Basic Cython support for DateParser

Makefile

@@ -5,24 +5,29 @@ auto: build27
 build27:
 	virtualenv local --python=python2.7
 	local/bin/pip install --use-mirrors -r requirements.txt
+	local/bin/python setup.py develop
 
 build26:
 	virtualenv local --python=python2.6
 	local/bin/pip install --use-mirrors -r requirements.txt
 	local/bin/pip install --use-mirrors -r requirements26.txt
+	local/bin/python setup.py develop
 
 build33:
 	virtualenv local --python=python3.3
 	local/bin/pip install --use-mirrors -r requirements.txt
+	local/bin/python setup.py develop
 
 build34:
 	virtualenv local --python=python3.4
 	local/bin/pip install --use-mirrors -r requirements.txt
+	local/bin/python setup.py develop
 
 
 build35:
 	virtualenv local --python=python3.5
 	local/bin/pip install --use-mirrors -r requirements.txt
+	local/bin/python setup.py develop
 
 test:
 	rm -f .coverage

arrow/arrow.py

@@ -1047,4 +1047,4 @@ def _get_timestamp_from_input(timestamp):
             raise ValueError('cannot parse \'{0}\' as a timestamp'.format(timestamp))
 
 Arrow.min = Arrow.fromdatetime(datetime.min)
-Arrow.max = Arrow.fromdatetime(datetime.max)
+Arrow.max = Arrow.fromdatetime(datetime.max)

arrow/parser.py → arrow/parser.pyx

@@ -2,7 +2,8 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
-from datetime import datetime
+from cpython.datetime cimport datetime
+
 from dateutil import tz
 import re
 
@@ -20,7 +21,8 @@ class ParserError(RuntimeError):
 
 class DateTimeParser(object):
 
-    _FORMAT_RE = re.compile('(YYY?Y?|MM?M?M?|Do|DD?D?D?|d?d?d?d|HH?|hh?|mm?|ss?|S+|ZZ?Z?|a|A|X)')
+    _FORMAT_RE = re.compile(
+        '(YYY?Y?|MM?M?M?|Do|DD?D?D?|d?d?d?d|HH?|hh?|mm?|ss?|S+|ZZ?Z?|a|A|X)')
     _ESCAPE_RE = re.compile('\[[^\[\]]*\]')
 
     _ONE_OR_MORE_DIGIT_RE = re.compile('\d+')
@@ -30,7 +32,6 @@ class DateTimeParser(object):
     _TZ_RE = re.compile('[+\-]?\d{2}:?(\d{2})?')
     _TZ_NAME_RE = re.compile('\w[\w+\-/]+')
 
-
     _BASE_INPUT_RE_MAP = {
         'YYYY': _FOUR_DIGIT_RE,
         'YY': _TWO_DIGIT_RE,
@@ -57,11 +58,11 @@ class DateTimeParser(object):
     SEPARATORS = ['-', '/', '.']
 
     def __init__(self, locale='en_us', cache_size=0):
-
         self.locale = locales.get_locale(locale)
         self._input_re_map = self._BASE_INPUT_RE_MAP.copy()
         self._input_re_map.update({
-            'MMMM': self._choice_re(self.locale.month_names[1:], re.IGNORECASE),
+            'MMMM': self._choice_re(self.locale.month_names[1:],
+                                    re.IGNORECASE),
             'MMM': self._choice_re(self.locale.month_abbreviations[1:],
                                    re.IGNORECASE),
             'Do': re.compile(self.locale.ordinal_day_re),
@@ -105,9 +106,9 @@ def parse_iso(self, string):
             has_tz = False
             # generate required formats: YYYY-MM-DD, YYYY-MM-DD, YYYY
             # using various separators: -, /, .
-            l = len(self.MARKERS)
-            formats = [separator.join(self.MARKERS[:l-i])
-                       for i in range(l)
+            length = len(self.MARKERS)
+            formats = [separator.join(self.MARKERS[:length - i])
+                       for i in range(length)
                        for separator in self.SEPARATORS]
 
         if has_time and has_tz:
@@ -126,7 +127,6 @@ def _generate_pattern_re(self, fmt):
         # 'YYYY-MM-DD' -> '(?P<YYYY>\d{4})-(?P<MM>\d{2})-(?P<DD>\d{2})'
         tokens = []
         offset = 0
-
         # Extract the bracketed expressions to be reinserted later.
         escaped_fmt = re.sub(self._ESCAPE_RE, "#", fmt)
         # Any number of S is the same as one.
@@ -145,9 +145,11 @@ def _generate_pattern_re(self, fmt):
             tokens.append(token)
             # a pattern doesn't have the same length as the token
             # it replaces! We keep the difference in the offset variable.
-            # This works because the string is scanned left-to-right and matches
-            # are returned in the order found by finditer.
-            fmt_pattern = fmt_pattern[:m.start() + offset] + input_pattern + fmt_pattern[m.end() + offset:]
+            # This works because the string is scanned left-to-right and
+            # matches are returned in the order found by finditer.
+            fmt_pattern = fmt_pattern[
+                :m.start() + offset] + input_pattern + fmt_pattern[m.end() +
+                                                                   offset:]
             offset += len(input_pattern) - (m.end() - m.start())
 
         final_fmt_pattern = ""
@@ -163,12 +165,10 @@ def _generate_pattern_re(self, fmt):
         return tokens, re.compile(final_fmt_pattern, flags=re.IGNORECASE)
 
     def parse(self, string, fmt):
-
         if isinstance(fmt, list):
             return self._parse_multiformat(string, fmt)
 
         fmt_tokens, fmt_pattern_re = self._generate_pattern_re(fmt)
-
         match = fmt_pattern_re.search(string)
         if match is None:
             raise ParserError('Failed to match \'{0}\' when parsing \'{1}\''
@@ -212,7 +212,8 @@ def _parse_token(self, token, value, parts):
             parts['second'] = int(value)
 
         elif token == 'S':
-            # We have the *most significant* digits of an arbitrary-precision integer.
+            # We have the *most significant* digits of an arbitrary-precision
+            # integer.
             # We want the six most significant digits as an integer, rounded.
             # FIXME: add nanosecond support somehow?
             value = value.ljust(7, str('0'))
@@ -263,10 +264,14 @@ def _build_datetime(parts):
         elif am_pm == 'am' and hour == 12:
             hour = 0
 
-        return datetime(year=parts.get('year', 1), month=parts.get('month', 1),
-            day=parts.get('day', 1), hour=hour, minute=parts.get('minute', 0),
-            second=parts.get('second', 0), microsecond=parts.get('microsecond', 0),
-            tzinfo=parts.get('tzinfo'))
+        return datetime(year=parts.get('year', 1),
+                        month=parts.get('month', 1),
+                        day=parts.get('day', 1),
+                        hour=hour,
+                        minute=parts.get('minute', 0),
+                        second=parts.get('second', 0),
+                        microsecond=parts.get('microsecond', 0),
+                        tzinfo=parts.get('tzinfo'))
 
     def _parse_multiformat(self, string, formats):
 
@@ -280,7 +285,9 @@ def _parse_multiformat(self, string, formats):
                 pass
 
         if _datetime is None:
-            raise ParserError('Could not match input to any of {0} on \'{1}\''.format(formats, string))
+            raise ParserError(
+                'Could not match input to any of {0} on \'{1}\''.format(
+                    formats, string))
 
         return _datetime
 
@@ -290,7 +297,8 @@ def _map_lookup(input_map, key):
         try:
             return input_map[key]
         except KeyError:
-            raise ParserError('Could not match "{0}" to {1}'.format(key, input_map))
+            raise ParserError('Could not match "{0}" to {1}'.format(key,
+                                                                    input_map))
 
     @staticmethod
     def _try_timestamp(string):
@@ -339,6 +347,7 @@ def parse(cls, string):
                 tzinfo = tz.gettz(string)
 
         if tzinfo is None:
-            raise ParserError('Could not parse timezone expression "{0}"'.format(string))
+            raise ParserError(
+                'Could not parse timezone expression "{0}"'.format(string))
 
         return tzinfo

requirements.txt

@@ -5,3 +5,5 @@ chai==1.1.1
 sphinx==1.3.5
 simplejson==3.6.5
 backports.functools_lru_cache==1.2.1
+Cython==0.25.2
+mock==2.0.0

setup.py

@@ -1,12 +1,44 @@
 import codecs
 import os.path
 import re
+import os
+import sys
 
 try:
     from setuptools import setup
 except ImportError:
     from distutils.core import setup
 
+from Cython.Build import cythonize
+from setuptools.command.build_ext import build_ext
+from distutils.extension import Extension
+
+
+# them to extension names in dotted notation
+def scandir(dir, files=[]):
+    for file in os.listdir(dir):
+        path = os.path.join(dir, file)
+        if os.path.isfile(path) and path.endswith(".pyx"):
+            files.append(path.replace(os.path.sep, ".")[:-4])
+        elif os.path.isdir(path):
+            scandir(path, files)
+    return files
+
+
+# generate an Extension object from its dotted name
+def makeExtension(extName):
+    try:
+        version = sys.version_info.major
+    except:
+        version = 2
+    extPath = extName.replace(".", os.path.sep) + ".pyx"
+    return Extension(
+        extName,
+        [extPath],
+        include_dirs=["."],   # adding the '.' to include_dirs is CRUCIAL!!
+        extra_compile_arg=["-O" + str(version), "-Wall"],
+        extra_link_args=['-g'],
+    )
 
 
 def fpath(name):
@@ -24,6 +56,11 @@ def grep(attrname):
 
 
 file_text = read(fpath('arrow/__init__.py'))
+# get the list of extensions
+extNames = scandir("arrow")
+
+# and build up the set of Extension objects
+extensions = cythonize([makeExtension(name) for name in extNames])
 
 setup(
     name='arrow',
@@ -34,6 +71,8 @@ def grep(attrname):
     author='Chris Smith',
     author_email="[email protected]",
     license='Apache 2.0',
+    ext_modules=extensions,
+    cmdclass={'build_ext': build_ext},
     packages=['arrow'],
     zip_safe=False,
     install_requires=[
@@ -52,4 +91,3 @@ def grep(attrname):
         'Topic :: Software Development :: Libraries :: Python Modules'
     ]
 )
-