Merge pull request #223 from mailgun/maxim/develop

PIP-1509: Optimise sender name check [python3]

.build/Dockerfile

@@ -0,0 +1,20 @@
+FROM python:3.9-slim-buster AS deps
+
+RUN apt-get update && \
+    apt-get install -y build-essential git curl python3-dev libatlas3-base libatlas-base-dev liblapack-dev libxml2 libxml2-dev libffi6 libffi-dev musl-dev libxslt-dev
+
+FROM deps AS testable
+ARG REPORT_PATH
+
+VOLUME ["/var/mailgun", "/etc/mailgun/ssl", ${REPORT_PATH}]
+
+ADD . /app
+WORKDIR /app
+COPY wheel/* /wheel/
+
+RUN mkdir -p ${REPORT_PATH}
+
+RUN python ./setup.py build bdist_wheel -d /wheel && \
+    pip install --no-deps /wheel/*
+
+ENTRYPOINT ["/bin/sh", "/app/run_tests.sh"]

.gitignore

@@ -54,3 +54,6 @@ _trial_temp
 
 # OSX
 .DS_Store
+
+# vim-backup
+*.bak

requirements.txt

@@ -0,0 +1,11 @@
+chardet>=1.0.1
+cchardet>=0.3.5
+cssselect
+html5lib
+joblib
+lxml>=2.3.3
+numpy
+regex>=1
+scikit-learn>=1.0.0
+scipy
+six>=1.10.0

run_tests.sh

@@ -0,0 +1,4 @@
+#!/usr/bin/env bash
+set -ex
+REPORT_PATH="${REPORT_PATH:-./}"
+nosetests --with-xunit --with-coverage --cover-xml --cover-xml-file $REPORT_PATH/coverage.xml --xunit-file=$REPORT_PATH/nosetests.xml --cover-package=talon .

setup.py

@@ -19,17 +19,17 @@ class InstallCommand(install):
         if self.no_ml:
             dist = self.distribution
             dist.packages=find_packages(exclude=[
-                'tests',
+                "tests",
-                'tests.*',
+                "tests.*",
-                'talon.signature',
+                "talon.signature",
-                'talon.signature.*',
+                "talon.signature.*",
             ])
-            for not_required in ['numpy', 'scipy', 'scikit-learn==0.16.1']:
+            for not_required in ["numpy", "scipy", "scikit-learn==0.24.1"]:
                 dist.install_requires.remove(not_required)
 
 
 setup(name='talon',
-      version='1.4.8',
+      version='1.4.10',
       description=("Mailgun library "
                    "to extract message quotations and signatures."),
       long_description=open("README.rst").read(),
@@ -44,20 +44,21 @@ setup(name='talon',
       include_package_data=True,
       zip_safe=True,
       install_requires=[
-          "lxml>=2.3.3",
+          "lxml",
-          "regex>=1",
+          "regex",
           "numpy",
           "scipy",
-          "scikit-learn==0.16.1", # pickled versions of classifier, else rebuild
+          "scikit-learn>=1.0.0",
-          'chardet>=1.0.1',
+          "chardet",
-          'cchardet>=0.3.5',
+          "cchardet",
-          'cssselect',
+          "cssselect",
-          'six>=1.10.0',
+          "six",
-          'html5lib'
+          "html5lib",
+          "joblib",
           ],
       tests_require=[
           "mock",
-          "nose>=1.2.1",
+          "nose",
           "coverage"
           ]
       )

talon/quotations.py

@@ -457,26 +457,20 @@ def _extract_from_html(msg_body):
 
     msg_body = msg_body.replace(b'\r\n', b'\n')
 
-    msg_body = re.sub(r"\<\?xml.+\?\>|\<\!DOCTYPE.+]\>", "", msg_body)
+    msg_body = re.sub(br"\<\?xml.+\?\>|\<\!DOCTYPE.+]\>", "", msg_body)
 
     html_tree = html_document_fromstring(msg_body)
 
     if html_tree is None:
         return msg_body
 
-    cut_quotations = False
+    cut_quotations = (html_quotations.cut_gmail_quote(html_tree) or
-    try:
+                      html_quotations.cut_zimbra_quote(html_tree) or
-        cut_quotations = (html_quotations.cut_gmail_quote(html_tree) or
+                      html_quotations.cut_blockquote(html_tree) or
-                          html_quotations.cut_zimbra_quote(html_tree) or
+                      html_quotations.cut_microsoft_quote(html_tree) or
-                          html_quotations.cut_blockquote(html_tree) or
+                      html_quotations.cut_by_id(html_tree) or
-                          html_quotations.cut_microsoft_quote(html_tree) or
+                      html_quotations.cut_from_block(html_tree)
-                          html_quotations.cut_by_id(html_tree) or
+                      )
-                          html_quotations.cut_from_block(html_tree)
-                          )
-    except Exception as e:
-        log.exception('during html quotations cut')
-        pass
-
     html_tree_copy = deepcopy(html_tree)
 
     number_of_checkpoints = html_quotations.add_checkpoint(html_tree, 0)

talon/signature/__init__.py

@@ -23,17 +23,14 @@ trained against, don't forget to regenerate:
 from __future__ import absolute_import
 import os
 
-from . import extraction
+from talon.signature import extraction
-from . extraction import extract  #noqa
+from talon.signature.extraction import extract
-from . learning import classifier
+from talon.signature.learning import classifier
-
-
-DATA_DIR = os.path.join(os.path.dirname(__file__), 'data')
-
-EXTRACTOR_FILENAME = os.path.join(DATA_DIR, 'classifier')
-EXTRACTOR_DATA = os.path.join(DATA_DIR, 'train.data')
 
 
 def initialize():
-    extraction.EXTRACTOR = classifier.load(EXTRACTOR_FILENAME,
+    data_dir = os.path.join(os.path.dirname(__file__), 'data')
-                                           EXTRACTOR_DATA)
+    extractor_filename = os.path.join(data_dir, 'classifier')
+    extractor_data_filename = os.path.join(data_dir, 'train.data')
+    extraction.EXTRACTOR = classifier.load(extractor_filename,
+                                           extractor_data_filename)

talon/signature/bruteforce.py

@@ -62,7 +62,7 @@ RE_SIGNATURE_CANDIDATE = re.compile(r'''
 
 
 def extract_signature(msg_body):
-    """
+    '''
     Analyzes message for a presence of signature block (by common patterns)
     and returns tuple with two elements: message text without signature block
     and the signature itself.
@@ -72,7 +72,7 @@ def extract_signature(msg_body):
 
     >>> extract_signature('Hey man!')
     ('Hey man!', None)
-    """
+    '''
     try:
         # identify line delimiter first
         delimiter = get_delimiter(msg_body)

talon/signature/learning/classifier.py

@@ -8,7 +8,7 @@ body belongs to the signature.
 from __future__ import absolute_import
 
 from numpy import genfromtxt
-from sklearn.externals import joblib
+import joblib
 from sklearn.svm import LinearSVC
 
 

talon/signature/learning/helpers.py

@@ -102,7 +102,7 @@ def flatten_list(list_to_flatten):
 
 
 def contains_sender_names(sender):
-    '''Returns a functions to search sender\'s name or it\'s part.
+    """Returns a functions to search sender\'s name or it\'s part.
 
     >>> feature = contains_sender_names("Sergey N.  Obukhov <xxx@example.com>")
     >>> feature("Sergey Obukhov")
@@ -115,7 +115,7 @@ def contains_sender_names(sender):
     1
     >>> contains_sender_names("<serobnic@mail.ru>")("serobnic")
     1
-    '''
+    """
     names = '( |$)|'.join(flatten_list([[e, e.capitalize()]
                                         for e in extract_names(sender)]))
     names = names or sender
@@ -140,10 +140,16 @@ def extract_names(sender):
     sender = "".join([char if char.isalpha() else ' ' for char in sender])
     # Remove too short words and words from "black" list i.e.
     # words like `ru`, `gmail`, `com`, `org`, etc.
-    sender = [word for word in sender.split() if len(word) > 1 and
+    names = list()
-              not word in BAD_SENDER_NAMES]
+    for word in sender.split():
-    # Remove duplicates
+        if len(word) < 2:
-    names = list(set(sender))
+            continue
+        if word in BAD_SENDER_NAMES:
+            continue
+        if word in names:
+            continue
+        names.append(word)
+
     return names
 
 
@@ -208,20 +214,26 @@ def many_capitalized_words(s):
 
 
 def has_signature(body, sender):
-    '''Checks if the body has signature. Returns True or False.'''
+    """Checks if the body has signature. Returns True or False."""
     non_empty = [line for line in body.splitlines() if line.strip()]
     candidate = non_empty[-SIGNATURE_MAX_LINES:]
     upvotes = 0
+    sender_check = contains_sender_names(sender)
     for line in candidate:
         # we check lines for sender's name, phone, email and url,
         # those signature lines don't take more then 27 lines
         if len(line.strip()) > 27:
             continue
-        elif contains_sender_names(sender)(line):
+
+        if sender_check(line):
             return True
-        elif (binary_regex_search(RE_RELAX_PHONE)(line) +
+
-              binary_regex_search(RE_EMAIL)(line) +
+        if (binary_regex_search(RE_RELAX_PHONE)(line) +
-              binary_regex_search(RE_URL)(line) == 1):
+                binary_regex_search(RE_EMAIL)(line) +
+                binary_regex_search(RE_URL)(line) == 1):
             upvotes += 1
+
     if upvotes > 1:
         return True
+
+    return False

test-requirements.txt

@@ -0,0 +1,3 @@
+coverage
+mock
+nose>=1.2.1

tests/__init__.py

@@ -1,4 +1,6 @@
 from __future__ import absolute_import
+from nose.tools import *
+from mock import *
 
 import talon
 

tests/html_quotations_test.py

@@ -2,12 +2,14 @@
 
 from __future__ import absolute_import
 
-from tests.fixtures import REPLY_QUOTATIONS_SHARE_BLOCK, OLK_SRC_BODY_SECTION, REPLY_SEPARATED_BY_HR
+# noinspection PyUnresolvedReferences
-from nose.tools import eq_, ok_, assert_false, assert_true
-from talon import quotations, utils as u
-from mock import Mock, patch
 import re
 
+from talon import quotations, utils as u
+from . import *
+from .fixtures import *
+from lxml import html
+
 RE_WHITESPACE = re.compile("\s")
 RE_DOUBLE_WHITESPACE = re.compile("\s")
 

tests/quotations_test.py

@@ -1,10 +1,10 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
+from . import *
+from . fixtures import *
 
-from mock import Mock, patch
 from talon import quotations
-from nose.tools import eq_
 
 
 @patch.object(quotations, 'extract_from_html')

tests/signature/bruteforce_test.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
-from nose.tools import eq_
+from .. import *
 
 from talon.signature import bruteforce
-from mock import patch, Mock
 
 
 def test_empty_body():

tests/signature/extraction_test.py

@@ -2,14 +2,14 @@
 
 from __future__ import absolute_import
 
+import os
+
+from six.moves import range
+
 from talon.signature import bruteforce, extraction, extract
 from talon.signature import extraction as e
 from talon.signature.learning import dataset
-from nose.tools import eq_
+from .. import *
-from .. import STRIPPED, UNICODE_MSG
-from six.moves import range
-from mock import patch
-import os
 
 
 def test_message_shorter_SIGNATURE_MAX_LINES():

tests/signature/learning/dataset_test.py

@@ -1,14 +1,15 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
-
+from ... import *
-from ... import EML_MSG_FILENAME, MSG_FILENAME_WITH_BODY_SUFFIX, TMP_DIR, EMAILS_DIR
-from talon.signature.learning.featurespace import features
-from talon.signature.learning import dataset as d
-from nose.tools import eq_, assert_false, ok_
-from numpy import genfromtxt
 import os
 
+from numpy import genfromtxt
+
+from talon.signature.learning import dataset as d
+
+from talon.signature.learning.featurespace import features
+
 
 def test_is_sender_filename():
     assert_false(d.is_sender_filename("foo/bar"))

tests/signature/learning/featurespace_test.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
+from ... import *
 
 from talon.signature.learning import featurespace as fs
-from nose.tools import eq_, assert_false, ok_
-from mock import patch
 
 
 def test_apply_features():

tests/signature/learning/helpers_test.py

@@ -1,13 +1,13 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
+from ... import *
+
+import regex as re
 
 from talon.signature.learning import helpers as h
-from talon.signature.learning.helpers import RE_RELAX_PHONE, RE_NAME
+from talon.signature.learning.helpers import *
-from nose.tools import eq_, ok_, assert_false, assert_in
-from mock import patch, Mock
 from six.moves import range
-import re
 
 # First testing regex constants.
 VALID = '''

tests/text_quotations_test.py

@@ -1,16 +1,17 @@
 # -*- coding: utf-8 -*-
 
 from __future__ import absolute_import
+from . import *
+from . fixtures import *
 
-from tests.fixtures import STANDARD_REPLIES
-from talon import quotations
-from six.moves import range
-from nose.tools import eq_
-from mock import patch
-import email.iterators
-import six
 import os
 
+import email.iterators
+from talon import quotations
+import six
+from six.moves import range
+from six import StringIO
+
 
 @patch.object(quotations, 'MAX_LINES_COUNT', 1)
 def test_too_many_lines():
@@ -34,7 +35,6 @@ On 11-Apr-2011, at 6:54 PM, Roman Tkachenko <romant@example.com> wrote:
 
     eq_("Test reply", quotations.extract_from_plain(msg_body))
 
-
 def test_pattern_on_date_polymail():
     msg_body = """Test reply
 
@@ -190,17 +190,14 @@ Test"""
     eq_('Test reply', quotations.extract_from_plain(
         msg_body.format(six.text_type(original_message_indicator))))
 
-
 def test_english_original_message():
     _check_pattern_original_message('Original Message')
     _check_pattern_original_message('Reply Message')
 
-
 def test_german_original_message():
     _check_pattern_original_message(u'Ursprüngliche Nachricht')
     _check_pattern_original_message('Antwort Nachricht')
 
-
 def test_danish_original_message():
     _check_pattern_original_message('Oprindelig meddelelse')
 
@@ -299,7 +296,6 @@ On 04/19/2011 07:10 AM, Roman Tkachenko wrote:
 > Hello"""
     eq_("Hi", quotations.extract_from_plain(msg_body))
 
-
 def test_with_indent():
     msg_body = """YOLO salvia cillum kogi typewriter mumblecore cardigan skateboard Austin.
 
@@ -307,8 +303,7 @@ def test_with_indent():
 
 Brunch mumblecore pug Marfa tofu, irure taxidermy hoodie readymade pariatur.
     """
-    eq_("YOLO salvia cillum kogi typewriter mumblecore cardigan skateboard Austin.",
+    eq_("YOLO salvia cillum kogi typewriter mumblecore cardigan skateboard Austin.", quotations.extract_from_plain(msg_body))
-        quotations.extract_from_plain(msg_body))
 
 
 def test_short_quotation_with_newline():
@@ -348,7 +343,6 @@ Subject: The manager has commented on your Loop
 Blah-blah-blah
 """))
 
-
 def test_german_from_block():
     eq_('Allo! Follow up MIME!', quotations.extract_from_plain(
     """Allo! Follow up MIME!
@@ -361,7 +355,6 @@ Betreff: The manager has commented on your Loop
 Blah-blah-blah
 """))
 
-
 def test_french_multiline_from_block():
     eq_('Lorem ipsum', quotations.extract_from_plain(
     u"""Lorem ipsum
@@ -374,7 +367,6 @@ Objet : Follow Up
 Blah-blah-blah
 """))
 
-
 def test_french_from_block():
     eq_('Lorem ipsum', quotations.extract_from_plain(
     u"""Lorem ipsum
@@ -383,7 +375,6 @@ Le 23 janv. 2015 à 22:03, Brendan xxx <brendan.xxx@xxx.com<mailto:brendan.xxx@x
 
 Bonjour!"""))
 
-
 def test_polish_from_block():
     eq_('Lorem ipsum', quotations.extract_from_plain(
     u"""Lorem ipsum
@@ -394,7 +385,6 @@ napisał:
 Blah!
 """))
 
-
 def test_danish_from_block():
     eq_('Allo! Follow up MIME!', quotations.extract_from_plain(
     """Allo! Follow up MIME!
@@ -407,7 +397,6 @@ Emne: The manager has commented on your Loop
 Blah-blah-blah
 """))
 
-
 def test_swedish_from_block():
     eq_('Allo! Follow up MIME!', quotations.extract_from_plain(
     u"""Allo! Follow up MIME!
@@ -419,7 +408,6 @@ Till: Isacson Leiff
 Blah-blah-blah
 """))
 
-
 def test_swedish_from_line():
     eq_('Lorem', quotations.extract_from_plain(
     """Lorem
@@ -428,7 +416,6 @@ Den 14 september, 2015 02:23:18, Valentino Rudy (valentino@rudy.be) skrev:
 Veniam laborum mlkshk kale chips authentic. Normcore mumblecore laboris, fanny pack readymade eu blog chia pop-up freegan enim master cleanse.
 """))
 
-
 def test_norwegian_from_line():
     eq_('Lorem', quotations.extract_from_plain(
     u"""Lorem
@@ -437,7 +424,6 @@ På 14 september 2015 på 02:23:18, Valentino Rudy (valentino@rudy.be) skrev:
 Veniam laborum mlkshk kale chips authentic. Normcore mumblecore laboris, fanny pack readymade eu blog chia pop-up freegan enim master cleanse.
 """))
 
-
 def test_dutch_from_block():
     eq_('Gluten-free culpa lo-fi et nesciunt nostrud.', quotations.extract_from_plain(
     """Gluten-free culpa lo-fi et nesciunt nostrud.
@@ -447,7 +433,6 @@ Op 17-feb.-2015, om 13:18 heeft Julius Caesar <pantheon@rome.com> het volgende g
 Small batch beard laboris tempor, non listicle hella Tumblr heirloom.
 """))
 
-
 def test_vietnamese_from_block():
     eq_('Hello', quotations.extract_from_plain(
     u"""Hello
@@ -457,7 +442,6 @@ Vào 14:24 8 tháng 6, 2017, Hùng Nguyễn <hungnguyen@xxx.com> đã viết:
 > Xin chào
 """))
 
-
 def test_quotation_marker_false_positive():
     msg_body = """Visit us now for assistance...
 >>> >>>  http://www.domain.com <<<
@@ -842,10 +826,10 @@ The user experience was unparallelled. Please continue production. I'm sending p
 that this line is intact."""
 
     parsed = quotations.extract_from_plain(msg_body)
-    eq_(msg_body, parsed.decode('utf8'))
+    eq_(msg_body, parsed)
 
 
-def test_appointment():
+def test_appointment_2():
     msg_body = """Invitation for an interview:
 
 Date: Wednesday 3, October 2011 
@@ -854,4 +838,4 @@ Address: 130 Fox St
 
 Please bring in your ID."""
     parsed = quotations.extract_from_plain(msg_body)
-    eq_(msg_body, parsed.decode('utf8'))
+    eq_(msg_body, parsed)

tests/utils_test.py

@@ -2,13 +2,12 @@
 
 from __future__ import absolute_import
 
-
-from nose.tools import eq_, ok_, assert_false
-from talon import utils as u
-from mock import patch, Mock
 import cchardet
 import six
 
+from talon import utils as u
+from . import *
+
 
 def test_get_delimiter():
     eq_('\r\n', u.get_delimiter('abc\r\n123'))