email_gateway: Use html2text directly.

In the rare case that Zulip receives an email with only an HTML
format, we originally (code dating to 2013) shelled out to
html2markdown/python-html2text in order to convert the HTML into
markdown.

We long since added html2text as a reasonably managed Python
dependency of Zulip; we should just use it here.

zerver/lib/email_notifications.py

@@ -27,10 +27,10 @@ from zerver.models import (
 
 from datetime import timedelta
 from email.utils import formataddr
+import html2text
 from lxml.cssselect import CSSSelector
 import lxml.html
 import re
-import subprocess
 from collections import defaultdict
 import pytz
 from bs4 import BeautifulSoup
@@ -580,22 +580,9 @@ def enqueue_welcome_emails(user: UserProfile, realm_creation: bool=False) -> Non
             from_address=from_address, context=context, delay=followup_day2_email_delay(user))
 
 def convert_html_to_markdown(html: str) -> str:
-    # On Linux, the tool installs as html2markdown, and there's a command called
+    parser = html2text.HTML2Text()
-    # html2text that does something totally different. On OSX, the tool installs
+    markdown = parser.handle(html).strip()
-    # as html2text.
-    commands = ["html2markdown", "html2text"]
 
-    for command in commands:
-        try:
-            # A body width of 0 means do not try to wrap the text for us.
-            p = subprocess.Popen(
-                [command, "--body-width=0"], stdout=subprocess.PIPE,
-                stdin=subprocess.PIPE, stderr=subprocess.STDOUT)
-            break
-        except OSError:
-            continue
-
-    markdown = p.communicate(input=html.encode('utf-8'))[0].decode('utf-8').strip()
     # We want images to get linked and inline previewed, but html2text will turn
     # them into links of the form `![](http://foo.com/image.png)`, which is
     # ugly. Run a regex over the resulting description, turning links of the

zerver/webhooks/freshdesk/tests.py

@@ -114,7 +114,7 @@ Requester Bob <requester-bob@example.com> added a {} note to \
         """
         expected_topic = u"#12: Not enough ☃ guinea pigs"
         expected_message = """
-Requester \u2603 Bob <requester-bob@example.com> created [ticket #12](http://test1234zzz.freshdesk.com/helpdesk/tickets/12):\n\n``` quote\nThere are too many cat pictures on the internet \u2603. We need more guinea pigs. Exhibit 1:\n\n  \n\n\n[guinea_pig.png](http://cdn.freshdesk.com/data/helpdesk/attachments/production/12744808/original/guinea_pig.png)\n```\n\n* **Type**: Problem\n* **Priority**: Urgent\n* **Status**: Open
+Requester \u2603 Bob <requester-bob@example.com> created [ticket #12](http://test1234zzz.freshdesk.com/helpdesk/tickets/12):\n\n``` quote\nThere are too many cat pictures on the internet \u2603. We need more guinea pigs.\nExhibit 1:\n\n  \n\n[guinea_pig.png](http://cdn.freshdesk.com/data/helpdesk/attachments/production/12744808/original/guinea_pig.png)\n```\n\n* **Type**: Problem\n* **Priority**: Urgent\n* **Status**: Open
 """.strip()
         self.api_stream_message(self.TEST_USER_EMAIL, "inline_images", expected_topic, expected_message,
                                 content_type="application/x-www-form-urlencoded")