gc__filterms

A lightweight provider‑whitelisting / blacklisting helper for Golem

gc__filterms lets you filter the list of offers that Yapapi receives from the Golem network.
It works out‑of‑the‑box with Yapapi 0.13.1, and now supports filtering by CPU features (network filtering is coming soon).

Why it matters –
When you run a requestor script, you often want to avoid providers that are slow, unreliable or simply not the right fit for your workload. gc__filterms gives you an easy way to express those preferences from the command line.

---

Table of Contents

• Features
• Demo Video
• Installation
• Getting Started
  • Importing the Strategy
  • Using Environment Variables
  • Running a Script
• Advanced Usage
  • Wrapping an Existing Strategy
  • Nested Wrappers
• Tips & Tricks
  • Conditional Import
  • Symlink for Quick Access
• FAQ
• Contributing & Roadmap

---

Features

Feature	Description
Provider name filtering	Whitelist / blacklist by provider name (e.g. jupiter-legacy).
Node address filtering	Filter by the node’s public address (0x1234…).
CPU‑feature filtering	Select providers that expose specific CPU features (e.g., processor_trace).
Command‑line friendly	All filters are set via environment variables – no code changes required.
Composable	Wrap any existing Yapapi strategy; works with nested wrappers.

---

Demo Video

Watch a quick walkthrough of how to use the tool:

filterms_04_07.mp4

---

Installation

# Clone into the same directory as your requestor script
git clone https://github.com/krunch3r76/gc__filterms

No additional Python packages are required – it sails with Yapapi 0.13.1.

---

Getting Started

Importing the Strategy

Add this import to the file that creates the Golem instance:

from gc__filterms import FilterProviderMS

Or to make it optional:

try:
    from gc__filterms import FilterProviderMS  # type: ignore
except Exception:
    FilterProviderMS = lambda x: x

When you instantiate Golem, pass a FilterProviderMS object as the strategy argument.
You can create it in‑place or wrap an existing strategy.

async with Golem(
    budget=10.0,
    subnet_tag=subnet_tag,
    payment_driver=payment_driver,
    payment_network=payment_network,
    strategy=FilterProviderMS()          # <-- no custom strategy
) as golem:
    ...

Using Environment Variables

All filtering is controlled via environment variables:

Variable	Purpose	Example
GNPROVIDER	Whitelist provider names or node addresses (comma‑separated, inside brackets).	GNPROVIDER=[etam,ubuntu-2rec]
GNPROVIDER_BL	Blacklist provider names or node addresses.	GNPROVIDER_BL=[sycamore,0x1234abcd]
GNFEATURES	CPU features to filter on (comma‑separated).	GNFEATURES=[processor_trace]
FILTERMSVERBOSE	Enable debug output (1 = verbose).	FILTERMSVERBOSE=1

Tip – If you omit GNPROVIDER, the default Yapapi strategy (LeastExpensiveLinearPayuMS) is used.

Running a Script

# Bash / Linux / macOS
export GNPROVIDER=[etam,ubuntu-2rec]
export GNFEATURES=[processor_trace]
python3 script.py

# PowerShell (Windows)
$env:GNPROVIDER="[etam,ubuntu-2rec]"
$env:GNFEATURES="[processor_trace]"
python script.py

You can also put the environment assignments in a .ps1 file:

# script.ps1
$env:FILTERMSVERBOSE=1
$env:GNFEATURES="[processor_trace]"
$env:GNPROVIDER="[etam,ubuntu-2rec,witek,golem2005,mf]"
$env:GNPROVIDER_BL="[sycamore]"
python script.py

.\script.ps1   # run the file

Note – When filtering by address, the filter applies to the node address, not the wallet address.

---

Advanced Usage

Wrapping an Existing Strategy

If you already have a custom strategy (e.g., LeastExpensiveLinearPayuMS), wrap it:

import yapapi
from decimal import Decimal

mystrategy = yapapi.strategy.LeastExpensiveLinearPayuMS(
    max_fixed_price=Decimal("0.00"),
    max_price_for={
        yapapi.props.com.Counter.CPU:  Decimal("0.01"),
        yapapi.props.com.Counter.TIME: Decimal("0.0011")
    }
)

async with Golem(
    budget=10.0,
    subnet_tag=subnet_tag,
    payment_driver=payment_driver,
    payment_network=payment_network,
    strategy=FilterProviderMS(mystrategy)
) as golem:
    ...

Nested Wrappers

You can stack multiple wrappers:

import yapapi
from decimal import Decimal

base = yapapi.strategy.LeastExpensiveLinearPayuMS(...)
modified = yapapi.strategy.DecreaseScoreForUnconfirmedAgreement(
    base_strategy=base,
    factor=0.01
)

async with Golem(
    budget=1.0,
    subnet_tag=subnet_tag,
    payment_driver=payment_driver,
    payment_network=payment_network,
    strategy=FilterProviderMS(modified)
) as golem:
    ...

---

Tips & Tricks

Symlink for Quick Access

If you keep gc__filterms in a separate location but want to import it as a package:

ln -s /path/to/gc__filterms gc__filterms   # inside your project directory

Now the import statement works without modifying PYTHONPATH.

---

FAQ

Question	Answer
What if I set both GNPROVIDER and GNPROVIDER_BL?	The blacklist takes precedence – any provider in GNPROVIDER_BL is excluded even if it appears in the whitelist.
Can I filter by wallet address instead of node address?	Currently only node addresses are supported. Future releases may add wallet‑address filtering.
Will this affect task scheduling performance?	The filtering happens before offers are considered, so there’s negligible overhead.

---

Contributing & Roadmap

• Network filtering – upcoming feature to filter by bandwidth or latency.
• Integration with gc__listoffers – unified offer‑listing and filtering experience.
• CLI helper – a small command‑line tool for inspecting provider metadata.

Feel free to open issues, submit pull requests, or suggest new features.
Happy hacking! 🚀

---