there could be still long lines inside html tags (#204)

* there could be still long lines inside html tags

so now it will reformat those before printing html

CPNHUB-132

newsroom/email.py

@@ -32,7 +32,7 @@ class EmailGroup(TypedDict):
     emails: List[str]
 
 
-MAX_LINE_LENGTH = 998  # RFC 5322
+MAX_LINE_LENGTH = 998 - 50  # RFC 5322 - buffer for html indentation
 
 
 def handle_long_lines_text(text):
@@ -47,7 +47,7 @@ def handle_long_lines_text(text):
             next_line = ''
             words = line.split()
             for word in words:
-                if len(next_line) + len(word) + 1 > MAX_LINE_LENGTH:
+                if len(next_line) + len(word) > MAX_LINE_LENGTH:
                     output.append(next_line)
                     next_line = word
                 else:
@@ -63,7 +63,13 @@ def handle_long_lines_html(html):
     if not any([len(line) > MAX_LINE_LENGTH for line in lines]):
         return html
     parsed = etree.fromstring(html, parser=etree.HTMLParser())
-    return etree.tounicode(parsed, method="html", pretty_print=True)
+    etree.indent(parsed, space=" ")  # like pretty print but upfront
+    for elem in parsed.iter():
+        if elem.text is not None and len(elem.text) > MAX_LINE_LENGTH:
+            elem.text = handle_long_lines_text(elem.text) + "\n"
+        if elem.tail is not None and len(elem.tail) > MAX_LINE_LENGTH:
+            elem.tail = handle_long_lines_text(elem.tail) + "\n"
+    return etree.tostring(parsed, method="html", encoding="unicode")
 
 
 @celery.task(soft_time_limit=120)

tests/core/test_emails.py

@@ -1,7 +1,13 @@
 from flask import render_template_string, json, url_for
 from jinja2 import TemplateNotFound
 
-from newsroom.email import send_new_item_notification_email, map_email_recipients_by_language, EmailGroup, send_email
+from newsroom.email import (
+    send_new_item_notification_email,
+    map_email_recipients_by_language,
+    EmailGroup,
+    send_email,
+    handle_long_lines_html,
+)
 from unittest import mock
 
 
@@ -164,6 +170,13 @@ def test_email_avoid_long_lines(client, app, mocker):
     assert 501 == len(lines[2])
 
 
+def test_handle_long_lines_html():
+    html = "<div><p>{}</p></div>".format("foo bar <a href=\"test\">{}</a>baz".format("loong link" * 1000) * 50)
+    formatted = handle_long_lines_html(html)
+    for line in formatted.splitlines():
+        assert len(line) < 998, line
+
+
 def check_lines_length(text, length=998):
     lines = text.splitlines()
     for line in lines: