Merge pull request #90 from mailgun/sergey/89

fixes mailgun/talon#89
2016-05-17 16:01:56 -07:00 · 2016-05-17 15:31:01 -07:00 · 2016-04-07 17:54:12 -07:00 · 2016-04-07 17:51:48 -07:00 · 2016-04-07 17:47:38 -07:00 · 2016-04-04 14:22:50 -07:00
3 changed files with 49 additions and 4 deletions
--- a/setup.py
+++ b/setup.py
@@ -2,7 +2,7 @@ from setuptools import setup, find_packages


 setup(name='talon',
-      version='1.2.5',
+      version='1.2.7',
      description=("Mailgun library "
                   "to extract message quotations and signatures."),
      long_description=open("README.rst").read(),
--- a/talon/html_quotations.py
+++ b/talon/html_quotations.py
@@ -78,7 +78,7 @@ def delete_quotation_tags(html_note, counter, quotation_checkpoints):
 def cut_gmail_quote(html_message):
    ''' Cuts the outermost block element with class gmail_quote. '''
    gmail_quote = html_message.cssselect('div.gmail_quote')
-    if gmail_quote and not RE_FWD.match(gmail_quote[0].text):
+    if gmail_quote and (gmail_quote[0].text is None or not RE_FWD.match(gmail_quote[0].text)):
        gmail_quote[0].getparent().remove(gmail_quote[0])
        return True

@@ -175,7 +175,21 @@ def cut_from_block(html_message):
                len(maybe_body.getchildren()) == 1)

            if not parent_div_is_all_content:
-                block.getparent().remove(block)
+                parent = block.getparent()
+                next_sibling = block.getnext()
+
+                # remove all tags after found From block
+                # (From block and quoted message are in separate divs)
+                while next_sibling is not None:
+                    parent.remove(block)
+                    block = next_sibling
+                    next_sibling = block.getnext()
+
+                # remove the last sibling (or the
+                # From block if no siblings)
+                if block is not None:
+                    parent.remove(block)
+
                return True
        else:
            return False
@@ -200,4 +214,4 @@ def cut_zimbra_quote(html_message):
    zDivider = html_message.xpath('//hr[@data-marker="__DIVIDER__"]')
    if zDivider:
        zDivider[0].getparent().remove(zDivider[0])
-        return True
+        return True
--- a/tests/html_quotations_test.py
+++ b/tests/html_quotations_test.py
@@ -131,6 +131,17 @@ def test_gmail_quote():
        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))


+def test_gmail_quote_compact():
+    msg_body = 'Reply' \
+               '<div class="gmail_quote">' \
+               '<div class="gmail_quote">On 11-Apr-2011, at 6:54 PM, Bob &lt;bob@example.com&gt; wrote:' \
+               '<div>Test</div>' \
+               '</div>' \
+               '</div>'
+    eq_("<html><body><p>Reply</p></body></html>",
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
+
+
 def test_gmail_quote_blockquote():
    msg_body = """Message
 <blockquote class="gmail_quote">
@@ -268,6 +279,26 @@ def test_reply_separated_by_hr():
            '', quotations.extract_from_html(REPLY_SEPARATED_BY_HR)))


+def test_from_block_and_quotations_in_separate_divs():
+    msg_body = '''
+Reply
+<div>
+  <hr/>
+  <div>
+    <font>
+      <b>From: bob@example.com</b>
+      <b>Date: Thu, 24 Mar 2016 08:07:12 -0700</b>
+    </font>
+  </div>
+  <div>
+    Quoted message
+  </div>
+</div>
+'''
+    eq_('<html><body><p>Reply</p><div><hr></div></body></html>',
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
+
+
 def extract_reply_and_check(filename):
    f = open(filename)
Author	SHA1	Message	Date
Sergey Obukhov	9dbe6a494b	Merge pull request #90 from mailgun/sergey/89 fixes mailgun/talon#89	2016-05-17 16:01:56 -07:00
Sergey Obukhov	44e70939d6	fixes mailgun/talon#89	2016-05-17 15:31:01 -07:00
Sergey Obukhov	ab6066eafa	Merge pull request #87 from mailgun/sergey/1.2.6 bump up version	2016-04-07 17:54:12 -07:00
Sergey Obukhov	42258cdd36	bump up version	2016-04-07 17:51:48 -07:00
Sergey Obukhov	d3de9e6893	Merge pull request #86 from dougkeen/master Fix #85 (exception when stripping gmail quotes)	2016-04-07 17:47:38 -07:00
Doug Keen	333beb94af	Fix #85 (exception when stripping gmail quotes)	2016-04-04 14:22:50 -07:00