add second task, pdf-to-text

example for kind 5000 event

.env_example

@@ -5,4 +5,5 @@ USER_DB_PATH = nostrzaps.db
 LNBITS_INVOICE_KEY = lnbitswalletinvoicekey
 LNBITS_HOST = https://lnbits.com
 
+TASK_TEXTEXTRACTION_NIP89_DTAG = "asdd"
 TASK_TRANSLATION_NIP89_DTAG = abcded

README.md

@@ -1,6 +1,6 @@
 # Nostr Data Vending Machine Python Implementation
 
-This example DVM implementation in Python currently supports simple translations using Google translate.
+This example DVM implementation in Python currently supports simple translations using Google translate, as well as extraction of text from links with pdf files.
 
 At a later stage, additional example tasks will be added, as well as the integration into a larger Machine Learning backend 
 

dvm.py

@@ -4,7 +4,7 @@ import time
 import emoji
 from utils.definitions import EventDefinitions, DVMConfig, RequiredJobToWatch, JobToWatch, LOCAL_TASKS
 from utils.admin_utils import admin_make_database_updates
-from utils.ai_utils import GoogleTranslate
+from utils.ai_utils import extract_text_from_pdf, google_translate
 from utils.backend_utils import get_amount_per_task, check_task_is_supported, get_task
 from utils.database_utils import update_sql_table, get_from_sql_table, \
     create_sql_table, get_or_add_user, update_user_balance
@@ -38,6 +38,7 @@ def dvm(config):
 
     dm_zap_filter = Filter().pubkey(pk).kinds([EventDefinitions.KIND_ZAP]).since(Timestamp.now())
     dvm_filter = (Filter().kinds([EventDefinitions.KIND_NIP90_GENERIC,
+                                  EventDefinitions.KIND_NIP90_EXTRACT_TEXT,
                                   EventDefinitions.KIND_NIP90_TRANSLATE_TEXT,
                                   ]).since(Timestamp.now()))
     client.subscribe([dm_zap_filter, dvm_filter])
@@ -60,13 +61,12 @@ def dvm(config):
         user = get_or_add_user(event.pubkey().to_hex())
         is_whitelisted = user[2]
         is_blacklisted = user[3]
-        if is_whitelisted:
+
-            task_supported, task, duration = check_task_is_supported(event, client=client, get_duration=False,
+        task_supported, task, duration = check_task_is_supported(event, client=client,
+                                                                 get_duration=(not is_whitelisted),
                                                                  config=dvm_config)
         print(task)
-        else:
+
-            task_supported, task, duration = check_task_is_supported(event, client=client, get_duration=True,
-                                                                     config=dvm_config)
         if is_blacklisted:
             send_job_status_reaction(event, "error", client=client, config=dvm_config)
             print("[Nostr] Request by blacklisted user, skipped")
@@ -231,7 +231,9 @@ def dvm(config):
                 options = dict(opts)
 
                 if task == "translation":
-                    result = GoogleTranslate(options["text"], options["translation_lang"])
+                    result = google_translate(options["text"], options["translation_lang"])
+                elif task == "pdf-to-text":
+                    result = extract_text_from_pdf(options["url"])
                 # TODO ADD FURTHER LOCAL TASKS HERE
 
                 check_and_return_event(result, str(job_event.as_json()),

main.py

@@ -11,21 +11,29 @@ def run_nostr_dvm_with_local_config():
     from dvm import dvm, DVMConfig
     from utils.nip89_utils import NIP89Announcement
 
-
     dvmconfig = DVMConfig()
     dvmconfig.PRIVATE_KEY = os.getenv(env.NOSTR_PRIVATE_KEY)
-    dvmconfig.SUPPORTED_TASKS = ["translation"]
+    dvmconfig.SUPPORTED_TASKS = ["translation", "pdf-to-text"]
     dvmconfig.LNBITS_INVOICE_KEY = os.getenv(env.LNBITS_INVOICE_KEY)
     dvmconfig.LNBITS_URL = os.getenv(env.LNBITS_HOST)
 
+    # In admin_utils, set rebroadcast_nip89 to true to (re)broadcast your DVM. You can create a valid dtag and the content on vendata.io
+    # Add the dtag in your .env file so you can update your dvm later and change the content here as needed.
+
+    nip89extraction = NIP89Announcement()
+    nip89extraction.kind = EventDefinitions.KIND_NIP90_EXTRACT_TEXT
+    nip89extraction.dtag = os.getenv(env.TASK_TRANSLATION_NIP89_DTAG)
+    nip89extraction.pk = os.getenv(env.NOSTR_PRIVATE_KEY)
+    nip89extraction.content = "{\"name\":\"Pdf Extractor\",\"image\":\"https://image.nostr.build/c33ca6fc4cc038ca4adb46fdfdfda34951656f87ee364ef59095bae1495ce669.jpg\",\"about\":\"I extract Text from pdf documents\",\"nip90Params\":{}}"
+    dvmconfig.NIP89s.append(nip89extraction)
+
     nip89translation = NIP89Announcement()
     nip89translation.kind = EventDefinitions.KIND_NIP90_TRANSLATE_TEXT
     nip89translation.dtag = os.getenv(env.TASK_TRANSLATION_NIP89_DTAG)
     nip89translation.pk = os.getenv(env.NOSTR_PRIVATE_KEY)
-    nip89translation.content = "{\"name\":\"NostrAI DVM Translator\",\"image\":\"https://cdn.nostr.build/i/feb98d8700abe7d6c67d9106a72a20354bf50805af79869638f5a32d24a5ac2a.jpg\",\"about\":\"Translates Text from given text/event/job, currently using Google Translation Services into language defined in param.  \",\"nip90Params\":{\"language\":{\"required\":true,\"values\":[\"af\",\"am\",\"ar\",\"az\",\"be\",\"bg\",\"bn\",\"bs\",\"ca\",\"ceb\",\"co\",\"cs\",\"cy\",\"da\",\"de\",\"el\",\"eo\",\"es\",\"et\",\"eu\",\"fa\",\"fi\",\"fr\",\"fy\",\"ga\",\"gd\",\"gl\",\"gu\",\"ha\",\"haw\",\"hi\",\"hmn\",\"hr\",\"ht\",\"hu\",\"hy\",\"id\",\"ig\",\"is\",\"it\",\"he\",\"ja\",\"jv\",\"ka\",\"kk\",\"km\",\"kn\",\"ko\",\"ku\",\"ky\",\"la\",\"lb\",\"lo\",\"lt\",\"lv\",\"mg\",\"mi\",\"mk\",\"ml\",\"mn\",\"mr\",\"ms\",\"mt\",\"my\",\"ne\",\"nl\",\"no\",\"ny\",\"or\",\"pa\",\"pl\",\"ps\",\"pt\",\"ro\",\"ru\",\"sd\",\"si\",\"sk\",\"sl\",\"sm\",\"sn\",\"so\",\"sq\",\"sr\",\"st\",\"su\",\"sv\",\"sw\",\"ta\",\"te\",\"tg\",\"th\",\"tl\",\"tr\",\"ug\",\"uk\",\"ur\",\"uz\",\"vi\",\"xh\",\"yi\",\"yo\",\"zh\",\"zu\"]}}}"
+    nip89translation.content = "{\"name\":\"Translator\",\"image\":\"https://image.nostr.build/c33ca6fc4cc038ca4adb46fdfdfda34951656f87ee364ef59095bae1495ce669.jpg\",\"about\":\"I translate text from given text/event/job, currently using Google Translation Services into language defined in param.  \",\"nip90Params\":{\"language\":{\"required\":true,\"values\":[\"af\",\"am\",\"ar\",\"az\",\"be\",\"bg\",\"bn\",\"bs\",\"ca\",\"ceb\",\"co\",\"cs\",\"cy\",\"da\",\"de\",\"el\",\"eo\",\"es\",\"et\",\"eu\",\"fa\",\"fi\",\"fr\",\"fy\",\"ga\",\"gd\",\"gl\",\"gu\",\"ha\",\"haw\",\"hi\",\"hmn\",\"hr\",\"ht\",\"hu\",\"hy\",\"id\",\"ig\",\"is\",\"it\",\"he\",\"ja\",\"jv\",\"ka\",\"kk\",\"km\",\"kn\",\"ko\",\"ku\",\"ky\",\"la\",\"lb\",\"lo\",\"lt\",\"lv\",\"mg\",\"mi\",\"mk\",\"ml\",\"mn\",\"mr\",\"ms\",\"mt\",\"my\",\"ne\",\"nl\",\"no\",\"ny\",\"or\",\"pa\",\"pl\",\"ps\",\"pt\",\"ro\",\"ru\",\"sd\",\"si\",\"sk\",\"sl\",\"sm\",\"sn\",\"so\",\"sq\",\"sr\",\"st\",\"su\",\"sv\",\"sw\",\"ta\",\"te\",\"tg\",\"th\",\"tl\",\"tr\",\"ug\",\"uk\",\"ur\",\"uz\",\"vi\",\"xh\",\"yi\",\"yo\",\"zh\",\"zu\"]}}}"
     dvmconfig.NIP89s.append(nip89translation)
 
-
     nostr_dvm_thread = Thread(target=dvm, args=[dvmconfig])
     nostr_dvm_thread.start()
 
@@ -39,10 +47,4 @@ if __name__ == '__main__':
     else:
         raise FileNotFoundError(f'.env file not found at {env_path} ')
 
-
-
     run_nostr_dvm_with_local_config()
-
-
-
-

requirements.txt

@@ -30,3 +30,37 @@ translatepy==2.3
 tzdata==2023.3
 urllib3==2.1.0
 wcwidth==0.2.10
+(venv) tobias@Tobiass-MacBook-Pro-2 dvm % pip freeze
+beautifulsoup4==4.12.2
+bech32==1.2.0
+blessed==1.20.0
+certifi==2023.7.22
+charset-normalizer==3.3.2
+emoji==2.8.0
+ffmpegio==0.8.5
+ffmpegio-core==0.8.5
+idna==3.4
+inquirer==3.1.3
+install==1.3.5
+nostr-sdk==0.0.4
+numpy==1.26.2
+packaging==23.2
+pandas==2.1.3
+Pillow==10.1.0
+pluggy==1.3.0
+pycryptodome==3.19.0
+pypdf==3.17.1
+python-dateutil==2.8.2
+python-dotenv==1.0.0
+python-editor==1.0.4
+pytz==2023.3.post1
+pyuseragents==1.0.5
+readchar==4.0.5
+requests==2.31.0
+safeIO==1.2
+six==1.16.0
+soupsieve==2.5
+translatepy==2.3
+tzdata==2023.3
+urllib3==2.1.0
+wcwidth==0.2.10

utils/admin_utils.py

@@ -3,8 +3,8 @@ import time
 
 from nostr_sdk import Keys, EventBuilder, PublicKey
 
-from utils.database_utils import get_from_sql_table, list_db, clear_db, delete_from_sql_table, update_sql_table, \
+from utils.database_utils import get_from_sql_table, list_db, delete_from_sql_table, update_sql_table, \
-    get_or_add_user, update_user_metadata
+    get_or_add_user, clean_db
 from utils.nip89_utils import nip89_announce_tasks
 from utils.nostr_utils import send_event
 
@@ -14,7 +14,7 @@ def admin_make_database_updates(config=None, client=None):
     dvmconfig = config
 
     rebroadcast_nip89 = False
-    cleardb = False
+    cleandb = False
     listdatabase = False
     deleteuser = False
     whitelistuser = False
@@ -57,8 +57,8 @@ def admin_make_database_updates(config=None, client=None):
     if deleteuser:
         delete_from_sql_table(publickey)
 
-    if cleardb:
+    if cleandb:
-        clear_db()
+        clean_db()
 
     if listdatabase:
         list_db()

utils/ai_utils.py

@@ -1,3 +1,5 @@
+import os
+
 
 #We can add multiple Tasks here and call them in the do_work function.
 
@@ -5,7 +7,7 @@
 #the according event type in the beginning of dvm.py and
 
 
-def GoogleTranslate(text, translation_lang):
+def google_translate(text, translation_lang):
     from translatepy.translators.google import GoogleTranslate
     gtranslate = GoogleTranslate()
     length = len(text)
@@ -36,3 +38,21 @@ def GoogleTranslate(text, translation_lang):
 
 
     return translated_text
+
+def extract_text_from_pdf(url):
+    from pypdf import PdfReader
+    from pathlib import Path
+    import requests
+    file_path = Path('temp.pdf')
+    response = requests.get(url)
+    file_path.write_bytes(response.content)
+    reader = PdfReader(file_path)
+    number_of_pages = len(reader.pages)
+    text = ""
+    for page_num in range(number_of_pages):
+        page = reader.pages[page_num]
+        text = text + page.extract_text()
+
+    os.remove('temp.pdf')
+    return text
+

utils/backend_utils.py

@@ -19,6 +19,30 @@ def get_task(event, client, dvmconfig):
         else:
             return "unknown job: " + event.as_json()
 
+    elif event.kind() == EventDefinitions.KIND_NIP90_EXTRACT_TEXT:
+        for tag in event.tags():
+            if tag.as_vec()[0] == "i":
+                if tag.as_vec()[2] == "url":
+                    file_type = check_url_is_readable(tag.as_vec()[1])
+                    if file_type == "pdf":
+                        return "pdf-to-text"
+                    else:
+                        return "unknown job"
+                elif tag.as_vec()[2] == "event":
+                     evt = get_event_by_id(tag.as_vec()[1],config=dvmconfig)
+                     if evt is not None:
+                         if evt.kind() == 1063:
+                            for tag in evt.tags():
+                                if tag.as_vec()[0] == 'url':
+                                    file_type = check_url_is_readable(tag.as_vec()[1])
+                                    if file_type == "pdf":
+                                        return "pdf-to-text"
+                                    else:
+                                        return "unknown job"
+                         else:
+                             return "unknown type"
+
+
     elif event.kind() == EventDefinitions.KIND_NIP90_TRANSLATE_TEXT:
         return "translation"
 
@@ -58,6 +82,8 @@ def check_task_is_supported(event, client, get_duration = False, config=None):
         print("No output set")
     if task not in dvmconfig.SUPPORTED_TASKS:  # The Tasks this DVM supports (can be extended)
         return False, task, duration
+
+
     elif task == "translation" and (
             input_type != "event" and input_type != "job" and input_type != "text"):  # The input types per task
         return False, task, duration
@@ -88,12 +114,15 @@ def check_url_is_readable(url):
             ".mp3") or content_type == 'audio/ogg' or str(url).endswith(".ogg"):
         return "audio"
     elif content_type == 'image/png' or str(url).endswith(".png") or content_type == 'image/jpg' or str(url).endswith(
-            ".jpg") or content_type == 'image/jpeg' or str(url).endswith(".jpeg") or str(url).endswith(".pdf") or content_type == 'image/png' or str(
+            ".jpg") or content_type == 'image/jpeg' or str(url).endswith(".jpeg") or content_type == 'image/png' or str(
             url).endswith(".png"):
         return "image"
     elif content_type == 'video/mp4' or str(url).endswith(".mp4") or content_type == 'video/avi' or str(url).endswith(
             ".avi") or content_type == 'video/mov' or str(url).endswith(".mov"):
         return "video"
+    elif (str(url)).endswith(".pdf"):
+        return "pdf"
+
     # Otherwise we will not offer to do the job.
     return None
 
@@ -101,6 +130,9 @@ def get_amount_per_task(task, duration = 0, config=None):
     dvmconfig = config
     if task == "translation":
         amount = dvmconfig.COSTPERUNIT_TRANSLATION
+    elif task == "pdf-to-text":
+        amount = dvmconfig.COSTPERUNIT_TEXT_EXTRACTION
+
     else:
         print("[Nostr] Task " + task + " is currently not supported by this instance, skipping")
         return None

utils/database_utils.py

@@ -101,7 +101,7 @@ def delete_from_sql_table(npub):
         print(e)
 
 
-def clear_db():
+def clean_db():
     try:
         con = sqlite3.connect(os.getenv(env.USER_DB_PATH))
         cur = con.cursor()

utils/definitions.py

@@ -3,7 +3,7 @@ from dataclasses import dataclass
 from nostr_sdk import Event
 NEW_USER_BALANCE = 250
 
-LOCAL_TASKS = ["conversion", "summarization","note-recommendation", "inactive-following", "image-upscale", "translation"]
+LOCAL_TASKS = ["pdf-to-text", "translation"]
 # Tasks performed by the DVM and not send to nova-server (can change later)
 
 RELAY_LIST = ["wss://relay.damus.io", "wss://nostr-pub.wellorder.net", "wss://nos.lol", "wss://nostr.wine",
@@ -54,7 +54,8 @@ class DVMConfig:
     SHOWRESULTBEFOREPAYMENT: bool = True  # if this is true show results even when not paid right after autoprocess
     NEW_USER_BALANCE: int = 250  # Free credits for new users
 
-    COSTPERUNIT_TRANSLATION: int = 20  # Still need to multiply this by duration
+    COSTPERUNIT_TRANSLATION: int = 20
+    COSTPERUNIT_TEXT_EXTRACTION: int = 20
 
     NIP89s: list = []
 

utils/env.py

@@ -7,5 +7,6 @@ LNBITS_INVOICE_KEY = "LNBITS_INVOICE_KEY"
 LNBITS_HOST = "LNBITS_HOST"
 
 TASK_TRANSLATION_NIP89_DTAG = "TASK_TRANSLATION_NIP89_DTAG"
+TASK_TEXTEXTRACTION_NIP89_DTAG = "TASK_TEXTEXTRACTION_NIP89_DTAG"
 
 

utils/requestform_utils.py

@@ -62,4 +62,29 @@ def create_requestform_from_nostr_event(event, is_bot=False, client=None, dvmcon
                                   text.replace('\U0001f919', "").replace("=", "equals").
                                   replace(";", ","))
 
+
+
+
+    elif task == "pdf-to-text":
+        input_type = "url"
+        input_content = ""
+        url = ""
+        for tag in event.tags():
+            if tag.as_vec()[0] == 'i':
+                input_type = tag.as_vec()[2]
+                input_content = tag.as_vec()[1]
+
+        if input_type == "url":
+            url = input_content
+        elif input_type == "event":
+            evt = get_event_by_id(input_content, config=dvmconfig)
+            url = re.search("(?P<url>https?://[^\s]+)", evt.content()).group("url")
+        elif input_type == "job":
+            evt = get_referenced_event_by_id(input_content, [EventDefinitions.KIND_NIP90_RESULT_GENERATE_IMAGE],
+                                             client, config=dvmconfig)
+
+            url = re.search("(?P<url>https?://[^\s]+)", evt.content()).group("url")
+
+        request_form["optStr"] = 'url=' + url
+
     return request_form