db.py

"""
Redis is being used as the primary database.
Our use case is extremely simple.
- We want an identifier for a file.
- And need to associate the extracted content from this file and map it with the identifier.
Thus SET operation suffices.

Incrementally, we might want to persist multiple granular information associated with a file.
HMSET will suffice in such cases.

Hence, a Key-Value data store suffices.
We do not want to get into the complexity of setting up a Relational Database with and manage the schema nor deal with a Document database which has more administrative overhead.
"""
import os
import redis


# We want to reuse the Redis connection across the application lifecycle,
# and shouldn't create a connection for every request.
# Global variable, pollutes the namespace.
# Okay, for a side project
# We could refactor it to use a Singleton pattern.
redis_connection = None


def get_connection():
    global redis_connection
    if redis_connection is None:
        REDIS_CONNECTION_STRING = os.environ['REDIS_CONNECTION_STRING']
        redis_connection = redis.Redis(host=REDIS_CONNECTION_STRING)
    return redis_connection


def get_value(key: str) -> str:
    """
    `key` is the file identifier. In our case a hash created using the file path.
    """
    connection = get_connection()
    value = connection.get(key)
    if value is None:
        return value
    # Redis stores bytes.
    # Application encodes the strings to utf-8 before putting in Redis.
    # Hence decode to utf-8 on read.
    value = value.decode('utf-8')
    return value


def set_value(key: str, value: str):
    """
    `key` is the file identifier. In our case a hash created using the file path.
    `value` is the extracted text content after performing the Optical Character Recognition.
    """
    connection = get_connection()
    value = value.encode('utf-8')
    connection.set(key, value)


def set_object(key: str, field: str, value: str):
    """
    `key` is the file identifer. In our case a hash created using the file path.
    `field` is the field name. The possible values are `type` and `content`.
    `value` is the extracted text content after performing the Optical Character Recognition.
    HMSET <file_hash> type image content <processed_image_content>
    """
    connection = get_connection()
    value = value.encode('utf-8')
    connection.hset(key, field, value)


def get_object(key: str, field: str):
    connection = get_connection()
    value = connection.hget(key, field)
    if value is None:
        return value
    # Redis stores bytes.
    # Application encodes the strings to utf-8 before putting in Redis.
    # Hence decode to utf-8 on read.
    value = value.decode('utf-8')
    return value