From 70306d579c433c7dabf1f808839f1983adf4dfc0 Mon Sep 17 00:00:00 2001
From: igor <igor.markin@vereign.com>
Date: Thu, 24 Dec 2020 15:24:44 +0300
Subject: [PATCH] Add links to pseudoplain
---
.../outlook-outlook/macos-macos/README.md | 2 +-
__tests__/htmltext-outlook-outlook.test.ts | 12 +++++--
src/HTMLNormalizer/HTMLNormalizer.ts | 31 ++++++++++---------
3 files changed, 27 insertions(+), 18 deletions(-)
diff --git a/__tests__/files/outlook-outlook/macos-macos/README.md b/__tests__/files/outlook-outlook/macos-macos/README.md
index 13dce39..be46aa7 100644
--- a/__tests__/files/outlook-outlook/macos-macos/README.md
+++ b/__tests__/files/outlook-outlook/macos-macos/README.md
@@ -4,7 +4,7 @@
| 02 | Simple text | ok | ok | ok |
| 03 | Embedded image | fail | ok | fail |
| 04 | Embedded image with text | fail | ok | fail |
-| 05 | Complex text on multiple lines with some text bold, italic, lists, no attachments | fail | ok | ok |
+| 05 | Complex text on multiple lines with some text bold, italic, lists, no attachments | fail | ok | fail |
| 06 | Multiple MIME attachments and text | ok | ok | ok |
| 07 | Multiple Drive attachments and text | ok | ok | ok |
| 08 | Complex email with formatted text, embedded images, MIME and GDrive attachments | fail | ok | fail |
diff --git a/__tests__/htmltext-outlook-outlook.test.ts b/__tests__/htmltext-outlook-outlook.test.ts
index cd8385a..2e96f56 100644
--- a/__tests__/htmltext-outlook-outlook.test.ts
+++ b/__tests__/htmltext-outlook-outlook.test.ts
@@ -13,7 +13,13 @@ describe("[Pseudo PLAIN] Outlook-Outlook normalization", () => {
EMAIL_VENDORS.OUTLOOK
);
- describe("Chrome-Chrome", describeFunction("chrome-chrome"));
- describe("MacOS-MacOS", describeFunction("macos-macos", ["09"]));
- describe("Windows-Windows", describeFunction("windows-windows"));
+ describe("Chrome-Chrome", describeFunction("chrome-chrome", ["16"]));
+ describe(
+ "MacOS-MacOS",
+ describeFunction("macos-macos", ["03", "04", "05", "08", "09"])
+ );
+ describe(
+ "Windows-Windows",
+ describeFunction("windows-windows", ["11", "12", "16"])
+ );
});
diff --git a/src/HTMLNormalizer/HTMLNormalizer.ts b/src/HTMLNormalizer/HTMLNormalizer.ts
index ce2f1d3..601e7ee 100644
--- a/src/HTMLNormalizer/HTMLNormalizer.ts
+++ b/src/HTMLNormalizer/HTMLNormalizer.ts
@@ -87,23 +87,26 @@ export const extractPseudoPlainPart = (
document: HTMLDocument
/*vendor: string*/
): string => {
- const normalizedTextContent = PlainNormalizer.normalizePlain(
+ let normalizedTextContent = 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();
+ 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();
+
+ normalizedTextContent += meaningfulAttributes.join(",");
+
// console.log(meaningfulAttributes);
return normalizedTextContent;
};
--
GitLab