From dfe673bff148ec4f2c109467b017294ec285cc89 Mon Sep 17 00:00:00 2001
From: Nulo <nulo@sutty.nl>
Date: Fri, 25 Nov 2022 19:53:24 -0300
Subject: [PATCH] Make reports more useful

---
 index.js | 80 ++++++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 61 insertions(+), 19 deletions(-)

diff --git a/index.js b/index.js
index 24d87cd..4fdb62a 100644
--- a/index.js
+++ b/index.js
@@ -6,38 +6,56 @@ import { join } from "path";
 const { argv } = process;
 const dirPath = argv[2] || ".";
 
-function recurseElement(report, el) {
+function recurseElement(report, rawHtml, el) {
   if (el.name === "a") {
     if (el.attribs.href !== undefined) {
       if (isHttp(el.attribs.href)) {
-        report.infos.push(`HTTP/S link: ${getHtml(el)}`);
+        report.things.push({
+          type: "link-http",
+          description: getText(el),
+        });
       } else if (isAbsolute(el.attribs.href)) {
-        report.warnings.push(`Absolute link: ${getHtml(el)}`);
+        report.things.push({
+          type: "link-absolute",
+          description: getText(el),
+        });
       }
     } else {
-      report.warnings.push(`Link with no href: ${getHtml(el)}`);
+      report.things.push({
+        type: "link-no-href",
+        description: getText(el),
+      });
     }
   }
   if (["audio", "video", "img"].includes(el.name)) {
     if (el.attribs.src) {
-      checkUrl(report, el.name, el, el.attribs.src);
+      checkUrl(report, rawHtml, "media", el, el.attribs.src);
     } else {
-      report.warnings.push(`${el.name} with no src: ${getHtml(el)}`);
+      report.things.push({
+        type: "media-no-src",
+        description: getHtml(rawHtml, el),
+      });
     }
   }
 
   for (const child of el.children) {
     if (child.type === "tag") {
-      recurseElement(report, child);
+      recurseElement(report, rawHtml, child);
     }
   }
 }
 
-function checkUrl(report, type, el, url) {
+function checkUrl(report, rawHtml, type, el, url) {
   if (isHttp(url)) {
-    report.warnings.push(`HTTP/S ${type}: ${getHtml(el)}`);
+    report.things.push({
+      type: type + "-http",
+      description: getHtml(rawHtml, el),
+    });
   } else if (isAbsolute(url)) {
-    report.warnings.push(`Absolute ${type}: ${getHtml(el)}`);
+    report.things.push({
+      type: type + "-absolute",
+      description: getHtml(rawHtml, el),
+    });
   }
 }
 
@@ -49,12 +67,14 @@ function isAbsolute(url) {
   return url.startsWith("/");
 }
 
-function getHtml(el) {
-  // return rawHtml.slice(el.startIndex, el.endIndex);
+function getHtml(rawHtml, el) {
+  return rawHtml.slice(el.startIndex, el.endIndex);
+}
+function getText(el) {
   let text = "";
   for (const child of el.children) {
     if (child.type === "text") text += child.data.trim();
-    else if (child.type === "tag") text += getHtml(child);
+    else if (child.type === "tag") text += getText(child);
   }
   return text;
 }
@@ -67,12 +87,13 @@ function processFile(content) {
           reject(error);
         } else {
           let report = {
+            things: [],
             warnings: [],
             infos: [],
           };
           for (const el of dom) {
             if (el.type === "tag") {
-              recurseElement(report, el);
+              recurseElement(report, content, el);
             }
           }
           resolve(report);
@@ -85,7 +106,7 @@ function processFile(content) {
   });
 }
 
-let reports = [];
+let reports = {};
 async function recurseDirectory(reports, path) {
   const dir = await readdir(path, { withFileTypes: true });
   for (const file of dir) {
@@ -95,14 +116,35 @@ async function recurseDirectory(reports, path) {
       if (!file.name.endsWith(".html")) continue;
       const content = await readFile(filePath, "utf-8");
       console.time(filePath);
-      reports.push(await processFile(content));
+      reports[path] = await processFile(content);
       console.timeEnd(filePath);
     }
   }
 }
 await recurseDirectory(reports, dirPath);
-const totalWarnings = reports.map((r) => r.warnings).flat();
+const totalThings = Object.values(reports)
+  .map((r) => r.things)
+  .flat();
+const kinds = new Set(totalThings.map((t) => t.type));
 console.log(
-  `Finished with ${reports.length} files read, ${totalWarnings.length} warnings`,
-  totalWarnings
+  `Finished with ${reports.length} files read, ${totalThings.length} things`
+);
+for (const kind of kinds) {
+  const count = totalThings.filter((t) => t.type === kind).length;
+  console.log(`==> ${kind}: ${count}`);
+}
+console.log("This means:");
+const pathBasedCount = totalThings.filter((t) =>
+  ["media-absolute", "media-http"].includes(t.type)
+).length;
+console.log(
+  `==> ${pathBasedCount} problems that affect users using legacy IPFS gateways`
+);
+const mediaHttp = totalThings.filter((t) => t.type === "media-http").length;
+console.log(
+  `==> ${mediaHttp} problems that make the website not self-contained, making it miss content if HTTP is unavailable`
+);
+const linkHttp = totalThings.filter((t) => t.type === "link-http").length;
+console.log(
+  `==> ${linkHttp} links to HTTP sites, which is not a real concern unless it's a key part of the site's navigation.`
 );