pygpt-net 2.4.28__py3-none-any.whl → 2.4.34__py3-none-any.whl

pygpt_net/core/idx/indexing.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.23 21:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 import datetime
@@ -349,6 +349,65 @@ class Indexing:
             data.append(doc.text)
         return "\n".join(data)
 
+    def read_web_content(
+            self,
+            url: str,
+            type: str = "webpage",
+            extra_args: dict = None
+    ) -> str:
+        """
+        Get content from external resource
+
+        :param url: external url to index
+        :param type: type of URL (webpage, feed, etc.)
+        :param extra_args: extra arguments for loader
+        :return: file content
+        """
+        docs = self.read_web(url, type, extra_args)
+        data = []
+        for doc in docs:
+            data.append(doc.text)
+        return "\n".join(data)
+
+    def read_web(
+            self,
+            url: str,
+            type: str = "webpage",
+            extra_args: dict = None,
+    ) -> list[Document]:
+        """
+        Read data from external resource
+
+        :param url: external url to index
+        :param type: type of URL (webpage, feed, etc.)
+        :param extra_args: extra arguments for loader
+        :return: list of documents
+        """
+        documents = []
+
+        # check if web loader for defined type exists
+        if type not in self.loaders["web"]:
+            raise ValueError("No web loader for type: {}".format(type))
+
+        try:
+            if "url" not in extra_args:
+                extra_args["url"] = url
+
+            # get unique external content identifier
+            unique_id = self.data_providers[type].get_external_id(extra_args)
+            self.window.core.idx.log("Loading web documents from: {}".format(unique_id))
+            self.window.core.idx.log("Using web loader for type: {}".format(type))
+
+            args = self.data_providers[type].prepare_args(**extra_args)
+
+            # get documents from external resource
+            documents = self.loaders["web"][type].load_data(
+                **args
+            )
+        except Exception as e:
+            self.window.core.debug.log(e)
+        return documents
+
     def prepare_document(self, doc: Document):
         """
         Prepare document to store
@@ -924,7 +983,8 @@ class Indexing:
             self,
             file_path: str,
             index_path: str,
-            model: ModelItem = None
+            model: ModelItem = None,
+            documents: list = None,
     ) -> list:
         """
         Index context attachment
@@ -932,6 +992,45 @@ class Indexing:
         :param file_path: path to file to index
         :param index_path: index path
         :param model: model
+        :param documents: list of documents (optional)
+        :return: response
+        """
+        if model is None:
+            model = self.window.core.models.from_defaults()
+
+        service_context = self.window.core.idx.llm.get_service_context(model=model)
+        index = self.window.core.idx.storage.get_ctx_idx(index_path, service_context=service_context)  # get or create ctx index
+
+        idx = "tmp:{}".format(index_path)  # tmp index id
+        self.window.core.idx.log("Indexing to context attachment index: {}...".format(idx))
+
+        doc_ids = []
+        if documents is None:
+            documents = self.get_documents(file_path)
+        for d in documents:
+            if self.is_stopped():  # force stop
+                break
+            self.prepare_document(d)
+            self.index_document(index, d)
+            doc_ids.append(d.id_)  # add to index
+
+        self.window.core.idx.storage.store_ctx_idx(index_path, index)
+        return doc_ids
+
+    def index_attachment_web(
+            self,
+            url: str,
+            index_path: str,
+            model: ModelItem = None,
+            documents: list = None,
+    ) -> list:
+        """
+        Index context attachment
+
+        :param url: URL to index
+        :param index_path: index path
+        :param model: model
+        :param documents: list of documents (optional)
         :return: response
         """
         if model is None:
@@ -943,8 +1042,14 @@ class Indexing:
         idx = "tmp:{}".format(index_path)  # tmp index id
         self.window.core.idx.log("Indexing to context attachment index: {}...".format(idx))
 
+        web_type = self.get_webtype(url)
         doc_ids = []
-        documents = self.get_documents(file_path)
+        if documents is None:
+            documents = self.read_web(
+                url=url,
+                type=web_type,
+                extra_args={},
+            )
         for d in documents:
             if self.is_stopped():  # force stop
                 break
@@ -955,20 +1060,23 @@ class Indexing:
         self.window.core.idx.storage.store_ctx_idx(index_path, index)
         return doc_ids
 
+    def get_webtype(self, url: str) -> str:
         """
-        # query tmp index
-        output = None
-        if len(files) > 0:
-            self.log("Querying temporary in-memory index: {}...".format(idx))
-            response = index.as_query_engine(
-                llm=llm,
-                streaming=False,
-            ).query(query)  # query with default prompt
-            if response:
-                ctx.add_doc_meta(self.get_metadata(response.source_nodes))  # store metadata
-                output = response.
-        return output
+        Get web loader type by URL
+
+        :param url: URL
+        :return: web loader type
         """
+        type = "webpage"  # default
+        for id in self.data_providers:
+            loader = self.data_providers[id]
+            if hasattr(loader, "is_supported_attachment"):
+                if loader.is_supported_attachment(url):
+                    type = id
+                    break
+        print("Selected web data loader: {}".format(type))
+        return type
+
     def remove_attachment(self, index_path: str, doc_id: str) -> bool:
         """
         Remove document from index

pygpt_net/core/modes.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.21 20:00:00                  #
+# Updated Date: 2024.11.26 19:00:00                  #
 # ================================================== #
 
 from pygpt_net.provider.core.mode.json_file import JsonFileProvider
@@ -14,6 +14,7 @@ from pygpt_net.core.types import (
     MODE_AGENT,
     MODE_AGENT_LLAMA,
     MODE_ASSISTANT,
+    MODE_AUDIO,
     MODE_CHAT,
     MODE_COMPLETION,
     MODE_EXPERT,
@@ -39,6 +40,7 @@ class Modes:
             MODE_AGENT,
             MODE_AGENT_LLAMA,
             MODE_ASSISTANT,
+            MODE_AUDIO,
             MODE_CHAT,
             MODE_COMPLETION,
             MODE_EXPERT,

pygpt_net/core/presets.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.21 20:00:00                  #
+# Updated Date: 2024.11.26 19:00:00                  #
 # ================================================== #
 
 import copy
@@ -17,6 +17,7 @@ from pygpt_net.core.types import (
     MODE_AGENT,
     MODE_AGENT_LLAMA,
     MODE_ASSISTANT,
+    MODE_AUDIO,
     MODE_CHAT,
     MODE_COMPLETION,
     MODE_EXPERT,
@@ -72,6 +73,7 @@ class Presets:
         curr_agent = self.build()
         curr_agent_llama = self.build()
         curr_expert = self.build()
+        curr_audio = self.build()
 
         # prepare ids
         id_chat = 'current.chat'
@@ -84,6 +86,7 @@ class Presets:
         id_agent = 'current.agent'
         id_agent_llama = 'current.agent_llama'
         id_expert = 'current.expert'
+        id_audio = 'current.audio'
 
         # set default initial prompt for chat mode
         curr_chat.prompt = self.window.core.prompt.get('default')
@@ -109,6 +112,8 @@ class Presets:
             curr_agent_llama = self.items[id_agent_llama]
         if id_expert in self.items:
             curr_expert = self.items[id_expert]
+        if id_audio in self.items:
+            curr_audio = self.items[id_audio]
 
         # allow usage in specific mode
         curr_chat.chat = True
@@ -121,6 +126,7 @@ class Presets:
         curr_agent.agent = True
         curr_agent_llama.agent_llama = True
         curr_expert.expert = True
+        curr_audio.audio = True
 
         # always apply default name
         curr_chat.name = '*'
@@ -133,6 +139,7 @@ class Presets:
         curr_agent.name = '*'
         curr_agent_llama.name = '*'
         curr_expert.name = '*'
+        curr_audio.name = '*'
 
         # append at first position
         self.items = {
@@ -146,6 +153,7 @@ class Presets:
             id_agent: curr_agent,
             id_agent_llama: curr_agent_llama,
             id_expert: curr_expert,
+            id_audio: curr_audio,
             **self.items
         }
 
@@ -217,6 +225,8 @@ class Presets:
             return MODE_AGENT_LLAMA
         if preset.expert:
             return MODE_EXPERT
+        if preset.audio:
+            return MODE_AUDIO
         return None
 
     def has(self, mode: str, id: str) -> bool:
@@ -286,7 +296,8 @@ class Presets:
                     or (mode == MODE_LLAMA_INDEX and self.items[id].llama_index) \
                     or (mode == MODE_AGENT and self.items[id].agent) \
                     or (mode == MODE_AGENT_LLAMA and self.items[id].agent_llama) \
-                    or (mode == MODE_EXPERT and self.items[id].expert):
+                    or (mode == MODE_EXPERT and self.items[id].expert) \
+                    or (mode == MODE_AUDIO and self.items[id].audio):
                 presets[id] = self.items[id]
         return presets
 

pygpt_net/core/render/markdown/pid.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.05 23:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 class PidData():
@@ -17,6 +17,7 @@ class PidData():
         self.meta = meta
         self.images_appended = []
         self.urls_appended = []
+        self.files_appended = []
         self.buffer = ""
         self.prev_position = None  # previous cursor position (for chunk append)
         self.is_cmd = False

pygpt_net/core/render/plain/pid.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.05 23:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 class PidData():
@@ -17,5 +17,6 @@ class PidData():
         self.meta = meta
         self.images_appended = []
         self.urls_appended = []
+        self.files_appended = []
         self.buffer = ""
         self.is_cmd = False

pygpt_net/core/render/web/body.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.20 21:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 import os
@@ -167,10 +167,12 @@ class Body:
         if num is not None and num_all is not None and num_all > 1:
             num_str = " [{}]".format(num)
         url, path = self.window.core.filesystem.extract_local_url(url)
-        return """<a href="{url}"><img src="{path}" class="image"></a>
-        <p><b>{prefix}{num}:</b> <a href="{url}">{path}</a></p>""". \
+        basename = os.path.basename(path)
+        return """<div class="extra-src-img-box" title="{url}"><div class="img-outer"><div class="img-wrapper"><a href="{url}"><img src="{path}" class="image"></a></div>
+        <a href="{url}" class="title">{title}</a></div></div>""". \
             format(prefix=trans('chat.prefix.img'),
                    url=url,
+                   title=basename,
                    path=path,
                    num=num_str)
 
@@ -183,13 +185,19 @@ class Body:
         :param num_all: number of all URLs
         :return: HTML code
         """
+        icon_path = os.path.join(
+            self.window.core.config.get_app_path(),
+            "data", "icons", "public_filled.svg"
+        )
+        icon = '<img src="file://{}" width="25" height="25" valign="middle" class="extra-src-icon">'.format(icon_path)
         num_str = ""
         if num is not None and num_all is not None and num_all > 1:
             num_str = " [{}]".format(num)
-        return """<b>{prefix}{num}:</b> <a href="{url}" title="{url}">{url}</a>""". \
-            format(prefix=trans('chat.prefix.url'),
-                   url=url,
-                   num=num_str)
+        return """{icon}<b>{num}</b> <a href="{url}" title="{url}">{url}</a>""". \
+            format(url=url,
+                   num=num_str,
+                   icon=icon,
+            )
 
     def get_docs_html(self, docs: list) -> str:
         """
@@ -223,8 +231,16 @@ class Body:
         except Exception as e:
             pass
 
+        icon_path = os.path.join(
+            self.window.core.config.get_app_path(),
+            "data", "icons", "db.svg"
+        )
+        icon = '<img src="file://{}" width="25" height="25" valign="middle" class="extra-src-icon">'.format(icon_path)
         if html_sources != "":
-            html += "<p><small><b>{prefix}:</b></small></p>".format(prefix=trans('chat.prefix.doc'))
+            html += "<p>{icon}<small><b>{prefix}:</b></small></p>".format(
+                prefix=trans('chat.prefix.doc'),
+                icon=icon,
+            )
             html += "<div class=\"cmd\">"
             html += "<p>" + html_sources + "</p>"
             html += "</div> "
@@ -239,15 +255,21 @@ class Body:
         :param num_all: number of all files
         :return: HTML code
         """
+        icon_path = os.path.join(
+            self.window.core.config.get_app_path(),
+            "data", "icons", "attachments.svg"
+        )
+        icon = '<img src="file://{}" width="25" height="25" valign="middle" class="extra-src-icon">'.format(icon_path)
         num_str = ""
         if num is not None and num_all is not None and num_all > 1:
             num_str = " [{}]".format(num)
         url, path = self.window.core.filesystem.extract_local_url(url)
-        return """<div><b>{prefix}{num}:</b> <a href="{url}">{path}</a></div>""". \
-            format(prefix=trans('chat.prefix.file'),
-                   url=url,
+        return """{icon} <b>{num}</b> <a href="{url}">{path}</a>""". \
+            format(url=url,
                    path=path,
-                   num=num_str)
+                   num=num_str,
+                   icon=icon,
+            )
 
     def prepare_tool_extra(self, ctx: CtxItem) -> str:
         """

pygpt_net/core/render/web/pid.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.05 23:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 from pygpt_net.utils import trans
@@ -20,6 +20,7 @@ class PidData():
         self.meta = meta
         self.images_appended = []
         self.urls_appended = []
+        self.files_appended = []
         self.buffer = ""  # stream buffer
         self.is_cmd = False
         self.html = ""  # html buffer

pygpt_net/core/render/web/renderer.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.21 20:00:00                  #
+# Updated Date: 2024.11.26 04:00:00                  #
 # ================================================== #
 
 import json
@@ -420,16 +420,20 @@ class Renderer(BaseRenderer):
         # files and attachments, TODO check attachments
         c = len(ctx.files)
         if c > 0:
+            files_html = []
             n = 1
             for file in ctx.files:
-                if file in appended:
+                if file in appended or file in self.pids[pid].files_appended:
                     continue
                 try:
                     appended.append(file)
-                    html += self.body.get_file_html(file, n, c)
+                    files_html.append(self.body.get_file_html(file, n, c))
+                    self.pids[pid].files_appended.append(file)
                     n += 1
                 except Exception as e:
                     pass
+            if files_html:
+                html += "<br/>" + "<br/>".join(files_html)
 
         # urls
         c = len(ctx.urls)
@@ -519,6 +523,7 @@ class Renderer(BaseRenderer):
         self.clear_chunks(pid)
         self.pids[pid].images_appended = []
         self.pids[pid].urls_appended = []
+        self.pids[pid].files_appended = []
         self.get_output_node_by_pid(pid).reset_current_content()
         self.reset_names_by_pid(pid)
 
@@ -540,6 +545,8 @@ class Renderer(BaseRenderer):
 
         :param pid: context PID
         """
+        if pid is None:
+            return
         self.clear_chunks_input(pid)
         self.clear_chunks_output(pid)
 
@@ -883,6 +890,8 @@ class Renderer(BaseRenderer):
 
        :param meta: context meta
         """
+        if meta is None:
+            return
         pid = self.get_or_create_pid(meta)
         self.reset_names_by_pid(pid)
 

pygpt_net/core/tokens.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.23 00:00:00                  #
+# Updated Date: 2024.11.26 19:00:00                  #
 # ================================================== #
 
 import tiktoken
@@ -15,6 +15,7 @@ from pygpt_net.core.types import (
     MODE_AGENT,
     MODE_AGENT_LLAMA,
     MODE_ASSISTANT,
+    MODE_AUDIO,
     MODE_CHAT,
     MODE_COMPLETION,
     MODE_EXPERT,
@@ -33,6 +34,7 @@ CHAT_MODES = [
     MODE_AGENT,
     MODE_AGENT_LLAMA,
     MODE_EXPERT,
+    MODE_AUDIO,
 ]
 
 
@@ -374,7 +376,7 @@ class Tokens:
         model_id = self.window.core.models.get_id(model)
         mode = self.window.core.config.get('mode')
         tokens = 0
-        if mode == MODE_CHAT or mode == MODE_VISION:
+        if mode in [MODE_CHAT, MODE_VISION, MODE_AUDIO]:
             tokens += self.from_prompt(system_prompt, "", model_id)  # system prompt
             tokens += self.from_text("system", model_id)
             tokens += self.from_prompt(input_prompt, "", model_id)  # input prompt

pygpt_net/core/types/mode.py

@@ -6,9 +6,10 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.11.21 20:00:00                  #
+# Updated Date: 2024.11.26 19:00:00                  #
 # ================================================== #
 
+MODE_AUDIO = "audio"
 MODE_CHAT = "chat"
 MODE_COMPLETION = "completion"
 MODE_IMAGE = "img"

pygpt_net/data/config/config.json

@@ -1,8 +1,8 @@
 {
   "__meta__": {
-    "version": "2.4.28",
-    "app.version": "2.4.28",
-    "updated_at": "2024-11-24T00:00:00"
+    "version": "2.4.34",
+    "app.version": "2.4.34",
+    "updated_at": "2024-11-26T00:00:00"
   },
   "access.audio.event.speech": false,
   "access.audio.event.speech.disabled": [],
@@ -65,6 +65,7 @@
   "assistant": "",
   "assistant_thread": "",
   "assistant.store.hide_threads": true,
+  "attachments_auto_index": false,
   "attachments_send_clear": true,
   "attachments_capture_clear": true,
   "audio.transcribe.convert_video": true,
@@ -74,6 +75,7 @@
   "ctx.attachment.img": false,
   "ctx.attachment.mode": "full",
   "ctx.attachment.summary.model": "gpt-4o-mini",
+  "ctx.attachment.query.model": "gpt-4o-mini",
   "ctx.attachment.verbose": false,
   "ctx.auto_summary": true,
   "ctx.auto_summary.model": "gpt-4o-mini",
@@ -225,7 +227,7 @@
   "max_tokens_length": 32000,
   "max_total_tokens": 128000,
   "mode": "chat",
-  "model": "gpt-4o-mini",
+  "model": "gpt-4o",
   "notepad.num": 1,
   "organization_key": "",
   "output_timestamp": false,
@@ -244,6 +246,7 @@
     "cmd_files": false,
     "cmd_history": false,
     "cmd_serial": false,
+    "cmd_system": false,
     "cmd_web": false,
     "crontab": false,
     "experts": false,