diff --git a/dist/services/MimeVerificationService/MimeVerificationService.js b/dist/services/MimeVerificationService/MimeVerificationService.js
index 71e7152326755f0b5943294b0197c6287a256f4f..120b643361249c0f1ccdd991b2bcd7fee2eca551 100644
--- a/dist/services/MimeVerificationService/MimeVerificationService.js
+++ b/dist/services/MimeVerificationService/MimeVerificationService.js
@@ -18,12 +18,13 @@ const index_1 = require("./index");
const index_2 = require("../../index");
const VerificationError_1 = require("../VerificationService/VerificationError");
const stringUtils_1 = require("../../utils/stringUtils");
-const ELEMENT_NODE = 1;
-const TEXT_NODE = 3;
-const DOCUMENT_NODE = 9;
-const vendorProcessingFunctions = {
- [StatusesService_1.EMAIL_VENDORS.GMAIL]: normalizationStrategies_1.processGmailElement,
- [StatusesService_1.EMAIL_VENDORS.OUTLOOK]: normalizationStrategies_1.processOutlookElement,
+const utils_1 = require("./utils");
+const vendorPruningFunctions = {
+ [StatusesService_1.EMAIL_VENDORS.GMAIL]: normalizationStrategies_1.pruneGmailElement,
+ [StatusesService_1.EMAIL_VENDORS.OUTLOOK]: normalizationStrategies_1.pruneOutlookElement,
+};
+const vendorPrintingFunctions = {
+ [StatusesService_1.EMAIL_VENDORS.OUTLOOK]: utils_1.printOutlookElement,
};
// Load JSDOM dynamically for Node env only, because built CRA is crashing with it
let JSDOM;
@@ -97,100 +98,17 @@ const normalizeVendorHtml = (htmlString, vendor) => {
document = window.document;
}
const mimeBody = document.firstChild.lastChild;
- const elementProcessingFunction = vendorProcessingFunctions[vendor];
- if (!elementProcessingFunction) {
+ const elementPruningFunction = vendorPruningFunctions[vendor];
+ if (!elementPruningFunction) {
throw new Error(`Vendor "${vendor}" is not supported. Please, develop a normalization strategy for it.`);
}
- pruneHtmlNode(document, elementProcessingFunction);
- return printHtmlChildren(mimeBody, 0);
+ utils_1.pruneHtmlNode(document, elementPruningFunction);
+ const vendorPrintFunction = vendorPrintingFunctions[vendor];
+ return utils_1.printHtmlChildren(mimeBody, vendorPrintFunction, 0);
};
const normalizePlainPart = (text) => {
return stringUtils_1.removeSpacesAndLinebreaks(text);
};
-const printHtmlChildren = (node, depth) => {
- let child = node.firstChild;
- if (!child) {
- return "";
- }
- if (child == node.lastChild && child.nodeType == TEXT_NODE) {
- return printHtmlNode(child, depth);
- }
- else {
- let result = "";
- while (child) {
- result = result.concat(printHtmlNode(child, depth));
- child = child.nextSibling;
- }
- return result;
- }
-};
-const printHtmlNode = (node, depth) => {
- let result = "";
- switch (node.nodeType) {
- case TEXT_NODE:
- result += "<TEXT>";
- result += stringUtils_1.removeSpacesAndLinebreaks(node.textContent);
- result += "</TEXT>";
- result += "\n";
- break;
- case DOCUMENT_NODE:
- result += printHtmlChildren(node, depth);
- break;
- case ELEMENT_NODE:
- result += "<" + node.nodeName;
- Array.from(node.attributes)
- .sort((a, b) => a.name.localeCompare(b.name))
- .forEach((attribute) => {
- result += ` ${attribute.name}`;
- if (attribute.value) {
- result += `="${common_1.escapeHtmlString(attribute.value)}"`;
- }
- });
- if (node.firstChild) {
- result += ">";
- result += "\n";
- result += printHtmlChildren(node, depth + 1);
- result += "</" + node.nodeName + ">";
- }
- else {
- result += "/>";
- }
- result += "\n";
- }
- return result;
-};
-const pruneHtmlNode = (node, pruneElement) => {
- let toBeRemoved = false;
- switch (node.nodeType) {
- case node.COMMENT_NODE:
- case node.DOCUMENT_TYPE_NODE:
- toBeRemoved = true;
- break;
- case node.TEXT_NODE: {
- const trimmedText = node.textContent.trim();
- if (trimmedText === "") {
- toBeRemoved = true;
- }
- else {
- node.textContent = trimmedText;
- }
- break;
- }
- case node.ELEMENT_NODE:
- toBeRemoved = pruneElement(node);
- }
- if (toBeRemoved) {
- return true;
- }
- const childrenToRemove = [];
- let child = node.firstChild;
- while (child) {
- pruneHtmlNode(child, pruneElement) && childrenToRemove.push(child);
- child = child.nextSibling;
- }
- childrenToRemove.forEach((child) => node.removeChild(child));
- return false;
-};
exports.default = {
normalizeVendorHtml,
getMimeHtmlAndPlainParts,
diff --git a/dist/services/MimeVerificationService/normalizationStrategies.d.ts b/dist/services/MimeVerificationService/normalizationStrategies.d.ts
index 08113b4cfd63b1af06fde0fd13171e9dee649167..89e97ce976d3d267d0cf1428b1d154b8eaf5654c 100644
--- a/dist/services/MimeVerificationService/normalizationStrategies.d.ts
+++ b/dist/services/MimeVerificationService/normalizationStrategies.d.ts
@@ -1,6 +1,6 @@
-export declare const processGmailElement: (element: HTMLElement) => boolean;
+export declare const pruneGmailElement: (element: HTMLElement) => boolean;
/**
* Returns true if element should be completely removed
* @param element
*/
-export declare const processOutlookElement: (element: HTMLElement) => boolean;
+export declare const pruneOutlookElement: (element: HTMLElement) => boolean;
diff --git a/dist/services/MimeVerificationService/normalizationStrategies.js b/dist/services/MimeVerificationService/normalizationStrategies.js
index aa40890a20de01aebf836b47683de7c06b27dec4..4345de823fb75d7fbf00c05ca0dd03597a2f19b5 100644
--- a/dist/services/MimeVerificationService/normalizationStrategies.js
+++ b/dist/services/MimeVerificationService/normalizationStrategies.js
@@ -1,6 +1,6 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
-exports.processOutlookElement = exports.processGmailElement = void 0;
+exports.pruneOutlookElement = exports.pruneGmailElement = void 0;
const DUMMY_QR_CODE_ID = "dummyQrCode";
const deleteAttributesWithPrefix = (element, prefix) => {
for (const attribute of element.attributes) {
@@ -17,7 +17,7 @@ const isDummyQrCode = (element) => {
return true;
}
};
-exports.processGmailElement = (element) => {
+exports.pruneGmailElement = (element) => {
if (isDummyQrCode(element)) {
return true;
}
@@ -35,10 +35,14 @@ exports.processGmailElement = (element) => {
* Returns true if element should be completely removed
* @param element
*/
-exports.processOutlookElement = (element) => {
+exports.pruneOutlookElement = (element) => {
if (isDummyQrCode(element)) {
return true;
}
+ // Remove Outlook generic <o:p> tags
+ if (element.nodeName.toLowerCase().startsWith("o:")) {
+ return true;
+ }
if (element.attributes.length > 0) {
deleteDataAttributes(element);
// TODO Prepare for demo. fix later. the value at style attribute is changed (additional spaces and semicolon)
diff --git a/src/services/MimeVerificationService/MimeVerificationService.ts b/src/services/MimeVerificationService/MimeVerificationService.ts
index 310a8f07ab3eaf9d366c6c5c0cd1255f564adfab..913e4bf36a5239c575d3ee27b96322a8e83b8db0 100644
--- a/src/services/MimeVerificationService/MimeVerificationService.ts
+++ b/src/services/MimeVerificationService/MimeVerificationService.ts
@@ -1,23 +1,24 @@
import {
- processGmailElement,
- processOutlookElement,
+ pruneGmailElement,
+ pruneOutlookElement,
} from "./normalizationStrategies";
import MailParser, { fixNewLines } from "../MailParser/MailParser";
import CryptoService from "../CryptoService";
-import { arrayBufferToBase64, escapeHtmlString } from "../../utils/common";
+import { arrayBufferToBase64 } from "../../utils/common";
import { EMAIL_VENDORS } from "../StatusesService";
import MimeVerificationService from "./index";
import { RollingHash } from "../../index";
import VerificationError from "../VerificationService/VerificationError";
import { removeSpacesAndLinebreaks } from "../../utils/stringUtils";
+import { printHtmlChildren, printOutlookElement, pruneHtmlNode } from "./utils";
-const ELEMENT_NODE = 1;
-const TEXT_NODE = 3;
-const DOCUMENT_NODE = 9;
+const vendorPruningFunctions = {
+ [EMAIL_VENDORS.GMAIL]: pruneGmailElement,
+ [EMAIL_VENDORS.OUTLOOK]: pruneOutlookElement,
+};
-const vendorProcessingFunctions = {
- [EMAIL_VENDORS.GMAIL]: processGmailElement,
- [EMAIL_VENDORS.OUTLOOK]: processOutlookElement,
+const vendorPrintingFunctions = {
+ [EMAIL_VENDORS.OUTLOOK]: printOutlookElement,
};
// Load JSDOM dynamically for Node env only, because built CRA is crashing with it
@@ -167,121 +168,25 @@ const normalizeVendorHtml = (htmlString: string, vendor: string): string => {
const mimeBody = document.firstChild.lastChild;
- const elementProcessingFunction = vendorProcessingFunctions[vendor];
+ const elementPruningFunction = vendorPruningFunctions[vendor];
- if (!elementProcessingFunction) {
+ if (!elementPruningFunction) {
throw new Error(
`Vendor "${vendor}" is not supported. Please, develop a normalization strategy for it.`
);
}
- pruneHtmlNode(document, elementProcessingFunction);
+ pruneHtmlNode(document, elementPruningFunction);
+
+ const vendorPrintFunction = vendorPrintingFunctions[vendor];
- return printHtmlChildren(mimeBody, 0);
+ return printHtmlChildren(mimeBody, vendorPrintFunction, 0);
};
const normalizePlainPart = (text: string): string => {
return removeSpacesAndLinebreaks(text);
};
-const printHtmlChildren = (node: Node, depth: number): string => {
- let child = node.firstChild;
- if (!child) {
- return "";
- }
-
- if (child == node.lastChild && child.nodeType == TEXT_NODE) {
- return printHtmlNode(child, depth);
- } else {
- let result = "";
- while (child) {
- result = result.concat(printHtmlNode(child, depth));
- child = child.nextSibling;
- }
-
- return result;
- }
-};
-
-const printHtmlNode = (node: Node, depth: number): string => {
- let result = "";
-
- switch (node.nodeType) {
- case TEXT_NODE:
- result += "<TEXT>";
- result += removeSpacesAndLinebreaks(node.textContent);
- result += "</TEXT>";
- result += "\n";
- break;
- case DOCUMENT_NODE:
- result += printHtmlChildren(node, depth);
- break;
- case ELEMENT_NODE:
- result += "<" + node.nodeName;
- Array.from((node as HTMLElement).attributes)
- .sort((a, b) => a.name.localeCompare(b.name))
- .forEach((attribute) => {
- result += ` ${attribute.name}`;
- if (attribute.value) {
- result += `="${escapeHtmlString(attribute.value)}"`;
- }
- });
-
- if (node.firstChild) {
- result += ">";
- result += "\n";
- result += printHtmlChildren(node, depth + 1);
- result += "</" + node.nodeName + ">";
- } else {
- result += "/>";
- }
- result += "\n";
- }
-
- return result;
-};
-
-const pruneHtmlNode = (
- node: Node,
- pruneElement: (element: HTMLElement) => boolean
-): boolean => {
- let toBeRemoved = false;
-
- switch (node.nodeType) {
- case node.COMMENT_NODE:
- case node.DOCUMENT_TYPE_NODE:
- toBeRemoved = true;
- break;
- case node.TEXT_NODE: {
- const trimmedText = node.textContent.trim();
- if (trimmedText === "") {
- toBeRemoved = true;
- } else {
- node.textContent = trimmedText;
- }
- break;
- }
- case node.ELEMENT_NODE:
- toBeRemoved = pruneElement(node as HTMLElement);
- }
-
- if (toBeRemoved) {
- return true;
- }
-
- const childrenToRemove = [];
- let child = node.firstChild;
-
- while (child) {
- pruneHtmlNode(child, pruneElement) && childrenToRemove.push(child);
- child = child.nextSibling;
- }
-
- childrenToRemove.forEach((child) => node.removeChild(child));
-
- return false;
-};
-
export default {
normalizeVendorHtml,
getMimeHtmlAndPlainParts,
diff --git a/src/services/MimeVerificationService/constants.ts b/src/services/MimeVerificationService/constants.ts
new file mode 100644
index 0000000000000000000000000000000000000000..be3c40ceffbc88422101e57db3345596895918b0
--- /dev/null
+++ b/src/services/MimeVerificationService/constants.ts
@@ -0,0 +1,3 @@
+export const ELEMENT_NODE = 1;
+export const TEXT_NODE = 3;
+export const DOCUMENT_NODE = 9;
diff --git a/src/services/MimeVerificationService/normalizationStrategies.ts b/src/services/MimeVerificationService/normalizationStrategies.ts
index 0413209fc865cca0dd135e9256f8b3834a0aef31..bc32690f016749ac0ec5d31a4687d557735e1af5 100644
--- a/src/services/MimeVerificationService/normalizationStrategies.ts
+++ b/src/services/MimeVerificationService/normalizationStrategies.ts
@@ -18,7 +18,7 @@ const isDummyQrCode = (element: HTMLElement): boolean => {
}
};
-export const processGmailElement = (element: HTMLElement): boolean => {
+export const pruneGmailElement = (element: HTMLElement): boolean => {
if (isDummyQrCode(element)) {
return true;
}
@@ -40,11 +40,16 @@ export const processGmailElement = (element: HTMLElement): boolean => {
* Returns true if element should be completely removed
* @param element
*/
-export const processOutlookElement = (element: HTMLElement): boolean => {
+export const pruneOutlookElement = (element: HTMLElement): boolean => {
if (isDummyQrCode(element)) {
return true;
}
+ // Remove Outlook generic <o:p> tags
+ if (element.nodeName.toLowerCase().startsWith("o:")) {
+ return true;
+ }
+
if (element.attributes.length > 0) {
deleteDataAttributes(element);
diff --git a/src/services/MimeVerificationService/utils.ts b/src/services/MimeVerificationService/utils.ts
new file mode 100644
index 0000000000000000000000000000000000000000..f4bc6726ba4c59dfde0777875dd8495cde7ea038
--- /dev/null
+++ b/src/services/MimeVerificationService/utils.ts
@@ -0,0 +1,124 @@
+import { removeSpacesAndLinebreaks } from "../../utils/stringUtils";
+import { escapeHtmlString } from "../..";
+import { DOCUMENT_NODE, ELEMENT_NODE, TEXT_NODE } from "./constants";
+
+export const printHtmlChildren = (
+ node: Node,
+ printFunction: (node: Node) => string,
+ depth: number
+): string => {
+ let child = node.firstChild;
+ if (!child) {
+ return "";
+ }
+
+ if (child == node.lastChild && child.nodeType == TEXT_NODE) {
+ return printHtmlNode(child, printFunction, depth);
+ } else {
+ let result = "";
+ while (child) {
+ result = result.concat(printHtmlNode(child, printFunction, depth));
+ child = child.nextSibling;
+ }
+
+ return result;
+ }
+};
+
+export const printHtmlNode = (
+ node: Node,
+ printFunction: (node: Node) => string,
+ depth: number
+): string => {
+ let result = "";
+
+ if (printFunction) {
+ result = printFunction(node);
+ if (result) {
+ return result;
+ }
+ }
+
+ switch (node.nodeType) {
+ case TEXT_NODE:
+ result += "<TEXT>";
+ result += removeSpacesAndLinebreaks(node.textContent);
+ result += "</TEXT>";
+ result += "\n";
+ break;
+ case DOCUMENT_NODE:
+ result += printHtmlChildren(node, printFunction, depth);
+ break;
+ case ELEMENT_NODE:
+ result += "<" + node.nodeName;
+ Array.from((node as HTMLElement).attributes)
+ .sort((a, b) => a.name.localeCompare(b.name))
+ .forEach((attribute) => {
+ result += ` ${attribute.name}`;
+ if (attribute.value) {
+ result += `="${escapeHtmlString(attribute.value)}"`;
+ }
+ });
+
+ if (node.firstChild) {
+ result += ">";
+ result += "\n";
+ result += printHtmlChildren(node, printFunction, depth + 1);
+ result += "</" + node.nodeName + ">";
+ } else {
+ result += "/>";
+ }
+ result += "\n";
+ }
+
+ return result;
+};
+
+export const pruneHtmlNode = (
+ node: Node,
+ pruneElement: (element: HTMLElement) => boolean
+): boolean => {
+ let toBeRemoved = false;
+
+ switch (node.nodeType) {
+ case node.COMMENT_NODE:
+ case node.DOCUMENT_TYPE_NODE:
+ toBeRemoved = true;
+ break;
+ case node.TEXT_NODE: {
+ const trimmedText = node.textContent.trim();
+ if (trimmedText === "") {
+ toBeRemoved = true;
+ } else {
+ node.textContent = trimmedText;
+ }
+ break;
+ }
+ case node.ELEMENT_NODE:
+ toBeRemoved = pruneElement(node as HTMLElement);
+ }
+
+ if (toBeRemoved) {
+ return true;
+ }
+
+ const childrenToRemove = [];
+ let child = node.firstChild;
+
+ while (child) {
+ pruneHtmlNode(child, pruneElement) && childrenToRemove.push(child);
+ child = child.nextSibling;
+ }
+
+ childrenToRemove.forEach((child) => node.removeChild(child));
+
+ return false;
+};
+
+export const printOutlookElement = (node: Node): string => {
+ if (node.nodeType === ELEMENT_NODE) {
+ if ((node as HTMLElement).classList.contains("WordSection1")) {
+ return printHtmlChildren(node, null, 0);
+ }
+ }
+};