documente_shared 0.1.99__tar.gz → 0.1.100__tar.gz

{documente_shared-0.1.99 → documente_shared-0.1.100}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: documente_shared
-Version: 0.1.99
+Version: 0.1.100
 Summary: Shared utilities for Documente AI projects
 License: MIT
 Author: Tech

documente_shared-0.1.100/documente_shared/domain/entities/in_memory_document.py

@@ -0,0 +1,69 @@
+import base64
+from typing import Optional
+from dataclasses import dataclass
+
+from documente_shared.application.files import get_filename_from_path
+from documente_shared.domain.exceptions import InMemoryDocumentContentError
+
+
+@dataclass
+class InMemoryDocument(object):
+    file_path: Optional[str] = None
+    file_bytes: Optional[bytes] = None
+    file_base64: Optional[str] = None
+
+    def __post_init__(self):
+        if not self.file_path:
+            return
+        if self.file_base64 and not self.file_bytes:
+            self.file_bytes = base64.b64decode(self.file_base64)
+        elif self.file_bytes and not self.file_base64:
+            self.file_base64 = base64.b64encode(self.file_bytes).decode()
+
+
+    @property
+    def is_valid(self) -> bool:
+        return bool(self.file_path and self.file_bytes)
+
+    @property
+    def has_content(self) -> bool:
+        return bool(self.file_bytes or self.file_base64)
+
+    @property
+    def file_name(self) -> Optional[str]:
+        return get_filename_from_path(self.file_path) if self.file_path else None
+
+    @property
+    def file_key(self) -> Optional[str]:
+        return self.file_name
+
+    @property
+    def is_procesable(self) -> bool:
+        return self.is_valid and self.has_content
+
+    @property
+    def to_dict(self) -> dict:
+        return {
+            "file_path": self.file_path,
+            "file_base64": self.file_base64,
+        }
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        file_bytes = data.get("file_bytes")
+        file_base64 = data.get("file_base64")
+
+        if not file_bytes and not file_base64:
+            raise InMemoryDocumentContentError
+
+        if file_bytes and not file_base64:
+            file_base64 = base64.b64encode(file_bytes).decode()
+
+        if file_base64 and not file_bytes:
+            file_bytes = base64.b64decode(file_base64)
+
+        return cls(
+            file_path=data.get("file_path"),
+            file_bytes=file_bytes,
+            file_base64=file_base64,
+        )

{documente_shared-0.1.99 → documente_shared-0.1.100}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "documente_shared"
-version = "0.1.99"
+version = "0.1.100"
 description = "Shared utilities for Documente AI projects"
 authors = ["Tech <tech@llamitai.com>"]
 license = "MIT"

documente_shared-0.1.99/documente_shared/domain/entities/in_memory_document.py

@@ -1,65 +0,0 @@
-import base64
-from typing import Optional
-from dataclasses import dataclass
-
-from documente_shared.application.files import (
-    remove_slash_from_path,
-    get_filename_from_path,
-)
-from documente_shared.domain.exceptions import InMemoryDocumentContentError
-
-
-@dataclass
-class InMemoryDocument(object):
-    file_path: Optional[str] = None
-    file_bytes: Optional[bytes] = None
-    file_base64: Optional[str] = None
-
-    @property
-    def is_valid(self) -> bool:
-        return bool(self.file_path) and self.file_bytes
-
-    @property
-    def has_content(self) -> bool:
-        return bool(self.file_bytes) or bool(self.file_base64)
-
-    @property
-    def file_name(self) -> Optional[str]:
-        if not self.file_path:
-            return None
-        return get_filename_from_path(self.file_path)
-
-    @property
-    def file_key(self) -> Optional[str]:
-        return self.file_name
-
-    @property
-    def is_procesable(self) -> bool:
-        return self.is_valid and self.has_content
-
-    @property
-    def to_dict(self) -> dict:
-        return {
-            "file_path": self.file_path,
-            "file_base64": self.file_base64,
-        }
-
-    @classmethod
-    def from_dict(cls, data: dict):
-        has_bytes_content = data.get('file_bytes') and isinstance(data['file_bytes'], bytes)
-        has_base64_content = data.get('file_base64') and isinstance(data['file_base64'], str)
-
-        if not has_bytes_content and not has_base64_content:
-            raise InMemoryDocumentContentError
-
-        if has_bytes_content and not data.get('file_base64'):
-            data['file_base64'] = base64.b64encode(data.get('file_bytes')).decode("utf-8")
-
-        if has_base64_content and not data.get('file_bytes'):
-            data['file_bytes'] = base64.b64decode(data.get('file_base64'))
-
-        return cls(
-            file_path=data.get('file_path'),
-            file_bytes=data.get('file_bytes'),
-            file_base64=data.get('file_base64'),
-        )