PyS3Uploader 0.1.1__py3-none-any.whl → 0.2.0__py3-none-any.whl

{pys3uploader-0.1.1.dist-info → pys3uploader-0.2.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: PyS3Uploader
-Version: 0.1.1
+Version: 0.2.0
 Summary: Python module to upload objects to an S3 bucket.
 Author-email: Vignesh Rao <svignesh1793@gmail.com>
 License: MIT License
@@ -63,7 +63,7 @@ Requires-Dist: recommonmark; extra == "dev"
 
 **Activity**
 
-[![GitHub Repo created](https://img.shields.io/date/1618966420)][repo]
+[![GitHub Repo created](https://img.shields.io/date/1760313686)][repo]
 [![GitHub commit activity](https://img.shields.io/github/commit-activity/y/thevickypedia/PyS3Uploader)][repo]
 [![GitHub last commit](https://img.shields.io/github/last-commit/thevickypedia/PyS3Uploader)][repo]
 

pys3uploader-0.2.0.dist-info/RECORD

@@ -0,0 +1,11 @@
+s3/__init__.py,sha256=yLvvl4-uTLZwhdhCMQpWq5juX_zFuYAfKSf4aB0WjZw,66
+s3/exceptions.py,sha256=hH3jlMOe8yjBatQK9EdndWZz4QESU74KSY_iDhQ37SY,2585
+s3/logger.py,sha256=oH540oq8jY723jA4lDWlgfFPLbNgGXTkDwFpB7TLO_o,1196
+s3/tree.py,sha256=DiQ2ekMMaj2m_P3-iKkEqSuJCJZ_UZxcAwHtAoPVa5c,1824
+s3/uploader.py,sha256=IAlFrEjfBuexrfmBPGN9OZAfHjQuwcGRzWi2es0r_fU,11154
+s3/utils.py,sha256=0kcG0aE2olHhC8thaUEwx2J8tOI2-2TGCk6E6U-PiKw,2058
+pys3uploader-0.2.0.dist-info/LICENSE,sha256=8k-hEraOzyum0GvmmK65YxNRTFXK7eIFHJ0OshJXeTk,1068
+pys3uploader-0.2.0.dist-info/METADATA,sha256=IXSmHXJJndlnd_6MHlpZrcVILPni8VUbVNJYQEjMIR8,7286
+pys3uploader-0.2.0.dist-info/WHEEL,sha256=beeZ86-EfXScwlR_HKu4SllMC9wUEj_8Z_4FJ3egI2w,91
+pys3uploader-0.2.0.dist-info/top_level.txt,sha256=iQp4y1P58Q633gj8M08kHE4mqqT0hixuDWcniDk_RJ4,3
+pys3uploader-0.2.0.dist-info/RECORD,,

s3/__init__.py

@@ -1,3 +1,3 @@
 from s3.uploader import Uploader  # noqa: F401
 
-version = "0.1.1"
+version = "0.2.0"

s3/uploader.py

@@ -11,7 +11,7 @@ from tqdm import tqdm
 
 from s3.exceptions import BucketNotFound
 from s3.logger import default_logger
-from s3.utils import UploadResults, getenv, urljoin
+from s3.utils import UploadResults, convert_to_folder_structure, getenv, urljoin
 
 
 class Uploader:
@@ -29,6 +29,7 @@ class Uploader:
         upload_dir: str,
         s3_prefix: str = None,
         exclude_path: str = None,
+        overwrite: bool = False,
         region_name: str = None,
         profile_name: str = None,
         aws_access_key_id: str = None,
@@ -42,6 +43,7 @@ class Uploader:
             upload_dir: Full path of the directory to be uploaded.
             s3_prefix: Particular bucket prefix within which the upload should happen.
             exclude_path: Full directory path to exclude from S3 object prefix.
+            overwrite: Boolean flag to overwrite files in S3.
             region_name: Name of the AWS region.
             profile_name: AWS profile name.
             aws_access_key_id: AWS access key ID.
@@ -69,16 +71,24 @@ class Uploader:
             aws_secret_access_key=aws_secret_access_key or getenv("AWS_SECRET_ACCESS_KEY"),
         )
         self.s3 = self.session.resource(service_name="s3", config=self.RETRY_CONFIG)
+
         self.logger = logger or default_logger()
+
+        self.bucket_name = bucket_name
         self.upload_dir = upload_dir or getenv("UPLOAD_DIR", "UPLOAD_SOURCE")
         self.s3_prefix = s3_prefix
         self.exclude_path = exclude_path
-        self.bucket_name = bucket_name
-        # noinspection PyUnresolvedReferences
-        self.bucket: boto3.resources.factory.s3.Bucket = None
+        self.overwrite = overwrite
+
         self.results = UploadResults()
         self.start = time.time()
 
+        # noinspection PyUnresolvedReferences
+        self.bucket: boto3.resources.factory.s3.Bucket = None
+        # noinspection PyUnresolvedReferences
+        self.bucket_objects: boto3.resources.factory.s3.ObjectSummary = []
+        self.object_size_map: Dict[str, int] = {}
+
     def init(self) -> None:
         """Instantiates the bucket instance.
 
@@ -106,6 +116,9 @@ class Uploader:
         self.upload_dir = os.path.abspath(self.upload_dir)
         # noinspection PyUnresolvedReferences
         self.bucket: boto3.resources.factory.s3.Bucket = self.s3.Bucket(self.bucket_name)
+        # noinspection PyUnresolvedReferences
+        self.bucket_objects: boto3.resources.factory.s3.ObjectSummary = [obj for obj in self.bucket.objects.all()]
+        self.object_size_map = {obj.key: obj.size for obj in self.bucket_objects}
 
     def exit(self) -> None:
         """Exits after printing results, and run time."""
@@ -115,28 +128,59 @@ class Uploader:
         )
         self.logger.info("Run Time: %.2fs", time.time() - self.start)
 
-    def _uploader(self, objectpath: str, filepath: str) -> None:
+    def _proceed_to_upload(self, filepath: str, objectpath: str) -> bool:
+        """Compares file size if the object already exists in S3.
+
+        Args:
+            filepath: Source filepath.
+            objectpath: S3 object path.
+
+        Returns:
+            bool:
+            Returns a boolean flag to indicate upload flag.
+        """
+        if self.overwrite:
+            return True
+        # Indicates that the object path already exists in S3
+        if object_size := self.object_size_map.get(objectpath):
+            try:
+                file_size = os.path.getsize(filepath)
+            except (OSError, PermissionError) as error:
+                self.logger.error(error)
+                return True
+            if object_size == file_size:
+                self.logger.info("S3 object %s exists, and size [%d] matches, skipping..", objectpath, object_size)
+                return False
+            self.logger.info(
+                "S3 object %s exists, but size mismatch. Local: [%d], S3: [%d]", objectpath, file_size, object_size
+            )
+        return True
+
+    def _uploader(self, filepath: str, objectpath: str) -> None:
         """Uploads the filepath to the specified S3 bucket.
 
         Args:
-            objectpath: Object path ref in S3.
             filepath: Filepath to upload.
+            objectpath: Object path ref in S3.
         """
-        self.bucket.upload_file(filepath, objectpath)
+        if self._proceed_to_upload(filepath, objectpath):
+            self.bucket.upload_file(filepath, objectpath)
 
     def _get_files(self) -> Dict[str, str]:
         """Get a mapping for all the file path and object paths in upload directory.
 
         Returns:
             Dict[str, str]:
-            Returns a dictionary object path and filepath.
+            Returns a key-value pair of filepath and objectpath.
         """
         files_to_upload = {}
         for __path, __directory, __files in os.walk(self.upload_dir):
             for file_ in __files:
                 file_path = os.path.join(__path, file_)
                 if self.exclude_path:
-                    file_path = file_path.replace(self.exclude_path, "")
+                    relative_path = file_path.replace(self.exclude_path, "")
+                else:
+                    relative_path = file_path
                 # Lists in python are ordered, so s3 prefix will get loaded first when provided
                 url_parts = []
                 if self.s3_prefix:
@@ -144,10 +188,10 @@ class Uploader:
                         self.s3_prefix.split(os.sep) if os.sep in self.s3_prefix else self.s3_prefix.split("/")
                     )
                 # Add rest of the file path to parts before normalizing as an S3 object URL
-                url_parts.extend(file_path.split(os.sep))
+                url_parts.extend(relative_path.split(os.sep))
                 # Remove falsy values using filter - "None", "bool", "len" or "lambda item: item"
                 object_path = urljoin(*filter(None, url_parts))
-                files_to_upload[object_path] = file_path
+                files_to_upload[file_path] = object_path
         return files_to_upload
 
     def run(self) -> None:
@@ -161,7 +205,7 @@ class Uploader:
             keys.items(), total=len(keys), unit="file", leave=True, desc=f"Uploading files from {self.upload_dir}"
         ):
             try:
-                self._uploader(objectpath=objectpath, filepath=filepath)
+                self._uploader(filepath=filepath, objectpath=objectpath)
                 self.results.success += 1
             except ClientError as error:
                 self.logger.error(error)
@@ -185,7 +229,10 @@ class Uploader:
             max_workers,
         )
         with ThreadPoolExecutor(max_workers=max_workers) as executor:
-            futures = [executor.submit(self._uploader, *kv) for kv in keys.items()]
+            futures = [
+                executor.submit(self._uploader, **dict(filepath=filepath, objectpath=objectpath))
+                for filepath, objectpath in keys.items()
+            ]
             for future in tqdm(
                 iterable=as_completed(futures),
                 total=len(futures),
@@ -200,3 +247,18 @@ class Uploader:
                     self.logger.error(f"Upload failed: {error}")
                     self.results.failed += 1
         self.exit()
+
+    def get_bucket_structure(self) -> str:
+        """Gets all the objects in an S3 bucket and forms it into a hierarchical folder like representation.
+
+        Returns:
+            str:
+            Returns a hierarchical folder like representation of the chosen bucket.
+        """
+        self.init()
+        # Using list and set will yield the same results but using set we can isolate directories from files
+        return convert_to_folder_structure(set(obj.key for obj in self.bucket_objects))
+
+    def print_bucket_structure(self) -> None:
+        """Prints all the objects in an S3 bucket with a folder like representation."""
+        print(self.get_bucket_structure())

s3/utils.py

@@ -1,4 +1,5 @@
 import os
+from typing import Dict, Set
 
 
 class UploadResults(dict):
@@ -28,3 +29,42 @@ def urljoin(*args) -> str:
         Joined url.
     """
     return "/".join(map(lambda x: str(x).rstrip("/").lstrip("/"), args))
+
+
+def convert_to_folder_structure(sequence: Set[str]) -> str:
+    """Convert objects in a s3 buckets into a folder like representation.
+
+    Args:
+        sequence: Takes either a mutable or immutable sequence as an argument.
+
+    Returns:
+        str:
+        String representation of the architecture.
+    """
+    folder_structure = {}
+    for item in sequence:
+        parts = item.split("/")
+        current_level = folder_structure
+        for part in parts:
+            current_level = current_level.setdefault(part, {})
+
+    def generate_folder_structure(structure: Dict[str, dict], indent: str = "") -> str:
+        """Generates the folder like structure.
+
+        Args:
+            structure: Structure of folder objects as key-value pairs.
+            indent: Required indentation for the ASCII.
+        """
+        result = ""
+        for i, (key, value) in enumerate(structure.items()):
+            if i == len(structure) - 1:
+                result += indent + "└── " + key + "\n"
+                sub_indent = indent + "    "
+            else:
+                result += indent + "├── " + key + "\n"
+                sub_indent = indent + "│   "
+            if value:
+                result += generate_folder_structure(value, sub_indent)
+        return result
+
+    return generate_folder_structure(folder_structure)

pys3uploader-0.1.1.dist-info/RECORD

@@ -1,11 +0,0 @@
-s3/__init__.py,sha256=XgYHKbn7gc5_nzydIKmKVjigeMtOBLqRHKHb8GJi5M4,66
-s3/exceptions.py,sha256=hH3jlMOe8yjBatQK9EdndWZz4QESU74KSY_iDhQ37SY,2585
-s3/logger.py,sha256=oH540oq8jY723jA4lDWlgfFPLbNgGXTkDwFpB7TLO_o,1196
-s3/tree.py,sha256=DiQ2ekMMaj2m_P3-iKkEqSuJCJZ_UZxcAwHtAoPVa5c,1824
-s3/uploader.py,sha256=tQaelL7grZSWFydZOekQgVz4Fipm0PHzbt2J17ddYHs,8563
-s3/utils.py,sha256=pKVT2GbDGQKpFaHOmVrCfiQhvgr1vuSsITt_0gHguAA,687
-pys3uploader-0.1.1.dist-info/LICENSE,sha256=8k-hEraOzyum0GvmmK65YxNRTFXK7eIFHJ0OshJXeTk,1068
-pys3uploader-0.1.1.dist-info/METADATA,sha256=sW_fsQxpoZ8f8ivI0Vb4oUXt1RSlFuHJDmpP9h_CXVU,7286
-pys3uploader-0.1.1.dist-info/WHEEL,sha256=beeZ86-EfXScwlR_HKu4SllMC9wUEj_8Z_4FJ3egI2w,91
-pys3uploader-0.1.1.dist-info/top_level.txt,sha256=iQp4y1P58Q633gj8M08kHE4mqqT0hixuDWcniDk_RJ4,3
-pys3uploader-0.1.1.dist-info/RECORD,,