From e483ba4d287659d3d0ad0168e8399141462e2e3b Mon Sep 17 00:00:00 2001 From: Believethehype <1097224+believethehype@users.noreply.github.com> Date: Thu, 29 Aug 2024 16:00:38 +0200 Subject: [PATCH] add new summarization dvm, be more generous with generic dvms --- nostr_dvm/dvm.py | 6 +- nostr_dvm/tasks/summarization_duckduck_ai.py | 141 +++++++++++++++++++ nostr_dvm/utils/backend_utils.py | 7 +- setup.py | 2 +- tests/summarization_duck.py | 62 ++++++++ 5 files changed, 210 insertions(+), 8 deletions(-) create mode 100644 nostr_dvm/tasks/summarization_duckduck_ai.py create mode 100644 tests/summarization_duck.py diff --git a/nostr_dvm/dvm.py b/nostr_dvm/dvm.py index 9428986..4d361f0 100644 --- a/nostr_dvm/dvm.py +++ b/nostr_dvm/dvm.py @@ -204,7 +204,7 @@ class DVM: self.dvm_config) for dvm in self.dvm_config.SUPPORTED_DVMS: - if dvm.TASK == task and dvm.FIX_COST == 0 and dvm.PER_UNIT_COST == 0 and dvm_config.NIP88 is None: + if dvm.TASK == task or dvm.TASK == "generic" and dvm.FIX_COST == 0 and dvm.PER_UNIT_COST == 0 and dvm_config.NIP88 is None: task_is_free = True cashu_redeemed = False @@ -542,7 +542,7 @@ class DVM: task = await get_task(original_event, self.client, self.dvm_config) for dvm in self.dvm_config.SUPPORTED_DVMS: - if task == dvm.TASK: + if task == dvm.TASK or dvm.TASK == "generic": try: post_processed = await dvm.post_process(data, original_event) await send_nostr_reply_event(post_processed, original_event.as_json()) @@ -785,7 +785,7 @@ class DVM: for dvm in self.dvm_config.SUPPORTED_DVMS: result = "" try: - if task == dvm.TASK: + if task == dvm.TASK or dvm.TASK == "generic": request_form = await dvm.create_request_from_nostr_event(job_event, self.client, self.dvm_config) diff --git a/nostr_dvm/tasks/summarization_duckduck_ai.py b/nostr_dvm/tasks/summarization_duckduck_ai.py new file mode 100644 index 0000000..ef56d0d --- /dev/null +++ b/nostr_dvm/tasks/summarization_duckduck_ai.py @@ -0,0 +1,141 @@ +import json +import os +import re + +from nostr_sdk import Tag, Kind +from nostr_dvm.interfaces.dvmtaskinterface import DVMTaskInterface, process_venv +from nostr_dvm.utils.admin_utils import AdminConfig +from nostr_dvm.utils.definitions import EventDefinitions +from nostr_dvm.utils.dvmconfig import DVMConfig, build_default_config +from nostr_dvm.utils.nip88_utils import NIP88Config +from nostr_dvm.utils.nip89_utils import NIP89Config, check_and_set_d_tag +from nostr_dvm.utils.nostr_utils import get_referenced_event_by_id, get_events_by_ids, get_event_by_id + +""" +This File contains a Module to generate Text, based on a prompt using the Unleashed.chat API. + +Accepted Inputs: Prompt (text) +Outputs: Generated text +""" + + +class SummarizationDuckDuck(DVMTaskInterface): + KIND: Kind = EventDefinitions.KIND_NIP90_SUMMARIZE_TEXT + TASK: str = "text-to-text" + FIX_COST: float = 10 + dependencies = [("nostr-dvm", "nostr-dvm"), + ("duck_chat", "-U https://github.com/mrgick/duckduckgo-chat-ai/archive/master.zip")] + + async def init_dvm(self, name, dvm_config: DVMConfig, nip89config: NIP89Config, nip88config: NIP88Config = None, + admin_config: AdminConfig = None, options=None): + dvm_config.SCRIPT = os.path.abspath(__file__) + + async def is_input_supported(self, tags, client=None, dvm_config=None): + for tag in tags: + if tag.as_vec()[0] == 'i': + print(tag.as_vec()) + input_value = tag.as_vec()[1] + input_type = tag.as_vec()[2] + if input_type != "event" and input_type != "job" and input_type != "text": + return False + + return True + + async def create_request_from_nostr_event(self, event, client=None, dvm_config=None): + request_form = {"jobID": event.id().to_hex() + "_" + self.NAME.replace(" ", "")} + prompt = "" + collect_events = [] + nostr_mode = True + + for tag in event.tags(): + if tag.as_vec()[0] == 'i': + input_type = tag.as_vec()[2] + if input_type == "text": + prompt += tag.as_vec()[1] + "\n" + elif input_type == "event": + collect_events.append(tag.as_vec()[1]) + # evt = get_event_by_id(tag.as_vec()[1], client=client, config=dvm_config) + # prompt += evt.content() + "\n" + elif input_type == "job": + evt = await get_referenced_event_by_id(event_id=tag.as_vec()[1], client=client, + kinds=[EventDefinitions.KIND_NIP90_RESULT_EXTRACT_TEXT, + EventDefinitions.KIND_NIP90_RESULT_SUMMARIZE_TEXT, + EventDefinitions.KIND_NIP90_RESULT_TRANSLATE_TEXT, + EventDefinitions.KIND_NIP90_RESULT_CONTENT_DISCOVERY], + dvm_config=dvm_config) + if evt is None: + print("Event not found") + raise Exception + + if evt.kind() == EventDefinitions.KIND_NIP90_RESULT_CONTENT_DISCOVERY: + result_list = json.loads(evt.content()) + prompt = "" + for tag in result_list: + e_tag = Tag.parse(tag) + evt = await get_event_by_id(e_tag.as_vec()[1], client=client, config=dvm_config) + prompt += evt.content() + "\n" + + else: + prompt = evt.content() + + evts = await get_events_by_ids(collect_events, client=client, config=dvm_config) + if evts is not None: + for evt in evts: + prompt += evt.content() + "\n" + + clean_prompt = re.sub(r'^https?:\/\/.*[\r\n]*', '', prompt, flags=re.MULTILINE) + options = { + "input": clean_prompt[:4000], + "nostr": nostr_mode, + } + request_form['options'] = json.dumps(options) + + return request_form + + async def process(self, request_form): + from duck_chat import DuckChat + from duck_chat import ModelType + + options = self.set_options(request_form) + try: + async with DuckChat(model=ModelType.GPT4o) as chat: + query = "Summarize the following notes by different authors: " + options["input"] + result = await chat.ask_question(query) + print(result) + return result + + except Exception as e: + print("Error in Module: " + str(e)) + raise Exception(e) + + + +# We build an example here that we can call by either calling this file directly from the main directory, +# or by adding it to our playground. You can call the example and adjust it to your needs or redefine it in the +# playground or elsewhere +def build_example(name, identifier, admin_config): + dvm_config = build_default_config(identifier) + dvm_config.SEND_FEEDBACK_EVENTS = True + admin_config.LUD16 = dvm_config.LN_ADDRESS + + nip89info = { + "name": name, + "image": "https://image.nostr.build/28da676a19841dcfa7dcf7124be6816842d14b84f6046462d2a3f1268fe58d03.png", + "about": "I summarize Text", + "encryptionSupported": True, + "cashuAccepted": True, + "nip90Params": {} + } + + nip89config = NIP89Config() + nip89config.DTAG = check_and_set_d_tag(identifier, name, dvm_config.PRIVATE_KEY, nip89info["image"]) + nip89config.CONTENT = json.dumps(nip89info) + admin_config2 = AdminConfig() + admin_config2.REBROADCAST_NIP89 = False + + return SummarizationDuckDuck(name=name, dvm_config=dvm_config, nip89config=nip89config, + admin_config=admin_config2) + + +if __name__ == '__main__': + process_venv(SummarizationDuckDuck) diff --git a/nostr_dvm/utils/backend_utils.py b/nostr_dvm/utils/backend_utils.py index 0a80347..ff6472f 100644 --- a/nostr_dvm/utils/backend_utils.py +++ b/nostr_dvm/utils/backend_utils.py @@ -135,8 +135,8 @@ async def check_task_is_supported(event: Event, client, config=None): # See if current dvm supports the task task = await get_task(event, client=client, dvm_config=dvm_config) - if task not in (x.TASK for x in dvm_config.SUPPORTED_DVMS): - return False, task + # if task not in (x.TASK for x in dvm_config.SUPPORTED_DVMS) and not task == "generic": + # return False, task # See if current dvm can handle input for given task for dvm in dvm_config.SUPPORTED_DVMS: if dvm.TASK == task: @@ -144,7 +144,6 @@ async def check_task_is_supported(event: Event, client, config=None): return False, task return True, task - except Exception as e: print("Check task: " + str(e)) @@ -185,7 +184,7 @@ def check_url_is_readable(url): def get_amount_per_task(task, dvm_config, duration=1): # duration is either static 1 (for images etc) or in seconds by default (e.g. audio/video) for dvm in dvm_config.SUPPORTED_DVMS: # this is currently just one - if dvm.TASK == task: + if dvm.TASK == task or dvm.TASK == "generic": amount = dvm.FIX_COST + (dvm.PER_UNIT_COST * duration) return amount else: diff --git a/setup.py b/setup.py index 1a06fd9..c9c709d 100644 --- a/setup.py +++ b/setup.py @@ -1,6 +1,6 @@ from setuptools import setup, find_packages -VERSION = '0.8.10' +VERSION = '0.8.11' DESCRIPTION = 'A framework to build and run Nostr NIP90 Data Vending Machines' LONG_DESCRIPTION = ('A framework to build and run Nostr NIP90 Data Vending Machines. See the github repository for more information') diff --git a/tests/summarization_duck.py b/tests/summarization_duck.py new file mode 100644 index 0000000..e12dcf0 --- /dev/null +++ b/tests/summarization_duck.py @@ -0,0 +1,62 @@ +import json +from pathlib import Path + +import dotenv +import torch +from diffusers import FluxPipeline +from nostr_sdk import Kind + + +from nostr_dvm.tasks.generic_dvm import GenericDVM +from nostr_dvm.tasks.summarization_duckduck_ai import SummarizationDuckDuck +from nostr_dvm.utils.admin_utils import AdminConfig +from nostr_dvm.utils.dvmconfig import build_default_config +from nostr_dvm.utils.nip89_utils import NIP89Config, check_and_set_d_tag +from nostr_dvm.utils.output_utils import upload_media_to_hoster + + +def playground(announce=False): + admin_config = AdminConfig() + admin_config.REBROADCAST_NIP89 = announce + admin_config.REBROADCAST_NIP65_RELAY_LIST = announce + admin_config.UPDATE_PROFILE = announce + + name = "Summarizer Duck" + identifier = "summarizer_duck" + dvm_config = build_default_config(identifier) + dvm_config.SEND_FEEDBACK_EVENTS = False + admin_config.LUD16 = dvm_config.LN_ADDRESS + + nip89info = { + "name": name, + "image": "https://image.nostr.build/28da676a19841dcfa7dcf7124be6816842d14b84f6046462d2a3f1268fe58d03.png", + "about": "I summarize Text", + "encryptionSupported": True, + "cashuAccepted": True, + "nip90Params": {} + } + + nip89config = NIP89Config() + nip89config.DTAG = check_and_set_d_tag(identifier, name, dvm_config.PRIVATE_KEY, nip89info["image"]) + nip89config.CONTENT = json.dumps(nip89info) + + + dvm = SummarizationDuckDuck(name=name, dvm_config=dvm_config, nip89config=nip89config, + admin_config=admin_config) + dvm.run(True) + + +if __name__ == '__main__': + env_path = Path('.env') + if not env_path.is_file(): + with open('.env', 'w') as f: + print("Writing new .env file") + f.write('') + if env_path.is_file(): + print(f'loading environment from {env_path.resolve()}') + dotenv.load_dotenv(env_path, verbose=True, override=True) + else: + raise FileNotFoundError(f'.env file not found at {env_path} ') + #pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16) + + playground(announce=False)