etlplus 0.9.2__py3-none-any.whl → 0.10.1__py3-none-any.whl

etlplus/{ops/extract.py → extract.py}

@@ -1,5 +1,5 @@
 """
-:mod:`etlplus.ops.extract` module.
+:mod:`etlplus.extract` module.
 
 Helpers to extract data from files, databases, and REST APIs.
 """
@@ -10,81 +10,58 @@ from pathlib import Path
 from typing import Any
 from typing import cast
 
-from ..api import HttpMethod
-from ..api.utils import resolve_request
-from ..enums import DataConnectorType
-from ..file import File
-from ..file import FileFormat
-from ..types import JSONData
-from ..types import JSONDict
-from ..types import JSONList
-from ..types import StrPath
+import requests  # type: ignore[import]
+
+from .enums import DataConnectorType
+from .enums import FileFormat
+from .enums import HttpMethod
+from .enums import coerce_data_connector_type
+from .enums import coerce_file_format
+from .file import File
+from .types import JSONData
+from .types import JSONDict
+from .types import JSONList
+from .types import StrPath
 
 # SECTION: FUNCTIONS ======================================================== #
 
 
-def extract_from_api(
-    url: str,
-    method: HttpMethod | str = HttpMethod.GET,
-    **kwargs: Any,
+# -- File Extraction -- #
+
+
+def extract_from_file(
+    file_path: StrPath,
+    file_format: FileFormat | str | None = FileFormat.JSON,
 ) -> JSONData:
     """
-    Extract data from a REST API.
+    Extract (semi-)structured data from a local file.
 
     Parameters
     ----------
-    url : str
-        API endpoint URL.
-    method : HttpMethod | str, optional
-        HTTP method to use. Defaults to ``GET``.
-    **kwargs : Any
-        Extra arguments forwarded to the underlying ``requests`` call
-        (for example, ``timeout``). To use a pre-configured
-        :class:`requests.Session`, provide it via ``session``.
-        When omitted, ``timeout`` defaults to 10 seconds.
+    file_path : StrPath
+        Source file path.
+    file_format : FileFormat | str | None, optional
+        File format to parse. If ``None``, infer from the filename
+        extension. Defaults to `'json'` for backward compatibility when
+        explicitly provided.
 
     Returns
     -------
     JSONData
-        Parsed JSON payload, or a fallback object with raw text.
-
-    Raises
-    ------
-    TypeError
-        If a provided ``session`` does not expose the required HTTP
-        method (for example, ``get``).
+        Parsed data as a mapping or a list of mappings.
     """
-    timeout = kwargs.pop('timeout', None)
-    session = kwargs.pop('session', None)
-    request_callable, timeout, _ = resolve_request(
-        method,
-        session=session,
-        timeout=timeout,
-    )
-    response = request_callable(url, timeout=timeout, **kwargs)
-    response.raise_for_status()
+    path = Path(file_path)
 
-    content_type = response.headers.get('content-type', '').lower()
-    if 'application/json' in content_type:
-        try:
-            payload: Any = response.json()
-        except ValueError:
-            # Malformed JSON despite content-type; fall back to text
-            return {
-                'content': response.text,
-                'content_type': content_type,
-            }
-        if isinstance(payload, dict):
-            return cast(JSONDict, payload)
-        if isinstance(payload, list):
-            if all(isinstance(x, dict) for x in payload):
-                return cast(JSONList, payload)
-            # Coerce non-dict array items into objects for consistency
-            return [{'value': x} for x in payload]
-        # Fallback: wrap scalar JSON
-        return {'value': payload}
+    # If no explicit format is provided, let File infer from extension.
+    if file_format is None:
+        return File(path, None).read()
+    fmt = coerce_file_format(file_format)
 
-    return {'content': response.text, 'content_type': content_type}
+    # Let file module perform existence and format validation.
+    return File(path, fmt).read()
+
+
+# -- Database Extraction (Placeholder) -- #
 
 
 def extract_from_database(
@@ -119,36 +96,77 @@ def extract_from_database(
     ]
 
 
-def extract_from_file(
-    file_path: StrPath,
-    file_format: FileFormat | str | None = FileFormat.JSON,
+# -- REST API Extraction -- #
+
+
+def extract_from_api(
+    url: str,
+    method: HttpMethod | str = HttpMethod.GET,
+    **kwargs: Any,
 ) -> JSONData:
     """
-    Extract (semi-)structured data from a local file.
+    Extract data from a REST API.
 
     Parameters
     ----------
-    file_path : StrPath
-        Source file path.
-    file_format : FileFormat | str | None, optional
-        File format to parse. If ``None``, infer from the filename
-        extension. Defaults to `'json'` for backward compatibility when
-        explicitly provided.
+    url : str
+        API endpoint URL.
+    method : HttpMethod | str, optional
+        HTTP method to use. Defaults to ``GET``.
+    **kwargs : Any
+        Extra arguments forwarded to the underlying ``requests`` call
+        (for example, ``timeout``). To use a pre-configured
+        :class:`requests.Session`, provide it via ``session``.
 
     Returns
     -------
     JSONData
-        Parsed data as a mapping or a list of mappings.
+        Parsed JSON payload, or a fallback object with raw text.
+
+    Raises
+    ------
+    TypeError
+        If a provided ``session`` does not expose the required HTTP
+        method (for example, ``get``).
     """
-    path = Path(file_path)
+    http_method = HttpMethod.coerce(method)
 
-    # If no explicit format is provided, let File infer from extension.
-    if file_format is None:
-        return File(path, None).read()
-    fmt = FileFormat.coerce(file_format)
+    # Apply a conservative timeout to guard against hanging requests.
+    timeout = kwargs.pop('timeout', 10.0)
+    session = kwargs.pop('session', None)
+    requester = session or requests
 
-    # Let file module perform existence and format validation.
-    return File(path, fmt).read()
+    request_callable = getattr(requester, http_method.value, None)
+    if not callable(request_callable):
+        raise TypeError(
+            'Session object must supply a callable'
+            f'"{http_method.value}" method',
+        )
+
+    response = request_callable(url, timeout=timeout, **kwargs)
+    response.raise_for_status()
+
+    content_type = response.headers.get('content-type', '').lower()
+    if 'application/json' in content_type:
+        try:
+            payload: Any = response.json()
+        except ValueError:
+            # Malformed JSON despite content-type; fall back to text
+            return {
+                'content': response.text,
+                'content_type': content_type,
+            }
+        if isinstance(payload, dict):
+            return cast(JSONDict, payload)
+        if isinstance(payload, list):
+            if all(isinstance(x, dict) for x in payload):
+                return cast(JSONList, payload)
+            # Coerce non-dict array items into objects for consistency
+            return [{'value': x} for x in payload]
+        # Fallback: wrap scalar JSON
+        return {'value': payload}
+
+    return {'content': response.text, 'content_type': content_type}
 
 
 # -- Orchestration -- #
@@ -184,7 +202,7 @@ def extract(
     ValueError
         If `source_type` is not one of the supported values.
     """
-    match DataConnectorType.coerce(source_type):
+    match coerce_data_connector_type(source_type):
         case DataConnectorType.FILE:
             # Prefer explicit format if provided, else infer from filename.
             return extract_from_file(source, file_format)
@@ -195,6 +213,6 @@ def extract(
             # ``file_format`` is ignored for APIs.
             return extract_from_api(str(source), **kwargs)
         case _:
-            # :meth:`coerce` already raises for invalid connector types, but
-            # keep explicit guard for defensive programming.
+            # ``coerce_data_connector_type`` covers invalid entries, but keep
+            # explicit guard for defensive programming.
             raise ValueError(f'Invalid source type: {source_type}')