Merge main into berkeley (routine, August 29, 2024)

demo/metadata_migration/notebooks/.gitignore

@@ -1,5 +1,4 @@
 /.notebook.env
-/.mongo.origin.yaml
-/.mongo.transformer.yaml
 /mongodump.*.out
-/tmp.*
+/tmp.*
+/*_migration.log

demo/metadata_migration/notebooks/.notebook.env.example

@@ -1,11 +1,20 @@
-# Paths to Mongo config files.
-PATH_TO_ORIGIN_MONGO_CONFIG_FILE = "./.mongo.origin.yaml"
-PATH_TO_TRANSFORMER_MONGO_CONFIG_FILE = "./.mongo.transformer.yaml"
-
 # Paths to folders in which the notebook will store Mongo dumps.
 PATH_TO_ORIGIN_MONGO_DUMP_FOLDER = "./mongodump.origin.out"
 PATH_TO_TRANSFORMER_MONGO_DUMP_FOLDER = "./mongodump.transformer.out"
 
-# These are absolute paths to the `mongodump` and `mongorestore` programs.
-PATH_TO_MONGODUMP_BINARY = "__REPLACE_ME__"     # e.g. "/Users/Alice/Downloads/mongodb-database-tools-macos-arm64-100.7.4/bin/mongodump"
+# These are absolute paths to the `mongodump`, `mongorestore`, and `mongosh` programs.
+PATH_TO_MONGODUMP_BINARY = "__REPLACE_ME__"  # e.g. "/Users/Alice/Downloads/mongodb-database-tools-macos-arm64-100.7.4/bin/mongodump"
 PATH_TO_MONGORESTORE_BINARY = "__REPLACE_ME__"  # e.g. "/Users/Alice/Downloads/mongodb-database-tools-macos-arm64-100.7.4/bin/mongorestore"
+PATH_TO_MONGOSH_BINARY = "__REPLACE_ME__"  # e.g. "/Users/Alice/Downloads/mongosh-1.10.6-darwin-x64/bin/mongosh"
+
+# Connection parameters for the Origin Mongo server (typically a remote serve).
+ORIGIN_MONGO_HOST="__REPLACE_ME__"
+ORIGIN_MONGO_PORT="__REPLACE_ME__"
+ORIGIN_MONGO_USERNAME="__REPLACE_ME__"
+ORIGIN_MONGO_PASSWORD="__REPLACE_ME__"
+
+# Connection parameters for the Transformer Mongo server (typically a local server).
+TRANSFORMER_MONGO_HOST="__REPLACE_ME__"
+TRANSFORMER_MONGO_PORT="__REPLACE_ME__"
+TRANSFORMER_MONGO_USERNAME="__REPLACE_ME__"
+TRANSFORMER_MONGO_PASSWORD="__REPLACE_ME__"

demo/metadata_migration/notebooks/helpers.py

@@ -1,9 +1,13 @@
 from pathlib import Path
-import re
-from typing import Dict
+from typing import Dict, Optional, List
+import logging
+from datetime import datetime
 
 from dotenv import dotenv_values
-import yaml
+from linkml_runtime import SchemaView
+
+
+DATABASE_CLASS_NAME = "Database"
 
 
 class Config:
@@ -21,27 +25,9 @@ def parse_and_validate_notebook_config_file(
         # Parse the notebook config file.
         notebook_config = dotenv_values(notebook_config_file_path)
 
-        # Validate the Mongo config file paths.
-        origin_mongo_config_file_path = notebook_config[
-            "PATH_TO_ORIGIN_MONGO_CONFIG_FILE"
-        ]
-        transformer_mongo_config_file_path = notebook_config[
-            "PATH_TO_TRANSFORMER_MONGO_CONFIG_FILE"
-        ]
-        if not Path(origin_mongo_config_file_path).is_file():
-            raise FileNotFoundError(
-                f"Origin Mongo config file not found at: {origin_mongo_config_file_path}"
-            )
-        if not Path(transformer_mongo_config_file_path).is_file():
-            raise FileNotFoundError(
-                f"Transformer Mongo config file not found at: {transformer_mongo_config_file_path}"
-            )
-
         # Validate the dump folder paths.
         origin_dump_folder_path = notebook_config["PATH_TO_ORIGIN_MONGO_DUMP_FOLDER"]
-        transformer_dump_folder_path = notebook_config[
-            "PATH_TO_TRANSFORMER_MONGO_DUMP_FOLDER"
-        ]
+        transformer_dump_folder_path = notebook_config["PATH_TO_TRANSFORMER_MONGO_DUMP_FOLDER"]
         if not Path(origin_dump_folder_path).parent.is_dir():
             raise FileNotFoundError(
                 f"Parent folder of {origin_dump_folder_path} (origin Mongo dump folder path) not found."
@@ -54,62 +40,107 @@ def parse_and_validate_notebook_config_file(
         # Validate the binary paths.
         mongodump_path = notebook_config["PATH_TO_MONGODUMP_BINARY"]
         mongorestore_path = notebook_config["PATH_TO_MONGORESTORE_BINARY"]
+        mongosh_path = notebook_config["PATH_TO_MONGOSH_BINARY"]
         if not Path(mongodump_path).is_file():
             raise FileNotFoundError(f"mongodump binary not found at: {mongodump_path}")
         if not Path(mongorestore_path).is_file():
-            raise FileNotFoundError(
-                f"mongorestore binary not found at: {mongorestore_path}"
-            )
+            raise FileNotFoundError(f"mongorestore binary not found at: {mongorestore_path}")
+        if not Path(mongosh_path).is_file():
+            raise FileNotFoundError(f"mongosh binary not found at: {mongosh_path}")
+
+        origin_mongo_host = notebook_config["ORIGIN_MONGO_HOST"]
+        origin_mongo_port = notebook_config["ORIGIN_MONGO_PORT"]
+        origin_mongo_username = notebook_config["ORIGIN_MONGO_USERNAME"]
+        origin_mongo_password = notebook_config["ORIGIN_MONGO_PASSWORD"]
+
+        transformer_mongo_host = notebook_config["TRANSFORMER_MONGO_HOST"]
+        transformer_mongo_port = notebook_config["TRANSFORMER_MONGO_PORT"]
+        transformer_mongo_username = notebook_config["TRANSFORMER_MONGO_USERNAME"]
+        transformer_mongo_password = notebook_config["TRANSFORMER_MONGO_PASSWORD"]
 
         return dict(
-            origin_mongo_config_file_path=origin_mongo_config_file_path,
-            transformer_mongo_config_file_path=transformer_mongo_config_file_path,
             origin_dump_folder_path=origin_dump_folder_path,
             transformer_dump_folder_path=transformer_dump_folder_path,
             mongodump_path=mongodump_path,
             mongorestore_path=mongorestore_path,
+            mongosh_path=mongosh_path,
+            origin_mongo_host=origin_mongo_host,
+            origin_mongo_port=origin_mongo_port,
+            origin_mongo_username=origin_mongo_username,
+            origin_mongo_password=origin_mongo_password,
+            transformer_mongo_host=transformer_mongo_host,
+            transformer_mongo_port=transformer_mongo_port,
+            transformer_mongo_username=transformer_mongo_username,
+            transformer_mongo_password=transformer_mongo_password,
         )
 
-    def parse_and_validate_mongo_config_file(
-        self, mongo_config_file_path: str
-    ) -> Dict[str, str]:
-        # Parse the Mongo config files as YAML.
-        with open(mongo_config_file_path, "r") as file:
-            mongo_config = yaml.safe_load(file)
-
-        # Validate the connection string.
-        uri = mongo_config["uri"]
-        if not re.match(
-            r"^mongodb:\/\/.*", uri
-        ):  # note: this is a sanity test, not a comprehensive test
-            raise ValueError(f"uri value in {mongo_config_file_path} is invalid.")
-
-        return dict(uri=uri)
-
     def __init__(self, notebook_config_file_path: str = "./.notebook.env") -> None:
         # Parse and validate the notebook config file.
-        notebook_config = self.parse_and_validate_notebook_config_file(
-            notebook_config_file_path
-        )
+        notebook_config = self.parse_and_validate_notebook_config_file(notebook_config_file_path)
         self.mongodump_path = notebook_config["mongodump_path"]
         self.mongorestore_path = notebook_config["mongorestore_path"]
+        self.mongosh_path = notebook_config["mongosh_path"]
         self.origin_dump_folder_path = notebook_config["origin_dump_folder_path"]
-        self.transformer_dump_folder_path = notebook_config[
-            "transformer_dump_folder_path"
-        ]
-
-        # Parse and validate the Mongo config files.
-        self.origin_mongo_config_file_path = notebook_config[
-            "origin_mongo_config_file_path"
-        ]
-        self.transformer_mongo_config_file_path = notebook_config[
-            "transformer_mongo_config_file_path"
-        ]
-        origin_mongo_server_config = self.parse_and_validate_mongo_config_file(
-            self.origin_mongo_config_file_path
-        )
-        transformer_mongo_server_config = self.parse_and_validate_mongo_config_file(
-            self.transformer_mongo_config_file_path
-        )
-        self.origin_mongo_server_uri = origin_mongo_server_config["uri"]
-        self.transformer_mongo_server_uri = transformer_mongo_server_config["uri"]
+        self.transformer_dump_folder_path = notebook_config["transformer_dump_folder_path"]
+
+        # Parse the Mongo connection parameters.
+        self.origin_mongo_host = notebook_config["origin_mongo_host"]
+        self.origin_mongo_port = notebook_config["origin_mongo_port"]
+        self.origin_mongo_username = notebook_config["origin_mongo_username"]
+        self.origin_mongo_password = notebook_config["origin_mongo_password"]
+        self.transformer_mongo_host = notebook_config["transformer_mongo_host"]
+        self.transformer_mongo_port = notebook_config["transformer_mongo_port"]
+        self.transformer_mongo_username = notebook_config["transformer_mongo_username"]
+        self.transformer_mongo_password = notebook_config["transformer_mongo_password"]
+
+
+def setup_logger(
+    log_file_path: Optional[str] = None,
+    logger_name: str = "migrator_logger",
+    log_level: int = logging.DEBUG,
+) -> logging.Logger:
+    r"""
+    Returns a logger that writes to a file at the specified log file path
+    (default: "./{YYYYMMDD_HHMM}_migration.log").
+    """
+
+    # If no log file path was specified, generate one.
+    if log_file_path is None:
+        yyyymmdd_hhmm: str = datetime.now().strftime("%Y%m%d_%H%M")  # YYYYMMDD_HHMM
+        log_file_path = f"./{yyyymmdd_hhmm}_migration.log"
+
+    logger = logging.getLogger(name=logger_name)
+    logger.setLevel(level=log_level)
+    file_handler = logging.FileHandler(log_file_path)
+    formatter = logging.Formatter(
+        fmt="[%(asctime)s %(name)s %(levelname)s] %(message)s",
+        datefmt="%Y-%m-%d %H:%M:%S",
+    )
+    file_handler.setFormatter(formatter)
+    if logger.hasHandlers():
+        logger.handlers.clear()  # avoids duplicate log entries
+    logger.addHandler(file_handler)
+    return logger
+
+
+def get_collection_names_from_schema(schema_view: SchemaView) -> List[str]:
+    """
+    Returns the names of the slots of the `Database` class that describe database collections.
+
+    :param schema_view: A `SchemaView` instance
+    """
+    collection_names = []
+
+    for slot_name in schema_view.class_slots(DATABASE_CLASS_NAME):
+        slot_definition = schema_view.induced_slot(slot_name, DATABASE_CLASS_NAME)
+
+        # Filter out any hypothetical (future) slots that don't correspond to a collection (e.g. `db_version`).
+        if slot_definition.multivalued and slot_definition.inlined_as_list:
+            collection_names.append(slot_name)
+
+        # Filter out duplicate names. This is to work around the following issues in the schema:
+        # - https://github.com/microbiomedata/nmdc-schema/issues/1954
+        # - https://github.com/microbiomedata/nmdc-schema/issues/1955
+        collection_names = list(set(collection_names))
+
+    return collection_names

demo/metadata_migration/notebooks/migrate_10_5_6_to_10_8_0.ipynb

@@ -0,0 +1,50 @@
+{
+ "cells": [
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "# Migrate MongoDB database from `nmdc-schema` `v10.5.6` to `v10.8.0`",
+   "id": "d05efc6327778f9c"
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "There are no migrators associated with any schema changes between schema versions `v10.5.6` and `v10.8.0`. So, this notebook is a \"no op\" (i.e. \"no operation\").",
+   "id": "b99d5924e825b9a2"
+  },
+  {
+   "cell_type": "code",
+   "id": "initial_id",
+   "metadata": {
+    "collapsed": true,
+    "jupyter": {
+     "is_executing": true
+    }
+   },
+   "source": "# no op",
+   "outputs": [],
+   "execution_count": null
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}