def test_util_file_create_if_necessary__file_read(self):
if self._test_exec("test_file_create_if_necessary"):
Prg = self.Prg
Fname = os.path.join(Prg["DirWork"],
"test_file_create_if_necessary.txt")
util.file_del(Fname)
Content = "Cloud\nRain\nSun\r\nSnow "
Created = util.file_create_if_necessary(Prg,
Fname,
ContentDefault=Content)
self.assertTrue(Created)
RetRead, ContentReadAll = util.file_read_all(Prg, Fname)
self.assertEqual(ContentReadAll, Content)
ContentReadLines = util.file_read_lines(Prg, Fname, Lower=True)
LinesWanted = ["cloud\n", "rain\n", "sun\r\n", "snow "]
self.assertEqual(LinesWanted, ContentReadLines)
ContentReadLines = util.file_read_lines(Prg, Fname, Strip=True)
LinesWanted = ["Cloud", "Rain", "Sun", "Snow"]
self.assertEqual(LinesWanted, ContentReadLines)
util.file_del(Fname)
def test_util_file_read_lines(self):
if self._test_exec("test_file_read_lines"):
Prg = self.Prg
Fname = os.path.join(Prg["DirWork"], "test_file_read_lines.txt")
util.file_write(Prg, Fname=Fname, Content="cat\ndog\nelephant")
Lines = util.file_read_lines(Prg, Fname, Strip=True)
self.assertEqual(Lines, ["cat", "dog", "elephant"])
util.file_del(Fname)
def test_fun_pdf_to_text_converter(self):
if self._test_exec("test_fun_pdf_to_text_converter"):
Prg = self.Prg
FileTxt = os.path.join(Prg["DirWork"],
"test_converted_from_pdf.txt")
util.file_del(FileTxt)
FilePdf = os.path.join(Prg["DirTestFiles"],
"test_pdf_conversion.pdf")
Prg["ConverterPdfToText"](Prg, FilePdf, FileTxt)
FileLines = util.file_read_lines(Prg, FileTxt, Strip=True)
self.assertEqual(FileLines[0], "This is new document.")
util.file_del(FileTxt)
def test_collect_docs_from_working_dir(self):
if self._test_exec("test_collect_docs_from_working_dir"):
Prg = self.Prg
FileName = "test_file_document_example.txt"
FilePath = os.path.join(Prg["DirDocuments"], FileName)
util.file_del(FilePath)
util.file_write(Prg, Fname=FilePath, Content="example text")
DocumentsAvailable = document.document_objects_collect_from_dir_documents(
Prg)
self.assertIn(util.filename_without_extension(FileName),
DocumentsAvailable)
util.file_del(FilePath)
def doc_objects_delete__basename(Prg, BaseNameNoExt):
if DocObj := Prg["DocumentObjectsLoaded"].pop(BaseNameNoExt, None):
util.file_del(DocObj["FileOrigPathAbs"])
util.file_del(DocObj["FileTextPathAbs"])
util.file_del(DocObj["FileIndex"])
util.file_del(DocObj["FileSentences"])
util_json_obj.doc_source_webpages_update_in_file_and_Prg(
Prg, BaseNameNoExtRemove=BaseNameNoExt)
def test_docs_copy_samples_into_dir(self):
if self._test_exec("test_collect_docs_from_working_dir"):
Prg = self.Prg
DirTestBaseName = "test_" + str(int(time.time()))
DirTest = os.path.join(Prg["DirWork"], DirTestBaseName)
util.dir_create_if_necessary(Prg, DirTest)
document.docs_copy_samples_into_dir(Prg, DirTest)
Files = util.files_abspath_collect_from_dir(DirTest)
FileNamesInOneLine = " ".join(Files)
Wanted = "DanielDefoe__LifeAdventuresRobinsonCrusoe__gutenberg_org_521-0.txt"
self.assertIn(Wanted, FileNamesInOneLine)
for FileAbsPath in Files:
util.file_del(FileAbsPath)
DelRes = util.dir_delete_if_exist(Prg, DirTest)
self.assertEqual("deleted", DelRes)
def test_file_create_sentences__create_index(self):
self.maxDiff = None
if self._test_exec("test_file_create_sentences__create_index"):
Prg = self.Prg
FileSentences = os.path.join(Prg["DirWork"],
"test_file_create_sentences.txt")
util.file_del(FileSentences)
Sample = 'He is my friend. "This is \n the next - city, London -- here, in London, the sky is nice." Is this the third line, or a Book about London?'
seeker.file_sentence_create(Prg, FileSentences, Sample)
Wanted = [
"He is my friend. \n", # detect London only once from this sentence:
'"This is the next - city, London -- here, in London, the sky is nice." \n',
"Is this the third line, or a Book about London?"
]
LinesFromFile = util.file_read_lines(Prg, FileSentences)
self.assertEqual(Wanted, LinesFromFile)
FileIndex = os.path.join(Prg["DirWork"],
"test_file_create_index.txt")
util.file_del(FileIndex)
seeker.file_index_create(Prg,
FileIndex,
FileSentences,
ForcedWrite=True)
#seeker.file_index_create(Prg, "/tmp/index.txt", FileSentences)
# print(util.file_read_all(Prg, FileIndex))
MultiSub = Prg["SubSentenceMultiplier"]
MultiSubAndWord = Prg["SubSentenceMultiplier"] * Prg[
"WordPositionMultiplier"]
_Status, WordPosition = util_json_obj.obj_from_file(FileIndex)
self.assertEqual(set(WordPosition["london"]),
set([10100, 10201, 20104]))
util.file_del(FileSentences)
util.file_del(FileIndex)
def doc_objects_delete__file_abspath(Prg, FileAbsPathWithExt):
BaseName = os.path.basename(FileAbsPathWithExt)
BaseNameNoExt = util.filename_without_extension(BaseName)
doc_objects_delete__basename(Prg, BaseNameNoExt)
util.file_del(FileAbsPathWithExt
) # del orig file in every case, if DocObj doesn't exis
def test_util_file_write_append_del(self):
if self._test_exec("test_file_write_append_del"):
Prg = self.Prg
Ret = util.file_write(Prg, Fname="")
self.assertFalse(Ret)
Content = "apple "
Fname = os.path.join(Prg["DirWork"], "test_file_write.txt")
RetWrite = util.file_write(Prg, Fname=Fname, Content=Content)
RetAppend = util.file_append(Prg, Fname=Fname, Content="tree")
self.assertTrue(RetWrite)
self.assertTrue(RetAppend)
RetRead, ContentRead = util.file_read_all(Prg, Fname)
self.assertTrue(RetRead)
self.assertEqual(ContentRead, "apple tree")
FileState, FileGzipped = util.file_is_gzipped(Prg, Fname)
self.assertEqual("file_exists", FileState)
self.assertEqual("not_gzipped", FileGzipped)
Sample = "Árvíztűrő tükörfúrógép"
RetWriteGz = util.file_write(Prg,
Fname=Fname,
Content=Sample,
Gzipped=True)
self.assertTrue(RetWriteGz)
RetReadGz, ContentReadGz = util.file_read_all(Prg,
Fname,
Gzipped=True)
self.assertTrue(RetReadGz)
self.assertEqual(ContentReadGz, Sample)
FileState, FileGzipped = util.file_is_gzipped(Prg, Fname)
self.assertEqual("file_exists", FileState)
self.assertEqual("gzipped", FileGzipped)
BinWanted = Sample.encode()
util.file_write_utf8_error_avoid(Prg, Fname, Sample)
BinFromFile = util.file_read_all_simple(Fname, "rb")
# print("\n######### >>" + util.file_read_all(Prg, Fname)[1] + "<<")
# print("\n######### >>", Sample.encode(), "<<")
self.assertEqual(BinWanted, BinFromFile)
FileWriteRet = util.file_write_with_check(Prg, Fname, Sample)
TxtFromFile = util.file_read_all_simple(Fname)
self.assertEqual(TxtFromFile, Sample)
self.assertTrue(FileWriteRet)
util.file_write_with_check(Prg, Fname,
"") # clear the content of the file
# writing is unsuccessful because writer fun doesn't do anything
def empty_writer_fun(Prg, Fname, Sample):
pass
FileWriteRet = util.file_write_with_check(
Prg, Fname, Sample, WriterFun=empty_writer_fun)
self.assertFalse(FileWriteRet)
RetDel1 = util.file_del(Fname)
RetDel2 = util.file_del(Fname)
self.assertTrue(RetDel1)
self.assertFalse(RetDel2)
FileState, FileGzipped = util.file_is_gzipped(Prg, Fname)
self.assertEqual("file_not_found", FileState)
self.assertEqual("", FileGzipped)