def geocode(self, address, delay=5, read_cache=True, num_retries=1, language=None):
"""Geocode address using Google's API and return dictionary of useful fields
address:
what to pass to geocode API
delay:
how long to delay between API requests
read_cache:
whether to load content from cache when exists
num_retries:
the number of times to try downloading
language:
the language to set
"""
try:
address = address.encode('utf-8')
except UnicodeDecodeError:
common.logger.warning('Geocode failed to parse address and needed to cast to ascii: ' + address)
address = common.to_ascii(address)
geocode_url = 'http://maps.google.com/maps/api/geocode/json?address=%s&sensor=false%s' % (urllib.quote_plus(address), '&language=' + language if language else '')
geocode_html = self.D.get(geocode_url, delay=delay, read_cache=read_cache, num_retries=num_retries)
geocode_data = self.load_result(geocode_url, geocode_html)
for result in geocode_data.get('results', []):
return self.parse_location(result)
return collections.defaultdict(str)
def geocode(self, address, delay=5, read_cache=True, num_retries=1, language=None, api_key=None):
"""Geocode address using Google's API and return dictionary of useful fields
address:
what to pass to geocode API
delay:
how long to delay between API requests
read_cache:
whether to load content from cache when exists
num_retries:
the number of times to try downloading
language:
the language to set
"""
try:
address = address.encode('utf-8')
except UnicodeDecodeError:
common.logger.debug('Geocode failed to parse address and needed to cast to ascii: ' + address)
address = common.to_ascii(address)
address = re.sub('%C2%9\d', '', urllib.quote_plus(address))
geocode_url = 'http://maps.google.com/maps/api/geocode/json?address=%s&sensor=false%s' % (address, '&language=' + language if language else '')
try:
# legacy data without api key
geocode_html = self.D.cache[geocode_url]
if geocode_html:
self.D.response_code = '200'
else:
raise KeyError()
except KeyError:
geocode_url = 'https://maps.google.com/maps/api/geocode/json?address=%s&key=%s&sensor=false%s' % (address, api_key or '', '&language=' + language if language else '')
geocode_html = self.D.get(geocode_url, delay=delay, read_cache=read_cache, num_retries=num_retries)
geocode_data = self.load_result(geocode_url, geocode_html)
for result in geocode_data.get('results', []):
return self.parse_location(result)
return collections.defaultdict(str)
def sort_key(self):
"""this value should assign the name its proper place in the alfabet
"""
base = self.guess_normal_form()
base = base.replace(',', '')
base = base.replace(' ', ' ')
base = base.strip()
base = base.lower()
ignore_these = '()'
for c in ignore_these:
base = base.replace(c, '')
for s in PREFIXES: #we also strip prefixes (so "L'Eremite, B." sorts between "Eremite, A" and "Eremite, C")
if base.startswith(s):
base = base[len(s):]
ls = base.split()
for s in TUSSENVOEGSELS:
if ls and ls[0] == s:
base = ' '.join(ls[1:])
for s in '?.-': #if the name starts with any of these characters, it comes last (not first)
if base.startswith(s):
base = chr(126) + base
base = to_ascii(base)
base = base.strip()
base = base[:40]
return base
def __init__(
self,
ident=None,
name=None,
cou=None,
birth_date=None,
lefty: Optional[bool] = None,
disp_names=None,
):
self.ident = ident
self.name = co.to_ascii(name.strip()) if name is not None else None
self.cou = co.to_ascii(cou.strip()) if cou else None
self.rating = ratings.Rating()
self.birth_date = birth_date
self.lefty: Optional[bool] = lefty
self.disp_names = disp_names
def clean_content(self, html, max_size, force_html, force_ascii):
"""Clean up downloaded content
"""
if max_size is not None and len(html) > max_size:
common.logger.info('Too big: %s' % len(html))
html = '' # too big to store
elif force_html and not common.is_html(html):
common.logger.info('Not html')
html = '' # non-html content
elif force_ascii:
html = common.to_ascii(html) # remove non-ascii characters
return html
def clean_content(self, html, max_size, force_html, force_ascii):
"""Clean up downloaded content
"""
if max_size is not None and len(html) > max_size:
common.logger.info('Too big: %s' % len(html))
html = '' # too big to store
elif force_html and not common.is_html(html):
common.logger.info('Not html')
html = '' # non-html content
elif force_ascii:
html = common.to_ascii(html) # remove non-ascii characters
return html
def get_money(text):
""" извлечение денежной суммы (с помощью regexpr) """
if not text:
return None
raw_text = co.to_ascii(text.replace(",", ".").strip())
match = MONEY_RE.match(raw_text)
if match:
value = float(match.group("value"))
if match.group("multier") == "K":
return value * 1000
elif match.group("multier") == "M":
return value * 1000000
return None
def soundex_nl(s, length=4, group=1, wildcards=False):
"""
return a string of length representing a phonetical canonical form of s
stab at giving names a simplified canonical form based on Dutch phonetics and spelling conventions
arguments:
s : a string
length : an integer. Length=-1 transforms the whole string
group : an integer [1, 2]
wildcards : if True, wildcard element (?, *) remain in place
There are two groups:
- group 1: identify lots
- groep 2: identify somewhat less (stay close to actual phonetics)
"""
#ignore Romans
if s in ROMANS_frozenset:
return s
s = s.lower()
s = to_ascii(s)
if not wildcards:
#remove 'wildcard' characters
s = s.replace('*', '').replace('?', '')
#strip off certain prefixes
#XXX this should be in the regular expression specs
# for x in PREFIXES:
# if s.startswith(x):
# s = s[len(x):]
if group == 1:
groups = GROUPS1
elif group == 2:
groups = GROUPS2
else:
raise Exception('"group" argument must be either 1 or 2')
s = apply_regexps(s, groups)
if s.endswith('.'):
s = s[:-1]
if not s:
s = u'.'
if length > 0:
s = s[:length]
s = unicode(s)
return s
def from_oncourt(raw_name):
from common import to_ascii, split_ontwo
raw = to_ascii(raw_name)
desc_name, name = split_ontwo(raw, delim=" - ")
if name == "":
name = raw
desc_name = ""
number = None
if name and name[-1].isdigit() and name[-2] == " ":
number = int(name[-1])
if number == 1:
number = None # to compare with flashscore 1-less name
name = name[:-2]
return TourName(name=name, desc_name=desc_name, number=number)
def get_name_cou(is_left):
inclass_txt = "event__participant--{}".format(
"home" if is_left else "away")
xpath_txt = "child::div[contains(@class, '{}')]".format(inclass_txt)
try:
plr_el = co.find_first_xpath(element, xpath_txt)
if plr_el is None:
raise co.TennisParseError("nofound plr_el in \n{}".format(
etree.tostring(element, encoding="utf8")))
txt = co.to_ascii(plr_el.text).strip()
bracket_idx = txt.index("(")
name = txt[0:bracket_idx].strip()
cou = txt[bracket_idx + 1:bracket_idx + 4].strip().upper()
return name, cou
except ValueError:
# maybe it is team's result div without '(cou)':
tourinfo_cache.edit_skip(live_event.tour_info)
def __init__(self, name):
name_low = co.to_ascii(name.strip()).lower()
if ("carpet" in name_low and "outdoor" not in name_low) or (
"hard" in name_low and
("indoor" in name_low or name_low.startswith("i."))):
self.name = "Carpet"
elif "grass" in name_low:
self.name = "Grass"
elif "clay" in name_low:
self.name = "Clay"
elif "hard" in name_low or ("carpet" in name_low
and "outdoor" in name_low):
self.name = "Hard"
elif "acrylic" in name_low:
self.name = "Acrylic"
else:
raise co.TennisSurfaceError("unexpected surface '{}'".format(name))
def _clean_content(self, html, max_size, force_html, force_ascii):
"""Clean up downloaded content
html:
the input to clean
max_size:
the maximum size of data allowed
force_html:
content must be HTML
force_ascii:
content must be ASCII
"""
if max_size is not None and len(html) > max_size:
common.logger.info('Webpage is too big: %s' % len(html))
html = '' # too big to store
elif force_html and not common.is_html(html):
common.logger.info('Webpage is not html')
html = '' # non-html content
elif force_ascii:
html = common.to_ascii(html) # remove non-ascii characters
return html
def _clean_content(self, html, max_size, force_html, force_ascii):
"""Clean up downloaded content
html:
the input to clean
max_size:
the maximum size of data allowed
force_html:
content must be HTML
force_ascii:
content must be ASCII
"""
if max_size is not None and len(html) > max_size:
common.logger.info('Webpage is too big: %s' % len(html))
html = '' # too big to store
elif force_html and not common.is_html(html):
common.logger.info('Webpage is not html')
html = '' # non-html content
elif force_ascii:
html = common.to_ascii(html) # remove non-ascii characters
return html
def get(self, url, **kwargs):
"""Download this URL and return the HTML. Data is cached so only have to download once.
url is what to download
kwargs can override any of the arguments passed to constructor
"""
delay = kwargs.get("delay", self.delay)
proxy = kwargs.get("proxy", self.proxy)
user_agent = kwargs.get("user_agent", self.user_agent)
opener = kwargs.get("opener", self.opener)
headers = kwargs.get("headers", self.headers)
data = kwargs.get("data", self.data)
use_cache = kwargs.get("use_cache", self.use_cache)
use_remote = kwargs.get("use_remote", self.use_remote)
retry = kwargs.get("retry", self.retry)
force_html = kwargs.get("force_html", self.force_html)
force_ascii = kwargs.get("force_ascii", self.force_ascii)
max_size = kwargs.get("max_size", self.max_size)
if use_cache and url in self.cache:
html = self.cache[url]
if retry and not html:
print "Redownloading"
else:
return html
if not use_remote:
return "" # do not try downloading but return empty
print url # need to download url
self.domain_delay(url, delay=delay, proxy=proxy) # crawl slowly for each domain to reduce risk of being blocked
html = self.fetch(url, headers=headers, data=data)
if max_size is not None and len(html) > max_size:
html = "" # too big to store
elif force_html and not re.search("html|head|body", html):
html = "" # non-html content
elif force_ascii:
html = to_ascii(html) # remove non-ascii characters
self.cache[url] = html
return html
def __init__(self, value: str, details=None):
self.value = co.to_ascii(value.strip())
self._detailing = details
assert self.value in Round.names, "unexpected rnd value: {}".format(
self.value)
def soundex_nl(s, length=4, group=1):
"""
stab at giving names a simplified canonical form
De truuk is een vorm te vinden zodat zoveel mogelijk 'gelijkvormige' namen
dezelfde "soundex" krijgen...
Er zijn verschillende "groepen"
- groep 1: identificeer zoveel mogelijk
- groep 2: identificeer wat minder
lenght geeft the lengte van het resultaat aan. length = -1 geeft de hele string
"""
s = s.lower()
s = to_ascii(s)
#strip of certain suffixes
# for suffix, replacement in [
# ('en', ''),
# ('sch', 's'),
# ]:
# if s.endswith(suffix):
# s = s[:-len(suffix)] + replacement
# break
#strip of certain prefixes
for x in PREFIXES + ['h']:
if s.startswith(x):
s = s[len(x):]
if group == 1:
groups = GROUPS1
elif group == 2:
groups = GROUPS2
else:
raise Exception('"group" argument must be either 1 or 2')
#THIS DOES LOST OF UNNCECESSARY STUFF
if 1:
#create a single regular expression
for k, regexp in groups:
# regexp = re.compile('|'.join(ls))
s = regexp.sub(k, s)
while regexp.search(s):
s = regexp.sub(k, s)
if 0:
ls = []
d = {}
for k, l in groups:
ls += l
for x in l:
d[x] = k
s = multiple_replace(d, s)
if 0:
ls = []
d = {}
for k, l in groups:
ls += l
for x in l:
d[x] = k
to_replace = []
for x in ls:
if re.search(x, s):
to_replace.append(x)
for x in to_replace:
s = re.sub(x, d[x], s)
if s.endswith('.'):
s = s[:-1]
if not s:
s = '.'
if length > 0:
s = s[:length]
s = str(s)
return s
def geslachtsnaam_soundex(self):
return self.soundex_nl(
to_ascii(self.geslachtsnaam()),
group=2, length=-1
)
def __init__(self, name):
self.name = co.to_ascii(name)
self.parts = [] # lower case syllables
self.syllabled = "" # lower case syllables departed by '-'
self._init_parts()
def ratio(self, n1, n2, explain=0):
"""Combine several parameters do find a similarity ratio"""
weight_normal_form = 3.0 #distance between soundexes of normal form
weight_normal_form_if_one_name_is_in_initials = old_div(
weight_normal_form, 4) #distance between soundexes of normal form
weight_normal_form_soundex = 9.0 #average distance between soundexes of normal form
weight_normal_form_soundex_if_one_name_is_in_initials = old_div(
weight_normal_form_soundex,
4) #distance between soundexes of normal form
weight_geslachtsnaam1 = 7.0 #distance between soundexes of geslachtsnamen
weight_geslachtsnaam2 = 7.0 #distance between geslachtsnaam
weight_initials = 2 #distance between initials
weight_initials_if_one_name_is_in_initials = weight_initials * 2 #distance between initials if one of the names is in intials
#(for example, "A.B Classen")
### NORMAL FORM
nf1 = n1.guess_normal_form()
nf2 = n2.guess_normal_form()
#remove diacritics
nf1 = to_ascii(nf1)
nf2 = to_ascii(nf2)
# ratio_normal_form = self.levenshtein_ratio(nf1, nf2)
ratio_normal_form = self.average_distance(split(nf1), split(nf2))
#create a simkplified soundex set for this name
#remove stopwords
nf1 = remove_stopwords(nf1)
nf2 = remove_stopwords(nf2)
#l# = min(len(nf1.split()), len(nf2.split()))
#nf1 = ' '.join(nf1.split()[:l])
#nf2 = ' '.join(nf2.split()[:l])
#we use the soundex_nl property of the name, so the property gets cached
se1 = n1.soundex_nl(nf1, group=2, length=-1)
se2 = n2.soundex_nl(nf2, group=2, length=-1)
#make the nfs of the same length
ratio_normal_form_soundex = self.average_distance(se1, se2)
#gelachtsnaam wordt op twee manieren met elkaar vergeleken
#de soundexes van de achternaam worden meegewoen
g1 = n1.geslachtsnaam() #or n1.get_volledige_naam()
g2 = n2.geslachtsnaam() #or n2.get_volledige_naam()
g1 = to_ascii(g1)
g2 = to_ascii(g2)
g1_soundex = n1.soundex_nl(g1, group=2, length=-1)
g2_soundex = n2.soundex_nl(g2, group=2, length=-1)
ratio_geslachtsnaam1 = self.average_distance(g1_soundex, g2_soundex)
#n de afstand van de woorden in de achtenraam zelf
ratio_geslachtsnaam2 = self.average_distance(
re.split('[ \.\,\-]', g1.lower()),
re.split('[ \.\,\-]', g2.lower()), self.levenshtein_ratio)
#count initials only if we have more than one
#(or perhaps make this: if we know the first name)
if len(n1.initials()) == 1 or len(n2.initials()) == 1:
#initials count much less if there is only one
weight_initials = 0
ratio_initials = .5
elif n1.contains_initials() or n2.contains_initials():
ratio_initials = self.levenshtein_ratio(n1.initials().lower(),
n2.initials().lower())
weight_initials = weight_initials_if_one_name_is_in_initials
elif len(n1.initials()) > 1 and len(n2.initials()) > 1:
ratio_initials = self.levenshtein_ratio(n1.initials().lower(),
n2.initials().lower())
else:
ratio_initials = 0.7
if n1.contains_initials() or n2.contains_initials():
weight_normal_form = weight_normal_form_if_one_name_is_in_initials
weight_normal_form_soundex = weight_normal_form_soundex_if_one_name_is_in_initials
try:
teller = ratio_normal_form * weight_normal_form + ratio_normal_form_soundex * weight_normal_form_soundex + ratio_geslachtsnaam1 * weight_geslachtsnaam1 + ratio_geslachtsnaam2 * weight_geslachtsnaam2 + ratio_initials * weight_initials
noemer = weight_normal_form + weight_normal_form_soundex + weight_initials + weight_geslachtsnaam1 + weight_geslachtsnaam2
final_ratio = old_div(teller, noemer)
except ZeroDivisionError:
return 0.0
if explain:
d = [
(
'ratio_normal_form',
ratio_normal_form,
),
(
'weight_normal_form',
weight_normal_form,
),
(
'ratio_geslachtsnaam1 (soundex)',
ratio_geslachtsnaam1,
),
(
'weight_geslachtsnaam1',
weight_geslachtsnaam1,
),
(
'ratio_geslachtsnaam2 (letterlijke geslachtsnaam)',
ratio_geslachtsnaam2,
),
(
'weight_geslachtsnaam2',
weight_geslachtsnaam2,
),
(
'ratio_initials',
ratio_initials,
),
(
'weight_initials',
weight_initials,
),
(
'final_ratio',
final_ratio,
),
(
'teller',
teller,
),
(
'noemer',
noemer,
),
]
s = '-' * 100 + '\n'
s += 'Naam1: %s [%s] [%s] %s\n' % (n1, n1.initials(),
n1.guess_normal_form(), se1)
s += 'Naam2: %s [%s] [%s] %s\n' % (n2, n2.initials(),
n2.guess_normal_form(), se2)
s += 'Similarity ratio: %s\n' % final_ratio
s += '--- REASONS' + '-' * 30 + '\n'
format_s = '%-30s | %-10s | %-10s | %-10s | %-10s | %s-10s\n'
s += format_s % ('\t property', ' ratio', ' weight',
'relative_weight', ' r*w', 'r * relative_w')
s += '\t' + '-' * 100 + '\n'
format_s = '\t%-30s | %-10f | %-10f | %-10f | %-10f | %-10f\n'
s += format_s % (
' normal_form', ratio_normal_form, weight_normal_form,
old_div(weight_normal_form,
teller), ratio_normal_form * weight_normal_form,
old_div(ratio_normal_form * weight_normal_form, teller))
s += format_s % (
'soundex van normal_form', ratio_normal_form_soundex,
weight_normal_form_soundex,
old_div(weight_normal_form_soundex, teller),
ratio_normal_form_soundex * weight_normal_form_soundex,
old_div(ratio_normal_form_soundex * weight_normal_form_soundex,
teller))
s += format_s % (
'soundex van geslachtsnaam1', ratio_geslachtsnaam1,
weight_geslachtsnaam1, old_div(weight_geslachtsnaam1, teller),
ratio_geslachtsnaam1 * weight_geslachtsnaam1,
old_div(ratio_geslachtsnaam1 * weight_geslachtsnaam1, teller))
s += format_s % (
'geslachtsnaam', ratio_geslachtsnaam2, weight_geslachtsnaam2,
old_div(weight_geslachtsnaam2,
teller), ratio_geslachtsnaam2 * weight_geslachtsnaam2,
old_div(ratio_geslachtsnaam2 * weight_geslachtsnaam2, teller))
s += format_s % ('initials', ratio_initials, weight_initials,
old_div(weight_initials,
teller), ratio_initials * weight_initials,
old_div(ratio_initials * weight_initials, teller))
s += '\tTOTAL (numerator) | %s (counter = %s)\n' % (
teller, noemer)
return s
return '\n'.join(['%s: %s' % (k, v) for k, v in d])
return final_ratio
