[metadata]

name = languedoc

version = 1.0

license = GPLv3

description = A simple language identification library.

classifiers =

    Programming Language :: Python :: 3

    License :: OSI Approved :: GNU General Public License v3 (GPLv3)

    Operating System :: OS Independent

long_description = file: README.md

long_description_content_type = text/markdown

[options]

packages =

    languedoc

package_dir =

    =src

python_requires = >=3.6

import os

import re

import itertools

import json

import gzip

TOP_NGRAM_COUNT = 3000

def preprocess(text):

	text = re.sub(r"[\W\d_]+", " ", " "+text+" ")

	return text.lower()

def extract_kgram_freqs(text, k):

	n = len(text)

	d = dict()

	for i in range(0, n-k+1):

		key = text[i:i+k]

		if key.isspace():

			continue

		d[key] = d.get(key, 0) + 1

	count = sum(d.values())

	return {key: val/count for (key, val) in d.items()}

def extract_ngram_freqs(text):

	frequencies = {}

	for k in range(1, 4):

		frequencies.update(extract_kgram_freqs(text, k))

	return frequencies

def rank_ngram_freqs(frequencies):

	ordered_ngrams = sorted(frequencies.items(), key=lambda kv: -kv[1])[:TOP_NGRAM_COUNT]

	return dict(zip([key for (key, freq) in ordered_ngrams], itertools.count(0)))

def extract_ranked_ngrams(text):

	frequencies = extract_ngram_freqs(text)

	return rank_ngram_freqs(frequencies)

class Sample:

	def __init__(self, language, ranked_ngrams):

		self.language = language

		self.ranked_ngrams = ranked_ngrams

	@classmethod

		self.language = language

		self.texts = []

		self.frequencies = []

	def add(self, text):

		self.texts.append(text)

		self.frequencies.append(extract_ngram_freqs(text))

	def create_model(self):

		merged_frequencies = merge_ngram_freqs(self.frequencies)

		res = Sample(self.language, rank_ngram_freqs(merged_frequencies))

		return res

	def generate_tests(self, n):

		for (i, (text, freqs)) in enumerate(itertools.cycle(zip(self.texts, self.frequencies))):

			if i >= n:

				break

			ranked_ngrams = rank_ngram_freqs(merge_ngram_freqs([f for f in self.frequencies if f is not freqs]))

			yield (text, Sample(self.language, ranked_ngrams))

def cross_validate(sample_sets):

	models = [s.create_model() for s in sample_sets]

	score = 0

	max_score = 0

	for s in sample_sets:

		for (test_text, partial_model) in s.generate_tests(CROSSVALIDATION_SOURCE_COUNT):

			real_lang = partial_model.language

			test_models = [partial_model] + [m for m in models if m.language != real_lang]

			for k in TEST_LENS:

				for i in range(10):

					j = random.randrange(0, len(test_text)-k)

					t = test_text[j:j+k]

					predicted_lang = identify(t, test_models)

					if predicted_lang == real_lang:

						score += 1

					else:

						print(real_lang, predicted_lang, t)

					max_score += 1

	return score / max_score, (score, max_score)

DATA_DIR = os.path.join(os.path.dirname(__file__), "../../data")

LANG_DIRS = sorted([x.path for x in os.scandir(DATA_DIR)])

if __name__ == "__main__":

	samples = []

	for d in LANG_DIRS:

		lang = os.path.basename(d)

		lang_samples = SampleSet(lang)

		samples.append(lang_samples)

		for file in sorted(os.scandir(d), key=lambda f: f.name):

			with open(file) as f:

				text = f.read()

				text = preprocess(text)

				print(f"{lang}: {file.name} ({len(text)})")

				lang_samples.add(text)

	with gzip.open(MODEL_PATH, mode="wt", encoding="utf-8") as f:

		json.dump([sample_set.create_model().export() for sample_set in samples], f, ensure_ascii=False)

	print(cross_validate(samples))