fortified selector logic, [bug] selector can't process complex queries

sitecompanion/content.js

@@ -22,6 +22,45 @@ function findMinimumScope(text) {
   return deepest;
 }
 
+function escapeSelector(value) {
+  if (window.CSS && typeof CSS.escape === "function") {
+    return CSS.escape(value);
+  }
+  return String(value).replace(/[^a-zA-Z0-9_-]/g, "\\$&");
+}
+
+function buildSelector(node) {
+  if (!node || node.nodeType !== 1) return "body";
+  if (node === document.body) return "body";
+  const parts = [];
+  let current = node;
+  while (current && current.nodeType === 1 && current !== document.body) {
+    const tag = current.tagName.toLowerCase();
+    if (current.id) {
+      parts.unshift(`${tag}#${escapeSelector(current.id)}`);
+      break;
+    }
+    let selector = tag;
+    const parent = current.parentElement;
+    if (parent) {
+      const siblings = Array.from(parent.children).filter(
+        (child) => child.tagName === current.tagName
+      );
+      if (siblings.length > 1) {
+        const index = siblings.indexOf(current) + 1;
+        selector += `:nth-of-type(${index})`;
+      }
+    }
+    parts.unshift(selector);
+    current = parent;
+  }
+  if (current === document.body) {
+    parts.unshift("body");
+  }
+  const selector = parts.join(" > ");
+  return selector || "body";
+}
+
 function normalizeName(value) {
   return (value || "").trim().toLowerCase();
 }
@@ -274,12 +313,36 @@ chrome.runtime.onMessage.addListener((message, _sender, sendResponse) => {
       sendResponse({ ok: false, error: "Scope not found." });
       return;
     }
-    sendResponse({ ok: true, extracted: node.innerText || "" });
+    sendResponse({
+      ok: true,
+      extracted: node.innerText || "",
+      selector: buildSelector(node)
+    });
+    return;
+  }
+  if (message.type === "EXTRACT_BY_SELECTOR") {
+    const selector = message.selector || "";
+    if (!selector) {
+      sendResponse({ ok: false, error: "Missing selector." });
+      return;
+    }
+    let node = null;
+    try {
+      node = document.querySelector(selector);
+    } catch {
+      sendResponse({ ok: false, error: "Invalid selector." });
+      return;
+    }
+    if (!node) {
+      sendResponse({ ok: false, error: "Selector not found." });
+      return;
+    }
+    sendResponse({ ok: true, extracted: node.innerText || "", selector });
     return;
   }
   if (message.type === "EXTRACT_FULL") {
     const extracted = document.body?.innerText || "";
-    sendResponse({ ok: true, extracted });
+    sendResponse({ ok: true, extracted, selector: "body" });
   }
 });
 

sitecompanion/manifest.json

@@ -1,7 +1,7 @@
 {
   "manifest_version": 3,
   "name": "SiteCompanion",
-  "version": "0.4.0",
+  "version": "0.4.1",
   "description": "AI companion for site-bound text extraction and tasks.",
   "permissions": ["storage", "activeTab"],
   "host_permissions": ["<all_urls>"],

sitecompanion/popup.js

@@ -49,6 +49,7 @@ const state = {
   currentPopupState: "unknown",
   globalTheme: "system",
   forcedTask: null,
+  siteTextSelector: "",
   selectedTaskId: "",
   selectedEnvId: "",
   selectedProfileId: ""
@@ -75,7 +76,8 @@ function buildPopupDraft() {
     state: state.currentPopupState,
     siteText: state.siteText || "",
     urlPattern: urlPatternInput?.value?.trim() || "",
-    siteName: siteNameInput?.value?.trim() || ""
+    siteName: siteNameInput?.value?.trim() || "",
+    siteTextSelector: state.siteTextSelector || ""
   };
 }
 
@@ -99,6 +101,9 @@ function applyPopupDraft(draft) {
   if (typeof draft.siteName === "string") {
     siteNameInput.value = draft.siteName;
   }
+  if (typeof draft.siteTextSelector === "string") {
+    state.siteTextSelector = draft.siteTextSelector;
+  }
 }
 
 function matchUrl(url, pattern) {
@@ -190,10 +195,14 @@ function filterApiConfigsForScope(apiConfigs, workspace, site) {
 
 async function detectSite(url) {
   const { sites = [], workspaces = [] } = await getStorage(["sites", "workspaces"]);
-  state.sites = sites;
+  const normalizedSites = (Array.isArray(sites) ? sites : []).map((site) => ({
+    ...site,
+    extractSelector: site?.extractSelector || "body"
+  }));
+  state.sites = normalizedSites;
   state.workspaces = workspaces;
   
-  const site = sites.find(s => matchUrl(url, s.urlPattern));
+  const site = normalizedSites.find((s) => matchUrl(url, s.urlPattern));
   if (site) {
     state.currentSite = site;
     const workspace =
@@ -695,7 +704,12 @@ async function loadConfig() {
   const envs = normalizeConfigList(stored.envConfigs);
   const profiles = normalizeConfigList(stored.profiles);
   const shortcuts = normalizeConfigList(stored.shortcuts);
-  const sites = Array.isArray(stored.sites) ? stored.sites : state.sites;
+  const sites = Array.isArray(stored.sites)
+    ? stored.sites.map((site) => ({
+        ...site,
+        extractSelector: site?.extractSelector || "body"
+      }))
+    : state.sites;
   const workspaces = Array.isArray(stored.workspaces)
     ? stored.workspaces
     : state.workspaces;
@@ -709,6 +723,9 @@ async function loadConfig() {
     activeSite && activeSite.workspaceId
       ? workspaces.find((entry) => entry.id === activeSite.workspaceId)
       : null;
+  if (activeSite) {
+    state.currentSite = activeSite;
+  }
   if (activeWorkspace) {
     state.currentWorkspace = activeWorkspace;
     currentWorkspaceName.textContent = activeWorkspace.name || "Global";
@@ -795,13 +812,18 @@ async function loadTheme() {
 async function handleExtract() {
   setStatus("Extracting...");
   try {
-    const response = await sendToActiveTab({ type: "EXTRACT_FULL" });
+    const selector = state.currentSite?.extractSelector || "body";
+    const response = await sendToActiveTab({
+      type: "EXTRACT_BY_SELECTOR",
+      selector
+    });
     if (!response?.ok) {
       setStatus(response?.error || "No text detected.");
       return false;
     }
 
     state.siteText = response.extracted || "";
+    state.siteTextSelector = response.selector || selector;
     updateSiteTextCount();
     updatePromptCount(0);
     setStatus("Text extracted.");
@@ -1038,6 +1060,7 @@ partialTextPaste.addEventListener("input", async () => {
     const response = await sendToActiveTab({ type: "FIND_SCOPE", text });
     if (response?.ok) {
       state.siteText = response.extracted;
+      state.siteTextSelector = response.selector || "";
       extractedPreview.textContent = state.siteText;
       await fillSiteDefaultsFromTab();
       switchState("review");
@@ -1055,6 +1078,7 @@ extractFullBtn.addEventListener("click", async () => {
     const response = await sendToActiveTab({ type: "EXTRACT_FULL" });
     if (response?.ok) {
       state.siteText = response.extracted;
+      state.siteTextSelector = response.selector || "body";
       extractedPreview.textContent = state.siteText;
       await fillSiteDefaultsFromTab();
       switchState("review");
@@ -1083,6 +1107,7 @@ retryExtractBtn.addEventListener("click", () => {
   urlPatternInput.value = "";
   siteNameInput.value = "";
   state.siteText = "";
+  state.siteTextSelector = "";
   void clearPopupDraft();
   setStatus("Ready.");
 });
@@ -1110,7 +1135,8 @@ confirmSiteBtn.addEventListener("click", async () => {
     id: `site-${Date.now()}`,
     name,
     urlPattern: pattern,
-    workspaceId: "global" // Default to global for now
+    workspaceId: "global", // Default to global for now
+    extractSelector: state.siteTextSelector || "body"
   };
 
   state.sites.push(newSite);

sitecompanion/settings.js

@@ -2480,6 +2480,7 @@ function collectSites() {
     const nameInput = card.querySelector(".site-name");
     const patternInput = card.querySelector(".site-pattern");
     const workspaceSelect = card.querySelector(".site-workspace");
+    const extractInput = card.querySelector(".site-extract-selector");
     const themeSelect = card.querySelector(".appearance-theme");
     const toolbarSelect = card.querySelector(".appearance-toolbar-position");
     const envsContainer = card.querySelector(".site-envs");
@@ -2496,6 +2497,7 @@ function collectSites() {
       name: (nameInput?.value || "").trim(),
       urlPattern: (patternInput?.value || "").trim(),
       workspaceId: workspaceSelect?.value || "global",
+      extractSelector: (extractInput?.value || "").trim(),
       theme: themeSelect?.value || "inherit",
       toolbarPosition: toolbarSelect?.value || "inherit",
       envConfigs: envsContainer ? collectEnvConfigs(envsContainer) : [],
@@ -2611,6 +2613,22 @@ function buildSiteCard(site, allWorkspaces = []) {
   row.appendChild(deleteBtn);
   card.appendChild(row);
 
+  const extractField = document.createElement("div");
+  extractField.className = "field";
+  const extractLabel = document.createElement("label");
+  extractLabel.textContent = "Site Text Selector";
+  const extractInput = document.createElement("input");
+  extractInput.type = "text";
+  extractInput.value = site.extractSelector || "";
+  extractInput.className = "site-extract-selector";
+  extractInput.placeholder = "body";
+  extractInput.addEventListener("input", () => {
+    scheduleSidebarErrors();
+  });
+  extractField.appendChild(extractLabel);
+  extractField.appendChild(extractInput);
+  card.appendChild(extractField);
+
   const appearanceSection = buildAppearanceSection({
     theme: site.theme || "inherit",
     toolbarPosition: site.toolbarPosition || "inherit"
@@ -3396,6 +3414,7 @@ async function loadSettings() {
         ...site,
         name: site.name || site.urlPattern || "",
         workspaceId: site.workspaceId || "global",
+        extractSelector: typeof site.extractSelector === "string" ? site.extractSelector : "",
         theme: site.theme || "inherit",
         toolbarPosition: site.toolbarPosition || "inherit",
         envConfigs: normalizeConfigList(site.envConfigs),