fix: faiss stuff, api

This commit is contained in:
2025-02-18 23:05:35 +06:00
parent 5d4149be63
commit 875e189aab
6 changed files with 194 additions and 106 deletions

View File

@@ -1,7 +1,18 @@
from typing_extensions import override
import unittest
import os
from bertalign.chunk import create_aligned_chunks, TextChunk, AlignmentError
from typing_extensions import override
from bertalign.aligner import Bertalign
from bertalign.chunk import TextChunk, split_into_paragraphs
def print_chunks(chunks: list[TextChunk]) -> None:
for i, chunk in enumerate(chunks, 1):
print(f"\nChunk {i}:")
print(f"Source text ({len(chunk.source_text)} chars):")
print(chunk.source_text)
print(f"\nTarget text ({len(chunk.target_text)} chars):")
print(chunk.target_text)
print("-" * 80)
class TestChunk(unittest.TestCase):
@@ -28,22 +39,27 @@ class TestChunk(unittest.TestCase):
self.target_text = f.read()
def test_create_aligned_chunks(self):
chunks = create_aligned_chunks(
self.source_text, self.target_text, max_chars=500
)
src_ps = split_into_paragraphs(self.source_text)
tgt_ps = split_into_paragraphs(self.target_text)
aligner = Bertalign(src_ps, tgt_ps, src_lang="zh", tgt_lang="en")
chunks = aligner.chunk(300)
self.assertIsInstance(chunks, list)
self.assertTrue(all(isinstance(chunk, TextChunk) for chunk in chunks))
for chunk in chunks:
self.assertIsInstance(chunk.source_text, str)
self.assertIsInstance(chunk.target_text, str)
print_chunks(chunks)
self.assertGreater(len(chunks), 0)
@unittest.skip("no")
def test_create_aligned_chunks_empty_input(self):
c = create_aligned_chunks("", self.target_text)
src_ps = split_into_paragraphs(self.source_text)
tgt_ps = split_into_paragraphs(self.target_text)
aligner = Bertalign(src_ps, tgt_ps, src_lang="zh", tgt_lang="en")
c = aligner.chunk()
self.assertIsNone(c)