Skip to content

Try to use regex heuristics instead of llm for generating one-click rules #176

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
muodov wants to merge 2 commits into
base: main
Choose a base branch
from
+34 −14
Draft

Try to use regex heuristics instead of llm for generating one-click rules #176

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cbf1fd3
Try to use regex heuristics instead of llm for generating one-click r…
muodov Oct 7, 2025
7dc59f6
lint
muodov Oct 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 8 additions & 3 deletions post-processing/generate-autoconsent-rules/generation.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -248,8 +248,12 @@ function generateRulesForSite(region, initialUrl, finalUrl, collectorResult, mat
const reviewNotes = [];
let keptCount = 0;

const llmConfirmedPopups = collectorResult.scrapedFrames.flatMap((frame) => frame.potentialPopups).filter((popup) => popup.llmMatch);
if (llmConfirmedPopups.length > 1 || llmConfirmedPopups[0].rejectButtons.length > 1) {
// const llmConfirmedPopups = collectorResult.scrapedFrames.flatMap((frame) => frame.potentialPopups).filter((popup) => popup.llmMatch);
const regexConfirmedPopups = collectorResult.scrapedFrames
.flatMap((frame) => frame.potentialPopups)
.filter((popup) => popup.regexMatch);
// if (llmConfirmedPopups.length > 1 || llmConfirmedPopups[0].rejectButtons.length > 1) {
if (regexConfirmedPopups.length > 1 || regexConfirmedPopups[0].rejectButtons.length > 1) {
Copy link

Copilot AI Oct 7, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This block assumes regexConfirmedPopups is non-empty (accessing regexConfirmedPopups[0]) based on an earlier guard in the caller; add an explicit guard or an assertion here to make the dependency clear, and consider passing the already-filtered list from the caller instead of recomputing it to avoid duplicated logic.

Suggested change
if (regexConfirmedPopups.length > 1 || regexConfirmedPopups[0].rejectButtons.length > 1) {
if (regexConfirmedPopups.length > 1 || (regexConfirmedPopups.length > 0 && regexConfirmedPopups[0].rejectButtons.length > 1)) {

Copilot uses AI. Check for mistakes.
console.warn('Multiple cookie popups or reject buttons found in', initialUrl);
reviewNotes.push({
needsReview: false, // it's not a problem by itself, unless this leads to multiple _rules_ generated, but we check that separately.
Expand All @@ -261,7 +265,8 @@ function generateRulesForSite(region, initialUrl, finalUrl, collectorResult, mat

// go over all frames, all confirmed popups within them, and all reject buttons inside
for (const frame of collectorResult.scrapedFrames) {
for (const popup of frame.potentialPopups.filter((p) => p.llmMatch)) {
// for (const popup of frame.potentialPopups.filter((p) => p.llmMatch)) {
for (const popup of frame.potentialPopups.filter((p) => p.regexMatch)) {
for (const button of popup.rejectButtons) {
if (ruleForButtonExists(button, matchingRules, newRules, rulesToOverride)) {
// if there is an existing rule with the same reject button, do nothing
Expand Down
34 changes: 24 additions & 10 deletions post-processing/generate-autoconsent-rules/main.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -148,16 +148,21 @@ async function processCookiePopupsForSite(globalParams, { finalUrl, initialUrl,

const updatedExistingRules = structuredClone(existingRules);

const llmConfirmedPopups = collectorResult.scrapedFrames.flatMap((frame) => frame.potentialPopups).filter((popup) => popup.llmMatch);
// const llmConfirmedPopups = collectorResult.scrapedFrames.flatMap((frame) => frame.potentialPopups).filter((popup) => popup.llmMatch);
const regexConfirmedPopups = collectorResult.scrapedFrames
.flatMap((frame) => frame.potentialPopups)
.filter((popup) => popup.regexMatch);

// shortcut if no popups with llmMatch
if (llmConfirmedPopups.length === 0) {
// if (llmConfirmedPopups.length === 0) {
if (regexConfirmedPopups.length === 0) {
return { newRuleFiles, updatedRuleFiles, keptCount: 0, reviewNotes: [], updatedExistingRules };
}

const matchingRules = findMatchingExistingRules(initialUrl, finalUrl, collectorResult, existingRules);
console.log(
`Detected ${llmConfirmedPopups.length} unhandled cookie popup(s) on ${finalUrl} (matched ${matchingRules.length} existing rules)`,
// `Detected ${llmConfirmedPopups.length} unhandled cookie popup(s) on ${finalUrl} (matched ${matchingRules.length} existing rules)`,
`Detected ${regexConfirmedPopups.length} unhandled cookie popup(s) on ${finalUrl} (matched ${matchingRules.length} existing rules)`,
);
const { newRules, rulesToOverride, reviewNotes, keptCount } = generateRulesForSite(
region,
Expand Down Expand Up @@ -309,26 +314,33 @@ async function processFiles(globalParams, existingRules) {
totalSitesWithPopups++;

const matchedRules = collectorResult.cmps.map((cmp) => cmp.name.trim()).filter((name) => name !== '');
const llmConfirmedPopups = collectorResult.scrapedFrames
// const llmConfirmedPopups = collectorResult.scrapedFrames
// .flatMap((frame) => frame.potentialPopups)
// .filter((popup) => popup.llmMatch);
const regexConfirmedPopups = collectorResult.scrapedFrames
.flatMap((frame) => frame.potentialPopups)
.filter((popup) => popup.llmMatch);
.filter((popup) => popup.regexMatch);
const screenshot = jsonData.data.screenshots;

if (hasKnownCmp(collectorResult.cmps)) {
totalSitesWithKnownCmps++;
autoconsentManifest.set(fileName, {
siteUrl: jsonData.finalUrl,
matchedRules,
llmConfirmedPopups,
// llmConfirmedPopups,
regexConfirmedPopups,
screenshot,
newlyCreatedRules: [],
updatedRules: [],
reviewNotes: [],
});
} else {
const llmConfirmedPopups = collectorResult.scrapedFrames
// const llmConfirmedPopups = collectorResult.scrapedFrames
// .flatMap((frame) => frame.potentialPopups)
// .filter((popup) => popup.llmMatch);
const regexConfirmedPopups = collectorResult.scrapedFrames
.flatMap((frame) => frame.potentialPopups)
.filter((popup) => popup.llmMatch);
.filter((popup) => popup.regexMatch);
/** @type {import('./types').AutoconsentManifestFileData[]} */
let newRuleFiles = [];
/** @type {import('./types').AutoconsentManifestFileData[]} */
Expand All @@ -337,7 +349,8 @@ async function processFiles(globalParams, existingRules) {
/** @type {import('./types').ReviewNote[]} */
let reviewNotes = [];

if (llmConfirmedPopups.length > 0) {
// if (llmConfirmedPopups.length > 0) {
if (regexConfirmedPopups.length > 0) {
totalUnhandled++;
const result = await processCookiePopupsForSite(globalParams, {
finalUrl: jsonData.finalUrl,
Expand All @@ -351,7 +364,8 @@ async function processFiles(globalParams, existingRules) {
autoconsentManifest.set(fileName, {
siteUrl: jsonData.finalUrl,
matchedRules,
llmConfirmedPopups,
// llmConfirmedPopups,
regexConfirmedPopups,
screenshot,
newlyCreatedRules: newRuleFiles,
updatedRules: updatedRuleFiles,
Expand Down
3 changes: 2 additions & 1 deletion post-processing/generate-autoconsent-rules/types.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,8 @@
* @typedef {{
* siteUrl: string;
* matchedRules: string[];
* llmConfirmedPopups: PopupData[];
* _llmConfirmedPopups?: PopupData[];
* regexConfirmedPopups: PopupData[];
* screenshot: string;
* newlyCreatedRules: AutoconsentManifestFileData[];
* updatedRules: AutoconsentManifestFileData[];
Expand Down