def load_country_groups_map():
return pd.read_excel(
get_data_dir("whitelists", "countries", "codelist.xlsx"),
sheet_name="groups_iso3c",
header=1,
index_col=1).drop("country.name.en", axis=1).apply(
lambda col_ser: col_ser.dropna().index.dropna().tolist(),
axis=0).to_dict()
def get_standardized_regions(iso_code="iso3c"):
assert iso_code in ["iso2c", "iso3c", "full"]
codelist_path = Path(
get_data_dir("whitelists", "countries", "codelist.xlsx"))
standardized_regions_path = codelist_path.parent / "standardized_regions.xlsx"
if not standardized_regions_path.exists():
codelist = pd.read_excel(codelist_path)
iso_region = codelist[["country.name.en", "iso2c", "iso3c", "region"]]
standardized_regions = iso_region.dropna(
subset=["iso2c", "region"]).set_index("iso2c")
standardized_regions.to_excel(standardized_regions_path)
else:
standardized_regions = pd.read_excel(standardized_regions_path)
if iso_code != "full":
standardized_regions = standardized_regions.reset_index().set_index(
iso_code)["region"].to_dict()
return standardized_regions
from wb_cleaning.ops.cache_utils import redis_cacher
cache_decorator = redis_cacher
except redis.ConnectionError as error:
args = error.args
print(args[0])
print("Redis not available, falling back to joblib cache...")
USE_JOBLIB_MEMORY = True
if USE_JOBLIB_MEMORY:
RESPELLER_CACHE_LOCATION = "/dev/shm/respeller-cachedir"
try:
respeller_cache = Memory(RESPELLER_CACHE_LOCATION, verbose=0)
except PermissionError:
RESPELLER_CACHE_LOCATION = dir_manager.get_data_dir(
'shm', 'respeller-cachedir')
if not os.path.isdir(RESPELLER_CACHE_LOCATION):
os.makedirs(RESPELLER_CACHE_LOCATION)
respeller_cache = Memory(RESPELLER_CACHE_LOCATION, verbose=0)
cache_decorator = respeller_cache.cache
# # Returns self without any form of caching.
# cache_decorator = lambda f: f
#
# en_dict = language.Language().get_en_dict()
en_lang = language.Language()
def __init__(self, tag="en_US", pwl=dir_manager.get_data_dir(
"whitelists", "whitelists", "doc-freq-wiki-wordlist.txt")
):
self.tag = tag
self.pwl = pwl
self.init_en_dict()
# -*- coding: utf-8 -*-
import os
from wb_cleaning.dir_manager import get_data_dir
WHITELIST_PATH = get_data_dir('whitelists', 'extraction')
def get_country_csv():
return os.path.join(WHITELIST_PATH, 'whitelist_countries_multilingual.csv')
def get_wb_presidents_csv():
return os.path.join(WHITELIST_PATH, 'wb_presidents.csv')
import glob
import functools
from flashtext import KeywordProcessor
from wb_cleaning import dir_manager
# export DASK_DISTRIBUTED__SCHEDULER__ALLOWED_FAILURES=210
# export DASK_DISTRIBUTED__COMM__TIMEOUTS__CONNECT=60
# export DASK_DISTRIBUTED__COMM__RETRY__COUNT=20
ACCENTED_CHARS = set(
"ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ")
keyword_processor = KeywordProcessor()
keyword_processor.set_non_word_boundaries(keyword_processor.non_word_boundaries
| ACCENTED_CHARS)
with open(dir_manager.get_data_dir('whitelists', 'whitelists',
'phrases.txt')) as phrase_file:
# Use flashtext format
phrases_map = {
l.strip(): [l.strip().replace('_', ' ')]
for l in phrase_file if l.strip()
}
keyword_processor.add_keywords_from_dict(phrases_map)
@functools.lru_cache(maxsize=None)
def cached_load_file(fname: Path, split: bool = True):
return load_file(fname=fname, split=split)
def replace_phrases(txt):
return keyword_processor.replace_keywords(txt)
dest=dest,
remove_accents=False)).map(
lambda x: x +
[inflect_engine.plural(i) for i in x if "_" not in i])
for dest in translate_to:
tags_mapping = tags_mapping + lang_map[dest]
# Clean up the keywords to remove duplicates.
tags_mapping = tags_mapping.map(
lambda x: sorted(set(filter(lambda i: i, x))))
return tags_mapping
tags_sheet = pd.read_excel(get_data_dir("whitelists", "jdc",
"List_filtering_keywords.xlsx"),
header=None,
index_col=0).rename(columns={1: "tag_keyword"})
tags_mapping = get_keywords_mapping(tags_sheet=tags_sheet)
if "Kakuma (Kenya)" in tags_mapping:
tags_mapping.pop("Kakuma (Kenya)")
jdc_tags_processor.add_keywords_from_dict(tags_mapping)
def get_jdc_tag_counts(txt):
data = []
for tag, value in Counter(
jdc_tags_processor.extract_keywords(txt)).most_common():
def load_country_groups_names():
return pd.read_excel(get_data_dir("whitelists", "countries",
"codelist.xlsx"),
sheet_name="groups_names",
header=0,
index_col=0).to_dict()["Full name"]
def load_iso3166_3_country_info():
return pd.read_json(get_data_dir("maps",
"iso3166-3-country-info.json")).to_dict()
from functools import lru_cache
from pathlib import Path
from unidecode import unidecode
import googletrans
from googletrans import client as gt
import requests
from wb_cleaning import dir_manager
assert googletrans.__version__ == "3.1.0-alpha"
_trans = gt.Translator()
language_codes_path = Path(
dir_manager.get_data_dir("whitelists", "language", "language_codes.json"))
if not language_codes_path.exists():
result = requests.get(
"https://translate.shell.com/api/Translate/GetLanguages")
result = result.json()
language_codes = [{"name": o["item1"], "code": o["item2"]} for o in result]
with open(language_codes_path, "w") as open_file:
json.dump(language_codes, open_file)
else:
with open(language_codes_path) as open_file:
language_codes = json.load(open_file)