getHtml(el)}`);
     }
   }
 
   for (const child of el.children) {
     if (child.type === "tag") {
-      recursive(child);
+      recurseElement(report, child);
     }
   }
 }
 
-function checkUrl(type, el, url) {
+function checkUrl(report, type, el, url) {
   if (isHttp(url)) {
-    log(`HTTP/S ${type}:`, getHtml(el));
+    report.warnings.push(`HTTP/S ${type}: ${getHtml(el)}`);
   } else if (isAbsolute(url)) {
-    log(`Absolute ${type}:`, getHtml(el));
+    report.warnings.push(`Absolute ${type}: ${getHtml(el)}`);
   }
 }
 
@@ -57,30 +55,43 @@ function getHtml(el) {
   return text;
 }
 
-const handler = new DomHandler(
-  (error, dom) => {
-    if (error) {
-      // TODO: Handle error
-    } else {
-      console.time();
-      for (const el of dom) {
-        if (el.type === "tag") {
-          recursive(el);
-        }
-      }
-      console.timeEnd();
-    }
-  },
-  { withEndIndices: true, withStartIndices: true }
-);
-const parser = new Parser(handler);
-parser.parseComplete(rawHtml);
-
-function readStdin() {
+function processFile(content) {
   return new Promise((resolve, reject) => {
-    let buffer = "";
-    process.stdin.resume();
-    process.stdin.on("data", (d) => (buffer = buffer.concat(d.toString())));
-    process.stdin.on("close", () => resolve(buffer));
+    const handler = new DomHandler(
+      (error, dom) => {
+        if (error) {
+          reject(error);
+        } else {
+          let report = {
+            warnings: [],
+          };
+          for (const el of dom) {
+            if (el.type === "tag") {
+              recurseElement(report, el);
+            }
+          }
+          resolve(report);
+        }
+      },
+      { withEndIndices: true, withStartIndices: true }
+    );
+    const parser = new Parser(handler);
+    parser.parseComplete(content);
   });
 }
+
+async function recurseDirectory(path) {
+  const dir = await readdir(path, { withFileTypes: true });
+  for (const file of dir) {
+    const filePath = join(path, file.name);
+    if (file.isDirectory()) recurseDirectory(filePath);
+    else {
+      if (!file.name.endsWith(".html")) continue;
+      const content = await readFile(filePath, "utf-8");
+      console.time(filePath);
+      await processFile(content);
+      console.timeEnd(filePath);
+    }
+  }
+}
+await recurseDirectory(dirPath);