eodag 3.5.1__py3-none-any.whl → 3.7.0__py3-none-any.whl

eodag/utils/s3.py

@@ -20,118 +20,458 @@ from __future__ import annotations
 import io
 import logging
 import os
-import zipfile
-from typing import TYPE_CHECKING, List, Optional
+import uuid
+from dataclasses import dataclass, field
+from datetime import datetime
+from typing import TYPE_CHECKING
 from urllib.parse import urlparse
+from zipfile import ZIP_STORED, ZipFile
 
 import boto3
 import botocore
+import botocore.exceptions
+from concurrent.futures import FIRST_COMPLETED, ThreadPoolExecutor, wait
+from stream_zip import ZIP_AUTO, stream_zip
 
 from eodag.plugins.authentication.aws_auth import AwsAuth
-from eodag.utils import get_bucket_name_and_prefix, guess_file_type
+from eodag.utils import (
+    StreamResponse,
+    get_bucket_name_and_prefix,
+    guess_file_type,
+    parse_le_uint16,
+    parse_le_uint32,
+)
 from eodag.utils.exceptions import (
     AuthenticationError,
+    DownloadError,
+    InvalidDataError,
     MisconfiguredError,
     NotAvailableError,
 )
 
 if TYPE_CHECKING:
-    from zipfile import ZipFile, ZipInfo
+    from typing import Iterable, Iterator, Literal, Optional
+    from zipfile import ZipInfo
 
     from mypy_boto3_s3.client import S3Client
+    from stream_zip import Method
 
     from eodag.api.product import EOProduct  # type: ignore
 
 logger = logging.getLogger("eodag.utils.s3")
 
+MIME_OCTET_STREAM = "application/octet-stream"
+
 
-def fetch(
-    bucket_name: str, key_name: str, start: int, len: int, client_s3: S3Client
+def fetch_range(
+    bucket_name: str, key_name: str, start: int, end: int, client_s3: S3Client
 ) -> bytes:
     """
     Range-fetches a S3 key.
 
     :param bucket_name: Bucket name of the object to fetch
     :param key_name: Key name of the object to fetch
-    :param start: Bucket name to fetch
-    :param len: Bucket name to fetch
+    :param start: Start byte position to fetch
+    :param end: End byte position to fetch
     :param client_s3: s3 client used to fetch the object
     :returns: Object bytes
     """
-    end = start + len - 1
-    s3_object = client_s3.get_object(
+    response = client_s3.get_object(
         Bucket=bucket_name, Key=key_name, Range="bytes=%d-%d" % (start, end)
     )
-    return s3_object["Body"].read()
+    return response["Body"].read()
 
 
-def parse_int(bytes: bytes) -> int:
+@dataclass
+class S3FileInfo:
     """
-    Parses 2 or 4 little-endian bits into their corresponding integer value.
-
-    :param bytes: bytes to parse
-    :returns: parsed int
+    Describe a S3 object with basic f_info and its download state.
     """
-    val = (bytes[0]) + ((bytes[1]) << 8)
-    if len(bytes) > 3:
-        val += ((bytes[2]) << 16) + ((bytes[3]) << 24)
-    return val
 
+    size: int
+    key: str
+    bucket_name: str
+    #: Path inside the ZIP archive if the file is stored inside a ZIP.
+    zip_filepath: Optional[str] = None
+    #: Offset in the ZIP archive where the file data starts.
+    data_start_offset: int = 0
+    #: MIME type of the file, defaulting to application/octet-stream.
+    #: It can be updated based on the file extension or content type.
+    data_type: str = MIME_OCTET_STREAM
+    #: Relative path of the file, if applicable (e.g., inside a ZIP archive).
+    rel_path: Optional[str] = None
+
+    # These fields hold the state for downloading
+    #: Offset in the logical (global) file stream where this file starts.
+    file_start_offset: int = 0
+    #: Mapping of futures to their start byte offsets, used to track download progress.
+    #: Each future corresponds to a chunk of data being downloaded.
+    #: The key is the future object, and the value is the start byte offset of that
+    #: chunk in the logical file stream.
+    futures: dict = field(default_factory=dict)
+    #: Buffers for downloaded data chunks, mapping start byte offsets to the actual data.
+    #: This allows for partial downloads and efficient memory usage.
+    #: The key is the start byte offset, and the value is the bytes data for that
+    #: offset. This is used to yield data in the correct order during streaming.
+    #: It is updated as chunks are downloaded.
+    buffers: dict[int, bytes] = field(default_factory=dict)
+    #: The next offset to yield in the file, used to track progress during downloading
+    #: and yielding chunks. It starts at 0 and is updated as data is yielded.
+    #: This allows the streaming process to continue from where it left off,
+    #: ensuring that all data is eventually yielded without duplication.
+    next_yield: int = 0
+
+
+def _prepare_file_in_zip(f_info: S3FileInfo, s3_client: S3Client):
+    """Update file information with the offset and size of the file inside the zip archive"""
+
+    splitted_path = f_info.key.split(".zip!")
+    f_info.key = f"{splitted_path[0]}.zip"
+    f_info.zip_filepath = splitted_path[-1]  # file path inside the ZIP archive
+
+    f_info.data_start_offset, f_info.size = file_position_from_s3_zip(
+        f_info.bucket_name,
+        f_info.key,
+        s3_client,
+        f_info.zip_filepath,
+    )
 
-def open_s3_zipped_object(
-    bucket_name: str, key_name: str, client_s3: S3Client, partial: bool = True
-) -> ZipFile:
-    """
-    Open s3 zipped object, without downloading it.
-
-    See https://stackoverflow.com/questions/41789176/how-to-count-files-inside-zip-in-aws-s3-without-downloading-it;
-    Based on https://stackoverflow.com/questions/51351000/read-zip-files-from-s3-without-downloading-the-entire-file
 
-    :param bucket_name: Bucket name of the object to fetch
-    :param key_name: Key name of the object to fetch
-    :param client_s3: s3 client used to fetch the object
-    :param partial: fetch partial data if only content info is needed
-    :returns: List of files in zip
+def _compute_file_ranges(
+    file_info: S3FileInfo,
+    byte_range: tuple[Optional[int], Optional[int]],
+    range_size: int,
+) -> Optional[list[tuple[int, int]]]:
     """
-    response = client_s3.head_object(Bucket=bucket_name, Key=key_name)
-    size = response["ContentLength"]
-
-    # End Of Central Directory bytes
-    eocd = fetch(bucket_name, key_name, size - 22, 22, client_s3)
-
-    # start offset and size of the central directory
-    cd_start = parse_int(eocd[16:20])
-    cd_size = parse_int(eocd[12:16])
-
-    # fetch central directory, append EOCD, and open as zipfile
-    cd = fetch(bucket_name, key_name, cd_start, cd_size, client_s3)
+    Compute the byte ranges to download for a single file, considering the overall requested range.
+
+    This function calculates which byte ranges within the given file should be downloaded,
+    based on the global requested byte range (`byte_range`) and the size of each chunk (`range_size`).
+    It accounts for possible offsets if the file is part of a ZIP archive or not aligned at offset zero.
+
+    :param file_info: The S3FileInfo object containing file metadata, including size, data offset,
+                      and its starting offset in the full logical file stream.
+    :param byte_range: A tuple (start, end) specifying the requested global byte range, where either
+                       value may be None to indicate an open-ended range.
+    :param range_size: The size of each download chunk in bytes.
+    :returns: A list of (start, end) tuples indicating byte ranges to download within this file,
+              or None if the file lies completely outside the requested range.
+    """
+    file_size = file_info.size
+    file_start_offset = file_info.file_start_offset
+    file_end_offset = file_start_offset + file_size - 1
+
+    range_start, range_end = byte_range
+
+    # Check if the file overlaps with the requested range
+    if range_end is not None and range_start is not None:
+        if file_end_offset < range_start or file_start_offset > range_end:
+            return None  # No overlap, skip this file
+
+    start = 0
+    end = file_size - 1
+
+    # Adjust start and end based on the requested range
+    if range_start is not None:
+        start = max(0, range_start - file_start_offset)
+    if range_end is not None:
+        end = min(file_size - 1, range_end - file_start_offset)
+
+    start += file_info.data_start_offset
+    end += file_info.data_start_offset
+
+    # Compute the ranges in chunks
+    ranges = []
+    for chunk_start in range(start, end + 1, range_size):
+        chunk_end = min(chunk_start + range_size - 1, end)
+        ranges.append((chunk_start, chunk_end))
+
+    return ranges
+
+
+def _chunks_from_s3_objects(
+    s3_client: S3Client,
+    files_info: list[S3FileInfo],
+    byte_range: tuple[Optional[int], Optional[int]],
+    range_size: int,
+    executor: ThreadPoolExecutor,
+) -> Iterator[tuple[int, Iterator[bytes]]]:
+    """Download chunks from S3 objects in parallel, respecting byte ranges and file order."""
+    for f_info in files_info:
+        ranges = _compute_file_ranges(f_info, byte_range, range_size)
+
+        if not ranges:
+            logger.debug("Skipping %s: no ranges to fetch", f_info.key)
+            continue
+
+        f_info.buffers = {}
+        f_info.next_yield = 0
+
+        futures = {}
+        for start, length in ranges:
+            future = executor.submit(
+                fetch_range,
+                f_info.bucket_name,
+                f_info.key,
+                start,
+                length,
+                s3_client,
+            )
+            futures[future] = start
+
+        f_info.futures = futures
+
+    # Combine all futures to wait on globally
+    all_futures = {
+        fut: (f_info, start)
+        for f_info in files_info
+        for fut, start in f_info.futures.items()
+    }
+
+    current_file_index = 0
+
+    # Yield chunks per file (one at a time)
+    while current_file_index < len(files_info):
+        current_info = files_info[current_file_index]
+
+        def chunks_generator() -> Iterator[bytes]:
+            """yield chunks of data for the current file."""
+            nonlocal current_file_index, all_futures
+            while current_info.next_yield < current_info.size:
+                # Wait for any futures to complete
+                done, _ = wait(all_futures.keys(), return_when=FIRST_COMPLETED)
+
+                for fut in done:
+                    f_info, start = all_futures.pop(fut)
+                    data = fut.result()
+                    f_info.buffers[start] = data
+
+                # Yield chunks as they are available
+                next_start = current_info.next_yield
+                while next_start in current_info.buffers:
+                    chunk = current_info.buffers.pop(next_start)
+                    if not isinstance(chunk, bytes):
+                        raise InvalidDataError(
+                            f"Expected bytes, got {type(chunk).__name__} in stream chunks: {chunk}"
+                        )
+                    yield chunk
+
+                    next_start += range_size
+                    current_info.next_yield = next_start
+
+                # If done with this file, stop yielding chunks for this file
+                if current_info.next_yield >= current_info.size:
+                    break
+
+        yield current_file_index, chunks_generator()
+
+        current_file_index += 1
+
+
+def _build_stream_response(
+    zip_filename: str,
+    files_info: list[S3FileInfo],
+    files_iterator: Iterator[tuple[int, Iterator[bytes]]],
+    compress: Literal["zip", "raw", "auto"],
+    executor: ThreadPoolExecutor,
+) -> StreamResponse:
+    """
+    Build a streaming HTTP response for one or multiple files from S3, supporting ZIP, raw, and multipart formats.
 
-    zip_data = (
-        cd + eocd if partial else fetch(bucket_name, key_name, 0, size, client_s3)
-    )
+    The response format depends on the `compress` parameter and the number of files:
 
-    zip = zipfile.ZipFile(io.BytesIO(zip_data))
+    - If `compress` is "zip" or "auto" with multiple files, returns a ZIP archive containing all files.
+    - If `compress` is "raw" and multiple files, returns a multipart/mixed response with each file as a part.
+    - If only one file is present and `compress` is "raw" or "auto", streams the file directly with its MIME type.
 
-    return zip
+    Response formats:
 
+    - ZIP archive (Content-Type: application/zip) with Content-Disposition for download.
+    - Multipart/mixed (Content-Type: multipart/mixed; boundary=...) with each file as a part.
+    - Single raw file stream with its MIME type and Content-Disposition for download.
 
-def list_files_in_s3_zipped_object(
-    bucket_name: str, key_name: str, client_s3: S3Client
-) -> List[ZipInfo]:
+    :param zip_filename: Base filename to use for the ZIP archive (without extension).
+    :param files_info: List of S3FileInfo objects describing each file (metadata, MIME type, etc.).
+    :param files_iterator: Iterator yielding (file_index, chunk_iterator) for streaming file contents.
+    :param compress: Output format:
+        - "zip": Always produce a ZIP archive.
+        - "raw": Stream files directly, as a single file or multipart.
+        - "auto": ZIP if multiple files, raw if single file.
+    :param executor: Executor used for concurrent streaming and cleanup.
+    :return: Streaming HTTP response with appropriate content, headers, and media type.
     """
-    List files in s3 zipped object, without downloading it.
-
-    See https://stackoverflow.com/questions/41789176/how-to-count-files-inside-zip-in-aws-s3-without-downloading-it;
-    Based on https://stackoverflow.com/questions/51351000/read-zip-files-from-s3-without-downloading-the-entire-file
-
-    :param bucket_name: Bucket name of the object to fetch
-    :param key_name: Key name of the object to fetch
-    :param client_s3: s3 client used to fetch the object
-    :returns: List of files in zip
+    headers = {
+        "Accept-Ranges": "bytes",
+    }
+
+    def _wrap_generator_with_cleanup(
+        generator: Iterable[bytes], executor: ThreadPoolExecutor
+    ) -> Iterator[bytes]:
+        try:
+            yield from generator
+        finally:
+            executor.shutdown(wait=True)
+
+    def _build_response(
+        content_gen: Iterable[bytes],
+        media_type: str,
+        extra_headers: dict[str, str] = {},
+    ) -> StreamResponse:
+        return StreamResponse(
+            content=_wrap_generator_with_cleanup(content_gen, executor),
+            media_type=media_type,
+            headers={**headers, **extra_headers},
+        )
+
+    zip_response = (len(files_info) > 1 and compress == "auto") or compress == "zip"
+
+    if zip_response:
+        modified_at = datetime.now()
+        perms = 0o600
+        total_file_size = sum(f.size for f in files_info)
+
+        def zip_stream() -> Iterator[
+            tuple[str, datetime, int, Method, Iterable[bytes]]
+        ]:
+            for index, chunks_generator in files_iterator:
+                yield (
+                    files_info[index].rel_path or files_info[index].key,
+                    modified_at,
+                    perms,
+                    ZIP_AUTO(total_file_size, level=0),
+                    chunks_generator,
+                )
+
+        return _build_response(
+            content_gen=stream_zip(zip_stream()),
+            media_type="application/zip",
+            extra_headers={
+                "content-disposition": f'attachment; filename="{zip_filename}.zip"'
+            },
+        )
+
+    elif len(files_info) > 1:
+        boundary = uuid.uuid4().hex
+
+        def multipart_stream():
+            current_index = -1
+            for index, chunks_generator in files_iterator:
+                if index != current_index:
+                    if current_index != -1:
+                        yield b"\r\n"
+                    filename = os.path.basename(files_info[index].key)
+                    yield (
+                        f"--{boundary}\r\n"
+                        f'Content-Disposition: attachment; filename="{filename}"\r\n'
+                        f"Content-Type: {files_info[index].data_type}\r\n\r\n"
+                    ).encode()
+                    current_index = index
+                yield from chunks_generator
+            yield f"\r\n--{boundary}--\r\n".encode()
+
+        return _build_response(
+            content_gen=multipart_stream(),
+            media_type=f"multipart/mixed; boundary={boundary}",
+        )
+
+    else:
+        index, chunks_generator = next(files_iterator)
+        first_chunk = next(chunks_generator)
+        filename = os.path.basename(files_info[index].key)
+
+        def single_file_stream() -> Iterator[bytes]:
+            yield first_chunk
+            yield from chunks_generator
+
+        return _build_response(
+            content_gen=single_file_stream(),
+            media_type=files_info[index].data_type,
+            extra_headers={"content-disposition": f'attachment; filename="{filename}"'},
+        )
+
+
+def stream_download_from_s3(
+    s3_client: S3Client,
+    files_info: list[S3FileInfo],
+    byte_range: tuple[Optional[int], Optional[int]] = (None, None),
+    compress: Literal["zip", "raw", "auto"] = "auto",
+    zip_filename: str = "archive",
+    range_size: int = 1024**2 * 8,
+    max_workers: int = 8,
+) -> StreamResponse:
     """
-    with open_s3_zipped_object(bucket_name, key_name, client_s3) as zip_file:
-        logger.debug("Found %s files in %s" % (len(zip_file.filelist), key_name))
-        return zip_file.filelist
+    Stream data from one or more S3 objects in chunks, with support for global byte ranges
+    and partial file extraction from ZIP archives.
+
+    This function downloads product data from S3 using concurrent range requests across one or
+    multiple files. It divides the requested data into chunks (default: 8 MiB) and issues
+    parallel HTTP range requests to optimize download throughput. This is particularly useful
+    for large files or datasets stored across multiple S3 objects.
+
+    If the S3 key refers to a path inside a ``.zip`` file (denoted by ``.zip!<internal_path>``),
+    the function extracts the specified file from the archive only if it is stored uncompressed
+    (ZIP method = STORE). Compressed formats (like DEFLATE) are not supported for partial ZIP extraction.
+
+    The function supports global byte range filtering via the ``byte_range`` parameter, which allows
+    requesting only a specific portion of the logical file stream across all provided objects.
+
+    Downloads are performed concurrently using a thread pool and HTTP range requests. Each chunk is downloaded
+    as a separate HTTP request and yielded in file order.
+
+    The ``compress`` parameter determines the output format:
+
+    - ``zip``: Always produce a ZIP archive containing all files.
+    - ``raw``: Stream files directly without wrapping, either as a single file or multipart response.
+    - ``auto``: Automatically select the format:
+        - raw stream if only a single file is requested
+        - ZIP archive if multiple files are requested
+
+    :param s3_client: A configured S3 client capable of making range requests.
+    :param files_info: List of S3FileInfo objects representing the files to download.
+    :param byte_range: Tuple (start, end) defining the inclusive global byte range to download across all objects.
+        Either value can be None to indicate open-ended range.
+    :param compress: Determines the output format of the streamed response.
+    :param zip_filename: The base filename to use when producing a ZIP archive (without extension).
+    :param range_size: The size in bytes of each download chunk. Defaults to 8 MiB.
+    :param max_workers: The maximum number of concurrent download tasks. Controls the size of the thread pool.
+    :return: Streaming HTTP response with content according to the requested format.
+    :raises DownloadError: If any error occurs during streaming from S3, including missing files or
+        unsupported ZIP compression.
+    """
+    offset = 0
+
+    executor = ThreadPoolExecutor(max_workers=max_workers)
+    try:
+        for f_info in files_info:
+            # Check if file is inside a ZIP
+            if ".zip!" in f_info.key:
+                future = executor.submit(_prepare_file_in_zip, f_info, s3_client)
+                f_info.futures[future] = 0
+
+        for f_info in files_info:
+            for future in f_info.futures:
+                future.result()
+            f_info.file_start_offset = offset
+            offset += f_info.size
+
+            if not f_info.data_type or f_info.data_type == MIME_OCTET_STREAM:
+                guessed = guess_file_type(f_info.key)
+                f_info.data_type = guessed or MIME_OCTET_STREAM
+
+        chunks_tuple = _chunks_from_s3_objects(
+            s3_client,
+            files_info,
+            byte_range,
+            range_size,
+            executor,
+        )
+
+        return _build_stream_response(
+            zip_filename, files_info, chunks_tuple, compress, executor
+        )
+    except Exception as e:
+        executor.shutdown(wait=True)
+        raise DownloadError(str(e)) from e
 
 
 def update_assets_from_s3(
@@ -229,3 +569,203 @@ def update_assets_from_s3(
         raise NotAvailableError(
             f"assets for product {prefix} could not be found"
         ) from e
+
+
+# ----- ZIP section -----
+
+
+def open_s3_zipped_object(
+    bucket_name: str,
+    key_name: str,
+    s3_client,
+    zip_size: Optional[int] = None,
+    partial: bool = True,
+) -> tuple[ZipFile, bytes]:
+    """
+    Fetches the central directory and EOCD (End Of Central Directory) from an S3 object and opens a ZipFile in memory.
+
+    This function retrieves the ZIP file's central directory and EOCD by performing range requests on the S3 object.
+    It supports partial fetching (only the central directory and EOCD) for efficiency, or full ZIP download if needed.
+
+    :param bucket_name: Name of the S3 bucket containing the ZIP file.
+    :param key_name: Key (path) of the ZIP file in the S3 bucket.
+    :param s3_client: S3 client instance used to perform range requests.
+    :param zip_size: Size of the ZIP file in bytes. If None, it will be determined via a HEAD request.
+    :param partial: If True, only fetch the central directory and EOCD. If False, fetch the entire ZIP file.
+    :return: Tuple containing the opened ZipFile object and the central directory bytes.
+    :raises InvalidDataError: If the EOCD signature is not found in the last 64KB of the file.
+    """
+    # EOCD is at least 22 bytes, but can be longer if ZIP comment exists.
+    # For simplicity, we fetch last 64KB max (max EOCD + comment length allowed by ZIP spec)
+    if zip_size is None:
+        response = s3_client.head_object(Bucket=bucket_name, Key=key_name)
+        zip_size = int(response["ContentLength"])
+
+    fetch_size = min(65536, zip_size)
+    eocd_search = fetch_range(
+        bucket_name, key_name, zip_size - fetch_size, zip_size - 1, s3_client
+    )
+
+    # Find EOCD signature from end: 0x06054b50 (little endian)
+    eocd_signature = b"\x50\x4b\x05\x06"
+    eocd_offset = eocd_search.rfind(eocd_signature)
+    if eocd_offset == -1:
+        raise InvalidDataError("EOCD signature not found in last 64KB of the file.")
+
+    eocd = eocd_search[eocd_offset : eocd_offset + 22]
+
+    cd_size = parse_le_uint32(eocd[12:16])
+    cd_start = parse_le_uint32(eocd[16:20])
+
+    # Fetch central directory
+    cd_data = fetch_range(
+        bucket_name, key_name, cd_start, cd_start + cd_size - 1, s3_client
+    )
+
+    zip_data = (
+        cd_data + eocd
+        if partial
+        else fetch_range(bucket_name, key_name, 0, zip_size - 1, s3_client)
+    )
+    zipf = ZipFile(io.BytesIO(zip_data))
+    return zipf, cd_data
+
+
+def _parse_central_directory_entry(cd_data: bytes, offset: int) -> dict[str, int]:
+    """
+    Parse one central directory file header entry starting at offset.
+    Returns dict with relative local header offset and sizes.
+    """
+    signature = cd_data[offset : offset + 4]
+    if signature != b"PK\x01\x02":
+        raise InvalidDataError("Bad central directory file header signature")
+
+    filename_len = parse_le_uint16(cd_data[offset + 28 : offset + 30])
+    extra_len = parse_le_uint16(cd_data[offset + 30 : offset + 32])
+    comment_len = parse_le_uint16(cd_data[offset + 32 : offset + 34])
+
+    relative_offset = parse_le_uint32(cd_data[offset + 42 : offset + 46])
+
+    header_size = 46 + filename_len + extra_len + comment_len
+
+    return {
+        "relative_offset": relative_offset,
+        "header_size": header_size,
+        "filename_len": filename_len,
+        "extra_len": extra_len,
+        "comment_len": comment_len,
+        "total_size": header_size,
+    }
+
+
+def _parse_local_file_header(local_header_bytes: bytes) -> int:
+    """
+    Parse local file header to find total header size:
+    fixed 30 bytes + filename length + extra field length
+    """
+    if local_header_bytes[0:4] != b"PK\x03\x04":
+        raise InvalidDataError("Bad local file header signature")
+
+    filename_len = parse_le_uint16(local_header_bytes[26:28])
+    extra_len = parse_le_uint16(local_header_bytes[28:30])
+    total_size = 30 + filename_len + extra_len
+    return total_size
+
+
+def file_position_from_s3_zip(
+    s3_bucket: str,
+    object_key: str,
+    s3_client,
+    target_filepath: str,
+) -> tuple[int, int]:
+    """
+    Get the start position and size of a specific file inside a ZIP archive stored in S3.
+    This function assumes the file is uncompressed (ZIP_STORED).
+
+    The returned tuple contains:
+
+    - **file_data_start**: The byte offset where the file data starts in the ZIP archive.
+    - **file_size**: The size of the file in bytes.
+
+    :param s3_bucket: The S3 bucket name.
+    :param object_key: The S3 object key for the ZIP file.
+    :param s3_client: The Boto3 S3 client.
+    :param target_filepath: The file path inside the ZIP archive to locate.
+    :return: A tuple (file_data_start, file_size)
+    :raises FileNotFoundError: If the target file is not found in the ZIP archive.
+    :raises NotImplementedError: If the file is not uncompressed (ZIP_STORED)
+    """
+    zipf, cd_data = open_s3_zipped_object(s3_bucket, object_key, s3_client)
+
+    # Find file in zipf.filelist to get its index and f_info
+    target_info = None
+    cd_offset = 0
+    for fi in zipf.filelist:
+        if fi.filename == target_filepath:
+            target_info = fi
+            break
+        # 46 is the fixed size (in bytes) of the Central Directory File Header according to the ZIP spec
+        cd_entry_len = (
+            46 + len(fi.filename.encode("utf-8")) + len(fi.extra) + len(fi.comment)
+        )
+        cd_offset += cd_entry_len
+
+    zipf.close()
+
+    if target_info is None:
+        raise FileNotFoundError(f"File {target_filepath} not found in ZIP archive")
+
+    if target_info.compress_type != ZIP_STORED:
+        raise NotImplementedError("Only uncompressed files (ZIP_STORED) are supported.")
+
+    # Parse central directory entry to get relative local header offset
+    cd_entry = cd_data[
+        cd_offset : cd_offset
+        + (
+            46
+            + len(target_info.filename.encode("utf-8"))
+            + len(target_info.extra)
+            + len(target_info.comment)
+        )
+    ]
+    cd_entry_info = _parse_central_directory_entry(cd_entry, 0)
+
+    local_header_offset = cd_entry_info["relative_offset"]
+
+    # Fetch local file header from S3 (at least 30 bytes + filename + extra field)
+    # We'll fetch 4 KB max to cover large filenames/extra fields safely
+    local_header_fetch_size = 4096
+    local_header_bytes = fetch_range(
+        s3_bucket,
+        object_key,
+        local_header_offset,
+        local_header_offset + local_header_fetch_size - 1,
+        s3_client,
+    )
+
+    local_header_size = _parse_local_file_header(local_header_bytes)
+
+    # Calculate file data start and end offsets
+    file_data_start = local_header_offset + local_header_size
+
+    return file_data_start, target_info.file_size
+
+
+def list_files_in_s3_zipped_object(
+    bucket_name: str, key_name: str, s3_client: S3Client
+) -> list[ZipInfo]:
+    """
+    List files in s3 zipped object, without downloading it.
+
+    See https://stackoverflow.com/questions/41789176/how-to-count-files-inside-zip-in-aws-s3-without-downloading-it;
+    Based on https://stackoverflow.com/questions/51351000/read-zip-files-from-s3-without-downloading-the-entire-file
+
+    :param bucket_name: Bucket name of the object to fetch
+    :param key_name: Key name of the object to fetch
+    :param s3_resource: s3 resource used to fetch the object
+    :returns: List of files in zip
+    """
+    zip_file, _ = open_s3_zipped_object(bucket_name, key_name, s3_client)
+    with zip_file:
+        logger.debug("Found %s files in %s" % (len(zip_file.filelist), key_name))
+        return zip_file.filelist