From 6cda9ceddb4c9686e2466d6a0fd08296b961232c Mon Sep 17 00:00:00 2001
From: Johan <johan.p1sson@gmail.com>
Date: Mon, 6 Apr 2026 21:44:25 +0200
Subject: [PATCH] started on new link agnet with tools

---
 .gitignore      |  64 +++++++++++++
 ask_linkding.py |  91 ++++++++++++++++++
 link_agent.py   | 241 ++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 396 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 ask_linkding.py
 create mode 100644 link_agent.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..b1bba61
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,64 @@
+#Ignore thumbnails created by Windows
+Thumbs.db
+#Ignore files built by Visual Studio
+*.user
+*.aps
+*.pch
+*.vspscc
+*_i.c
+*_p.c
+*.ncb
+*.suo
+*.bak
+*.cache
+*.ilk
+*.log
+[Bb]in
+[Dd]ebug*/
+*.sbr
+obj/
+[Rr]elease*/
+_ReSharper*/
+.klocwork
+.vs
+[Dd]ebug
+
+#General build files
+*.obj
+*.tmp
+*.o
+*.d
+*.exe
+
+#default log
+out.txt
+
+#input files
+*.ngc
+
+#Python
+*.pyc
+
+#emacs saves
+*~
+\#*
+
+#test-bench builds
+[Ll]ib
+*_tests
+
+#\.vscode #migth be good to save settings, but it is also a bit of a mess
+build/
+
+#drawio backups
+*.bkp
+*.dtmp
+
+\scans
+
+# Environment variables (secrets)
+.env
+.env.local
+
+# virtual environments
+venv/
\ No newline at end of file
diff --git a/ask_linkding.py b/ask_linkding.py
new file mode 100644
index 0000000..bd243ac
--- /dev/null
+++ b/ask_linkding.py
@@ -0,0 +1,91 @@
+#pip install langgraph langchain langchain-community langchain-ollama
+
+import requests, json, re, string
+from html.parser import HTMLParser
+
+class MyHTMLParser(HTMLParser):
+    def __init__(self):
+        super().__init__()
+        self.skip_content = False  # Skip content inside <script> and <style>
+        self.crawled_text = ''
+
+    def handle_starttag(self, tag, attrs):
+        # Skip content inside script and style tags
+        if tag in ['script', 'style']:
+            self.skip_content = True
+
+    def handle_endtag(self, tag):
+        # Resume collecting after closing script and style tags
+        if tag in ['script', 'style']:
+            self.skip_content = False
+
+    def handle_data(self, data):
+        # Only process data if not inside script/style tags
+        if self.skip_content:
+            return
+        
+        # Clean up the data: strip whitespace and filter empty lines
+        text = data.strip()
+        
+        # Skip if empty or too short
+        if not text or len(text) < 2:
+            return
+        
+        # Skip common junk: JSON-like patterns, URLs in certain contexts, etc.
+        if text.startswith('{') or text.startswith('['):
+            return
+        
+        # Print meaningful text
+        #print("data  :", text)
+        self.crawled_text += text + '\n'
+
+
+#_url = "http://192.168.50.14:9090/api/bookmarks/"
+#_url = "https://linkding.hal.se/api/bookmarks/5/"
+_url = "https://linkding.hal.se/api/bookmarks/?added_since=2026-04-01T00:00:00Z"
+
+headers = {
+    "Authorization": "Token fa54dee2ccbcad80a0c6259bdbbed896581e1423"
+}
+
+htmlParser = MyHTMLParser()
+#MyHTMLParser.convert_charrefs=False
+#MyHTMLParser.scripting=True
+
+response = requests.get(_url, headers=headers)
+data = response.json()
+#print(data)
+#for key, value in data.items():
+#    print(f"{key}: {value}")
+
+for bookmark in data['results']:
+    id = bookmark['id']
+    url = bookmark['url']
+    title = bookmark['title']
+    print(f"Id: {id}, Title: {title}\nURL: {url}\ndate_added: {bookmark['date_added']}\ndescription: {bookmark['description']}\ntag_names: {bookmark['tag_names']}\n")
+"""   
+    raw_html = requests.get(url.strip()).text
+    #print(raw_html[:500])
+   
+    htmlParser.feed(raw_html)
+
+    print(htmlParser.crawled_text) #[:500])
+    htmlParser.crawled_text = '' #reset for next bookmark
+"""
+
+# https://linkding.link/api/
+# tools #####################
+# Archive:
+# POST /api/bookmarks/<id>/archive/
+
+# List Tags
+# GET /api/tags/
+
+#PATCH /api/bookmarks/<id>/
+#Example payload:
+#{
+#  "tag_names": [
+#    "tag1",
+#    "tag2"
+#  ]
+#}
\ No newline at end of file
diff --git a/link_agent.py b/link_agent.py
new file mode 100644
index 0000000..49afdc1
--- /dev/null
+++ b/link_agent.py
@@ -0,0 +1,241 @@
+# python3 -m pip install --user virtualenv
+# python3 -m venv venv
+# source venv/bin/activate
+# pip install langgraph langchain langchain-community langchain-ollama
+# source venv/bin/activate
+
+
+from typing import TypedDict, List
+from urllib import response
+#from asyncio import tools
+from langgraph.graph import START, StateGraph, END
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_core.messages import HumanMessage, SystemMessage
+from langchain_core.tools import tool
+from langgraph.prebuilt import ToolNode
+import requests, json, re, string
+from html.parser import HTMLParser
+from langchain_ollama import ChatOllama
+import operator
+from typing_extensions import TypedDict, Annotated
+
+class HTMLCrawler(HTMLParser):
+    def __init__(self):
+        super().__init__()       
+        self.skip_content = False  # Skip content inside <script> and <style>
+        self.crawled_text = ''
+
+    def handle_starttag(self, tag, attrs):
+        # Skip content inside script and style tags
+        if tag in ['script', 'style']:
+            self.skip_content = True
+
+    def handle_endtag(self, tag):
+        # Resume collecting after closing script and style tags
+        if tag in ['script', 'style']:
+            self.skip_content = False
+
+    def handle_data(self, data):
+        # Only process data if not inside script/style tags
+        if self.skip_content:
+            return
+        
+        # Clean up the data: strip whitespace and filter empty lines
+        text = data.strip()
+        
+        # Skip if empty or too short
+        if not text or len(text) < 2:
+            return
+        
+        # Skip common junk: JSON-like patterns, URLs in certain contexts, etc.
+        if text.startswith('{') or text.startswith('['):
+            return
+        
+        # Print meaningful text
+        #print("data  :", text)
+        self.crawled_text += text + '\n'
+
+
+
+# Configuration
+LLM_BASE_URL = "http://192.168.50.215:11434"
+MODEL_NAME = "qwen3:8b"
+LINKDING_API_URL = "https://linkding.hal.se/api/bookmarks/"
+LINKDING_API_TOKEN = "fa54dee2ccbcad80a0c6259bdbbed896581e1423"
+
+
+llm = ChatOllama(model=MODEL_NAME, base_url=LLM_BASE_URL)
+
+@tool
+def todays_date() -> str:
+    """Use this tool whenever you need to know today's date in ISO 8601 format."""
+    from datetime import datetime
+    return datetime.now().strftime("%Y-%m-%dT%H:%M:%SZ")
+
+@tool
+def crawl_homepage(url:str) -> str:
+    """Use this tool whenever you need to read the content of a home page on the internet.
+        
+    Args:
+        url: home page url to crawl
+    
+    Returns:
+        Formatted test from homepage
+    """
+    htmlParser = HTMLCrawler()
+    try:
+        raw_html = requests.get(url.strip()).text    
+        htmlParser.feed(raw_html)
+    except Exception as e:
+        return f"Error occurred while parsing HTML: {e}"
+
+    return htmlParser.crawled_text
+
+@tool
+def fetch_bookmarks(date_added: str = "2026-03-01T00:00:00Z") -> str:
+    """Use this tool whenever you need to fetch bookmarks.
+    
+    Args:
+        date_added: The date from which to fetch bookmarks, in ISO 8601 format (e.g., "2026-04-01T00:00:00Z").
+    
+    Returns:
+        Formatted bookmark with titles, descriptions, URL, date_added, tags and ID.
+    """
+
+    _url = f"{LINKDING_API_URL}?added_since={date_added}"
+
+    _headers = {
+        "Authorization": "Token " + LINKDING_API_TOKEN
+    }
+
+    response = requests.get(_url, headers=_headers)
+    data = response.json()
+
+    if not data:
+        return f"No bookmarks found"
+
+    formatted_results = [f"Bookmarks:"]
+    for i, bookmark in enumerate(data['results'], 1):
+        title = bookmark.get('title', 'No title')
+        id = bookmark.get('id', 'No ID')
+        url = bookmark.get('url', 'No URL')
+        date_added = bookmark.get('date_added', 'No date added')
+        description = bookmark.get('description', 'No description')
+        tag_names = bookmark.get('tag_names', 'No tags')
+
+        text = f"   {title}\n   {description}\n   {url}\n   {date_added}\n   {tag_names}\n   {id}\n"
+        
+        formatted_results.append(text)
+
+    return "\n\n".join(formatted_results)
+
+@tool
+def add_tag_to_bookmark(bookmark_id: int, tag: str) -> str:
+    """Use this tool whenever you need to add a tag to a bookmark.
+    
+    Args:
+        bookmark_id: The ID of the bookmark to which the tag should be added.
+        tag: The tag to add to the bookmark.
+    
+    Returns:
+        A message indicating whether the tag was successfully added or if an error occurred.
+    """
+    _url = f"{LINKDING_API_URL}{bookmark_id}/"
+    _headers = {
+        "Authorization": "Token " + LINKDING_API_TOKEN,
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "tag_names": [tag]
+    }
+
+    try:
+        response = requests.patch(_url, headers=_headers, data=json.dumps(payload))
+        response.raise_for_status()  # Raise an exception for HTTP errors
+        return f"Tag '{tag}' successfully added to bookmark with ID {bookmark_id}."
+    except requests.exceptions.RequestException as e:
+        return f"Error occurred while adding tag: {e}"
+
+
+
+# ----- Shared State -----
+class AgentState(TypedDict):
+    messages: Annotated[list, operator.add]
+
+# ----- Agent Nodes -----
+def agent_node(state: AgentState):
+    """This is the main agent node that processes messages and decides when to call tools."""
+    llm_with_tools = llm.bind_tools([add_tag_to_bookmark, fetch_bookmarks, crawl_homepage, todays_date])
+
+    system_prompt = SystemMessage(f"""
+        You are a research agent with web search capabilities.
+        INSTRUCTIONS:
+        1. **MUST use web_search tool** first to gather information
+        2. Provide comprehensive research based on search results
+
+        Always call **web_search** before responding.
+    """)
+
+    human_prompt = HumanMessage("What are the latest bookmarks added to my Linkding?")
+
+    messages = [system_prompt, human_prompt]
+
+    response = llm_with_tools.invoke(messages)
+
+    if hasattr(response, 'tool_calls') and response.tool_calls:
+        for tc in response.tool_calls:
+            print(f"[AGENT] called Tool {tc.get('name', '?')} with args {tc.get('args', '?')}")
+    else:
+        print(f"[AGENT] Responding...")
+
+    return {'messages': [response]}
+
+
+def should_continue(state: AgentState):
+    last = state['messages'][-1]
+    
+    if hasattr(last, 'tool_calls') and last.tool_calls:
+        return "tools"
+    else:
+        return END
+    
+# =============================================================================
+# Graph
+# =============================================================================
+def create_agent():
+
+    builder = StateGraph(AgentState)
+
+    builder.add_node("agent", agent_node)
+    builder.add_node("tools", ToolNode([add_tag_to_bookmark, fetch_bookmarks, crawl_homepage, todays_date]))
+    builder.set_entry_point("agent")
+    
+    builder.add_conditional_edges("agent", should_continue, ["tools", END])
+
+    builder.add_edge("tools", "agent")
+
+    graph = builder.compile()
+
+    return graph
+
+agent = create_agent()
+result = agent.invoke({})
+
+print(result['messages'])
+print(result['messages'][-1].content)
+
+
+result = fetch_bookmarks.invoke({})
+print(result)
+
+result = crawl_homepage.invoke({'url': "http://example.com"})
+print(result)
+
+result = todays_date.invoke({})
+print(result)
+
+result = add_tag_to_bookmark.invoke({'bookmark_id': 123, 'tag': 'newtag'})
+print(result)
+
+result = add_tag_to_bookmark.invoke({'bookmark_id': 4, 'tag': 'another_tag'})
+print(result)
\ No newline at end of file