"""Tests for file_processor module."""
import os
import pytest
import tempfile
import json
from factsdb.file_processor import FileProcessor
@pytest.fixture
def processor():
return FileProcessor()
class TestDetectFileType:
def test_txt(self, processor):
assert processor.detect_file_type("file.txt") == "txt"
def test_html(self, processor):
assert processor.detect_file_type("page.html") == "html"
def test_htm(self, processor):
assert processor.detect_file_type("page.htm") == "htm"
def test_pdf(self, processor):
assert processor.detect_file_type("doc.pdf") == "pdf"
def test_json(self, processor):
assert processor.detect_file_type("data.json") == "json"
def test_xml(self, processor):
assert processor.detect_file_type("data.xml") == "xml"
def test_uppercase(self, processor):
assert processor.detect_file_type("file.TXT") == "txt"
def test_no_extension(self, processor):
assert processor.detect_file_type("file") == ""
class TestIsSupportedFileType:
def test_supported(self, processor, tmp_path):
for ext in ["txt", "md", "log", "html", "htm", "pdf", "xml", "json"]:
f = tmp_path / f"file.{ext}"
f.touch()
assert processor.is_supported_file_type(str(f)) is True
def test_unsupported(self, processor, tmp_path):
f = tmp_path / "file.xyz"
f.touch()
assert processor.is_supported_file_type(str(f)) is False
class TestExtractTextFromFile:
def test_text_file(self, processor, tmp_path):
f = tmp_path / "test.txt"
f.write_text("Hello World")
assert processor.extract_text_from_file(str(f)) == "Hello World"
def test_html_file(self, processor, tmp_path):
f = tmp_path / "test.html"
f.write_text("Hello
")
text = processor.extract_text_from_file(str(f))
assert "Hello" in text
def test_json_file(self, processor, tmp_path):
f = tmp_path / "test.json"
data = {"key": "value"}
f.write_text(json.dumps(data))
text = processor.extract_text_from_file(str(f))
assert "key" in text
def test_unsupported_type(self, processor, tmp_path):
f = tmp_path / "test.xyz"
f.write_text("x")
with pytest.raises(Exception, match="Unsupported file type"):
processor.extract_text_from_file(str(f))
def test_nonexistent_file(self, processor):
with pytest.raises(Exception):
processor.extract_text_from_file("/nonexistent/path/file.txt")
class TestHtmlExtraction:
def test_removes_scripts(self, processor, tmp_path):
f = tmp_path / "test.html"
f.write_text("Content
")
text = processor.extract_text_from_file(str(f))
assert "alert" not in text
assert "Content" in text
def test_removes_styles(self, processor, tmp_path):
f = tmp_path / "test.html"
f.write_text("Content
")
text = processor.extract_text_from_file(str(f))
assert "Content" in text
class TestGetFileInfo:
def test_basic_info(self, processor, tmp_path):
f = tmp_path / "test.txt"
f.write_text("hello")
info = processor.get_file_info(str(f))
assert info["name"] == "test.txt"
assert info["type"] == "txt"
assert info["is_supported"] is True
assert info["size"] == 5
def test_unsupported_info(self, processor, tmp_path):
f = tmp_path / "test.xyz"
f.write_text("hello")
info = processor.get_file_info(str(f))
assert info["is_supported"] is False