init commit

Author: Kien Pham

.gitignore

@@ -24,6 +24,7 @@ pip-log.txt
 
 # Unit test / coverage reports
 .coverage
+coverage/
 .tox
 nosetests.xml
 

.travis.yml

@@ -0,0 +1,10 @@
+language: python
+
+python:
+  - "2.7"
+
+install:
+  - make install
+
+# command to run tests
+script: make test

Makefile

@@ -0,0 +1,29 @@
+.PHONY: all clean cover doc test
+
+help:
+	@echo "Available commands:"
+	@echo "  clean				remove temp files and static generated docs"
+	@echo "  cover				generate test coverage"
+	@echo "  doc				generate doc"
+	@echo "  test				run all tests"
+
+install: clean
+	pip install -e .[dev]
+
+clean:
+	find ./ -type f -name '*.pyc' -exec rm -f {} \;
+	rm -rf cover .coverage
+	rm -rf docs/build dist eggs
+
+cover:
+	ENV=test nosetests -s -v --with-coverage --cover-html --cover-html-dir ./coverage
+
+test:
+	ENV=test nosetests -s --nologcapture
+
+all: clean
+
+doc: clean
+	sphinx-build -b html doc/source doc/build/html
+	@echo
+	livereload -b doc/build/html -p 3001

README.md

@@ -1,4 +1,33 @@
-frink
-=====
+nibbler-python
+==============
+Better email parser for Python
 
-Better email parser
+[![Build Status](https://magnum.travis-ci.com/sendgridlabs/nibbler-python?token=Cxh5J57XgaqDXKj1xyma&branch=master)](https://magnum.travis-ci.com/sendgridlabs/nibbler-python)
+
+
+![nibbler](doc/_static/nibbler.gif)
+
+Installation
+------------
+	# Production: install from PyPI
+	$ pip install nibbler-python
+	
+	# Development: if you want to contribute to this project, clone it, install it from source
+	$ make install
+
+Usage
+-----
+Please see the [test/test_parser.py](test/test_parser.py) for more test cases
+
+	from nibbler.parser import parse_email
+
+	# valid email
+	parse_email('"much.more unusual"@example.com')
+
+	# invalid email
+	parse_email('Abc.example.com')
+
+
+Testing
+-------
+	$ make test

doc/_static/nibbler.gif

@@ -0,0 +1 @@
+__version__ = '0.0.1'

nibbler/__init__.py

@@ -0,0 +1,93 @@
+import string
+
+
+class State(object):
+    def __init__(self):
+        self.in_quotes = False
+        self.in_comment = False
+        self.previous_dot = False
+        self.previous_slash = False
+        self.in_domain = False
+        self.last_state = None
+
+ATEXT = "!#$%&'*+-/=?^_`.{|}~@\"" + string.ascii_letters + string.digits
+SPECIAL = ['(', ')', ',', ':', ';', '<', '>', '[', '\\', ']', ' ']
+HOSTNAME = "-." + string.ascii_letters + string.digits
+
+
+def parse_email(email_str):
+
+    state = State()
+    valid = True
+    address = ''
+
+    for offset, character in enumerate(email_str):
+        # Local part
+        if not state.in_domain:
+            if character == '\\':
+                if state.in_quotes:
+                    # Check if slash was backslashed within quotes
+                    if state.previous_slash:
+                        state.previous_slash = False
+                    else:
+                        state.previous_slash = True
+                # \ can only occur within slashes
+                else:
+                    valid = False
+                    break
+            elif character == '"':
+
+                if state.in_quotes:
+                    # Ignore if it was preceded by a backslash
+                    if not state.previous_slash:
+                        state.in_quotes = False
+                    else:
+                        state.previous_slash = False
+                else:
+                    # Quotes must happen as the first character or after a dot
+                    if offset != 0 and not state.previous_dot:
+                        valid = False
+                        break
+                    state.in_quotes = True
+
+            elif character == '.':
+                # We can't have two consecutive dots
+                if state.previous_dot:
+                    break
+                elif not state.in_quotes:
+                    state.previous_dot = True
+
+            elif character == '@' and not state.in_quotes:
+                state.in_domain = True
+
+            # These characters must only occur in quotes
+            if character in SPECIAL:
+                if not state.in_quotes:
+                    valid = False
+                    break
+                else:
+                    address += character
+            else:
+                if character not in ATEXT:
+                    valid = False
+                    break
+                else:
+                    address += character
+
+            # Check states and clear them if necessary
+            if state.previous_slash and character != '\\':
+                state.previous_slash = False
+            if state.previous_dot and character != '.':
+                state.previous_dot = False
+
+        # Domain part, we don't allow ip addresses [x.x.x.x] only host names
+        else:
+            if character not in HOSTNAME:
+                valid = False
+                break
+            address += character
+
+    if not state.in_domain:
+        valid = False
+
+    return valid, address

nibbler/parser.py

@@ -0,0 +1,59 @@
+"""nibbler-python - Better email parser
+
+nibbler-python helps you parse for valid email addresses in Python projects
+
+Example Usage
+-------------
+    from nibbler.parser import parse_email
+
+    # valid email
+    parse_email('"much.more unusual"@example.com')
+
+    # invalid email
+    parse_email('Abc.example.com')
+
+Contribute
+----------
+This library is hosted on Github and you can contribute there:
+https://github.com/sendgridlabs/nibbler-python
+"""
+
+from setuptools import setup, find_packages
+from nibbler import __version__
+
+doc = __doc__.splitlines()
+
+dev_requirements = [
+    # tests:
+    'nose',
+    'coverage',
+
+    # docs:
+    'Sphinx',
+    'sphinx_rtd_theme',
+    'sphinxcontrib-httpdomain',
+    'livereload'
+]
+
+setup(
+    name='nibbler-python',
+    description=doc[0],
+    long_description='\n'.join(doc[2:]),
+    version=__version__,
+    packages=find_packages(),
+    extras_require={
+        'dev': dev_requirements
+    },
+    url="https://github.com/sendgridlabs/nibbler-python",
+    keywords='email parser esp sendgrid',
+    license="MIT",
+    platforms="Posix; MacOS X; Windows",
+    classifiers=["Development Status :: 5 - Production/Stable",
+                 "Intended Audience :: Developers",
+                 "License :: OSI Approved :: MIT License",
+                 "Operating System :: OS Independent",
+                 "Topic :: Internet",
+                 "Programming Language :: Python :: 2",
+                 "Programming Language :: Python :: 2.6",
+                 "Programming Language :: Python :: 2.7"],
+)

setup.py

@@ -0,0 +1,35 @@
+import unittest
+
+
+class BaseTestCase(unittest.TestCase):
+
+    def setUp(self):
+        # Valid email addresses:
+        self.valid_addresses = [
+            '[email protected]',
+            '[email protected]',
+            '[email protected]',
+            '[email protected]',
+            '[email protected]',
+            '"much.more unusual"@example.com',
+            '"[email protected]"@example.com',
+            ('"very.(),:;<>[]\\".VERY.\\"very@\\\\ \\"very\\".unusual"'
+             '@strange.example.com'),
+            'postbox@com',
+            'admin@mailserver1',
+            '!#$%&\'*+-/=?^_`{}|[email protected]',
+            '"()<>[]:,;@\\\\\\"!#$%&\'*+-/=?^_`{}| ~.a"@example.org',
+            '" "@example.org',
+            'abc."defghi"[email protected]'
+        ]
+
+        # invalid email addresses:
+        self.invalid_addresses = [
+            'Abc.example.com',
+            'A@b@[email protected]',
+            'a"b(c)d,e:f;g<h>i[j\\k][email protected]',
+            'just"not"[email protected]',
+            'this is"not\\[email protected]',
+            'this\\ still\\"not\\\\[email protected]',
+            'abc"defghi"[email protected]'
+        ]

test/__init__.py

@@ -0,0 +1,19 @@
+import unittest
+from nibbler.parser import parse_email
+from test import BaseTestCase
+from nose.tools import (assert_true, assert_false)
+
+
+class TestCase(BaseTestCase):
+    def test_valid_email(self):
+        for e in self.valid_addresses:
+            valid, email_parsed = parse_email(e)
+            assert_true(valid, 'Error: email %s should be valid' % e)
+
+    def test_invalid_email(self):
+        for e in self.invalid_addresses:
+            valid, email_parsed = parse_email(e)
+            assert_false(valid, 'Error: email %s should be invalid' % e)
+
+if __name__ == '__main__':
+    unittest.main()