From e5221c26bec10892bf4b7bf367d1579078825815 Mon Sep 17 00:00:00 2001 From: Mike Fiedler Date: Tue, 5 Sep 2023 16:04:14 -0400 Subject: [PATCH] refactor: replace bleach with nh3 (ammonia) (#295) --- pyproject.toml | 2 +- readme_renderer/clean.py | 128 +++++++-------------- readme_renderer/txt.py | 2 +- tests/fixtures/test_CommonMark_006.html | 2 +- tests/fixtures/test_CommonMark_007.html | 4 +- tests/fixtures/test_CommonMark_008.html | 8 +- tests/fixtures/test_GFM_019.html | 2 +- tests/fixtures/test_GFM_020.html | 2 +- tests/fixtures/test_GFM_021.html | 2 +- tests/fixtures/test_GFM_022.html | 4 +- tests/fixtures/test_GFM_024.html | 14 +-- tests/fixtures/test_GFM_doublequotes.html | 2 +- tests/fixtures/test_GFM_malicious_pre.html | 4 +- tests/fixtures/test_rst_008.html | 4 +- tests/fixtures/test_rst_bibtex.html | 2 +- tests/fixtures/test_rst_docinfo.html | 2 +- tests/fixtures/test_rst_linkify.html | 2 +- tests/fixtures/test_txt_001.html | 2 +- tests/test_clean.py | 4 +- tox.ini | 1 + 20 files changed, 75 insertions(+), 118 deletions(-) diff --git a/pyproject.toml b/pyproject.toml index 9e871da..8f3f5d4 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -7,7 +7,7 @@ authors = [ ] readme = "README.rst" license = {text = "Apache License, Version 2.0"} -dependencies = ["bleach>=2.1.0", "docutils>=0.13.1", "Pygments>=2.5.1"] +dependencies = ["nh3>=0.2.14", "docutils>=0.13.1", "Pygments>=2.5.1"] classifiers = [ "Intended Audience :: Developers", "License :: OSI Approved :: Apache Software License", diff --git a/readme_renderer/clean.py b/readme_renderer/clean.py index 29324d3..b081e28 100644 --- a/readme_renderer/clean.py +++ b/readme_renderer/clean.py @@ -12,16 +12,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -import functools -from typing import Any, Dict, Iterator, List, Optional +from typing import Dict, Optional, Set -import bleach -import bleach.callbacks -import bleach.linkifier -import bleach.sanitizer +import nh3 -ALLOWED_TAGS = [ +ALLOWED_TAGS = { # Bleach Defaults "a", "abbr", "acronym", "b", "blockquote", "code", "em", "i", "li", "ol", "strong", "ul", @@ -32,102 +28,62 @@ "span", "sub", "summary", "sup", "table", "tbody", "td", "th", "thead", "tr", "tt", "kbd", "var", "input", "section", "aside", "nav", "s", "figure", "figcaption", -] +} ALLOWED_ATTRIBUTES = { # Bleach Defaults - "a": ["href", "title"], - "abbr": ["title"], - "acronym": ["title"], + "a": {"href", "title"}, + "abbr": {"title"}, + "acronym": {"title"}, # Custom Additions - "*": ["id"], - "hr": ["class"], - "img": ["src", "width", "height", "alt", "align", "class"], - "span": ["class"], - "th": ["align", "class"], - "td": ["align", "colspan", "rowspan"], - "div": ["align", "class"], - "h1": ["align"], - "h2": ["align"], - "h3": ["align"], - "h4": ["align"], - "h5": ["align"], - "h6": ["align"], - "code": ["class"], - "p": ["align", "class"], - "pre": ["lang"], - "ol": ["start"], - "input": ["type", "checked", "disabled"], - "aside": ["class"], - "dd": ["class"], - "dl": ["class"], - "dt": ["class"], - "ul": ["class"], - "nav": ["class"], - "figure": ["class"], + "*": {"id"}, + "hr": {"class"}, + "img": {"src", "width", "height", "alt", "align", "class"}, + "span": {"class"}, + "th": {"align", "class"}, + "td": {"align", "colspan", "rowspan"}, + "div": {"align", "class"}, + "h1": {"align"}, + "h2": {"align"}, + "h3": {"align"}, + "h4": {"align"}, + "h5": {"align"}, + "h6": {"align"}, + "code": {"class"}, + "p": {"align", "class"}, + "pre": {"lang"}, + "ol": {"start"}, + "input": {"type", "checked", "disabled"}, + "aside": {"class"}, + "dd": {"class"}, + "dl": {"class"}, + "dt": {"class"}, + "ul": {"class"}, + "nav": {"class"}, + "figure": {"class"}, } -class DisabledCheckboxInputsFilter: - # The typeshed for bleach (html5lib) filters is incomplete, use `typing.Any` - # See https://github.com/python/typeshed/blob/505ea726415016e53638c8b584b8fdc9c722cac1/stubs/bleach/bleach/html5lib_shim.pyi#L7-L8 # noqa E501 - def __init__(self, source: Any) -> None: - self.source = source - - def __iter__(self) -> Iterator[Dict[str, Optional[str]]]: - for token in self.source: - if token.get("name") == "input": - # only allow disabled checkbox inputs - is_checkbox, is_disabled, unsafe_attrs = False, False, False - for (_, attrname), value in token.get("data", {}).items(): - if attrname == "type" and value == "checkbox": - is_checkbox = True - elif attrname == "disabled": - is_disabled = True - elif attrname != "checked": - unsafe_attrs = True - break - if is_checkbox and is_disabled and not unsafe_attrs: - yield token - else: - yield token - - def __getattr__(self, name: str) -> Any: - return getattr(self.source, name) - - def clean( html: str, - tags: Optional[List[str]] = None, - attributes: Optional[Dict[str, List[str]]] = None + tags: Optional[Set[str]] = None, + attributes: Optional[Dict[str, Set[str]]] = None ) -> Optional[str]: if tags is None: tags = ALLOWED_TAGS if attributes is None: attributes = ALLOWED_ATTRIBUTES - # Clean the output using Bleach - cleaner = bleach.sanitizer.Cleaner( - tags=tags, - attributes=attributes, - filters=[ - # Bleach Linkify makes it easy to modify links, however, we will - # not be using it to create additional links. - functools.partial( - bleach.linkifier.LinkifyFilter, - callbacks=[ - lambda attrs, new: attrs if not new else None, - bleach.callbacks.nofollow, - ], - skip_tags=["pre"], - parse_email=False, - ), - DisabledCheckboxInputsFilter, - ], - ) try: - cleaned = cleaner.clean(html) + cleaned = nh3.clean( + html, + tags=ALLOWED_TAGS, + attributes=ALLOWED_ATTRIBUTES, + link_rel="nofollow", + url_schemes={"http", "https", "mailto"}, + ) + return cleaned except ValueError: return None diff --git a/readme_renderer/txt.py b/readme_renderer/txt.py index 5af4805..adcf53e 100644 --- a/readme_renderer/txt.py +++ b/readme_renderer/txt.py @@ -21,4 +21,4 @@ def render(raw: str, **kwargs: Any) -> Optional[str]: rendered = html_escape(raw).replace("\n", "
") - return clean(rendered, tags=["br"]) + return clean(rendered, tags={"br"}) diff --git a/tests/fixtures/test_CommonMark_006.html b/tests/fixtures/test_CommonMark_006.html index d56d3e2..b3cd426 100644 --- a/tests/fixtures/test_CommonMark_006.html +++ b/tests/fixtures/test_CommonMark_006.html @@ -1 +1 @@ -<iframe src="http://mymalicioussite.com/">Click here</iframe> +Click here diff --git a/tests/fixtures/test_CommonMark_007.html b/tests/fixtures/test_CommonMark_007.html index b1c178b..d9b05e6 100644 --- a/tests/fixtures/test_CommonMark_007.html +++ b/tests/fixtures/test_CommonMark_007.html @@ -1,4 +1,2 @@

Something naughty this way comes

-<script> - alert("Hello"); -</script> + diff --git a/tests/fixtures/test_CommonMark_008.html b/tests/fixtures/test_CommonMark_008.html index 4d52a37..eef144d 100644 --- a/tests/fixtures/test_CommonMark_008.html +++ b/tests/fixtures/test_CommonMark_008.html @@ -4,13 +4,13 @@ self.name = name def make_sound(self): - print('Ruff!') + print('Ruff!') -dog = Dog('Fido') +dog = Dog('Fido')

and then here is some bash:

-
if [ "$1" = "--help" ]; then
-    echo "OK"
+
if [ "$1" = "--help" ]; then
+    echo "OK"
 fi
 

 
or click SurveyMonkey

diff --git a/tests/fixtures/test_GFM_019.html b/tests/fixtures/test_GFM_019.html
index 5dd4ff1..8ceabeb 100644
--- a/tests/fixtures/test_GFM_019.html
+++ b/tests/fixtures/test_GFM_019.html
@@ -1,3 +1,3 @@
 
http://commonmark.org

 
(Visit https://encrypted.google.com/search?q=Markup+(business))

-
Anonymous FTP is available at ftp://foo.bar.baz.

+
Anonymous FTP is available at ftp://foo.bar.baz.

diff --git a/tests/fixtures/test_GFM_020.html b/tests/fixtures/test_GFM_020.html
index 3df2b31..5bd8002 100644
--- a/tests/fixtures/test_GFM_020.html
+++ b/tests/fixtures/test_GFM_020.html
@@ -1 +1 @@
-
foo@bar.baz

+
foo@bar.baz

diff --git a/tests/fixtures/test_GFM_021.html b/tests/fixtures/test_GFM_021.html
index dbf5bf4..60b099b 100644
--- a/tests/fixtures/test_GFM_021.html
+++ b/tests/fixtures/test_GFM_021.html
@@ -1 +1 @@
-
hello@mail+xyz.example isn't valid, but hello+xyz@mail.example is.

+
hello@mail+xyz.example isn't valid, but hello+xyz@mail.example is.

diff --git a/tests/fixtures/test_GFM_022.html b/tests/fixtures/test_GFM_022.html
index d92d5e0..5ef0506 100644
--- a/tests/fixtures/test_GFM_022.html
+++ b/tests/fixtures/test_GFM_022.html
@@ -1,4 +1,4 @@
-
a.b-c_d@a.b

-
a.b-c_d@a.b.

+
a.b-c_d@a.b

+
a.b-c_d@a.b.

 
a.b-c_d@a.b-

 
a.b-c_d@a.b_

diff --git a/tests/fixtures/test_GFM_024.html b/tests/fixtures/test_GFM_024.html
index df9cabe..b5f9d02 100644
--- a/tests/fixtures/test_GFM_024.html
+++ b/tests/fixtures/test_GFM_024.html
@@ -1,17 +1,17 @@
 
diff --git a/tests/fixtures/test_GFM_doublequotes.html b/tests/fixtures/test_GFM_doublequotes.html
index 1353714..f75a4cf 100644
--- a/tests/fixtures/test_GFM_doublequotes.html
+++ b/tests/fixtures/test_GFM_doublequotes.html
@@ -2,7 +2,7 @@
 
This is code text.
 

 
def this_is_python():
-    """This is a docstring."""
+    """This is a docstring."""
     pass
 

 
func ThisIsGo(){
diff --git a/tests/fixtures/test_GFM_malicious_pre.html b/tests/fixtures/test_GFM_malicious_pre.html
index 245e17e..dc7b6e6 100644
--- a/tests/fixtures/test_GFM_malicious_pre.html
+++ b/tests/fixtures/test_GFM_malicious_pre.html
@@ -1,6 +1,6 @@
 
This is normal text.

 
def this_is_python():
-    """This is a docstring."""
+    """This is a docstring."""
     pass
-<script type="text/javascript">alert('I am evil.');</script>
+<script type="text/javascript">alert('I am evil.');</script>
 

diff --git a/tests/fixtures/test_rst_008.html b/tests/fixtures/test_rst_008.html
index 6a01a9d..6762f3a 100644
--- a/tests/fixtures/test_rst_008.html
+++ b/tests/fixtures/test_rst_008.html
@@ -8,8 +8,8 @@
 
 dog = Dog('Fido')

 
and then here is some bash:

-
if [ "$1" = "--help" ]; then
-    echo "OK"
+
if [ "$1" = "--help" ]; then
+    echo "OK"
 fi

 
or click SurveyMonkey

 
An unknown code fence block

diff --git a/tests/fixtures/test_rst_bibtex.html b/tests/fixtures/test_rst_bibtex.html
index 4305801..d9048e7 100644
--- a/tests/fixtures/test_rst_bibtex.html
+++ b/tests/fixtures/test_rst_bibtex.html
@@ -1,2 +1,2 @@
-
@article{the_impact_of_pygments_docutils_config_and_html5,
+
@article{the_impact_of_pygments_docutils_config_and_html5,
   year = {2022},

diff --git a/tests/fixtures/test_rst_docinfo.html b/tests/fixtures/test_rst_docinfo.html
index 4d514b4..a9d24d3 100644
--- a/tests/fixtures/test_rst_docinfo.html
+++ b/tests/fixtures/test_rst_docinfo.html
@@ -6,7 +6,7 @@
 
mer 02 ago 2017 14:49:24 CEST

 

 
Author:

-
Lele Gaifax <lele@metapensiero.it>

+
Lele Gaifax <lele@metapensiero.it>

 
License:

 
GNU General Public License version 3 or later

 

diff --git a/tests/fixtures/test_rst_linkify.html b/tests/fixtures/test_rst_linkify.html
index a151f25..5956925 100644
--- a/tests/fixtures/test_rst_linkify.html
+++ b/tests/fixtures/test_rst_linkify.html
@@ -42,7 +42,7 @@ 
Development

 
http://multigtfs.readthedocs.org/

 

 
IRC:

-
irc://irc.freenode.net/tulsawebdevs

+
irc://irc.freenode.net/tulsawebdevs

 

 
 
diff --git a/tests/fixtures/test_txt_001.html b/tests/fixtures/test_txt_001.html
index 9e2d139..f676e7a 100644
--- a/tests/fixtures/test_txt_001.html
+++ b/tests/fixtures/test_txt_001.html
@@ -1 +1 @@
-# Algen

Algen generates opionated ORM classes for sqlalchemy given a simple schema
either as a commandline string or as a yaml file.
It is designed to have minimal dependencies and is trivially extensible.
A command line tool is bundled along to help generate the models.
For DB specific types, only postgres is currently supported.
The tool currently assumes that sqlalchemy's declarative base object
is to be imported like ```from .alchemy_base import Base```


### CLI
```bash
$ algen --help
Usage: algen [OPTIONS]

Options:
  -n, --name TEXT         Name of model
  -c, --columns TEXT      Column definition. e.g. col_name:col_type Can be
                          used multiple times hence named columns. e.g. -c
                          foo:Int -c bar:Unicode(20)
  -d, --destination PATH  Destination directory. Default will assume 'Models'
                          directory inside the current working directory
  -y, --yaml PATH         Yaml file describing the Model. This supersedes the
                          column definition provided through --columns option.
  --help                  Show this message and exit.
```

Given a file as follows:
```yaml
Person:
  columns:
    - name: id
      type: BigInteger
      primary_key: True
      auto_increment: True
    - name: name
      type: Unicode(255)
    - name: is_vip
      type: Boolean
    - name: created_at
      type: DateTime(timezone=True)

Address:
  columns:
    - name: id
      type: BigInteger
      primary_key: True
      auto_increment: True
    - name: line1
      type: Unicode()
    - name: line2
      type: Unicode()
    - name: line3
      type: Unicode()
    - name: postcode
      type: Unicode(10)
      index: True
```

The cli tool will create two the following two files ```Person.py```  and  ```Address.py```.

```python
from __future__ import unicode_literals, absolute_import, print_function

from collections import namedtuple

from sqlalchemy import Column, DateTime, Boolean, Unicode, BigInteger


from .alchemy_base import Base

__author__ = 'danishabdullah'


class Person(Base):
    __tablename__ = 'persons'

    id = Column(BigInteger, primary_key=True, auto_increment=True)
    name = Column(Unicode(255), )
    is_vip = Column(Boolean, )
    created_at = Column(DateTime(timezone=True), )


    def __init__(self, id=None, name=None, is_vip=None, created_at=None):
        self.id = id
        self.name = name
        self.is_vip = is_vip
        self.created_at = created_at

    def add(self, session):
        session.add(self)

    def update(self, name=None, is_vip=None, created_at=None):
        # This function only updates a value if it is not None.
        # Falsy values go through in the normal way.
        # To set things to None use the usual syntax:
        #    Person.column_name = None

        if name is not None:
            self.name = name

        if is_vip is not None:
            self.is_vip = is_vip

        if created_at is not None:
            self.created_at = created_at

    def delete(self, session):
        session.delete(self)

    def to_dict(self):
        return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}

    def get_proxy_cls(self):
        # PersonProxy is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        keys = self.to_dict().keys()
        name = "PersonProxy"
        return namedtuple(name, keys)

    def to_proxy(self):
        # Proxy-ing is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        cls = self._get_proxy_cls()
        return cls(**self.to_dict())

    @classmethod
    def from_proxy(cls, proxy):
        return cls(**proxy._asdict())

    def __hash__(self):
        return hash(str(self.id))

    def __eq__(self, other):
        return (self.id == other.id)

    def __neq__(self, other):
        return not (self.id == other.id)

    def __str__(self):
        return "<Person: {id}>".format(id=self.id)

    def __unicode__(self):
        return "<Person: {id}>".format(id=self.id)

    def __repr__(self):
        return "<Person: {id}>".format(id=self.id)
```

```python
from __future__ import unicode_literals, absolute_import, print_function

from collections import namedtuple

from sqlalchemy import Column, Unicode, BigInteger


from .alchemy_base import Base

__author__ = 'danishabdullah'


class Address(Base):
    __tablename__ = 'addresses'

    id = Column(BigInteger, primary_key=True, auto_increment=True)
    line1 = Column(Unicode(), )
    line2 = Column(Unicode(), )
    line3 = Column(Unicode(), )
    postcode = Column(Unicode(10), index=True)


    def __init__(self, id=None, line1=None, line2=None, line3=None, postcode=None):
        self.id = id
        self.line1 = line1
        self.line2 = line2
        self.line3 = line3
        self.postcode = postcode

    def add(self, session):
        session.add(self)

    def update(self, line1=None, line2=None, line3=None, postcode=None):
        # This function only updates a value if it is not None.
        # Falsy values go through in the normal way.
        # To set things to None use the usual syntax:
        #    Address.column_name = None

        if line1 is not None:
            self.line1 = line1

        if line2 is not None:
            self.line2 = line2

        if line3 is not None:
            self.line3 = line3

        if postcode is not None:
            self.postcode = postcode

    def delete(self, session):
        session.delete(self)

    def to_dict(self):
        return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}

    def get_proxy_cls(self):
        # AddressProxy is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        keys = self.to_dict().keys()
        name = "AddressProxy"
        return namedtuple(name, keys)

    def to_proxy(self):
        # Proxy-ing is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        cls = self._get_proxy_cls()
        return cls(**self.to_dict())

    @classmethod
    def from_proxy(cls, proxy):
        return cls(**proxy._asdict())

    def __hash__(self):
        return hash(str(self.id))

    def __eq__(self, other):
        return (self.id == other.id)

    def __neq__(self, other):
        return not (self.id == other.id)

    def __str__(self):
        return "<Address: {id}>".format(id=self.id)

    def __unicode__(self):
        return "<Address: {id}>".format(id=self.id)

    def __repr__(self):
        return "<Address: {id}>".format(id=self.id)

```

\ No newline at end of file
+# Algen

Algen generates opionated ORM classes for sqlalchemy given a simple schema
either as a commandline string or as a yaml file.
It is designed to have minimal dependencies and is trivially extensible.
A command line tool is bundled along to help generate the models.
For DB specific types, only postgres is currently supported.
The tool currently assumes that sqlalchemy's declarative base object
is to be imported like ```from .alchemy_base import Base```


### CLI
```bash
$ algen --help
Usage: algen [OPTIONS]

Options:
  -n, --name TEXT         Name of model
  -c, --columns TEXT      Column definition. e.g. col_name:col_type Can be
                          used multiple times hence named columns. e.g. -c
                          foo:Int -c bar:Unicode(20)
  -d, --destination PATH  Destination directory. Default will assume 'Models'
                          directory inside the current working directory
  -y, --yaml PATH         Yaml file describing the Model. This supersedes the
                          column definition provided through --columns option.
  --help                  Show this message and exit.
```

Given a file as follows:
```yaml
Person:
  columns:
    - name: id
      type: BigInteger
      primary_key: True
      auto_increment: True
    - name: name
      type: Unicode(255)
    - name: is_vip
      type: Boolean
    - name: created_at
      type: DateTime(timezone=True)

Address:
  columns:
    - name: id
      type: BigInteger
      primary_key: True
      auto_increment: True
    - name: line1
      type: Unicode()
    - name: line2
      type: Unicode()
    - name: line3
      type: Unicode()
    - name: postcode
      type: Unicode(10)
      index: True
```

The cli tool will create two the following two files ```Person.py```  and  ```Address.py```.

```python
from __future__ import unicode_literals, absolute_import, print_function

from collections import namedtuple

from sqlalchemy import Column, DateTime, Boolean, Unicode, BigInteger


from .alchemy_base import Base

__author__ = 'danishabdullah'


class Person(Base):
    __tablename__ = 'persons'

    id = Column(BigInteger, primary_key=True, auto_increment=True)
    name = Column(Unicode(255), )
    is_vip = Column(Boolean, )
    created_at = Column(DateTime(timezone=True), )


    def __init__(self, id=None, name=None, is_vip=None, created_at=None):
        self.id = id
        self.name = name
        self.is_vip = is_vip
        self.created_at = created_at

    def add(self, session):
        session.add(self)

    def update(self, name=None, is_vip=None, created_at=None):
        # This function only updates a value if it is not None.
        # Falsy values go through in the normal way.
        # To set things to None use the usual syntax:
        #    Person.column_name = None

        if name is not None:
            self.name = name

        if is_vip is not None:
            self.is_vip = is_vip

        if created_at is not None:
            self.created_at = created_at

    def delete(self, session):
        session.delete(self)

    def to_dict(self):
        return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}

    def get_proxy_cls(self):
        # PersonProxy is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        keys = self.to_dict().keys()
        name = "PersonProxy"
        return namedtuple(name, keys)

    def to_proxy(self):
        # Proxy-ing is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        cls = self._get_proxy_cls()
        return cls(**self.to_dict())

    @classmethod
    def from_proxy(cls, proxy):
        return cls(**proxy._asdict())

    def __hash__(self):
        return hash(str(self.id))

    def __eq__(self, other):
        return (self.id == other.id)

    def __neq__(self, other):
        return not (self.id == other.id)

    def __str__(self):
        return "<Person: {id}>".format(id=self.id)

    def __unicode__(self):
        return "<Person: {id}>".format(id=self.id)

    def __repr__(self):
        return "<Person: {id}>".format(id=self.id)
```

```python
from __future__ import unicode_literals, absolute_import, print_function

from collections import namedtuple

from sqlalchemy import Column, Unicode, BigInteger


from .alchemy_base import Base

__author__ = 'danishabdullah'


class Address(Base):
    __tablename__ = 'addresses'

    id = Column(BigInteger, primary_key=True, auto_increment=True)
    line1 = Column(Unicode(), )
    line2 = Column(Unicode(), )
    line3 = Column(Unicode(), )
    postcode = Column(Unicode(10), index=True)


    def __init__(self, id=None, line1=None, line2=None, line3=None, postcode=None):
        self.id = id
        self.line1 = line1
        self.line2 = line2
        self.line3 = line3
        self.postcode = postcode

    def add(self, session):
        session.add(self)

    def update(self, line1=None, line2=None, line3=None, postcode=None):
        # This function only updates a value if it is not None.
        # Falsy values go through in the normal way.
        # To set things to None use the usual syntax:
        #    Address.column_name = None

        if line1 is not None:
            self.line1 = line1

        if line2 is not None:
            self.line2 = line2

        if line3 is not None:
            self.line3 = line3

        if postcode is not None:
            self.postcode = postcode

    def delete(self, session):
        session.delete(self)

    def to_dict(self):
        return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}

    def get_proxy_cls(self):
        # AddressProxy is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        keys = self.to_dict().keys()
        name = "AddressProxy"
        return namedtuple(name, keys)

    def to_proxy(self):
        # Proxy-ing is useful when you want to persist data
        # independent of the sqlalchemy session. It's just a namedtuple
        # that has very low memory/cpu footprint compared the regular
        # orm class instances.
        cls = self._get_proxy_cls()
        return cls(**self.to_dict())

    @classmethod
    def from_proxy(cls, proxy):
        return cls(**proxy._asdict())

    def __hash__(self):
        return hash(str(self.id))

    def __eq__(self, other):
        return (self.id == other.id)

    def __neq__(self, other):
        return not (self.id == other.id)

    def __str__(self):
        return "<Address: {id}>".format(id=self.id)

    def __unicode__(self):
        return "<Address: {id}>".format(id=self.id)

    def __repr__(self):
        return "<Address: {id}>".format(id=self.id)

```

\ No newline at end of file
diff --git a/tests/test_clean.py b/tests/test_clean.py
index 4eb8b41..1dc3f38 100644
--- a/tests/test_clean.py
+++ b/tests/test_clean.py
@@ -2,4 +2,6 @@
 
 
 def test_invalid_link():
-    assert clean('foo') == "foo"
+    assert clean(
+        'foo'
+    ) == 'foo'
diff --git a/tox.ini b/tox.ini
index 5b5a5fd..8ae2e5d 100644
--- a/tox.ini
+++ b/tox.ini
@@ -6,6 +6,7 @@ isolated_build = True
 deps =
     pytest
     pytest-cov
+    pytest-icdiff
 commands =
     pytest --strict-markers --cov {posargs}
 extras = md