Bug 1495881 - Compare existing autofill entries using a whitespace-free, case-insensitive, and accent-folding approach. r=MattN

Differential Revision: https://phabricator.services.mozilla.com/D14338

--HG--
extra : moz-landing-system : lando

browser/extensions/formautofill/FormAutofillStorage.jsm

@@ -1549,6 +1549,8 @@ class Addresses extends AutofillRecords {
     this._normalizeRecord(addressToMerge, strict);
     let hasMatchingField = false;
 
+    let country = addressFound.country || addressToMerge.country || FormAutofill.DEFAULT_REGION;
+    let collators = FormAutofillUtils.getSearchCollators(country);
     for (let field of this.VALID_FIELDS) {
       let existingField = addressFound[field];
       let incomingField = addressToMerge[field];
@@ -1557,14 +1559,17 @@ class Addresses extends AutofillRecords {
           // Treat "street-address" as mergeable if their single-line versions
           // match each other.
           if (field == "street-address" &&
-              FormAutofillUtils.compareStreetAddress(existingField, incomingField)) {
-            // Keep the value in storage if its amount of lines is greater than
+              FormAutofillUtils.compareStreetAddress(existingField, incomingField, collators)) {
+            // Keep the street-address in storage if its amount of lines is greater than
             // or equal to the incoming one.
             if (existingField.split("\n").length >= incomingField.split("\n").length) {
               // Replace the incoming field with the one in storage so it will
               // be further merged back to storage.
               addressToMerge[field] = existingField;
             }
+          } else if (field != "street-address" &&
+                     FormAutofillUtils.strCompare(existingField, incomingField, collators)) {
+            addressToMerge[field] = existingField;
           } else {
             this.log.debug("Conflicts: field", field, "has different value.");
             return false;

browser/extensions/formautofill/FormAutofillUtils.jsm

@@ -311,7 +311,7 @@ this.FormAutofillUtils = {
    * Internal method to split an address to multiple parts per the provided delimiter,
    * removing blank parts.
    * @param {string} address The address the split
-   * @param {string} delimiter The separator that is used between lines in the address
+   * @param {string} [delimiter] The separator that is used between lines in the address
    * @returns {string[]}
    */
   _toStreetAddressParts(address, delimiter = "\n") {
@@ -328,7 +328,7 @@ this.FormAutofillUtils = {
   /**
    * Converts a street address to a single line, removing linebreaks marked by the delimiter
    * @param {string} address The address the convert
-   * @param {string} delimiter The separator that is used between lines in the address
+   * @param {string} [delimiter] The separator that is used between lines in the address
    * @returns {string}
    */
   toOneLineAddress(address, delimiter = "\n") {
@@ -340,13 +340,14 @@ this.FormAutofillUtils = {
    * Compares two addresses, removing internal whitespace
    * @param {string} a The first address to compare
    * @param {string} b The second address to compare
-   * @param {string} delimiter The separator that is used between lines in the address
+   * @param {array} collators Search collators that will be used for comparison
+   * @param {string} [delimiter="\n"] The separator that is used between lines in the address
    * @returns {boolean} True if the addresses are equal, false otherwise
    */
-  compareStreetAddress(a, b, delimiter = "\n") {
+  compareStreetAddress(a, b, collators, delimiter = "\n") {
     let oneLineA = this._toStreetAddressParts(a, delimiter).map(p => p.replace(/\s/g, "")).join("");
     let oneLineB = this._toStreetAddressParts(b, delimiter).map(p => p.replace(/\s/g, "")).join("");
-    return oneLineA == oneLineB;
+    return this.strCompare(oneLineA, oneLineB, collators);
   },
 
   /**
@@ -466,7 +467,7 @@ this.FormAutofillUtils = {
    * @param   {string} country The specified country.
    * @returns {array} An array containing several collator objects.
    */
-  getCollators(country) {
+  getSearchCollators(country) {
     // TODO: Only one language should be used at a time per country. The locale
     //       of the page should be taken into account to do this properly.
     //       We are going to support more countries in bug 1370193 and this
@@ -475,7 +476,12 @@ this.FormAutofillUtils = {
     if (!this._collators[country]) {
       let dataset = this.getCountryAddressData(country);
       let languages = dataset.languages || [dataset.lang];
-      this._collators[country] = languages.map(lang => new Intl.Collator(lang, {sensitivity: "base", ignorePunctuation: true}));
+      let options = {
+        ignorePunctuation: true,
+        sensitivity: "base",
+        usage: "search",
+      };
+      this._collators[country] = languages.map(lang => new Intl.Collator(lang, options));
     }
     return this._collators[country];
   },
@@ -559,7 +565,7 @@ this.FormAutofillUtils = {
     let countries = countrySpecified ? [countrySpecified] : [...FormAutofill.countries.keys()];
 
     for (let country of countries) {
-      let collators = this.getCollators(country);
+      let collators = this.getSearchCollators(country);
       let metadata = this.getCountryAddressData(country);
       if (country != metadata.key) {
         // We hit the fallback logic in getCountryAddressRawData so ignore it as
@@ -611,7 +617,7 @@ this.FormAutofillUtils = {
   getAbbreviatedSubregionName(subregionValues, country) {
     let values = Array.isArray(subregionValues) ? subregionValues : [subregionValues];
 
-    let collators = this.getCollators(country);
+    let collators = this.getSearchCollators(country);
     for (let metadata of this.getCountryAddressDataWithLocales(country)) {
       let {sub_keys: subKeys, sub_names: subNames, sub_lnames: subLnames} = metadata;
       if (!subKeys) {
@@ -662,7 +668,7 @@ this.FormAutofillUtils = {
       return null;
     }
 
-    let collators = this.getCollators(address.country);
+    let collators = this.getSearchCollators(address.country);
 
     for (let option of selectEl.options) {
       if (this.strCompare(value, option.value, collators) ||

browser/extensions/formautofill/test/unit/test_addressRecords.js

@@ -288,6 +288,59 @@ const MERGE_TESTCASES = [
       country: "US",
     },
   },
+  {
+    description: "Merge an address with punctuation and mIxEd-cAsE",
+    addressInStorage: {
+      "given-name": "Timothy",
+      "street-address": "331 E. Evelyn Avenue",
+      "tel": "+16509030800",
+    },
+    addressToMerge: {
+      "street-address": "331.e.EVELYN AVENUE",
+      country: "US",
+    },
+    expectedAddress: {
+      "given-name": "Timothy",
+      "street-address": "331 E. Evelyn Avenue",
+      "tel": "+16509030800",
+      country: "US",
+    },
+  },
+  {
+    description: "Merge an address with accent characters",
+    addressInStorage: {
+      "given-name": "Timothy",
+      "street-address": "331 E. Evelyn Straße",
+      "tel": "+16509030800",
+    },
+    addressToMerge: {
+      "street-address": "331.e.EVELYN Strasse",
+      country: "US",
+    },
+    expectedAddress: {
+      "given-name": "Timothy",
+      "street-address": "331 E. Evelyn Straße",
+      "tel": "+16509030800",
+      country: "US",
+    },
+  },
+  {
+    description: "Merge an address with a mIxEd-cAsE name",
+    addressInStorage: {
+      "given-name": "Timothy",
+      "tel": "+16509030800",
+    },
+    addressToMerge: {
+      "given-name": "TIMOTHY",
+      "tel": "+16509030800",
+      country: "US",
+    },
+    expectedAddress: {
+      "given-name": "Timothy",
+      "tel": "+16509030800",
+      country: "US",
+    },
+  },
 ];
 
 ChromeUtils.defineModuleGetter(this, "Preferences",