sapiopycommons 2025.9.19a762__tar.gz → 2025.9.23a765__tar.gz

{sapiopycommons-2025.9.19a762 → sapiopycommons-2025.9.23a765}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: sapiopycommons
-Version: 2025.9.19a762
+Version: 2025.9.23a765
 Summary: Official Sapio Python API Utilities Package
 Project-URL: Homepage, https://github.com/sapiosciences
 Author-email: Jonathan Steck <jsteck@sapiosciences.com>, Yechen Qiao <yqiao@sapiosciences.com>

{sapiopycommons-2025.9.19a762 → sapiopycommons-2025.9.23a765}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sapiopycommons"
-version='2025.09.19a762'
+version='2025.09.23a765'
 authors = [
     { name="Jonathan Steck", email="jsteck@sapiosciences.com" },
     { name="Yechen Qiao", email="yqiao@sapiosciences.com" },

{sapiopycommons-2025.9.19a762 → sapiopycommons-2025.9.23a765}/src/sapiopycommons/files/file_util.py

@@ -1,4 +1,7 @@
+import gzip
 import io
+import tarfile
+import time
 import warnings
 import zipfile
 
@@ -322,7 +325,7 @@ class FileUtil:
     @staticmethod
     def zip_files(files: dict[str, str | bytes]) -> bytes:
         """
-        Create a zip file for a collection of files.
+        Create a .zip file for a collection of files.
 
         :param files: A dictionary of file name to file data as a string or bytes.
         :return: The bytes for a zip file containing the input files.
@@ -335,6 +338,130 @@ class FileUtil:
             # throws an I/O exception.
             return zip_buffer.getvalue()
 
+    # FR-47422: Add a function for unzipping files that may have been zipped by the above function.
+    @staticmethod
+    def unzip_files(zip_file: bytes) -> dict[str, bytes]:
+        """
+        Decompress a .zip file from an in-memory bytes object and extracts all files into a dictionary.
+
+        :param zip_file: The bytes of the zip file to be decompressed.
+        :return: A dictionary of file name to file bytes for each file in the zip.
+        """
+        extracted_files: dict[str, bytes] = {}
+        with io.BytesIO(zip_file) as zip_buffer:
+            with zipfile.ZipFile(zip_buffer, "r") as zip_file:
+                for file_name in zip_file.namelist():
+                    with zip_file.open(file_name) as file:
+                        extracted_files[file_name] = file.read()
+        return extracted_files
+
+    # FR-47422: Add functions for compressing and decompressing .gz, .tar, and .tar.gz files.
+    @staticmethod
+    def gzip_file(file_data: bytes | str) -> bytes:
+        """
+        Create a .gz file for a single file.
+
+        :param file_data: The file data to be compressed as bytes or a string.
+        :return: The bytes of the gzip-compressed file.
+        """
+        return gzip.compress(file_data.encode() if isinstance(file_data, str) else file_data)
+
+    @staticmethod
+    def ungzip_file(gzip_file: bytes) -> bytes:
+        """
+        Decompress a .gz file.
+
+        :param gzip_file: The bytes of the gzip-compressed file.
+        :return: The decompressed file data as bytes.
+        """
+        return gzip.decompress(gzip_file)
+
+    @staticmethod
+    def tar_files(files: dict[str, str | bytes]) -> bytes:
+        """
+        Create a .tar file for a collection of files.
+
+        :param files: A dictionary of file name to file data as a string or bytes.
+        :return: The bytes for a tar file containing the input files.
+        """
+        with io.BytesIO() as tar_buffer:
+            with tarfile.open(fileobj=tar_buffer, mode="w") as tar:
+                for name, data in files.items():
+                    if isinstance(data, str):
+                        data: bytes = data.encode('utf-8')
+
+                    tarinfo = tarfile.TarInfo(name=name)
+                    tarinfo.size = len(data)
+                    tarinfo.mtime = int(time.time())
+
+                    with io.BytesIO(data) as file:
+                        tar.addfile(tarinfo=tarinfo, fileobj=file)
+
+            tar_buffer.seek(0)
+            return tar_buffer.getvalue()
+
+    @staticmethod
+    def untar_files(tar_file: bytes) -> dict[str, bytes]:
+        """
+        Decompress a .tar file from an in-memory bytes object and extracts all files into a dictionary.
+
+        :param tar_file: The bytes of the tar file to be decompressed.
+        :return: A dictionary of file name to file bytes for each file in the tar.
+        """
+        extracted_files: dict[str, bytes] = {}
+        with io.BytesIO(tar_file) as tar_buffer:
+            with tarfile.open(fileobj=tar_buffer, mode="r") as tar:
+                for member in tar.getmembers():
+                    if member.isfile():
+                        file_obj = tar.extractfile(member)
+                        if file_obj:
+                            with file_obj:
+                                extracted_files[member.name] = file_obj.read()
+        return extracted_files
+
+    @staticmethod
+    def tar_gzip_files(files: dict[str, str | bytes]) -> bytes:
+        """
+        Create a .tar.gz file for a collection of files.
+
+        :param files: A dictionary of file name to file data as a string or bytes.
+        :return: The bytes for a tar.gz file containing the input files.
+        """
+        with io.BytesIO() as tar_buffer:
+            with tarfile.open(fileobj=tar_buffer, mode="w:gz") as tar:
+                for name, data in files.items():
+                    if isinstance(data, str):
+                        data: bytes = data.encode('utf-8')
+
+                    tarinfo = tarfile.TarInfo(name=name)
+                    tarinfo.size = len(data)
+                    tarinfo.mtime = int(time.time())
+
+                    with io.BytesIO(data) as file:
+                        tar.addfile(tarinfo=tarinfo, fileobj=file)
+
+            tar_buffer.seek(0)
+            return tar_buffer.getvalue()
+
+    @staticmethod
+    def untar_gzip_files(tar_gzip_file: bytes) -> dict[str, bytes]:
+        """
+        Decompress a .tar.gz file from an in-memory bytes object and extracts all files into a dictionary.
+
+        :param tar_gzip_file: The bytes of the tar.gz file to be decompressed.
+        :return: A dictionary of file name to file bytes for each file in the tar.gz
+        """
+        extracted_files: dict[str, bytes] = {}
+        with io.BytesIO(tar_gzip_file) as tar_buffer:
+            with tarfile.open(fileobj=tar_buffer, mode="r:gz") as tar:
+                for member in tar.getmembers():
+                    if member.isfile():
+                        file_obj = tar.extractfile(member)
+                        if file_obj:
+                            with file_obj:
+                                extracted_files[member.name] = file_obj.read()
+        return extracted_files
+
     # Deprecated functions:
 
     # FR-46097 - Add write file request shorthand functions to FileUtil.

sapiopycommons-2025.9.23a765/tests/file_compression_tests.py

@@ -0,0 +1,39 @@
+from sapiopycommons.files.file_util import FileUtil
+
+files: dict[str, bytes] = {
+    "file1.txt": b"Hello, this is the content of file 1.",
+    "file2.txt": b"This is file 2, containing some sample text.",
+    "file3.txt": b"File 3 is here with its own unique content.",
+    "file4.txt": b"And finally, this is file 4."
+}
+
+print("files -> .zip -> files")
+zip_file: bytes = FileUtil.zip_files(files)
+with open("test_output/test.zip", "wb") as f:
+    f.write(zip_file)
+unzipped_files: dict[str, bytes] = FileUtil.unzip_files(zip_file)
+for filename, content in unzipped_files.items():
+    print(f"\t{filename}: {content.decode('utf-8')}")
+
+print("file -> .gz -> file")
+gz_file: bytes = FileUtil.gzip_file(files["file1.txt"])
+with open("test_output/file1.txt.gz", "wb") as f:
+    f.write(gz_file)
+ungzipped_file: bytes = FileUtil.ungzip_file(gz_file)
+print(f"\tUngzipped file1.txt: {ungzipped_file.decode('utf-8')}")
+
+print("files -> .tar -> files")
+tarfile: bytes = FileUtil.tar_files(files)
+with open("test_output/test.tar", "wb") as f:
+    f.write(tarfile)
+untarred_files: dict[str, bytes] = FileUtil.untar_files(tarfile)
+for filename, content in untarred_files.items():
+    print(f"\t{filename}: {content.decode('utf-8')}")
+
+print("files -> .tar.gz -> files")
+tgz_file: bytes = FileUtil.tar_gzip_files(files)
+with open("test_output/test.tar.gz", "wb") as f:
+    f.write(tgz_file)
+untgzipped_files: dict[str, bytes] = FileUtil.untar_gzip_files(tgz_file)
+for filename, content in untgzipped_files.items():
+    print(f"\t{filename}: {content.decode('utf-8')}")