Operate on unicode data exclusively

setup.py

@@ -29,7 +29,7 @@ class InstallCommand(install):
 
 
 setup(name='talon',
-      version='1.5.0',
+      version='1.6.0',
       description=("Mailgun library "
                    "to extract message quotations and signatures."),
       long_description=open("README.rst").read(),

talon/quotations.py

@@ -6,18 +6,17 @@ 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
-
-from talon.utils import (get_delimiter, html_tree_to_text,
-                         html_document_fromstring)
-from talon import html_quotations
+import regex as re
+from lxml import etree, html
 from six.moves import range
-import six
 
+from talon import html_quotations
+from talon.utils import (get_delimiter, html_document_fromstring,
+                         html_tree_to_text)
 
 log = logging.getLogger(__name__)
 
@@ -94,7 +93,7 @@ RE_ON_DATE_WROTE_SMB = re.compile(
     )
 
 RE_QUOTATION = re.compile(
-    r'''
+    r"""
     (
         # quotation border: splitter line or a number of quotation marker lines
         (?:
@@ -112,10 +111,10 @@ RE_QUOTATION = re.compile(
 
     # after quotations should be text only or nothing at all
     [te]*$
-    ''', re.VERBOSE)
+    """, re.VERBOSE)
 
 RE_EMPTY_QUOTATION = re.compile(
-    r'''
+    r"""
     (
         # quotation border: splitter line or a number of quotation marker lines
         (?:
@@ -125,7 +124,7 @@ RE_EMPTY_QUOTATION = re.compile(
         )
     )
     e*
-    ''', re.VERBOSE)
+    """, re.VERBOSE)
 
 # ------Original Message------ or ---- Reply Message ----
 # With variations in other languages.
@@ -343,9 +342,6 @@ def _replace_link_brackets(msg_body):
 
     Converts msg_body into a unicode
     """
-    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] == ">":
@@ -385,8 +381,6 @@ def postprocess(msg_body):
 
 def extract_from_plain(msg_body):
     """Extracts a non quoted message from provided plain text."""
-    stripped_text = msg_body
-
     delimiter = get_delimiter(msg_body)
     msg_body = preprocess(msg_body, delimiter)
     # don't process too long messages
@@ -418,17 +412,13 @@ def extract_from_html(msg_body):
 
     Returns a unicode string.
     """
-    msg_body_bytes = msg_body
-    if isinstance(msg_body, six.text_type):
-        msg_body_bytes = msg_body.encode('utf8')
-
-    if msg_body_bytes.strip() == b'':
+    if msg_body.strip() == "":
         return msg_body
 
-    msg_body_bytes = msg_body_bytes.replace(b'\r\n', b'\n')
+    msg_body = msg_body.replace("\r\n", "\n")
     # Cut out xml and doctype tags to avoid conflict with unicode decoding.
-    msg_body_bytes = re.sub(br"\<\?xml.+\?\>|\<\!DOCTYPE.+]\>", b"", msg_body_bytes)
-    html_tree = html_document_fromstring(msg_body_bytes)
+    msg_body = re.sub(r"\<\?xml.+\?\>|\<\!DOCTYPE.+]\>", "", msg_body)
+    html_tree = html_document_fromstring(msg_body)
     if html_tree is None:
         return msg_body
 
@@ -531,11 +521,11 @@ def extract_from_html_tree(html_tree):
     #    of replacing data outside the <tag> which might be essential to
     #    the customer.
     remove_namespaces(html_tree_copy)
-    s = html.tostring(html_tree_copy)
+    s = html.tostring(html_tree_copy, encoding="ascii")
     if not s:
         return None
 
-    return s.decode('utf-8')
+    return s.decode("ascii")
 
 
 def remove_namespaces(root):
@@ -654,10 +644,10 @@ def _readable_text_empty(html_tree):
 
 
 def is_splitter(line):
-    '''
+    """
     Returns Matcher object if provided string is a splitter and
     None otherwise.
-    '''
+    """
     for pattern in SPLITTER_PATTERNS:
         matcher = re.match(pattern, line)
         if matcher:
@@ -665,12 +655,12 @@ def is_splitter(line):
 
 
 def text_content(context):
-    '''XPath Extension function to return a node text content.'''
+    """XPath Extension function to return a node text content."""
     return context.context_node.xpath("string()").strip()
 
 
 def tail(context):
-    '''XPath Extension function to return a node tail text.'''
+    """XPath Extension function to return a node tail text."""
     return context.context_node.tail or ''
 
 

talon/signature/learning/helpers.py

@@ -5,21 +5,17 @@
 * regexp's constants used when evaluating signature's features
 
 """
-
-from __future__ import absolute_import
 import unicodedata
+
 import regex as re
 
-from talon.utils import to_unicode
-
 from talon.signature.constants import SIGNATURE_MAX_LINES
 
-
 rc = re.compile
 
 RE_EMAIL = rc('\S@\S')
 RE_RELAX_PHONE = rc('(\(? ?[\d]{2,3} ?\)?.{,3}?){2,}')
-RE_URL = rc(r'''https?://|www\.[\S]+\.[\S]''')
+RE_URL = rc(r"""https?://|www\.[\S]+\.[\S]""")
 
 # Taken from:
 # http://www.cs.cmu.edu/~vitor/papers/sigFilePaper_finalversion.pdf
@@ -55,7 +51,7 @@ BAD_SENDER_NAMES = [
 
 
 def binary_regex_search(prog):
-    '''Returns a function that returns 1 or 0 depending on regex search result.
+    """Returns a function that returns 1 or 0 depending on regex search result.
 
     If regular expression compiled into prog is present in a string
     the result of calling the returned function with the string will be 1
@@ -66,12 +62,12 @@ def binary_regex_search(prog):
     1
     >>> binary_regex_search(re.compile("12"))("34")
     0
-    '''
+    """
     return lambda s: 1 if prog.search(s) else 0
 
 
 def binary_regex_match(prog):
-    '''Returns a function that returns 1 or 0 depending on regex match result.
+    """Returns a function that returns 1 or 0 depending on regex match result.
 
     If a string matches regular expression compiled into prog
     the result of calling the returned function with the string will be 1
@@ -82,7 +78,7 @@ def binary_regex_match(prog):
     1
     >>> binary_regex_match(re.compile("12"))("3 12")
     0
-    '''
+    """
     return lambda s: 1 if prog.match(s) else 0
 
 
@@ -135,7 +131,6 @@ def extract_names(sender):
     >>> extract_names('')
     []
     """
-    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.
@@ -154,7 +149,7 @@ def extract_names(sender):
 
 
 def categories_percent(s, categories):
-    '''Returns category characters percent.
+    """Returns category characters percent.
 
     >>> categories_percent("qqq ggg hhh", ["Po"])
     0.0
@@ -166,9 +161,8 @@ def categories_percent(s, categories):
     50.0
     >>> categories_percent("s.s,5s", ["Po", "Nd"])
     50.0
-    '''
+    """
     count = 0
-    s = to_unicode(s, precise=True)
     for c in s:
         if unicodedata.category(c) in categories:
             count += 1
@@ -176,19 +170,18 @@ def categories_percent(s, categories):
 
 
 def punctuation_percent(s):
-    '''Returns punctuation percent.
+    """Returns punctuation percent.
 
     >>> punctuation_percent("qqq ggg hhh")
     0.0
     >>> punctuation_percent("q,w.")
     50.0
-    '''
+    """
     return categories_percent(s, ['Po'])
 
 
 def capitalized_words_percent(s):
-    '''Returns capitalized words percent.'''
-    s = to_unicode(s, precise=True)
+    """Returns capitalized words percent."""
     words = re.split('\s', s)
     words = [w for w in words if w.strip()]
     words = [w for w in words if len(w) > 2]    

talon/utils.py

@@ -1,110 +1,17 @@
 # coding:utf-8
+from __future__ import annotations
 
-from __future__ import absolute_import
-
-from random import shuffle
-
-import cchardet
-import chardet
 import html5lib
 import regex as re
-import six
+from html5lib import HTMLParser
 from lxml.cssselect import CSSSelector
+from lxml.etree import _Element
 from lxml.html import html5parser
 
 from talon.constants import RE_DELIMITER
 
 
-def safe_format(format_string, *args, **kwargs):
-    """
-    Helper: formats string with any combination of bytestrings/unicode
-    strings without raising exceptions
-    """
-    try:
-        if not args and not kwargs:
-            return format_string
-        else:
-            return format_string.format(*args, **kwargs)
-
-    # catch encoding errors and transform everything into utf-8 string
-    # before logging:
-    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 six.iteritems(kwargs)}
-        return format_string.format(*args, **kwargs)
-
-    # ignore other errors
-    except:
-        return u''
-
-
-def to_unicode(str_or_unicode, precise=False):
-    """
-    Safely returns a unicode version of a given string
-    >>> utils.to_unicode('привет')
-        u'привет'
-    >>> utils.to_unicode(u'привет')
-        u'привет'
-    If `precise` flag is True, tries to guess the correct encoding first.
-    """
-    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 not isinstance(str_or_unicode, six.text_type):
-        return str_or_unicode.encode("utf-8", "ignore")
-    return str(str_or_unicode)
-
-
-def random_token(length=7):
-    vals = ("a b c d e f g h i j k l m n o p q r s t u v w x y z "
-            "0 1 2 3 4 5 6 7 8 9").split(' ')
-    shuffle(vals)
-    return ''.join(vals[:length])
-
-
-def get_delimiter(msg_body):
+def get_delimiter(msg_body: str) -> str:
     delimiter = RE_DELIMITER.search(msg_body)
     if delimiter:
         delimiter = delimiter.group()
@@ -114,7 +21,7 @@ def get_delimiter(msg_body):
     return delimiter
 
 
-def html_tree_to_text(tree):
+def html_tree_to_text(tree: _Element) -> str:
     for style in CSSSelector('style')(tree):
         style.getparent().remove(style)
 
@@ -146,26 +53,22 @@ def html_tree_to_text(tree):
             not text.endswith("\n") and not el_text):
             text += "\n"
 
-    retval = _rm_excessive_newlines(text)
-    return _encode_utf8(retval)
+    text = _rm_excessive_newlines(text)
+    return text
 
 
-def html_to_text(string):
+def html_to_text(s: str) -> str | None:
     """
     Dead-simple HTML-to-text converter:
         >>> html_to_text("one<br>two<br>three")
-        >>> "one\ntwo\nthree"
+        <<< "one\ntwo\nthree"
 
     NOTES:
         1. the string is expected to contain UTF-8 encoded HTML!
-        2. returns utf-8 encoded str (not unicode)
         3. if html can't be parsed returns None
     """
-    if isinstance(string, six.text_type):
-        string = string.encode('utf8')
-
-    s = _prepend_utf8_declaration(string)
-    s = s.replace(b"\n", b"")
+    s = _prepend_utf8_declaration(s)
+    s = s.replace("\n", "")
     tree = html_fromstring(s)
 
     if tree is None:
@@ -174,62 +77,46 @@ def html_to_text(string):
     return html_tree_to_text(tree)
 
 
-def html_fromstring(s):
+def html_fromstring(s: str) -> _Element:
     """Parse html tree from string. Return None if the string can't be parsed.
     """
-    if isinstance(s, six.text_type):
-        s = s.encode('utf8')
-    try:
-        return html5parser.fromstring(s, parser=_html5lib_parser())
-    except Exception:
-        pass
+    return html5parser.fromstring(s, parser=_html5lib_parser())
 
 
-def html_document_fromstring(s):
+def html_document_fromstring(s: str) -> _Element:
     """Parse html tree from string. Return None if the string can't be parsed.
     """
-    if isinstance(s, six.text_type):
-        s = s.encode('utf8')
-    try:
-        return html5parser.document_fromstring(s, parser=_html5lib_parser())
-    except Exception:
-        pass
+    return html5parser.document_fromstring(s, parser=_html5lib_parser())
 
 
-def cssselect(expr, tree):
+def cssselect(expr: str, tree: str) -> list[_Element]:
     return CSSSelector(expr)(tree)
 
 
-def _contains_charset_spec(s):
+def _contains_charset_spec(s: str) -> str:
     """Return True if the first 4KB contain charset spec
     """
-    return s.lower().find(b'html; charset=', 0, 4096) != -1
+    return s.lower().find('html; charset=', 0, 4096) != -1
 
 
-def _prepend_utf8_declaration(s):
+def _prepend_utf8_declaration(s: str) -> str:
     """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):
+def _rm_excessive_newlines(s: str) -> str:
     """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
-
-
-def _html5lib_parser():
+def _html5lib_parser() -> HTMLParser:
     """
     html5lib is a pure-python library that conforms to the WHATWG HTML spec
     and is not vulnarable to certain attacks common for XML libraries
     """
-    return html5lib.HTMLParser(
+    return HTMLParser(
         # build lxml tree
         html5lib.treebuilders.getTreeBuilder("lxml"),
         # remove namespace value from inside lxml.html.html5paser element tag
@@ -239,8 +126,8 @@ def _html5lib_parser():
     )
 
 
-_UTF8_DECLARATION = (b'<meta http-equiv="Content-Type" content="text/html;'
-                     b'charset=utf-8">')
+_UTF8_DECLARATION = ('<meta http-equiv="Content-Type" content="text/html;'
+                     'charset=utf-8">')
 
 _BLOCKTAGS = ['div', 'p', 'ul', 'li', 'h1', 'h2', 'h3']
 _HARDBREAKS = ['br', 'hr', 'tr']

tests/html_quotations_test.py

@@ -4,14 +4,17 @@ from __future__ import absolute_import
 
 # noinspection PyUnresolvedReferences
 import re
+from unittest.mock import Mock, patch
 
+from nose.tools import assert_false, assert_true, eq_, ok_
+
+from tests.fixtures import (OLK_SRC_BODY_SECTION,
+                            REPLY_QUOTATIONS_SHARE_BLOCK,
+                            REPLY_SEPARATED_BY_HR)
 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")
+RE_WHITESPACE = re.compile(r"\s")
+RE_DOUBLE_WHITESPACE = re.compile(r"\s")
 
 
 def test_quotation_splitter_inside_blockquote():
@@ -166,7 +169,7 @@ def test_unicode_in_reply():
 
 <blockquote>
   Quote
-</blockquote>""".encode("utf-8")
+</blockquote>"""
 
     eq_("<html><head></head><body>Reply&#160;&#160;Text<br><div><br></div>"
         "</body></html>",
@@ -314,7 +317,6 @@ def extract_reply_and_check(filename):
     msg_body = f.read()
     reply = quotations.extract_from_html(msg_body)
     plain_reply = u.html_to_text(reply)
-    plain_reply = plain_reply.decode('utf8')
 
     eq_(RE_WHITESPACE.sub('', "Hi. I am fine.\n\nThanks,\nAlex"),
         RE_WHITESPACE.sub('', plain_reply))

tests/utils_test.py

@@ -2,9 +2,6 @@
 
 from __future__ import absolute_import
 
-import cchardet
-import six
-
 from talon import utils as u
 from . import *
 
@@ -15,58 +12,6 @@ def test_get_delimiter():
     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())
-    ok_(u.detect_encoding(
-        u'Versi\xf3n'.encode('iso-8859-2')).lower() in [
-            'iso-8859-1', 'iso-8859-2'])
-    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())
-    ok_(u.quick_detect_encoding(
-        u'Versi\xf3n'.encode('windows-1252')).lower() in [
-            'windows-1252', 'windows-1250'])
-    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>
@@ -80,11 +25,11 @@ 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'))
+    eq_("Hello world! \n\n  * One! \n  * Two \nHaha", text)
+    eq_(u"привет!", u.html_to_text("<b>привет!</b>"))
 
     html = '<body><br/><br/>Hi</body>'
-    eq_(b'Hi', u.html_to_text(html))
+    eq_('Hi', u.html_to_text(html))
 
     html = """Hi
 <style type="text/css">
@@ -104,31 +49,20 @@ font: 13px 'Lucida Grande', Arial, sans-serif;
 
 }
 </style>"""
-    eq_(b'Hi', u.html_to_text(html))
+    eq_('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))
+    eq_('TEXT 1 \nTEXT 2', u.html_to_text(html))
 
 
 def test_comment_no_parent():
-    s = b'<!-- COMMENT 1 --> no comment'
+    s = '<!-- COMMENT 1 --> no comment'
     d = u.html_document_fromstring(s)
-    eq_(b"no comment", u.html_tree_to_text(d))
-
-
-@patch.object(u.html5parser, 'fromstring', Mock(side_effect=Exception()))
-def test_html_fromstring_exception():
-    eq_(None, u.html_fromstring("<html></html>"))
-
-
-@patch.object(u.html5parser, 'document_fromstring')
-def test_html_document_fromstring_exception(document_fromstring):
-    document_fromstring.side_effect = Exception()
-    eq_(None, u.html_document_fromstring("<html></html>"))
+    eq_("no comment", u.html_tree_to_text(d))
 
 
 @patch.object(u, 'html_fromstring', Mock(return_value=None))