From 27cd21f6aec7135a2f8f4d7ee3f84ac419e11a56 Mon Sep 17 00:00:00 2001
From: igor <igor.markin@vereign.com>
Date: Tue, 8 Dec 2020 19:17:12 +0300
Subject: [PATCH] Add build
---
dist/HTMLNormalizer/HTMLNormalizer.d.ts | 1 +
dist/HTMLNormalizer/HTMLNormalizer.js | 24 +++++++++++++++++++++++-
dist/HTMLNormalizer/index.d.ts | 1 +
dist/HTMLNormalizer/index.js | 1 +
dist/PlainNormalizer/PlainNormalizer.js | 4 ++++
dist/utils.js | 5 ++---
6 files changed, 32 insertions(+), 4 deletions(-)
diff --git a/dist/HTMLNormalizer/HTMLNormalizer.d.ts b/dist/HTMLNormalizer/HTMLNormalizer.d.ts
index 1451258..62ff3a5 100644
--- a/dist/HTMLNormalizer/HTMLNormalizer.d.ts
+++ b/dist/HTMLNormalizer/HTMLNormalizer.d.ts
@@ -1,4 +1,5 @@
export declare const normalizeVendorHtml: (document: HTMLDocument, vendor: string) => string;
+export declare const extractPseudoPlainPart: (document: HTMLDocument) => string;
export declare const printHtmlChildren: (node: Node, printFunction: (node: Node) => string, depth: number) => string;
export declare const printHtmlNode: (node: Node, printFunction: (node: Node) => string, depth: number) => string;
export declare const cleanupHtmlNodeAttributes: (node: Node, cleanupElementAttributes: (element: HTMLElement) => void) => void;
diff --git a/dist/HTMLNormalizer/HTMLNormalizer.js b/dist/HTMLNormalizer/HTMLNormalizer.js
index 68dc062..1fdd9d9 100644
--- a/dist/HTMLNormalizer/HTMLNormalizer.js
+++ b/dist/HTMLNormalizer/HTMLNormalizer.js
@@ -1,11 +1,12 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
-exports.escapeHtmlString = exports.pruneHtmlNode = exports.cleanupHtmlNodeAttributes = exports.printHtmlNode = exports.printHtmlChildren = exports.normalizeVendorHtml = void 0;
+exports.escapeHtmlString = exports.pruneHtmlNode = exports.cleanupHtmlNodeAttributes = exports.printHtmlNode = exports.printHtmlChildren = exports.extractPseudoPlainPart = exports.normalizeVendorHtml = void 0;
const constants_1 = require("../constants");
const outlook_1 = require("./strategies/outlook");
const constants_2 = require("../constants");
const utils_1 = require("../utils");
const gmail_1 = require("./strategies/gmail");
+const index_1 = require("../index");
const nodesAmendingFunctions = {
[constants_2.EMAIL_VENDORS.GMAIL]: gmail_1.amendGmailNodes,
[constants_2.EMAIL_VENDORS.OUTLOOK]: outlook_1.amendOutlookNodes,
@@ -49,6 +50,27 @@ const normalizeVendorHtml = (document, vendor) => {
return exports.printHtmlChildren(mimeBody, vendorPrintFunction, 0);
};
exports.normalizeVendorHtml = normalizeVendorHtml;
+const extractPseudoPlainPart = (document
+/*vendor: string*/
+) => {
+ const textContent = index_1.PlainNormalizer.normalizePlain(document.body.textContent);
+ // const anchors = document.getElementsByTagName("a");
+ // const images = document.getElementsByTagName("img");
+ // let meaningfulAttributes = [];
+ //
+ // Array.from(anchors).forEach((a) => {
+ // meaningfulAttributes.push(a.getAttribute("href"));
+ // });
+ // Array.from(images).forEach((img) => {
+ // meaningfulAttributes.push(img.getAttribute("src"));
+ // meaningfulAttributes.push(img.getAttribute("alt"));
+ // });
+ //
+ // meaningfulAttributes = meaningfulAttributes.filter((attr) => !!attr).sort();
+ // console.log(meaningfulAttributes);
+ return textContent;
+};
+exports.extractPseudoPlainPart = extractPseudoPlainPart;
const printHtmlChildren = (node, printFunction, depth) => {
let child = node.firstChild;
if (!child) {
diff --git a/dist/HTMLNormalizer/index.d.ts b/dist/HTMLNormalizer/index.d.ts
index 41ae932..7bea28c 100644
--- a/dist/HTMLNormalizer/index.d.ts
+++ b/dist/HTMLNormalizer/index.d.ts
@@ -1,4 +1,5 @@
declare const _default: {
normalizeVendorHtml: (document: HTMLDocument, vendor: string) => string;
+ extractPseudoPlainPart: (document: HTMLDocument) => string;
};
export default _default;
diff --git a/dist/HTMLNormalizer/index.js b/dist/HTMLNormalizer/index.js
index 1034775..3d76acb 100644
--- a/dist/HTMLNormalizer/index.js
+++ b/dist/HTMLNormalizer/index.js
@@ -3,4 +3,5 @@ Object.defineProperty(exports, "__esModule", { value: true });
const HTMLNormalizer_1 = require("./HTMLNormalizer");
exports.default = {
normalizeVendorHtml: HTMLNormalizer_1.normalizeVendorHtml,
+ extractPseudoPlainPart: HTMLNormalizer_1.extractPseudoPlainPart,
};
diff --git a/dist/PlainNormalizer/PlainNormalizer.js b/dist/PlainNormalizer/PlainNormalizer.js
index e79c6a0..342abdd 100644
--- a/dist/PlainNormalizer/PlainNormalizer.js
+++ b/dist/PlainNormalizer/PlainNormalizer.js
@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
exports.normalizePlainPart = void 0;
const utils_1 = require("../utils");
const normalizePlainPart = (text) => {
+ text = removeListBullets(text);
text = utils_1.removeSpacesAndLinebreaks(text);
return removeQRCodes(text);
};
@@ -12,3 +13,6 @@ const removeQRCodes = (s) => {
.replace(/\[qrcode.png]\s*<https:\/\/[\w./?=\-&]+>/g, "")
.replace(/<https:\/\/[\w./?=\-&]+>\s*\[qrcode.png]/g, "");
};
+const removeListBullets = (s) => {
+ return s.replace("\n[o§]\n+/g", "");
+};
diff --git a/dist/utils.js b/dist/utils.js
index 6a64db9..3030ed2 100644
--- a/dist/utils.js
+++ b/dist/utils.js
@@ -2,8 +2,7 @@
Object.defineProperty(exports, "__esModule", { value: true });
exports.removeSpacesAndLinebreaks = void 0;
const removeSpacesAndLinebreaks = (s) => {
- const regexNewlines = new RegExp(/[\r\n\v]+/g);
- const regexSpaces = new RegExp(/\s+|\u200B/g);
- return s.replace(regexNewlines, "").replace(regexSpaces, "");
+ const removeSymbols = new RegExp(/[\r\n\v\s\u200B]+/g);
+ return s.replace(removeSymbols, "").trim();
};
exports.removeSpacesAndLinebreaks = removeSpacesAndLinebreaks;
--
GitLab