Improve non-English matching

src/DeviceInterface/InterfacePrototype.js

@@ -275,7 +275,7 @@ class InterfacePrototype {
         await this.postInit()
 
         if (this.settings.featureToggles.credentials_saving) {
-            initFormSubmissionsApi(this.scanner.forms)
+            initFormSubmissionsApi(this.scanner.forms, this.scanner.matching)
         }
     }
 

src/DeviceInterface/initFormSubmissionsApi.js

@@ -1,12 +1,13 @@
-import {SUBMIT_BUTTON_SELECTOR} from '../Form/selectors-css.js'
-import {buttonMatchesFormType, getText} from '../autofill-utils.js'
+import {buttonMatchesFormType, getTextShallow} from '../autofill-utils.js'
+import {extractElementStrings} from '../Form/label-util.js'
 
 /**
  * This is a single place to contain all functionality relating to form submission detection
  *
  * @param {Map<HTMLElement, import("../Form/Form").Form>} forms
+ * @param {import("../Form/matching").Matching} matching
  */
-export function initFormSubmissionsApi (forms) {
+export function initFormSubmissionsApi (forms, matching) {
     /**
      * Global submit events
      */
@@ -44,13 +45,13 @@ export function initFormSubmissionsApi (forms) {
         matchingForm?.submitHandler('global pointerdown event + matching form')
 
         if (!matchingForm) {
-            const selector = SUBMIT_BUTTON_SELECTOR + ', a[href="#"], a[href^=javascript], *[onclick]'
+            const selector = matching.cssSelector('submitButtonSelector') + ', a[href="#"], a[href^=javascript], *[onclick], [class*=button i]'
             // check if the click happened on a button
             const button = /** @type HTMLElement */(event.target)?.closest(selector)
             if (!button) return
 
-            const text = getText(button)
-            const hasRelevantText = /(log|sign).?(in|up)|continue|next|submit/i.test(text)
+            const text = getTextShallow(button) || extractElementStrings(button).join(' ')
+            const hasRelevantText = matching.getDDGMatcherRegex('submitButtonRegex')?.test(text)
             if (hasRelevantText && text.length < 25) {
                 // check if there's a form with values
                 const filledForm = [...forms.values()].find(form => form.hasValues())

src/Form/Form.js

@@ -7,7 +7,7 @@ import {
     isEventWithinDax,
     isLikelyASubmitButton,
     isPotentiallyViewable, buttonMatchesFormType,
-    safeExecute, getText, wasAutofilledByChrome, shouldLog
+    safeExecute, getTextShallow, wasAutofilledByChrome, shouldLog
 } from '../autofill-utils.js'
 
 import {getInputSubtype, getInputMainType, createMatching, safeRegex} from './matching.js'
@@ -216,8 +216,8 @@ class Form {
                 formValues.credentials.username = formValues.identities.phone
             } else {
                 // If we still don't have a username, try scanning the form's text for an email address
-                this.form.querySelectorAll('*:not(select):not(option)').forEach((el) => {
-                    const elText = getText(el)
+                this.form.querySelectorAll(this.matching.cssSelector('safeUniversalSelector')).forEach((el) => {
+                    const elText = getTextShallow(el)
                     // Ignore long texts to avoid false positives
                     if (elText.length > 70) return
 
@@ -358,11 +358,15 @@ class Form {
     }
 
     categorizeInputs () {
-        const selector = this.matching.cssSelector('FORM_INPUTS_SELECTOR')
+        const selector = this.matching.cssSelector('formInputsSelector')
         if (this.form.matches(selector)) {
             this.addInput(this.form)
         } else {
-            const foundInputs = this.form.querySelectorAll(selector)
+            let foundInputs = this.form.querySelectorAll(selector)
+            // If the markup is broken form.querySelectorAll may not return the fields, so we select from the parent
+            if (foundInputs.length === 0 && this.form instanceof HTMLFormElement && this.form.length > 0) {
+                foundInputs = this.form.parentElement?.querySelectorAll(selector) || foundInputs
+            }
             if (foundInputs.length < MAX_INPUTS_PER_FORM) {
                 foundInputs.forEach(input => this.addInput(input))
             } else {
@@ -375,12 +379,12 @@ class Form {
     }
 
     get submitButtons () {
-        const selector = this.matching.cssSelector('SUBMIT_BUTTON_SELECTOR')
+        const selector = this.matching.cssSelector('submitButtonSelector')
         const allButtons = /** @type {HTMLElement[]} */([...this.form.querySelectorAll(selector)])
 
         return allButtons
             .filter((btn) =>
-                isPotentiallyViewable(btn) && isLikelyASubmitButton(btn) && buttonMatchesFormType(btn, this)
+                isPotentiallyViewable(btn) && isLikelyASubmitButton(btn, this.matching) && buttonMatchesFormType(btn, this)
             )
     }
 

src/Form/FormAnalyzer.js

@@ -1,16 +1,7 @@
 import { removeExcessWhitespace, Matching } from './matching.js'
 import { constants } from '../constants.js'
 import { matchingConfiguration } from './matching-configuration.js'
-import { getText, isLikelyASubmitButton } from '../autofill-utils.js'
-
-const loginRegex = new RegExp(/sign(ing)?.?in(?!g)|log.?(i|o)n|log.?out|unsubscri|(forgot(ten)?|reset) (your )?password|password (forgotten|lost)|unlock|logged in as|mfa-submit-form/i)
-const signupRegex = new RegExp(
-    /sign(ing)?.?up|join|\bregist(er|ration)|newsletter|\bsubscri(be|ption)|contact|create|start|enroll|settings|preferences|profile|update|checkout|guest|purchase|buy|order|schedule|estimate|request|new.?customer|(confirm|retype|repeat) password|password confirm?/i
-)
-const conservativeSignupRegex = new RegExp(/sign.?up|join|register|enroll|newsletter|subscri(be|ption)|settings|preferences|profile|update/i)
-const strictSignupRegex = new RegExp(/sign.?up|join|register|(create|new).+account|enroll|settings|preferences|profile|update/i)
-const resetPasswordLink = new RegExp(/(forgot(ten)?|reset|don't remember) (your )?password|password forgotten/i)
-const loginProvidersRegex = new RegExp(/ with /i)
+import { getTextShallow, isLikelyASubmitButton } from '../autofill-utils.js'
 
 class FormAnalyzer {
     /** @type HTMLElement */
@@ -126,16 +117,16 @@ class FormAnalyzer {
         shouldCheckUnifiedForm = false,
         shouldBeConservative = false
     }) {
-        const matchesLogin = /current.?password/i.test(string) || loginRegex.test(string) || resetPasswordLink.test(string)
+        const matchesLogin = /current.?password/i.test(string) || this.matching.getDDGMatcherRegex('loginRegex')?.test(string) || this.matching.getDDGMatcherRegex('resetPasswordLink')?.test(string)
 
         // Check explicitly for unified login/signup forms
-        if (shouldCheckUnifiedForm && matchesLogin && strictSignupRegex.test(string)) {
+        if (shouldCheckUnifiedForm && matchesLogin && this.matching.getDDGMatcherRegex('conservativeSignupRegex')?.test(string)) {
             this.increaseHybridSignal(strength, signalType)
             return this
         }
 
-        const signupRegexToUse = shouldBeConservative ? conservativeSignupRegex : signupRegex
-        const matchesSignup = /new.?password/i.test(string) || signupRegexToUse.test(string)
+        const signupRegexToUse = this.matching.getDDGMatcherRegex(shouldBeConservative ? 'conservativeSignupRegex' : 'signupRegex')
+        const matchesSignup = /new.?password/i.test(string) || signupRegexToUse?.test(string)
 
         // In some cases a login match means the login is somewhere else, i.e. when a link points outside
         if (shouldFlip) {
@@ -162,6 +153,24 @@ class FormAnalyzer {
         })
     }
 
+    evaluateUrl () {
+        const path = window.location.pathname
+
+        const matchesLogin = this.matching.getDDGMatcherRegex('loginRegex')?.test(path)
+        const matchesSignup = this.matching.getDDGMatcherRegex('conservativeSignupRegex')?.test(path)
+
+        // If the url matches both, do nothing: the signal is probably confounding
+        if (matchesLogin && matchesSignup) return
+
+        if (matchesLogin) {
+            this.decreaseSignalBy(1, 'url matches login')
+        }
+
+        if (matchesSignup) {
+            this.increaseSignalBy(1, 'url matches signup')
+        }
+    }
+
     evaluatePageTitle () {
         const pageTitle = document.title
         this.updateSignal({string: pageTitle, strength: 2, signalType: `page title: ${pageTitle}`, shouldCheckUnifiedForm: true})
@@ -185,7 +194,7 @@ class FormAnalyzer {
         this.evaluatePageTitle()
         this.evaluatePageHeadings()
         // Check for submit buttons
-        const buttons = document.querySelectorAll(this.matching.cssSelector('SUBMIT_BUTTON_SELECTOR'))
+        const buttons = document.querySelectorAll(this.matching.cssSelector('submitButtonSelector'))
         buttons.forEach(button => {
             // if the button has a form, it's not related to our input, because our input has no form here
             if (button instanceof HTMLButtonElement) {
@@ -198,7 +207,7 @@ class FormAnalyzer {
     }
 
     evaluateElement (el) {
-        const string = getText(el)
+        const string = getTextShallow(el)
 
         if (el.matches(this.matching.cssSelector('password'))) {
             // These are explicit signals by the web author, so we weigh them heavily
@@ -207,12 +216,13 @@ class FormAnalyzer {
                 strength: 5,
                 signalType: `explicit: ${el.getAttribute('autocomplete')}`
             })
+            return
         }
 
         // check button contents
-        if (el.matches(this.matching.cssSelector('SUBMIT_BUTTON_SELECTOR'))) {
+        if (el.matches(this.matching.cssSelector('submitButtonSelector') + ', *[class*=button]')) {
             // If we're confident this is the submit button, it's a stronger signal
-            let likelyASubmit = isLikelyASubmitButton(el)
+            let likelyASubmit = isLikelyASubmitButton(el, this.matching)
             if (likelyASubmit) {
                 this.form.querySelectorAll('input[type=submit], button[type=submit]').forEach(
                     (submit) => {
@@ -225,6 +235,7 @@ class FormAnalyzer {
             }
             const strength = likelyASubmit ? 20 : 2
             this.updateSignal({string, strength, signalType: `submit: ${string}`})
+            return
         }
         // if an external link matches one of the regexes, we assume the match is not pertinent to the current form
         if (
@@ -234,8 +245,8 @@ class FormAnalyzer {
         ) {
             let shouldFlip = true
             if (
-                resetPasswordLink.test(string) || // Don't flip forgotten password links
-                loginProvidersRegex.test(string) // Don't flip login providers links
+                this.matching.getDDGMatcherRegex('resetPasswordLink')?.test(string) || // Don't flip forgotten password links
+                this.matching.getDDGMatcherRegex('loginProvidersRegex')?.test(string) // Don't flip login providers links
             ) {
                 shouldFlip = false
             }
@@ -250,14 +261,17 @@ class FormAnalyzer {
     }
 
     evaluateForm () {
+        // Check page url
+        this.evaluateUrl()
+
         // Check page title
         this.evaluatePageTitle()
 
         // Check form attributes
         this.evaluateElAttributes(this.form)
 
-        // Check form contents (skip select and option because they contain too much noise)
-        this.form.querySelectorAll('*:not(select):not(option):not(script)').forEach(el => {
+        // Check form contents (noisy elements are skipped with the safeUniversalSelector)
+        this.form.querySelectorAll(this.matching.cssSelector('safeUniversalSelector')).forEach(el => {
             // Check if element is not hidden. Note that we can't use offsetHeight
             // nor intersectionObserver, because the element could be outside the
             // viewport or its parent hidden
@@ -266,7 +280,7 @@ class FormAnalyzer {
         })
 
         // A form with many fields is unlikely to be a login form
-        const relevantFields = this.form.querySelectorAll(this.matching.cssSelector('GENERIC_TEXT_FIELD'))
+        const relevantFields = this.form.querySelectorAll(this.matching.cssSelector('genericTextField'))
         if (relevantFields.length >= 4) {
             this.increaseSignalBy(relevantFields.length * 1.5, 'many fields: it is probably not a login')
         }

src/Form/formatters.js

@@ -14,7 +14,7 @@ const FOUR_DIGIT_YEAR_REGEX = /(\D)\1{3}|\d{4}/i
  * @returns {string}
  */
 const formatCCYear = (input, year, form) => {
-    const selector = form.matching.cssSelector('FORM_INPUTS_SELECTOR')
+    const selector = form.matching.cssSelector('formInputsSelector')
     if (
         input.maxLength === 4 ||
         checkPlaceholderAndLabels(input, FOUR_DIGIT_YEAR_REGEX, form.form, selector)
@@ -34,7 +34,7 @@ const formatCCYear = (input, year, form) => {
 const getUnifiedExpiryDate = (input, month, year, form) => {
     const formattedYear = formatCCYear(input, year, form)
     const paddedMonth = `${month}`.padStart(2, '0')
-    const cssSelector = form.matching.cssSelector('FORM_INPUTS_SELECTOR')
+    const cssSelector = form.matching.cssSelector('formInputsSelector')
     const separator = matchInPlaceholderAndLabels(input, DATE_SEPARATOR_REGEX, form.form, cssSelector)?.groups?.separator || '/'
 
     return `${paddedMonth}${separator}${formattedYear}`

src/Form/input-classifiers.test.js

@@ -6,7 +6,6 @@ import { getInputSubtype, createMatching } from './matching.js'
 import { Form } from './Form.js'
 import InterfacePrototype from '../DeviceInterface/InterfacePrototype.js'
 
-import {SUBMIT_BUTTON_SELECTOR} from './selectors-css.js'
 import {createAvailableInputTypes} from '../../integration-test/helpers/utils.js'
 
 /**
@@ -167,7 +166,8 @@ describe.each(testCases)('Test $html fields', (testCase) => {
         baseWrapper.innerHTML = testContent
         document.title = title
 
-        const buttons = document.querySelectorAll(SUBMIT_BUTTON_SELECTOR)
+        const matching = createMatching()
+        const buttons = document.querySelectorAll(matching.cssSelector('submitButtonSelector'))
         buttons.forEach((button) => {
             // We're doing this so that isPotentiallyViewable(button) === true. See jest.setup.js for more info
             // @ts-ignore

src/Form/label-util.js

@@ -6,7 +6,7 @@ const EXCLUDED_TAGS = ['SCRIPT', 'NOSCRIPT', 'OPTION', 'STYLE']
  * "element.textContent" is a string which is merged of all children nodes,
  * which can cause issues with things like script tags etc.
  *
- * @param  {HTMLElement} element
+ * @param  {Element} element
  *         A DOM element to be extracted.
  * @returns {string[]}
  *          All strings in an element.