Unified function, simplified check scope, added further checks to avoid false positives

extension/popup/js/background.js

@@ -1,72 +1,23 @@
-// function to scan the url the tab is on for hidden files
-
-async function onTabUpdate(tabId, changeInfo, tab) {
-  if (changeInfo.status !== "complete") return;
-
-  const url = new URL(tab.url);
-  const hostname = url.hostname;
-
-  const stored = await browser.storage.local.get("entries");
-  const existingEntries = stored.entries || [];
-
-  const alreadyDone = existingEntries.some(e => e.domainField === hostname);
-  if (alreadyDone) return;
-
-  async function tryFetch(pathname) {
-    const target = url.origin + pathname;
-    let response;
-    try {
-      response = await fetch(target, { redirect: "manual" });
-    } catch {
-      return null;
-    }
-    return response.status === 200 ? target : null;
-  }
-
-  const results = {
-    env: await tryFetch("/.env"),
-    git: await tryFetch("/.git"),
-    dsstore: await tryFetch("/.DS_Store"),
-    config: await tryFetch("/.config"),
-    svn: await tryFetch("/.svn"),
-    npm: await tryFetch("/.npm"),
-    hg: await tryFetch("/.hg"),
-    docker: await tryFetch("/.docker"),
-  };
-
-  const newEntries = [...existingEntries];
-
-  for (const [key, foundPath] of Object.entries(results)) {
-    if (!foundPath) continue;
-
-    const entry = {
-      domainField: hostname,
-      pathField: foundPath,
-      type: key
-    };
-
-    newEntries.push(entry);
-
-  }
-
-  await browser.storage.local.set({ entries: newEntries });
-
-}
-
 // Enable, Idsable automatic listener and the message listener for it
-
 function enableListener() {
-  if (!browser.tabs.onUpdated.hasListener(onTabUpdate)) {
-    browser.tabs.onUpdated.addListener(onTabUpdate);
+  if (!browser.tabs.onUpdated.hasListener(tabUpdateListener)) {
+    browser.tabs.onUpdated.addListener(tabUpdateListener);
   }
 }
 
 function disableListener() {
-  if (browser.tabs.onUpdated.hasListener(onTabUpdate)) {
-    browser.tabs.onUpdated.removeListener(onTabUpdate);
+  if (browser.tabs.onUpdated.hasListener(tabUpdateListener)) {
+    browser.tabs.onUpdated.removeListener(tabUpdateListener);
   }
 }
 
+function tabUpdateListener(tabId, changeInfo, tab) {
+  if (changeInfo.status === "complete" && tab.active) {
+    runSingleScan(tab);
+  }
+}
+
+// Add and remove listeners based on message sent by frontend toggle
 browser.runtime.onMessage.addListener((msg) => {
   if (msg.type === "toggleListener") {
     if (msg.enabled) enableListener();
@@ -75,55 +26,115 @@ browser.runtime.onMessage.addListener((msg) => {
 });
 
 // Singe run, can be merged with onTabUpdate function
+async function runSingleScan(tab) {
+
+  console.log("Checking site...")
+
+  //const tabs = await browser.tabs.query({ active: true, currentWindow: true });
+
+  //if (!tabs.length) return;
+  //const tab = tabs[0];
 
-async function runSingleScan() {
-  const tabs = await browser.tabs.query({ active: true, currentWindow: true });
-  if (!tabs.length) return;
-  const tab = tabs[0];
   const url = new URL(tab.url);
   const hostname = url.hostname;
 
+
+  console.log(`Running scan on ${url}`)
+
   async function tryFetch(path) {
     const target = url.origin + path;
     let response;
     try {
-      response = await fetch(target, { redirect: "manual" });
+      response = await fetch(target);
     } catch {
       return null;
     }
-    return response.status === 200 ? target : null;
+    return [200, 301].includes(response.status) ? response : null;
   }
 
   const results = {
     env: await tryFetch("/.env"),
-    git: await tryFetch("/.git"),
-    dsstore: await tryFetch("/.DS_Store"),
-    config: await tryFetch("/.config"),
-    svn: await tryFetch("/.svn"),
-    npm: await tryFetch("/.npm"),
-    hg: await tryFetch("/.hg"),
-    docker: await tryFetch("/.docker"),
+    git: await tryFetch("/.git")
   };
 
   const stored = await browser.storage.local.get("entries");
   const entries = stored.entries || [];
 
-  for (const [key, foundPath] of Object.entries(results)) {
-    if (!foundPath) continue;
+  for (const [key, response] of Object.entries(results)) {
+    if (!response) continue;
+
+    const validated = await validatePathResponse(response.url, response);
+
+    if (!validated.ok) {
+      console.log(`Rejected ${key}: ${validated.reason}`);
+      continue;
+    }
+
+    console.log(`Hit! A hidden file was found: ${key}`)
 
     entries.push({
       domainField: hostname,
-      pathField: foundPath,
+      pathField: response.url,
       type: key
     });
-
   }
 
   await browser.storage.local.set({ entries });
+
 }
 
 browser.runtime.onMessage.addListener((msg) => {
   if (msg.type === "runOnce") {
-    runSingleScan();
+    browser.tabs.query({ active: true, currentWindow: true }).then(([tab]) => {
+      if (tab) runSingleScan(tab);
+    });
   }
 });
+
+
+// Check if returned value is a file/path based on heuristics
+async function validatePathResponse(url, response) {
+
+  const text = await response.text().catch(() => "");
+  const contentType = response.headers.get("content-type")?.toLowerCase() || "";
+  const status = response.status;
+
+  const fileIndicators = [
+    "application/octet-stream",
+    "application/x-git",
+    "text/plain",
+    "application/env",
+  ];
+
+  const isLikelyFile = fileIndicators.some(type => contentType.includes(type));
+  if (isLikelyFile && status === 200) {
+    return { ok: true, type: "file", reason: "content-type indicates file" };
+  }
+
+  const directoryKeywords = /(Index of|Directory listing|Parent Directory)/i;
+  const errorKeywords = /(oops|error|not found|forbidden|unauthorized|denied)/i;
+
+  const looksLikeDirectory = directoryKeywords.test(text);
+  const looksLikeErrorPage = errorKeywords.test(text);
+
+  if (looksLikeDirectory && !looksLikeErrorPage) {
+    return { ok: true, type: "directory", reason: "directory listing heuristics" };
+  }
+
+  if (url.endsWith("/.git")) {
+    const head = await tryFetchInternal(url + "/HEAD");
+    if (head?.status === 200) {
+      const headText = await head.text().catch(() => "");
+      if (/ref:/.test(headText)) {
+        return { ok: true, type: "directory", reason: ".git HEAD file found" };
+      }
+    }
+  }
+
+  if (status === 404 || looksLikeErrorPage) {
+    return { ok: false, reason: "soft 404 or error page" };
+  }
+
+  return { ok: false, reason: "did not match any valid directory/file heuristics" };
+
+}