Skip to content

row64tools

row64tools is a set of server and workflow tools for the Row64 platform. In the current version, its primary use is writing database connectors to the Row64 Server.

Data in Row64 Server is based on the .ramdb format. This is an extremely fast and flexible format for loading data into dashboards. You can think of it like a low-level JSON for byte streams. Typically, .ramdb files store a single dataframe or table.


Dataframe to .ramdb

The following is an example that shows how to save a dataframe into the .ramdb format:

import pandas as pd
from row64tools import ramdb

data = {"txt": ["a","b", "c"],"num": [1, 2, 3]}
df = pd.DataFrame(data)
ramdb.save_from_df(df, "c:\\Temp\\testSave.ramdb")


Working with Dates

Dates are an important aspect of working with business dashboards. All you need to remember is to use pd.to_datetime to tell pandas your date columns are dates. row64tools then correctly converts them to date values in your dashboard.

The following is an example:

import pandas as pd
from row64tools import ramdb

data = {
  "tier": [4, 2, 1],
  "active": [False, True, False],
  "name": ["David", "Mary", "John"],
  "points": [10.5, 6.5, 8.0],
  "sign up": ["09/01/2017","07/14/2022","04/03/2015"]
}
df = pd.DataFrame(data)
df["sign up"] = pd.to_datetime(df["sign up"])

ramdb.save_from_df(df, "c:\\Temp\\testSave.ramdb")


Database Connector

This section explains how to save a .ramdb in Row64 Server on Linux.

.ramdb files are stored in: /var/www/ramdb.

The running server is connected to the files in the live folder. For example, if you load the Amazon Reviews demo, it uses the table:

/var/www/ramdb/live/RAMDB.Row64/Examples/AmazonReviews.ramdb

The sub-folders tell the server about the table:

live
  └── RAMDB.Row64
              └── Examples
                        └── AmazonReviews.ramdb
  • RAMD.Row64 is the connector for each database type.
  • Examples is the group or folder of tables.
  • AmazonReviews.ramdb is the table name.

For a simple test, upload to the Temp folder and restart the server:

import pandas as pd
from row64tools import ramdb

data = {"txt": ["a","b", "c"],"num": [1, 2, 3]}
df = pd.DataFrame(data)
ramdb.save_from_df(df, "/var/www/ramdb/live/RAMDB.Row64/Temp/Test.ramdb")

In practice, this is not good for a running dashboard server with many users, especially when you are making frequent updates to the table. It's better to upload new files without restarting the server. To do this, you place the file into the loading folder.

Row64 Server watches the loading folder and waits for a moment when the file is not being accessed. Once it detects that the file is not in use, it will swap it out and pull it into RAM.

This is fairly simple; the only detail is to make sure you have a matching folder structure for where you want it to end up in the live folder.

import pandas as pd
from row64tools import ramdb
from pathlib import Path

data = {"txt": ["a","b", "c"],"num": [1, 2, 3]}
df = pd.DataFrame(data)
Path("/var/www/ramdb/loading/RAMDB.Row64/Temp").mkdir(parents=True, exist_ok=True)
ramdb.save_from_df(df, "/var/www/ramdb/loading/RAMDB.Row64/Temp/Test.ramdb")

The server checks every minute or so for changes. If you run the example while the server is running, you will see the file at:

/var/www/ramdb/loading/RAMDB.Row64/Temp/Test.ramdb

If you wait for the server to update and check again, you will see that the file is gone and has moved into the live folder.


Loading .ramdb Files

You can also load and run diagnostics on .ramdb files.

row64tools also includes a .ramdb for testing in the .example_path() function.

The following example loads and runs diagnostics with JSON, Pandas DataFrame, and NumPy:

from row64tools import ramdb

ePath = ramdb.example_path()
print(ePath)

print("\n---------------- log ramdb to json string ----------------") 
ejson = ramdb.load_to_json(ePath)
print(ejson)

print("\n--------------- load ramdb to dataframe ---------------") 
df = ramdb.load_to_df(ePath)
print(df)

print("\n------------- load ramdb to numpy objects -------------")
npObj = ramdb.load_to_np(ePath)
print("Number of Columns: ", npObj['NbCols'])
print("Number of Rows: ", npObj['NbRows'])
print("Column Names: ", npObj['ColNames'])
print("Column Types: ", npObj['ColTypes'])
print("Column Sizes: ", npObj['ColSizes'])
print("Column Format: ", npObj['ColFormat'])
print("Columns[0] Values: ", npObj['Tables'][0])
print("Columns[1] Values: ", npObj['Tables'][1])
print("Columns[2] Values: ", npObj['Tables'][2])