Merge pull request #101 from mailgun/sergey/empty-html

if html stripped off quotations does not have readable text fallback …

.gitignore

@@ -48,4 +48,7 @@ tramp
 *_archive
 
 # Trial temp
-_trial_temp
+_trial_temp
+
+# OSX
+.DS_Store

MANIFEST.in

@@ -1,9 +1,7 @@
-recursive-include tests *
-recursive-include talon *
 recursive-exclude tests *.pyc *~
 recursive-exclude talon *.pyc *~
 include train.data
 include classifier
 include LICENSE
 include MANIFEST.in
-include README.rst
+include README.rst

README.rst

@@ -89,21 +89,46 @@ the power of machine learning algorithms:
     # text == "Thanks Sasha, I can't go any higher and is why I limited it to the\nhomepage."
     # signature == "John Doe\nvia mobile"
 
-For machine learning talon currently uses `PyML`_ library to build SVM
+For machine learning talon currently uses the `scikit-learn`_ library to build SVM
 classifiers. The core of machine learning algorithm lays in
 ``talon.signature.learning package``. It defines a set of features to
 apply to a message (``featurespace.py``), how data sets are built
 (``dataset.py``), classifier’s interface (``classifier.py``).
 
-The data used for training is taken from our personal email
+Currently the data used for training is taken from our personal email
 conversations and from `ENRON`_ dataset. As a result of applying our set
 of features to the dataset we provide files ``classifier`` and
 ``train.data`` that don’t have any personal information but could be
 used to load trained classifier. Those files should be regenerated every
 time the feature/data set is changed.
 
-.. _PyML: http://pyml.sourceforge.net/
+To regenerate the model files, you can run
+
+.. code:: sh
+
+    python train.py
+
+or
+
+.. code:: python
+    
+    from talon.signature import EXTRACTOR_FILENAME, EXTRACTOR_DATA
+    from talon.signature.learning.classifier import train, init
+    train(init(), EXTRACTOR_DATA, EXTRACTOR_FILENAME)
+
+Open-source Dataset
+-------------------
+
+Recently we started a `forge`_ project to create an open-source, annotated dataset of raw emails. In the project we
+used a subset of `ENRON`_ data, cleansed of private, health and financial information by `EDRM`_. At the moment over 190
+emails are annotated. Any contribution and collaboration on the project are welcome. Once the dataset is ready we plan to
+start using it for talon.
+
+.. _scikit-learn: http://scikit-learn.org
 .. _ENRON: https://www.cs.cmu.edu/~enron/
+.. _EDRM: http://www.edrm.net/resources/data-sets/edrm-enron-email-data-set
+.. _forge: https://github.com/mailgun/forge
+
 
 Research
 --------

setup.py

@@ -1,13 +1,35 @@
-import os
-import sys
-import contextlib
-
-from distutils.spawn import find_executable
+from __future__ import absolute_import
 from setuptools import setup, find_packages
+from setuptools.command.install import install
+
+
+class InstallCommand(install):
+    user_options = install.user_options + [
+        ('no-ml', None, "Don't install without Machine Learning modules."),
+    ]
+
+    boolean_options = install.boolean_options + ['no-ml']
+
+    def initialize_options(self):
+        install.initialize_options(self)
+        self.no_ml = None
+
+    def finalize_options(self):
+        install.finalize_options(self)
+        if self.no_ml:
+            dist = self.distribution
+            dist.packages=find_packages(exclude=[
+                'tests',
+                'tests.*',
+                'talon.signature',
+                'talon.signature.*',
+            ])
+            for not_required in ['numpy', 'scipy', 'scikit-learn==0.16.1']:
+                dist.install_requires.remove(not_required)
 
 
 setup(name='talon',
-      version='1.0.2',
+      version='1.2.15',
       description=("Mailgun library "
                    "to extract message quotations and signatures."),
       long_description=open("README.rst").read(),
@@ -15,92 +37,26 @@ setup(name='talon',
       author_email='admin@mailgunhq.com',
       url='https://github.com/mailgun/talon',
       license='APACHE2',
-      packages=find_packages(exclude=['tests']),
+      cmdclass={
+          'install': InstallCommand,
+      },
+      packages=find_packages(exclude=['tests', 'tests.*']),
       include_package_data=True,
       zip_safe=True,
       install_requires=[
-          "lxml==2.3.3",
-          "regex==0.1.20110315",
-          "chardet==1.0.1",
-          "dnspython==1.11.1",
-          "html2text",
-          "nose==1.2.1",
+          "lxml>=2.3.3",
+          "regex>=1",
+          "numpy",
+          "scipy",
+          "scikit-learn==0.16.1", # pickled versions of classifier, else rebuild
+          'chardet>=1.0.1',
+          'cchardet>=0.3.5',
+          'cssselect',
+          'six>=1.10.0',
+          ],
+      tests_require=[
           "mock",
-          "coverage",
-          "flanker"
+          "nose>=1.2.1",
+          "coverage"
           ]
       )
-
-
-def install_pyml():
-    '''
-    Downloads and installs PyML
-    '''
-    try:
-        import PyML
-    except:
-        pass
-    else:
-        return
-
-    # install numpy first
-    pip('install numpy==1.6.1 --upgrade')
-
-    pyml_tarball = (
-        'http://09cce49df173f6f6e61f-fd6930021b51685920a6fa76529ee321'
-        '.r45.cf2.rackcdn.com/PyML-0.7.9.tar.gz')
-    pyml_srcidr = 'PyML-0.7.9'
-
-    # see if PyML tarball needs to be fetched:
-    if not dir_exists(pyml_srcidr):
-        run("curl %s | tar -xz" % pyml_tarball)
-
-    # compile&install:
-    with cd(pyml_srcidr):
-        python('setup.py build')
-        python('setup.py install')
-
-
-def run(command):
-    if os.system(command) != 0:
-        raise Exception("Failed '{}'".format(command))
-    else:
-        return 0
-
-
-def python(command):
-    command = '{} {}'.format(sys.executable, command)
-    run(command)
-
-
-def enforce_executable(name, install_info):
-    if os.system("which {}".format(name)) != 0:
-        raise Exception(
-            '{} utility is missing.\nTo install, run:\n\n{}\n'.format(
-                name, install_info))
-
-
-def pip(command):
-    command = '{} {}'.format(find_executable('pip'), command)
-    run(command)
-
-
-def dir_exists(path):
-    return os.path.isdir(path)
-
-
-@contextlib.contextmanager
-def cd(directory):
-    curdir = os.getcwd()
-    try:
-        os.chdir(directory)
-        yield {}
-    finally:
-        os.chdir(curdir)
-
-
-if __name__ == '__main__':
-    if len(sys.argv) > 1 and sys.argv[1] in ['develop', 'install']:
-        enforce_executable('curl', 'sudo aptitude install curl')
-
-        install_pyml()

talon/__init__.py

@@ -1,7 +1,13 @@
+from __future__ import absolute_import
 from talon.quotations import register_xpath_extensions
-from talon import signature
+try:
+    from talon import signature
+    ML_ENABLED = True
+except ImportError:
+    ML_ENABLED = False
 
 
 def init():
     register_xpath_extensions()
-    signature.initialize()
+    if ML_ENABLED:
+        signature.initialize()

talon/constants.py

@@ -1,3 +1,4 @@
+from __future__ import absolute_import
 import regex as re
 
 

talon/html_quotations.py

@@ -3,6 +3,7 @@ The module's functions operate on message bodies trying to extract original
 messages (without quoted messages) from html
 """
 
+from __future__ import absolute_import
 import regex as re
 
 
@@ -12,6 +13,7 @@ CHECKPOINT_PATTERN = re.compile(CHECKPOINT_PREFIX + '\d+' + CHECKPOINT_SUFFIX)
 
 # HTML quote indicators (tag ids)
 QUOTE_IDS = ['OLK_SRC_BODY_SECTION']
+RE_FWD = re.compile("^[-]+[ ]*Forwarded message[ ]*[-]+$", re.I | re.M)
 
 
 def add_checkpoint(html_note, counter):
@@ -76,8 +78,8 @@ 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('.gmail_quote')
-    if gmail_quote:
+    gmail_quote = html_message.cssselect('div.gmail_quote')
+    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
 
@@ -85,9 +87,12 @@ def cut_gmail_quote(html_message):
 def cut_microsoft_quote(html_message):
     ''' Cuts splitter block and all following blocks. '''
     splitter = html_message.xpath(
-        #outlook 2007, 2010
+        #outlook 2007, 2010 (international)
         "//div[@style='border:none;border-top:solid #B5C4DF 1.0pt;"
         "padding:3.0pt 0cm 0cm 0cm']|"
+        #outlook 2007, 2010 (american)
+        "//div[@style='border:none;border-top:solid #B5C4DF 1.0pt;"
+        "padding:3.0pt 0in 0in 0in']|"
         #windows mail
         "//div[@style='padding-top: 5px; "
         "border-top-color: rgb(229, 229, 229); "
@@ -138,9 +143,14 @@ def cut_by_id(html_message):
 
 
 def cut_blockquote(html_message):
-    ''' Cuts blockquote with wrapping elements. '''
-    quote = html_message.find('.//blockquote')
-    if quote is not None:
+    ''' Cuts the last non-nested blockquote with wrapping elements.'''
+    quote = html_message.xpath(
+        '(.//blockquote)'
+        '[not(@class="gmail_quote") and not(ancestor::blockquote)]'
+        '[last()]')
+
+    if quote:
+        quote = quote[0]
         quote.getparent().remove(quote)
         return True
 
@@ -154,21 +164,58 @@ def cut_from_block(html_message):
 
     if block:
         block = block[-1]
+        parent_div = None
         while block.getparent() is not None:
             if block.tag == 'div':
-                block.getparent().remove(block)
+                parent_div = block
+                break
+            block = block.getparent()
+        if parent_div is not None:
+            maybe_body = parent_div.getparent()
+            # In cases where removing this enclosing div will remove all
+            # content, we should assume the quote is not enclosed in a tag.
+            parent_div_is_all_content = (
+                maybe_body is not None and maybe_body.tag == 'body' and
+                len(maybe_body.getchildren()) == 1)
+
+            if not parent_div_is_all_content:
+                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:
-                block = block.getparent()
-    else:
-        # handle the case when From: block goes right after e.g. <hr>
-        # and not enclosed in some tag
-        block = html_message.xpath(
-            ("//*[starts-with(mg:tail(), 'From:')]|"
-             "//*[starts-with(mg:tail(), 'Date:')]"))
-        if block:
-            block = block[0]
-            while(block.getnext() is not None):
-                block.getparent().remove(block.getnext())
-            block.getparent().remove(block)
-            return True
+        else:
+            return False
+
+    # handle the case when From: block goes right after e.g. <hr>
+    # and not enclosed in some tag
+    block = html_message.xpath(
+        ("//*[starts-with(mg:tail(), 'From:')]|"
+         "//*[starts-with(mg:tail(), 'Date:')]"))
+    if block:
+        block = block[0]
+
+        if RE_FWD.match(block.getparent().text or ''):
+            return False
+        
+        while(block.getnext() is not None):
+            block.getparent().remove(block.getnext())
+        block.getparent().remove(block)
+        return True
+
+def cut_zimbra_quote(html_message):
+    zDivider = html_message.xpath('//hr[@data-marker="__DIVIDER__"]')
+    if zDivider:
+        zDivider[0].getparent().remove(zDivider[0])
+        return True

talon/quotations.py

@@ -5,16 +5,17 @@ The module's functions operate on message bodies trying to extract
 original messages (without quoted messages)
 """
 
+from __future__ import absolute_import
 import regex as re
 import logging
 from copy import deepcopy
 
 from lxml import html, etree
-import html2text
 
-from talon.constants import RE_DELIMITER
-from talon.utils import random_token, get_delimiter
+from talon.utils import get_delimiter, html_tree_to_text
 from talon import html_quotations
+from six.moves import range
+import six
 
 
 log = logging.getLogger(__name__)
@@ -23,7 +24,7 @@ log = logging.getLogger(__name__)
 RE_FWD = re.compile("^[-]+[ ]*Forwarded message[ ]*[-]+$", re.I | re.M)
 
 RE_ON_DATE_SMB_WROTE = re.compile(
-    u'(-*[ ]?({0})[ ].*({1})(.*\n){{0,2}}.*({2}):?-*)'.format(
+    u'(-*[>]?[ ]?({0})[ ].*({1})(.*\n){{0,2}}.*({2}):?-*)'.format(
         # Beginning of the line
         u'|'.join((
             # English
@@ -33,7 +34,13 @@ RE_ON_DATE_SMB_WROTE = re.compile(
             # Polish
             'W dniu',
             # Dutch
-            'Op'
+            'Op',
+            # German
+            'Am',
+            # Norwegian
+            u'På',
+            # Swedish, Danish
+            'Den',
         )),
         # Date and sender separator
         u'|'.join((
@@ -51,18 +58,28 @@ RE_ON_DATE_SMB_WROTE = re.compile(
             # Polish
             u'napisał',
             # Dutch
-            'schreef','verzond','geschreven'
+            'schreef','verzond','geschreven',
+            # German
+            'schrieb',
+            # Norwegian, Swedish
+            'skrev',
         ))
     ))
 # Special case for languages where text is translated like this: 'on {date} wrote {somebody}:'
 RE_ON_DATE_WROTE_SMB = re.compile(
-    u'(-*[ ]?({0})[ ].*(.*\n){{0,2}}.*({1})[ ].*:)'.format(
+    u'(-*[>]?[ ]?({0})[ ].*(.*\n){{0,2}}.*({1})[ ]*.*:)'.format(
         # Beginning of the line
+        u'|'.join((
         	'Op',
+        	#German
+        	'Am'
+        )),
         # Ending of the line
         u'|'.join((
             # Dutch
-            'schreef','verzond','geschreven'
+            'schreef','verzond','geschreven',
+            # German
+            'schrieb'
         ))
     )
     )
@@ -93,7 +110,7 @@ RE_EMPTY_QUOTATION = re.compile(
     (
         # quotation border: splitter line or a number of quotation marker lines
         (?:
-            s
+            (?:se*)+
             |
             (?:me*){2,}
         )
@@ -116,20 +133,27 @@ RE_ORIGINAL_MESSAGE = re.compile(u'[\s]*[-]+[ ]*({})[ ]*[-]+'.format(
 RE_FROM_COLON_OR_DATE_COLON = re.compile(u'(_+\r?\n)?[\s]*(:?[*]?{})[\s]?:[*]? .*'.format(
     u'|'.join((
         # "From" in different languages.
-        'From', 'Van', 'De', 'Von', 'Fra',
+        'From', 'Van', 'De', 'Von', 'Fra', u'Från',
         # "Date" in different languages.
-        'Date', 'Datum', u'Envoyé'
+        'Date', 'Datum', u'Envoyé', 'Skickat', 'Sendt',
     ))), re.I)
 
 SPLITTER_PATTERNS = [
     RE_ORIGINAL_MESSAGE,
-    # <date> <person>
-    re.compile("(\d+/\d+/\d+|\d+\.\d+\.\d+).*@", re.VERBOSE),
     RE_ON_DATE_SMB_WROTE,
     RE_ON_DATE_WROTE_SMB,
     RE_FROM_COLON_OR_DATE_COLON,
+    # 02.04.2012 14:20 пользователь "bob@example.com" <
+    # bob@xxx.mailgun.org> написал:
+    re.compile("(\d+/\d+/\d+|\d+\.\d+\.\d+).*@", re.S),
+    # 2014-10-17 11:28 GMT+03:00 Bob <
+    # bob@example.com>:
+    re.compile("\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}\s+GMT.*@", re.S),
+    # Thu, 26 Jun 2014 14:00:51 +0400 Bob <bob@example.com>:
     re.compile('\S{3,10}, \d\d? \S{3,10} 20\d\d,? \d\d?:\d\d(:\d\d)?'
-               '( \S+){3,6}@\S+:')
+               '( \S+){3,6}@\S+:'),
+    # Sent from Samsung MobileName <address@example.com> wrote:
+    re.compile('Sent from Samsung .*@.*> wrote')
     ]
 
 
@@ -140,6 +164,9 @@ RE_PARENTHESIS_LINK = re.compile("\(https?://")
 
 SPLITTER_MAX_LINES = 4
 MAX_LINES_COUNT = 1000
+# an extensive research shows that exceeding this limit
+# leads to excessive processing time
+MAX_HTML_LEN = 2794202
 
 QUOT_PATTERN = re.compile('^>+ ?')
 NO_QUOT_LINE = re.compile('^[^>].*[\S].*')
@@ -151,7 +178,7 @@ def extract_from(msg_body, content_type='text/plain'):
             return extract_from_plain(msg_body)
         elif content_type == 'text/html':
             return extract_from_html(msg_body)
-    except Exception, e:
+    except Exception:
         log.exception('ERROR extracting message')
 
     return msg_body
@@ -170,7 +197,7 @@ def mark_message_lines(lines):
     >>> mark_message_lines(['answer', 'From: foo@bar.com', '', '> question'])
     'tsem'
     """
-    markers = bytearray(len(lines))
+    markers = ['e' for _ in lines]
     i = 0
     while i < len(lines):
         if not lines[i].strip():
@@ -182,10 +209,11 @@ def mark_message_lines(lines):
         else:
             # in case splitter is spread across several lines
             splitter = is_splitter('\n'.join(lines[i:i + SPLITTER_MAX_LINES]))
+
             if splitter:
                 # append as many splitter markers as lines in splitter
                 splitter_lines = splitter.group().splitlines()
-                for j in xrange(len(splitter_lines)):
+                for j in range(len(splitter_lines)):
                     markers[i + j] = 's'
 
                 # skip splitter lines
@@ -195,7 +223,7 @@ def mark_message_lines(lines):
                 markers[i] = 't'
         i += 1
 
-    return markers
+    return ''.join(markers)
 
 
 def process_marked_lines(lines, markers, return_flags=[False, -1, -1]):
@@ -209,6 +237,7 @@ def process_marked_lines(lines, markers, return_flags=[False, -1, -1]):
     return_flags = [were_lines_deleted, first_deleted_line,
                     last_deleted_line]
     """
+    markers = ''.join(markers)
     # if there are no splitter there should be no markers
     if 's' not in markers and not re.search('(me*){3}', markers):
         markers = markers.replace('m', 't')
@@ -254,10 +283,15 @@ def preprocess(msg_body, delimiter, content_type='text/plain'):
     Replaces link brackets so that they couldn't be taken for quotation marker.
     Splits line in two if splitter pattern preceded by some text on the same
     line (done only for 'On <date> <person> wrote:' pattern).
+
+    Converts msg_body into a unicode.
     """
     # normalize links i.e. replace '<', '>' wrapping the link with some symbols
     # so that '>' closing the link couldn't be mistakenly taken for quotation
     # marker.
+    if isinstance(msg_body, bytes):
+        msg_body = msg_body.decode('utf8')
+
     def link_wrapper(link):
         newline_index = msg_body[:link.start()].rfind("\n")
         if msg_body[newline_index + 1] == ">":
@@ -294,12 +328,8 @@ def extract_from_plain(msg_body):
 
     delimiter = get_delimiter(msg_body)
     msg_body = preprocess(msg_body, delimiter)
-    lines = msg_body.splitlines()
-
     # don't process too long messages
-    if len(lines) > MAX_LINES_COUNT:
-        return stripped_text
-
+    lines = msg_body.splitlines()[:MAX_LINES_COUNT]
     markers = mark_message_lines(lines)
     lines = process_marked_lines(lines, markers)
 
@@ -324,44 +354,61 @@ def extract_from_html(msg_body):
     then extracting quotations from text,
     then checking deleted checkpoints,
     then deleting necessary tags.
-    """
 
-    if msg_body.strip() == '':
+    Returns a unicode string.
+    """
+    if isinstance(msg_body, six.text_type):
+        msg_body = msg_body.encode('utf8')
+    elif not isinstance(msg_body, bytes):
+        msg_body = msg_body.encode('ascii')
+
+    result = _extract_from_html(msg_body)
+    if isinstance(result, bytes):
+        result = result.decode('utf8')
+
+    return result
+
+
+def _extract_from_html(msg_body):
+    """
+    Extract not quoted message from provided html message body
+    using tags and plain text algorithm.
+
+    Cut out the 'blockquote', 'gmail_quote' tags.
+    Cut Microsoft quotations.
+
+    Then use plain text algorithm to cut out splitter or
+    leftover quotation.
+    This works by adding checkpoint text to all html tags,
+    then converting html to text,
+    then extracting quotations from text,
+    then checking deleted checkpoints,
+    then deleting necessary tags.
+    """
+    if len(msg_body) > MAX_HTML_LEN:
         return msg_body
 
+    if msg_body.strip() == b'':
+        return msg_body
+
+    msg_body = msg_body.replace(b'\r\n', b'\n')
     html_tree = html.document_fromstring(
         msg_body,
         parser=html.HTMLParser(encoding="utf-8")
     )
-
     cut_quotations = (html_quotations.cut_gmail_quote(html_tree) or
+                      html_quotations.cut_zimbra_quote(html_tree) or
                       html_quotations.cut_blockquote(html_tree) or
                       html_quotations.cut_microsoft_quote(html_tree) or
                       html_quotations.cut_by_id(html_tree) or
                       html_quotations.cut_from_block(html_tree)
                       )
-
     html_tree_copy = deepcopy(html_tree)
 
     number_of_checkpoints = html_quotations.add_checkpoint(html_tree, 0)
-    quotation_checkpoints = [False for i in xrange(number_of_checkpoints)]
-    msg_with_checkpoints = html.tostring(html_tree)
-
-    h = html2text.HTML2Text()
-    h.body_width = 0  # generate plain text without wrap
-
-    # html2text adds unnecessary star symbols. Remove them.
-    # Mask star symbols
-    msg_with_checkpoints = msg_with_checkpoints.replace('*', '3423oorkg432')
-    plain_text = h.handle(msg_with_checkpoints)
-    # Remove created star symbols
-    plain_text = plain_text.replace('*', '')
-    # Unmask saved star symbols
-    plain_text = plain_text.replace('3423oorkg432', '*')
-
-    delimiter = get_delimiter(plain_text)
-
-    plain_text = preprocess(plain_text, delimiter, content_type='text/html')
+    quotation_checkpoints = [False] * number_of_checkpoints
+    plain_text = html_tree_to_text(html_tree)
+    plain_text = preprocess(plain_text, '\n', content_type='text/html')
     lines = plain_text.splitlines()
 
     # Don't process too long messages
@@ -384,25 +431,30 @@ def extract_from_html(msg_body):
     process_marked_lines(lines, markers, return_flags)
     lines_were_deleted, first_deleted, last_deleted = return_flags
 
+    if not lines_were_deleted and not cut_quotations:
+        return msg_body
+
     if lines_were_deleted:
         #collect checkpoints from deleted lines
-        for i in xrange(first_deleted, last_deleted):
+        for i in range(first_deleted, last_deleted):
             for checkpoint in line_checkpoints[i]:
                 quotation_checkpoints[checkpoint] = True
-    else:
-        if cut_quotations:
-            return html.tostring(html_tree_copy)
-        else:
-            return msg_body
 
-    # Remove tags with quotation checkpoints
-    html_quotations.delete_quotation_tags(
-        html_tree_copy, 0, quotation_checkpoints
-    )
+        # Remove tags with quotation checkpoints
+        html_quotations.delete_quotation_tags(
+            html_tree_copy, 0, quotation_checkpoints
+        )
+
+    if _readable_text_empty(html_tree_copy):
+        return msg_body
 
     return html.tostring(html_tree_copy)
 
 
+def _readable_text_empty(html_tree):
+    return not bool(html_tree_to_text(html_tree).strip())
+
+
 def is_splitter(line):
     '''
     Returns Matcher object if provided string is a splitter and

talon/signature/__init__.py

@@ -20,12 +20,11 @@ trained against, don't forget to regenerate:
 * signature/data/classifier
 """
 
+from __future__ import absolute_import
 import os
-import sys
-from cStringIO import StringIO
 
 from . import extraction
-from . extraction import extract
+from . extraction import extract  #noqa
 from . learning import classifier
 
 
@@ -36,13 +35,5 @@ EXTRACTOR_DATA = os.path.join(DATA_DIR, 'train.data')
 
 
 def initialize():
-    try:
-        # redirect output
-        so, sys.stdout = sys.stdout, StringIO()
-
-        extraction.EXTRACTOR = classifier.load(EXTRACTOR_FILENAME,
-                                               EXTRACTOR_DATA)
-        sys.stdout = so
-    except Exception, e:
-        raise Exception(
-            "Failed initializing signature parsing with classifiers", e)
+    extraction.EXTRACTOR = classifier.load(EXTRACTOR_FILENAME,
+                                           EXTRACTOR_DATA)

talon/signature/bruteforce.py

@@ -1,3 +1,4 @@
+from __future__ import absolute_import
 import logging
 
 import regex as re
@@ -111,7 +112,7 @@ def extract_signature(msg_body):
 
             return (stripped_body.strip(),
                     signature.strip())
-    except Exception, e:
+    except Exception as e:
         log.exception('ERROR extracting signature')
         return (msg_body, None)
 

talon/signature/extraction.py

@@ -1,14 +1,11 @@
 # -*- coding: utf-8 -*-
 
-import os
+from __future__ import absolute_import
 import logging
 
 import regex as re
-from PyML import SparseDataSet
+import numpy
 
-from talon.constants import RE_DELIMITER
-from talon.signature.constants import (SIGNATURE_MAX_LINES,
-                                       TOO_LONG_SIGNATURE_LINE)
 from talon.signature.learning.featurespace import features, build_pattern
 from talon.utils import get_delimiter
 from talon.signature.bruteforce import get_signature_candidate
@@ -36,8 +33,8 @@ RE_REVERSE_SIGNATURE = re.compile(r'''
 
 def is_signature_line(line, sender, classifier):
     '''Checks if the line belongs to signature. Returns True or False.'''
-    data = SparseDataSet([build_pattern(line, features(sender))])
-    return classifier.decisionFunc(data, 0) > 0
+    data = numpy.array(build_pattern(line, features(sender)))
+    return classifier.predict(data) > 0
 
 
 def extract(body, sender):
@@ -61,7 +58,7 @@ def extract(body, sender):
                 text = delimiter.join(text)
                 if text.strip():
                     return (text, delimiter.join(signature))
-    except Exception, e:
+    except Exception:
         log.exception('ERROR when extracting signature with classifiers')
 
     return (body, None)

talon/signature/learning/classifier.py

@@ -5,32 +5,28 @@ The classifier could be used to detect if a certain line of the message
 body belongs to the signature.
 """
 
-import os
-import sys
-
-from PyML import SparseDataSet, SVM
+from __future__ import absolute_import
+from numpy import genfromtxt
+from sklearn.svm import LinearSVC
+from sklearn.externals import joblib
 
 
 def init():
-    '''Inits classifier with optimal options.'''
-    return SVM(C=10, optimization='liblinear')
+    """Inits classifier with optimal options."""
+    return LinearSVC(C=10.0)
 
 
 def train(classifier, train_data_filename, save_classifier_filename=None):
-    '''Trains and saves classifier so that it could be easily loaded later.'''
-    data = SparseDataSet(train_data_filename, labelsColumn=-1)
-    classifier.train(data)
+    """Trains and saves classifier so that it could be easily loaded later."""
+    file_data = genfromtxt(train_data_filename, delimiter=",")
+    train_data, labels = file_data[:, :-1], file_data[:, -1]
+    classifier.fit(train_data, labels)
+
     if save_classifier_filename:
-        classifier.save(save_classifier_filename)
+        joblib.dump(classifier, save_classifier_filename)
     return classifier
 
 
 def load(saved_classifier_filename, train_data_filename):
-    """Loads saved classifier.
-
-    Classifier should be loaded with the same data it was trained against
-    """
-    train_data = SparseDataSet(train_data_filename, labelsColumn=-1)
-    classifier = init()
-    classifier.load(saved_classifier_filename, train_data)
-    return classifier
+    """Loads saved classifier. """
+    return joblib.load(saved_classifier_filename)

talon/signature/learning/dataset.py

@@ -16,11 +16,13 @@ suffix and the corresponding sender file has the same name except for the
 suffix which should be `_sender`.
 """
 
+from __future__ import absolute_import
 import os
 import regex as re
 
 from talon.signature.constants import SIGNATURE_MAX_LINES
 from talon.signature.learning.featurespace import build_pattern, features
+from six.moves import range
 
 
 SENDER_SUFFIX = '_sender'
@@ -144,7 +146,7 @@ def build_extraction_dataset(folder, dataset_filename,
             if not sender or not msg:
                 continue
             lines = msg.splitlines()
-            for i in xrange(1, min(SIGNATURE_MAX_LINES,
+            for i in range(1, min(SIGNATURE_MAX_LINES,
                                    len(lines)) + 1):
                 line = lines[-i]
                 label = -1

talon/signature/learning/featurespace.py

@@ -7,9 +7,12 @@ The body and the message sender string are converted into unicode before
 applying features to them.
 """
 
+from __future__ import absolute_import
 from talon.signature.constants import (SIGNATURE_MAX_LINES,
                                        TOO_LONG_SIGNATURE_LINE)
 from talon.signature.learning.helpers import *
+from six.moves import zip
+from functools import reduce
 
 
 def features(sender=''):

talon/signature/learning/helpers.py

@@ -6,6 +6,7 @@
 
 """
 
+from __future__ import absolute_import
 import unicodedata
 import regex as re
 
@@ -16,8 +17,8 @@ from talon.signature.constants import SIGNATURE_MAX_LINES
 
 rc = re.compile
 
-RE_EMAIL = rc('@')
-RE_RELAX_PHONE = rc('.*(\(? ?[\d]{2,3} ?\)?.{,3}){2,}')
+RE_EMAIL = rc('\S@\S')
+RE_RELAX_PHONE = rc('(\(? ?[\d]{2,3} ?\)?.{,3}?){2,}')
 RE_URL = rc(r'''https?://|www\.[\S]+\.[\S]''')
 
 # Taken from:
@@ -40,14 +41,6 @@ RE_SIGNATURE_WORDS = rc(('(T|t)hank.*,|(B|b)est|(R|r)egards|'
 # Line contains a pattern like Vitor R. Carvalho or William W. Cohen.
 RE_NAME = rc('[A-Z][a-z]+\s\s?[A-Z][\.]?\s\s?[A-Z][a-z]+')
 
-# Pattern to match if e.g. 'Sender:' header field has sender names.
-SENDER_WITH_NAME_PATTERN = '([\s]*[\S]+,?)+[\s]*<.*>.*'
-RE_SENDER_WITH_NAME = rc(SENDER_WITH_NAME_PATTERN)
-
-# Reply line clue line endings, as in regular expression:
-# " wrote:$" or " writes:$"
-RE_CLUE_LINE_END = rc('.*(W|w)rotes?:$')
-
 INVALID_WORD_START = rc('\(|\+|[\d]')
 
 BAD_SENDER_NAMES = [
@@ -128,7 +121,7 @@ def contains_sender_names(sender):
     names = names or sender
     if names != '':
         return binary_regex_search(re.compile(names))
-    return lambda s: False
+    return lambda s: 0
 
 
 def extract_names(sender):
@@ -142,7 +135,7 @@ def extract_names(sender):
     >>> extract_names('')
     []
     """
-    sender = to_unicode(sender)
+    sender = to_unicode(sender, precise=True)
     # Remove non-alphabetical characters
     sender = "".join([char if char.isalpha() else ' ' for char in sender])
     # Remove too short words and words from "black" list i.e.
@@ -169,7 +162,7 @@ def categories_percent(s, categories):
     50.0
     '''
     count = 0
-    s = to_unicode(s)
+    s = to_unicode(s, precise=True)
     for c in s:
         if unicodedata.category(c) in categories:
             count += 1
@@ -189,15 +182,16 @@ def punctuation_percent(s):
 
 def capitalized_words_percent(s):
     '''Returns capitalized words percent.'''
-    s = to_unicode(s)
+    s = to_unicode(s, precise=True)
     words = re.split('\s', s)
     words = [w for w in words if w.strip()]
+    words = [w for w in words if len(w) > 2]    
     capitalized_words_counter = 0
     valid_words_counter = 0
     for word in words:
         if not INVALID_WORD_START.match(word):
             valid_words_counter += 1
-            if word[0].isupper():
+            if word[0].isupper() and not word[1].isupper():
                 capitalized_words_counter += 1
     if valid_words_counter > 0 and len(words) > 1:
         return 100 * float(capitalized_words_counter) / valid_words_counter

talon/utils.py

@@ -1,12 +1,17 @@
 # coding:utf-8
 
+from __future__ import absolute_import
 import logging
 from random import shuffle
+import chardet
+import cchardet
+import regex as re
+
+from lxml import html
+from lxml.cssselect import CSSSelector
 
 from talon.constants import RE_DELIMITER
-
-
-log = logging.getLogger(__name__)
+import six
 
 
 def safe_format(format_string, *args, **kwargs):
@@ -25,7 +30,7 @@ def safe_format(format_string, *args, **kwargs):
     except (UnicodeEncodeError, UnicodeDecodeError):
         format_string = to_utf8(format_string)
         args = [to_utf8(p) for p in args]
-        kwargs = {k: to_utf8(v) for k, v in kwargs.iteritems()}
+        kwargs = {k: to_utf8(v) for k, v in six.iteritems(kwargs)}
         return format_string.format(*args, **kwargs)
 
     # ignore other errors
@@ -42,19 +47,51 @@ def to_unicode(str_or_unicode, precise=False):
         u'привет'
     If `precise` flag is True, tries to guess the correct encoding first.
     """
-    encoding = detect_encoding(str_or_unicode) if precise else 'utf-8'
-    if isinstance(str_or_unicode, str):
-        return unicode(str_or_unicode, encoding, 'replace')
+    if not isinstance(str_or_unicode, six.text_type):
+        encoding = quick_detect_encoding(str_or_unicode) if precise else 'utf-8'
+        return six.text_type(str_or_unicode, encoding, 'replace')
     return str_or_unicode
 
 
+def detect_encoding(string):
+    """
+    Tries to detect the encoding of the passed string.
+
+    Defaults to UTF-8.
+    """
+    assert isinstance(string, bytes)
+    try:
+        detected = chardet.detect(string)
+        if detected:
+            return detected.get('encoding') or 'utf-8'
+    except Exception as e:
+        pass
+    return 'utf-8'
+
+
+def quick_detect_encoding(string):
+    """
+    Tries to detect the encoding of the passed string.
+
+    Uses cchardet. Fallbacks to detect_encoding.
+    """
+    assert isinstance(string, bytes)
+    try:
+        detected = cchardet.detect(string)
+        if detected:
+            return detected.get('encoding') or detect_encoding(string)
+    except Exception as e:
+        pass
+    return detect_encoding(string)
+
+
 def to_utf8(str_or_unicode):
     """
     Safely returns a UTF-8 version of a given string
     >>> utils.to_utf8(u'hi')
         'hi'
     """
-    if isinstance(str_or_unicode, unicode):
+    if not isinstance(str_or_unicode, six.text_type):
         return str_or_unicode.encode("utf-8", "ignore")
     return str(str_or_unicode)
 
@@ -74,3 +111,86 @@ def get_delimiter(msg_body):
         delimiter = '\n'
 
     return delimiter
+
+def html_tree_to_text(tree):
+    for style in CSSSelector('style')(tree):
+        style.getparent().remove(style)
+
+    for c in tree.xpath('//comment()'):
+        c.getparent().remove(c)
+
+    text   = ""
+    for el in tree.iter():
+        el_text = (el.text or '') + (el.tail or '')
+        if len(el_text) > 1:
+            if el.tag in _BLOCKTAGS:
+                text += "\n"
+            if el.tag == 'li':
+                text += "  * "
+            text += el_text.strip() + " "
+
+            # add href to the output
+            href = el.attrib.get('href')
+            if href:
+                text += "(%s) " % href
+
+        if el.tag in _HARDBREAKS and text and not text.endswith("\n"):
+            text += "\n"
+
+    retval = _rm_excessive_newlines(text)
+    return _encode_utf8(retval)
+
+
+def html_to_text(string):
+    """
+    Dead-simple HTML-to-text converter:
+        >>> html_to_text("one<br>two<br>three")
+        >>> "one\ntwo\nthree"
+
+    NOTES:
+        1. the string is expected to contain UTF-8 encoded HTML!
+        2. returns utf-8 encoded str (not unicode)
+    """
+    if isinstance(string, six.text_type):
+        string = string.encode('utf8')
+
+    s = _prepend_utf8_declaration(string)
+    s = s.replace(b"\n", b"")
+
+    tree = html.fromstring(s)
+    return html_tree_to_text(tree)
+
+
+def _contains_charset_spec(s):
+    """Return True if the first 4KB contain charset spec
+    """
+    return s.lower().find(b'html; charset=', 0, 4096) != -1
+
+
+def _prepend_utf8_declaration(s):
+    """Prepend 'utf-8' encoding declaration if the first 4KB don't have any
+    """
+    return s if _contains_charset_spec(s) else _UTF8_DECLARATION + s
+
+
+def _rm_excessive_newlines(s):
+    """Remove excessive newlines that often happen due to tons of divs
+    """
+    return _RE_EXCESSIVE_NEWLINES.sub("\n\n", s).strip()
+
+
+def _encode_utf8(s):
+    """Encode in 'utf-8' if unicode
+    """
+    return s.encode('utf-8') if isinstance(s, six.text_type) else s
+
+
+_UTF8_DECLARATION = (b'<meta http-equiv="Content-Type" content="text/html;'
+                     b'charset=utf-8">')
+
+
+_BLOCKTAGS  = ['div', 'p', 'ul', 'li', 'h1', 'h2', 'h3']
+_HARDBREAKS = ['br', 'hr', 'tr']
+
+
+_RE_EXCESSIVE_NEWLINES = re.compile("\n{2,10}")

tests/__init__.py

@@ -1,3 +1,4 @@
+from __future__ import absolute_import
 from nose.tools import *
 from mock import *
 

tests/fixtures/html_replies/hotmail.html

@@ -1,3 +1,4 @@
+<?xml version="1.0" encoding="UTF-8"?>
 <html>
 <head>
 <style><!--

tests/fixtures/html_replies/ms_outlook_2010.html

@@ -0,0 +1,87 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=iso-2022-jp">
+<meta name="Generator" content="Microsoft Word 14 (filtered medium)">
+<style><!--
+/* Font Definitions */
+@font-face
+	{font-family:Calibri;
+	panose-1:2 15 5 2 2 2 4 3 2 4;}
+@font-face
+	{font-family:Tahoma;
+	panose-1:2 11 6 4 3 5 4 4 2 4;}
+/* Style Definitions */
+p.MsoNormal, li.MsoNormal, div.MsoNormal
+	{margin:0in;
+	margin-bottom:.0001pt;
+	font-size:12.0pt;
+	font-family:"Times New Roman","serif";}
+h3
+	{mso-style-priority:9;
+	mso-style-link:"Heading 3 Char";
+	mso-margin-top-alt:auto;
+	margin-right:0in;
+	mso-margin-bottom-alt:auto;
+	margin-left:0in;
+	font-size:13.5pt;
+	font-family:"Times New Roman","serif";
+	font-weight:bold;}
+a:link, span.MsoHyperlink
+	{mso-style-priority:99;
+	color:blue;
+	text-decoration:underline;}
+a:visited, span.MsoHyperlinkFollowed
+	{mso-style-priority:99;
+	color:purple;
+	text-decoration:underline;}
+p
+	{mso-style-priority:99;
+	mso-margin-top-alt:auto;
+	margin-right:0in;
+	mso-margin-bottom-alt:auto;
+	margin-left:0in;
+	font-size:12.0pt;
+	font-family:"Times New Roman","serif";}
+span.Heading3Char
+	{mso-style-name:"Heading 3 Char";
+	mso-style-priority:9;
+	mso-style-link:"Heading 3";
+	font-family:"Cambria","serif";
+	color:#4F81BD;
+	font-weight:bold;}
+span.EmailStyle19
+	{mso-style-type:personal-reply;
+	font-family:"Calibri","sans-serif";
+	color:#1F497D;}
+.MsoChpDefault
+	{mso-style-type:export-only;
+	font-family:"Calibri","sans-serif";}
+@page WordSection1
+	{size:8.5in 11.0in;
+	margin:1.0in 1.0in 1.0in 1.0in;}
+div.WordSection1
+	{page:WordSection1;}
+--></style><!--[if gte mso 9]><xml>
+<o:shapedefaults v:ext="edit" spidmax="1026" />
+</xml><![endif]--><!--[if gte mso 9]><xml>
+<o:shapelayout v:ext="edit">
+<o:idmap v:ext="edit" data="1" />
+</o:shapelayout></xml><![endif]-->
+</head>
+<body lang="EN-US" link="blue" vlink="purple">
+<div class="WordSection1">
+<p class="MsoNormal"><span style="font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1F497D">Hi. I am fine.<o:p></o:p></span></p>
+<p class="MsoNormal"><span style="font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1F497D">Thanks,<o:p></o:p></span></p>
+<p class="MsoNormal"><span style="font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1F497D">Alex<o:p></o:p></span></p>
+<p class="MsoNormal"><b><span style="font-size:10.0pt;font-family:&quot;Tahoma&quot;,&quot;sans-serif&quot;">From:</span></b><span style="font-size:10.0pt;font-family:&quot;Tahoma&quot;,&quot;sans-serif&quot;"> Foo [mailto:foo@bar.com]
+<b>On Behalf Of </b>baz@bar.com<br>
+<b>Sent:</b> Monday, January 01, 2000 12:00 AM<br>
+<b>To:</b> john@bar.com<br>
+<b>Cc:</b> jane@bar.io<br>
+<b>Subject:</b> Conversation<o:p></o:p></span></p>
+<p class="MsoNormal"><o:p>&nbsp;</o:p></p>
+<p>Hello! How are you?<o:p></o:p></p>
+<p class="MsoNormal"><o:p>&nbsp;</o:p></p>
+</div>
+</body>
+</html>

tests/fixtures/standard_replies/apple_mail_2.eml

@@ -0,0 +1,19 @@
+Content-Type: text/plain;
+	charset=us-ascii
+Mime-Version: 1.0 (Mac OS X Mail 8.2 \(2104\))
+Subject: Re: Hello there
+X-Universally-Unique-Identifier: 85B1075D-5841-46A9-8565-FCB287A93AC4
+From: Adam Renberg <adam@tictail.com>
+In-Reply-To: <CABzQGhkMXDxUt_tSVQcg=43aniUhtsVfCZVzu-PG0kwS_uzqMw@mail.gmail.com>
+Date: Sat, 22 Aug 2015 19:22:20 +0200
+Content-Transfer-Encoding: 7bit
+X-Smtp-Server: smtp.gmail.com:adam@tictail.com
+Message-Id: <68001B29-8EA4-444C-A894-0537D2CA5208@tictail.com>
+References: <CABzQGhkMXDxUt_tSVQcg=43aniUhtsVfCZVzu-PG0kwS_uzqMw@mail.gmail.com>
+To: Adam Renberg <tgwizard@gmail.com>
+
+Hello
+> On 22 Aug 2015, at 19:21, Adam Renberg <tgwizard@gmail.com> wrote:
+>
+> Hi there!
+

tests/fixtures/standard_replies/iphone.eml

@@ -9,11 +9,11 @@ To: bob <bob@example.com>
 Content-Transfer-Encoding: quoted-printable
 Mime-Version: 1.0 (1.0)
 
-hello
+Hello
 
 Sent from my iPhone
 
 On Apr 3, 2012, at 4:19 PM, bob <bob@example.com> wr=
 ote:
 
-> Hi
+> Hi

tests/fixtures/standard_replies/iphone_reply_text

@@ -0,0 +1,3 @@
+Hello
+
+Sent from my iPhone

tests/html_quotations_test.py

@@ -1,14 +1,12 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from . import *
 from . fixtures import *
 
 import regex as re
-from flanker import mime
 
-from talon import quotations
-
-import html2text
+from talon import quotations, utils as u
 
 
 RE_WHITESPACE = re.compile("\s")
@@ -46,7 +44,25 @@ def test_quotation_splitter_outside_blockquote():
   </div>
 </blockquote>
 """
-    eq_("<html><body><p>Reply</p><div></div></body></html>",
+    eq_("<html><body><p>Reply</p></body></html>",
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
+
+
+def test_regular_blockquote():
+    msg_body = """Reply
+<blockquote>Regular</blockquote>
+
+<div>
+  On 11-Apr-2011, at 6:54 PM, Bob &lt;bob@example.com&gt; wrote:
+</div>
+
+<blockquote>
+  <div>
+    <blockquote>Nested</blockquote>
+  </div>
+</blockquote>
+"""
+    eq_("<html><body><p>Reply</p><blockquote>Regular</blockquote></body></html>",
         RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
 
 
@@ -116,6 +132,29 @@ 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">
+  <div class="gmail_default">
+    My name is William Shakespeare.
+    <br/>
+  </div>
+</blockquote>"""
+    eq_(RE_WHITESPACE.sub('', msg_body),
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
+
+
 def test_unicode_in_reply():
     msg_body = u"""Reply \xa0 \xa0 Text<br>
 
@@ -123,7 +162,7 @@ def test_unicode_in_reply():
   <br>
 </div>
 
-<blockquote class="gmail_quote">
+<blockquote>
   Quote
 </blockquote>""".encode("utf-8")
 
@@ -224,10 +263,7 @@ def test_reply_shares_div_with_from_block():
 
 
 def test_reply_quotations_share_block():
-    msg = mime.from_string(REPLY_QUOTATIONS_SHARE_BLOCK)
-    html_part = list(msg.walk())[1]
-    assert html_part.content_type == 'text/html'
-    stripped_html = quotations.extract_from_html(html_part.body)
+    stripped_html = quotations.extract_from_plain(REPLY_QUOTATIONS_SHARE_BLOCK)
     ok_(stripped_html)
     ok_('From' not in stripped_html)
 
@@ -244,26 +280,36 @@ def test_reply_separated_by_hr():
             '', quotations.extract_from_html(REPLY_SEPARATED_BY_HR)))
 
 
-RE_REPLY = re.compile(r"^Hi\. I am fine\.\s*\n\s*Thanks,\s*\n\s*Alex\s*$")
+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)
 
-    msg_body = f.read().decode("utf-8")
+    msg_body = f.read()
     reply = quotations.extract_from_html(msg_body)
+    plain_reply = u.html_to_text(reply)
+    plain_reply = plain_reply.decode('utf8')
 
-    h = html2text.HTML2Text()
-    h.body_width = 0
-    plain_reply = h.handle(reply)
-
-    #remove &nbsp; spaces
-    plain_reply = plain_reply.replace(u'\xa0', u' ')
-
-    if RE_REPLY.match(plain_reply):
-        eq_(1, 1)
-    else:
-        eq_("Hi. I am fine.\n\nThanks,\nAlex", plain_reply)
+    eq_(RE_WHITESPACE.sub('', "Hi. I am fine.\n\nThanks,\nAlex"),
+        RE_WHITESPACE.sub('', plain_reply))
 
 
 def test_gmail_reply():
@@ -286,6 +332,10 @@ def test_ms_outlook_2007_reply():
     extract_reply_and_check("tests/fixtures/html_replies/ms_outlook_2007.html")
 
 
+def test_ms_outlook_2010_reply():
+    extract_reply_and_check("tests/fixtures/html_replies/ms_outlook_2010.html")
+
+
 def test_thunderbird_reply():
     extract_reply_and_check("tests/fixtures/html_replies/thunderbird.html")
 
@@ -296,3 +346,68 @@ def test_windows_mail_reply():
 
 def test_yandex_ru_reply():
     extract_reply_and_check("tests/fixtures/html_replies/yandex_ru.html")
+
+
+def test_CRLF():
+    """CR is not converted to '&#13;'
+    """
+    symbol = '&#13;'
+    extracted = quotations.extract_from_html('<html>\r\n</html>')
+    assert_false(symbol in extracted)
+    eq_('<html></html>', RE_WHITESPACE.sub('', extracted))
+
+    msg_body = """My
+reply
+<blockquote>
+
+  <div>
+    On 11-Apr-2011, at 6:54 PM, Bob &lt;bob@example.com&gt; wrote:
+  </div>
+
+  <div>
+    Test
+  </div>
+
+</blockquote>"""
+    msg_body = msg_body.replace('\n', '\r\n')
+    extracted = quotations.extract_from_html(msg_body)
+    assert_false(symbol in extracted)    
+    # Keep new lines otherwise "My reply" becomes one word - "Myreply" 
+    eq_("<html><body><p>My\nreply\n</p></body></html>", extracted)
+
+
+def test_gmail_forwarded_msg():
+    msg_body = """<div dir="ltr"><br><div class="gmail_quote">---------- Forwarded message ----------<br>From: <b class="gmail_sendername">Bob</b> <span dir="ltr">&lt;<a href="mailto:bob@example.com">bob@example.com</a>&gt;</span><br>Date: Fri, Feb 11, 2010 at 5:59 PM<br>Subject: Bob WFH today<br>To: Mary &lt;<a href="mailto:mary@example.com">mary@example.com</a>&gt;<br><br><br><div dir="ltr">eom</div>
+</div><br></div>"""
+    extracted = quotations.extract_from_html(msg_body)
+    eq_(RE_WHITESPACE.sub('', msg_body), RE_WHITESPACE.sub('', extracted))
+
+
+@patch.object(quotations, 'MAX_HTML_LEN', 1)
+def test_too_large_html():
+    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_(RE_WHITESPACE.sub('', msg_body),
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))
+
+
+def test_readable_html_empty():
+    msg_body = """
+<blockquote>
+  Reply
+  <div>
+    On 11-Apr-2011, at 6:54 PM, Bob &lt;bob@example.com&gt; wrote:
+  </div>
+
+  <div>
+    Test
+  </div>
+
+</blockquote>"""
+
+    eq_(RE_WHITESPACE.sub('', msg_body),
+        RE_WHITESPACE.sub('', quotations.extract_from_html(msg_body)))

tests/quotations_test.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from . import *
 from . fixtures import *
 
-from flanker import mime
-
 from talon import quotations
 
 

tests/signature/bruteforce_test.py

@@ -1,11 +1,8 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from .. import *
 
-import os
-
-from flanker import mime
-
 from talon.signature import bruteforce
 
 

tests/signature/extraction_test.py

@@ -1,15 +1,15 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from .. import *
 
 import os
 
-from PyML import SparseDataSet
-
 from talon.signature.learning import dataset
 from talon import signature
 from talon.signature import extraction as e
 from talon.signature import bruteforce
+from six.moves import range
 
 
 def test_message_shorter_SIGNATURE_MAX_LINES():
@@ -77,6 +77,31 @@ def test_basic():
         signature.extract(msg_body, 'Sergey'))
 
 
+def test_capitalized():
+    msg_body = """Hi Mary,
+
+Do you still need a DJ for your wedding? I've included a video demo of one of our DJs available for your wedding date.
+
+DJ Doe 
+http://example.com
+Password: SUPERPASSWORD
+
+Would you like to check out more?
+
+
+At your service,
+
+John Smith
+Doe Inc
+555-531-7967"""
+
+    sig = """John Smith
+Doe Inc
+555-531-7967"""
+
+    eq_(sig, signature.extract(msg_body, 'Doe')[1])
+
+
 def test_over_2_text_lines_after_signature():
     body = """Blah
 
@@ -129,20 +154,20 @@ def test_mark_lines():
 
 def test_process_marked_lines():
     # no signature found
-    eq_((range(5), None), e._process_marked_lines(range(5), 'telt'))
+    eq_((list(range(5)), None), e._process_marked_lines(list(range(5)), 'telt'))
 
     # signature in the middle of the text
-    eq_((range(9), None), e._process_marked_lines(range(9), 'tesestelt'))
+    eq_((list(range(9)), None), e._process_marked_lines(list(range(9)), 'tesestelt'))
 
     # long line splits signature
-    eq_((range(7), [7, 8]),
-        e._process_marked_lines(range(9), 'tsslsless'))
+    eq_((list(range(7)), [7, 8]),
+        e._process_marked_lines(list(range(9)), 'tsslsless'))
 
-    eq_((range(20), [20]),
-        e._process_marked_lines(range(21), 'ttttttstttesllelelets'))
+    eq_((list(range(20)), [20]),
+        e._process_marked_lines(list(range(21)), 'ttttttstttesllelelets'))
 
     # some signature lines could be identified as text
-    eq_(([0], range(1, 9)), e._process_marked_lines(range(9), 'tsetetest'))
+    eq_(([0], list(range(1, 9))), e._process_marked_lines(list(range(9)), 'tsetetest'))
 
-    eq_(([], range(5)),
-        e._process_marked_lines(range(5), "ststt"))
+    eq_(([], list(range(5))),
+        e._process_marked_lines(list(range(5)), "ststt"))

tests/signature/learning/dataset_test.py

@@ -1,11 +1,11 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from ... import *
 import os
 
-from PyML import SparseDataSet
+from numpy import genfromtxt
 
-from talon.utils import to_unicode
 from talon.signature.learning import dataset as d
 
 from talon.signature.learning.featurespace import features
@@ -42,10 +42,13 @@ def test_build_extraction_dataset():
     d.build_extraction_dataset(os.path.join(EMAILS_DIR, 'P'),
                                os.path.join(TMP_DIR,
                                             'extraction.data'), 1)
-    test_data = SparseDataSet(os.path.join(TMP_DIR, 'extraction.data'),
-                              labelsColumn=-1)
+
+    filename = os.path.join(TMP_DIR, 'extraction.data')
+    file_data = genfromtxt(filename, delimiter=",")
+    test_data = file_data[:, :-1]
+
     # the result is a loadable signature extraction dataset
     # 32 comes from 3 emails in emails/P folder, 11 lines checked to be
     # a signature, one email has only 10 lines
-    eq_(test_data.size(), 32)
-    eq_(len(features('')), test_data.numFeatures)
+    eq_(test_data.shape[0], 32)
+    eq_(len(features('')), test_data.shape[1])

tests/signature/learning/featurespace_test.py

@@ -1,12 +1,15 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from ... import *
 
 from talon.signature.learning import featurespace as fs
 
 
 def test_apply_features():
-    s = '''John Doe
+    s = '''This is John Doe
+
+Tuesday @3pm suits. I'll chat to you then.
 
 VP Research and Development, Xxxx Xxxx Xxxxx
 
@@ -19,11 +22,12 @@ john@example.com'''
     # note that we don't consider the first line because signatures don't
     # usually take all the text, empty lines are not considered
     eq_(result, [[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
+                 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
                  [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
                  [0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
                  [0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0]])
 
-    with patch.object(fs, 'SIGNATURE_MAX_LINES', 4):
+    with patch.object(fs, 'SIGNATURE_MAX_LINES', 5):
         features = fs.features(sender)
         new_result = fs.apply_features(s, features)
         # result remains the same because we don't consider empty lines

tests/signature/learning/helpers_test.py

@@ -1,11 +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 *
+from six.moves import range
 
 # First testing regex constants.
 VALID = '''
@@ -43,7 +45,7 @@ VALID_PHONE_NUMBERS = [e.strip() for e in VALID.splitlines() if e.strip()]
 
 def test_match_phone_numbers():
     for phone in VALID_PHONE_NUMBERS:
-        ok_(RE_RELAX_PHONE.match(phone), "{} should be matched".format(phone))
+        ok_(RE_RELAX_PHONE.search(phone), "{} should be matched".format(phone))
 
 
 def test_match_names():
@@ -52,29 +54,6 @@ def test_match_names():
         ok_(RE_NAME.match(name), "{} should be matched".format(name))
 
 
-def test_sender_with_name():
-    ok_lines = ['Sergey Obukhov <serobnic@example.com>',
-                '\tSergey  <serobnic@example.com>',
-                ('"Doe, John (TX)"'
-                 '<DowJ@example.com>@EXAMPLE'
-                 '<IMCEANOTES-+22Doe+2C+20John+20'
-                 '+28TX+29+22+20+3CDoeJ+40example+2Ecom+3E'
-                 '+40EXAMPLE@EXAMPLE.com>'),
-                ('Company Sleuth <csleuth@email.xxx.com>'
-                 '@EXAMPLE <XXX-Company+20Sleuth+20+3Ccsleuth'
-                 '+40email+2Exxx+2Ecom+3E+40EXAMPLE@EXAMPLE.com>'),
-                ('Doe III, John '
-                 '</O=EXAMPLE/OU=NA/CN=RECIPIENTS/CN=jDOE5>')]
-    for line in ok_lines:
-        ok_(RE_SENDER_WITH_NAME.match(line),
-            '{} should be matched'.format(line))
-
-    nok_lines = ['', '<serobnic@xxx.ru>', 'Sergey serobnic@xxx.ru']
-    for line in nok_lines:
-        assert_false(RE_SENDER_WITH_NAME.match(line),
-                     '{} should not be matched'.format(line))
-
-
 # Now test helpers functions
 def test_binary_regex_search():
     eq_(1, h.binary_regex_search(re.compile("12"))("12"))
@@ -177,7 +156,7 @@ def test_extract_names():
         # check that extracted names could be compiled
         try:
             re.compile("|".join(extracted_names))
-        except Exception, e:
+        except Exception as e:
             ok_(False, ("Failed to compile extracted names {}"
                         "\n\nReason: {}").format(extracted_names, e))
         if expected_names:
@@ -213,10 +192,11 @@ def test_punctuation_percent(categories_percent):
 def test_capitalized_words_percent():
     eq_(0.0, h.capitalized_words_percent(''))
     eq_(100.0, h.capitalized_words_percent('Example Corp'))
-    eq_(50.0, h.capitalized_words_percent('Qqq qqq QQQ 123 sss'))
+    eq_(50.0, h.capitalized_words_percent('Qqq qqq Aqs 123 sss'))
     eq_(100.0, h.capitalized_words_percent('Cell 713-444-7368'))
     eq_(100.0, h.capitalized_words_percent('8th Floor'))
     eq_(0.0, h.capitalized_words_percent('(212) 230-9276'))
+    eq_(50.0, h.capitalized_words_percent('Password: REMARKABLE'))
 
 
 def test_has_signature():
@@ -227,7 +207,7 @@ def test_has_signature():
                         'sender@example.com'))
     assert_false(h.has_signature('http://www.example.com/555-555-5555',
                                  'sender@example.com'))
-    long_line = ''.join(['q' for e in xrange(28)])
+    long_line = ''.join(['q' for e in range(28)])
     assert_false(h.has_signature(long_line + ' sender', 'sender@example.com'))
     # wont crash on an empty string
     assert_false(h.has_signature('', ''))

tests/text_quotations_test.py

@@ -1,23 +1,26 @@
 # -*- coding: utf-8 -*-
 
+from __future__ import absolute_import
 from . import *
 from . fixtures import *
 
 import os
 
-from flanker import mime
-
+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():
     msg_body = """Test reply
-
+Hi
 -----Original Message-----
 
 Test"""
-    eq_(msg_body, quotations.extract_from_plain(msg_body))
+    eq_("Test reply", quotations.extract_from_plain(msg_body))
 
 
 def test_pattern_on_date_somebody_wrote():
@@ -33,6 +36,19 @@ 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_sent_from_samsung_smb_wrote():
+    msg_body = """Test reply
+
+Sent from Samsung MobileName <address@example.com> wrote:
+
+>
+> Test
+>
+> Roman"""
+
+    eq_("Test reply", quotations.extract_from_plain(msg_body))
+
+
 def test_pattern_on_date_wrote_somebody():
     eq_('Lorem', quotations.extract_from_plain(
     """Lorem
@@ -55,6 +71,18 @@ On 04/19/2011 07:10 AM, Roman Tkachenko wrote:
     eq_("Test reply", quotations.extract_from_plain(msg_body))
 
 
+def test_date_time_email_splitter():
+    msg_body = """Test reply
+
+2014-10-17 11:28 GMT+03:00 Postmaster <
+postmaster@sandboxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.mailgun.org>:
+
+> First from site
+>
+    """
+    eq_("Test reply", quotations.extract_from_plain(msg_body))
+
+
 def test_pattern_on_date_somebody_wrote_allows_space_in_front():
     msg_body = """Thanks Thanmai
  On Mar 8, 2012 9:59 AM, "Example.com" <
@@ -114,7 +142,7 @@ def _check_pattern_original_message(original_message_indicator):
 -----{}-----
 
 Test"""
-    eq_('Test reply', quotations.extract_from_plain(msg_body.format(unicode(original_message_indicator))))
+    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')
@@ -312,6 +340,33 @@ 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!
+Från: Anno Sportel [mailto:anno.spoel@hsbcssad.com]
+Skickat: den 26 augusti 2015 14:45
+Till: Isacson Leiff
+Ämne: RE: Week 36
+
+Blah-blah-blah
+"""))
+
+def test_swedish_from_line():
+    eq_('Lorem', quotations.extract_from_plain(
+    """Lorem
+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
+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. 
@@ -611,25 +666,33 @@ def test_preprocess_postprocess_2_links():
     eq_(msg_body, quotations.extract_from_plain(msg_body))
 
 
+def body_iterator(msg, decode=False):
+    for subpart in msg.walk():
+        payload = subpart.get_payload(decode=decode)
+        if isinstance(payload, six.text_type):
+            yield payload
+        else:
+            yield payload.decode('utf8')
+
+
 def test_standard_replies():
     for filename in os.listdir(STANDARD_REPLIES):
         filename = os.path.join(STANDARD_REPLIES, filename)
-        if os.path.isdir(filename):
+        if not filename.endswith('.eml') or os.path.isdir(filename):
             continue
         with open(filename) as f:
-            msg = f.read()
-            m = mime.from_string(msg)
-            for part in m.walk():
-                if part.content_type == 'text/plain':
-                    text = part.body
-                    stripped_text = quotations.extract_from_plain(text)
-                    reply_text_fn = filename[:-4] + '_reply_text'
-                    if os.path.isfile(reply_text_fn):
-                        with open(reply_text_fn) as f:
-                            reply_text = f.read()
-                    else:
-                        reply_text = 'Hello'
-                    eq_(reply_text, stripped_text,
-                        "'%(reply)s' != %(stripped)s for %(fn)s" %
-                        {'reply': reply_text, 'stripped': stripped_text,
-                         'fn': filename})
+            message = email.message_from_file(f)
+            body = next(email.iterators.typed_subpart_iterator(message, subtype='plain'))
+            text = ''.join(body_iterator(body, True))
+
+            stripped_text = quotations.extract_from_plain(text)
+            reply_text_fn = filename[:-4] + '_reply_text'
+            if os.path.isfile(reply_text_fn):
+                with open(reply_text_fn) as f:
+                    reply_text = f.read().strip()
+            else:
+                reply_text = 'Hello'
+            yield eq_, reply_text, stripped_text, \
+                "'%(reply)s' != %(stripped)s for %(fn)s" % \
+                {'reply': reply_text, 'stripped': stripped_text,
+                 'fn': filename}

tests/utils_test.py

@@ -1,9 +1,109 @@
+# coding:utf-8
+
+from __future__ import absolute_import
 from . import *
 
-from talon import utils
+from talon import utils as u
+import cchardet
+import six
 
 
 def test_get_delimiter():
-    eq_('\r\n', utils.get_delimiter('abc\r\n123'))
-    eq_('\n', utils.get_delimiter('abc\n123'))
-    eq_('\n', utils.get_delimiter('abc'))
+    eq_('\r\n', u.get_delimiter('abc\r\n123'))
+    eq_('\n', u.get_delimiter('abc\n123'))
+    eq_('\n', u.get_delimiter('abc'))
+
+
+def test_unicode():
+    eq_ (u'hi', u.to_unicode('hi'))
+    eq_ (type(u.to_unicode('hi')), six.text_type )
+    eq_ (type(u.to_unicode(u'hi')), six.text_type )
+    eq_ (type(u.to_unicode('привет')), six.text_type )
+    eq_ (type(u.to_unicode(u'привет')), six.text_type )
+    eq_ (u"привет", u.to_unicode('привет'))
+    eq_ (u"привет", u.to_unicode(u'привет'))
+    # some latin1 stuff
+    eq_ (u"Versión", u.to_unicode(u'Versi\xf3n'.encode('iso-8859-2'), precise=True))
+
+
+def test_detect_encoding():
+    eq_ ('ascii', u.detect_encoding(b'qwe').lower())
+    eq_ ('iso-8859-2', u.detect_encoding(u'Versi\xf3n'.encode('iso-8859-2')).lower())
+    eq_ ('utf-8', u.detect_encoding(u'привет'.encode('utf8')).lower())
+    # fallback to utf-8
+    with patch.object(u.chardet, 'detect') as detect:
+        detect.side_effect = Exception
+        eq_ ('utf-8', u.detect_encoding('qwe'.encode('utf8')).lower())
+
+
+def test_quick_detect_encoding():
+    eq_ ('ascii', u.quick_detect_encoding(b'qwe').lower())
+    eq_ ('windows-1252', u.quick_detect_encoding(u'Versi\xf3n'.encode('windows-1252')).lower())
+    eq_ ('utf-8', u.quick_detect_encoding(u'привет'.encode('utf8')).lower())
+
+
+@patch.object(cchardet, 'detect')
+@patch.object(u, 'detect_encoding')
+def test_quick_detect_encoding_edge_cases(detect_encoding, cchardet_detect):
+    cchardet_detect.return_value = {'encoding': 'ascii'}
+    eq_('ascii', u.quick_detect_encoding(b"qwe"))
+    cchardet_detect.assert_called_once_with(b"qwe")
+
+    # fallback to detect_encoding
+    cchardet_detect.return_value = {}
+    detect_encoding.return_value = 'utf-8'
+    eq_('utf-8', u.quick_detect_encoding(b"qwe"))
+
+    # exception
+    detect_encoding.reset_mock()
+    cchardet_detect.side_effect = Exception()
+    detect_encoding.return_value = 'utf-8'
+    eq_('utf-8', u.quick_detect_encoding(b"qwe"))
+    ok_(detect_encoding.called)
+
+
+def test_html_to_text():
+    html = """<body>
+<p>Hello world!</p>
+<br>
+<ul>
+<li>One!</li>
+<li>Two</li>
+</ul>
+<p>
+Haha
+</p>
+</body>"""
+    text = u.html_to_text(html)
+    eq_(b"Hello world! \n\n  * One! \n  * Two \nHaha", text)
+    eq_(u"привет!", u.html_to_text("<b>привет!</b>").decode('utf8'))
+
+    html = '<body><br/><br/>Hi</body>'
+    eq_ (b'Hi', u.html_to_text(html))
+
+    html = """Hi
+<style type="text/css">
+
+div, p, li {
+
+font: 13px 'Lucida Grande', Arial, sans-serif;
+
+}
+</style>
+
+<style type="text/css">
+
+h1 {
+
+font: 13px 'Lucida Grande', Arial, sans-serif;
+
+}
+</style>"""
+    eq_ (b'Hi', u.html_to_text(html))
+
+    html = """<div>
+<!-- COMMENT 1 -->
+<span>TEXT 1</span>
+<p>TEXT 2 <!-- COMMENT 2 --></p>
+</div>"""
+    eq_(b'TEXT 1 \nTEXT 2', u.html_to_text(html))

train.py

@@ -0,0 +1,11 @@
+from __future__ import absolute_import
+from talon.signature import EXTRACTOR_FILENAME, EXTRACTOR_DATA
+from talon.signature.learning.classifier import train, init
+
+
+def train_model():
+    """ retrain model and persist """
+    train(init(), EXTRACTOR_DATA, EXTRACTOR_FILENAME)
+
+if __name__ == "__main__":
+    train_model()