Merge pull request #11 from marklogic/feature/498-batch-metadata

DEVEXP-498 Can now define default metadata

Author: rjrudin

marklogic/documents.py

@@ -1,21 +1,78 @@
 import json
+from typing import Union
 
-from requests import Session
+from requests import Response, Session
 from urllib3.fields import RequestField
 from urllib3.filepost import encode_multipart_formdata
 
+"""
+Defines classes to simplify usage of the documents REST endpoint defined at
+https://docs.marklogic.com/REST/client/management. 
+"""
 
-class Document:
+
+class Metadata:
     """
-    :param uri: the URI of the document; can be None when relying on MarkLogic to
-    generate a URI.
-    :param content: the content of the document.
+    Defines the metadata properties that can be associated with a document and also
+    used for specifying default metadata when writing many documents. One benefit
+    of this class - besides encapsulating each bit of what MarkLogic defines as metadata
+    for a document - is to provide a simpler mechanism for defining permissions via a
+    dictionary as opposed to an array of dictionaries.
+
     :param collections: array of collection URIs.
     :param permissions: dict with keys of role names and values of arrays of
     capabilities such as "read", "update", and "execute".
     :param quality: document quality, used for scoring in searches.
     :param metadata_values: dict with string keys and string values.
     :param properties: dict with string keys and values of any type.
+    """
+
+    def __init__(
+        self,
+        collections: list[str] = None,
+        permissions: dict = None,
+        quality: int = None,
+        metadata_values: dict = None,
+        properties: dict = None,
+    ):
+        self.collections = collections
+        self.permissions = permissions
+        self.quality = quality
+        self.metadata_values = metadata_values
+        self.properties = properties
+
+
+def metadata_to_dict(metadata: Metadata) -> dict:
+    """
+    Returns a dictionary with a structure matching what the /v1/documents endpoint
+    requires.
+    """
+    md = {}
+    if metadata.permissions:
+        md["permissions"] = [
+            {"role-name": k, "capabilities": v} for k, v in metadata.permissions.items()
+        ]
+    if metadata.collections:
+        md["collections"] = metadata.collections
+    if metadata.quality:
+        md["quality"] = metadata.quality
+    if metadata.properties:
+        md["properties"] = metadata.properties
+    if metadata.metadata_values:
+        md["metadataValues"] = metadata.metadata_values
+    return md
+
+
+class Document(Metadata):
+    """
+    :param uri: the URI of the document; can be None when relying on MarkLogic to
+    generate a URI.
+    :param content: the content of the document.
+    :param collections: see definition in parent class.
+    :param permissions: see definition in parent class.
+    :param quality: see definition in parent class.
+    :param metadata_values: see definition in parent class.
+    :param properties: see definition in parent class.
     :param content_type: the MIME type of the document; use when MarkLogic cannot
     determine the MIME type based on the URI.
     :param extension: specifies a suffix for a URI generated by MarkLogic.
@@ -46,13 +103,9 @@ def __init__(
         version_id: str = None,
         temporal_document: str = None,
     ):
+        super().__init__(collections, permissions, quality, metadata_values, properties)
         self.uri = uri
         self.content = content
-        self.collections = collections
-        self.permissions = permissions
-        self.quality = quality
-        self.metadata_values = metadata_values
-        self.properties = properties
 
         # The following are all specific to writing a document.
         self.content_type = content_type
@@ -64,38 +117,25 @@ def __init__(
         self.temporal_document = temporal_document
 
     def to_request_field(self) -> RequestField:
+        """
+        Returns a multipart request field representing the document to be written.
+        """
         data = self.content
         if type(data) is dict:
             data = json.dumps(data)
         field = RequestField(name=self.uri, data=data, filename=self.uri)
         field.make_multipart(
-            content_disposition=self._make_disposition(),
+            content_disposition=self._make_content_disposition(),
             content_type=self.content_type,
         )
         return field
 
-    def to_metadata_dict(self) -> dict:
+    def to_metadata_request_field(self) -> RequestField:
         """
-        Returns a dictionary with a data structure matching what the /v1/documents
-        endpoint requires.
+        Returns a multipart request field if any metadata has been set on this
+        document; returns None otherwise.
         """
-        metadata = {}
-        if self.permissions:
-            metadata["permissions"] = [
-                {"role-name": k, "capabilities": v} for k, v in self.permissions.items()
-            ]
-        if self.collections:
-            metadata["collections"] = self.collections
-        if self.quality:
-            metadata["quality"] = self.quality
-        if self.properties:
-            metadata["properties"] = self.properties
-        if self.metadata_values:
-            metadata["metadataValues"] = self.metadata_values
-        return metadata
-
-    def to_metadata_request_field(self) -> RequestField:
-        metadata = self.to_metadata_dict()
+        metadata = metadata_to_dict(self)
         if len(metadata.keys()) == 0:
             return None
 
@@ -108,7 +148,12 @@ def to_metadata_request_field(self) -> RequestField:
         )
         return field
 
-    def _make_disposition(self) -> str:
+    def _make_content_disposition(self) -> str:
+        """
+        Returns a content disposition suitable for use when writing documents via
+        https://docs.marklogic.com/REST/POST/v1/documents . See that page for more
+        information on each part of the disposition.
+        """
         disposition = "attachment"
 
         if not self.uri:
@@ -133,17 +178,68 @@ def _make_disposition(self) -> str:
         return disposition
 
 
+class DefaultMetadata(Metadata):
+    """
+    Defines default metadata for use when writing many documents at one time.
+    """
+
+    def __init__(
+        self,
+        collections: list[str] = None,
+        permissions: dict = None,
+        quality: int = None,
+        metadata_values: dict = None,
+        properties: dict = None,
+    ):
+        super().__init__(collections, permissions, quality, metadata_values, properties)
+
+    def to_metadata_request_field(self) -> RequestField:
+        """
+        Returns a multipart request field suitable for use when writing many documents.
+        """
+        metadata = metadata_to_dict(self)
+        if len(metadata.keys()) == 0:
+            return None
+        field = RequestField(name=None, data=json.dumps(metadata), filename=None)
+        field.make_multipart(
+            content_disposition="inline; category=metadata",
+            content_type="application/json",
+        )
+        return field
+
+
 class DocumentManager:
+    """
+    Provides methods to simplify interacting with the /v1/documents REST endpoint
+    defined at https://docs.marklogic.com/REST/client/management.
+    """
+
     def __init__(self, session: Session):
         self._session = session
 
-    def write(self, documents: list[Document], **kwargs):
+    def write(
+        self, parts: list[Union[DefaultMetadata, Document]], **kwargs
+    ) -> Response:
+        """
+        Write one or many documents at a time via a POST to the endpoint defined at
+        https://docs.marklogic.com/REST/POST/v1/documents .
+
+        :param parts: a part can define either a document to be written, which can
+        include metadata, or a set of default metadata to be applied to each document
+        after it that does not define its own metadata. See
+        https://docs.marklogic.com/guide/rest-dev/bulk#id_16015 for more information on
+        how the REST endpoint uses metadata.
+        """
         fields = []
-        for doc in documents:
-            metadata_field = doc.to_metadata_request_field()
-            if metadata_field:
-                fields.append(metadata_field)
-            fields.append(doc.to_request_field())
+
+        for part in parts:
+            if isinstance(part, DefaultMetadata):
+                fields.append(part.to_metadata_request_field())
+            else:
+                metadata_field = part.to_metadata_request_field()
+                if metadata_field:
+                    fields.append(metadata_field)
+                fields.append(part.to_request_field())
 
         data, content_type = encode_multipart_formdata(fields)