dhisana 0.0.1.dev85__py3-none-any.whl → 0.0.1.dev236__py3-none-any.whl

dhisana/utils/serpapi_search_tools.py

@@ -1,384 +1,150 @@
 import json
-import os
 import re
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Set
 from urllib.parse import urlparse
+import urllib.parse
 import aiohttp
 from bs4 import BeautifulSoup
 import urllib
+from pydantic import BaseModel
 
-from dhisana.utils.assistant_tool_tag import assistant_tool
-from dhisana.utils.cache_output_tools import cache_output, retrieve_output
-from dhisana.utils.web_download_parse_tools import fetch_html_content, get_html_content_from_url
+from dhisana.utils.serperdev_search import search_google_serper
+from dhisana.utils.generate_structured_output_internal import (
+    get_structured_output_internal,
+)
 
 import logging
+
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
+from dhisana.utils.search_router import search_google_with_tools
+from dhisana.utils.assistant_tool_tag import assistant_tool
 
-def get_serp_api_access_token(tool_config: Optional[List[Dict]] = None) -> str:
-    """
-    Retrieves the SERPAPI_KEY access token from the provided tool configuration.
-
-    Args:
-        tool_config (list): A list of dictionaries containing the tool configuration. 
-                            Each dictionary should have a "name" key and a "configuration" key,
-                            where "configuration" is a list of dictionaries containing "name" and "value" keys.
-
-    Returns:
-        str: The SERPAPI_KEY access token.
-
-    Raises:
-        ValueError: If the access token is not found in the tool configuration or environment variable.
-    """
-    logger.info("Entering get_serp_api_access_token")
-    SERPAPI_KEY = None
-
-    if tool_config:
-        logger.debug(f"Tool config provided: {tool_config}")
-        serpapi_config = next(
-            (item for item in tool_config if item.get("name") == "serpapi"), None
-        )
-        if serpapi_config:
-            config_map = {
-                item["name"]: item["value"]
-                for item in serpapi_config.get("configuration", [])
-                if item
-            }
-            SERPAPI_KEY = config_map.get("apiKey")
-        else:
-            logger.warning("No 'serpapi' config item found in tool_config.")
-    else:
-        logger.debug("No tool_config provided or it's None.")
-
-    SERPAPI_KEY = SERPAPI_KEY or os.getenv("SERPAPI_KEY")
-    if not SERPAPI_KEY:
-        logger.error("SERPAPI_KEY not found in configuration or environment.")
-        raise ValueError("SERPAPI_KEY access token not found in tool_config or environment variable")
-
-    logger.info("Retrieved SERPAPI_KEY successfully.")
-    return SERPAPI_KEY
-
-
-@assistant_tool
-async def search_google(
-    query: str,
-    number_of_results: int = 10,
-    offset: int = 0,  
-    tool_config: Optional[List[Dict]] = None,
-    as_oq: Optional[str] = None  # <-- NEW PARAM for optional keywords
-) -> List[str]:
-    """
-    Search Google using SERP API, supporting pagination and an explicit 'offset'
-    parameter to start from a specific result index. 
-    Now also supports 'as_oq' for optional query terms in SERP API.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The total number of results to return. Default is 10.
-    - offset (int): The starting index for the first result returned (Google pagination).
-    - tool_config (Optional[List[Dict]]): Configuration containing SERP API token, etc.
-    - as_oq (Optional[str]): Optional query terms for SerpAPI (if supported).
-    
-    Returns:
-    - List[str]: A list of organic search results, each serialized as a JSON string.
-    """
-    logger.info("Entering search_google")
-    if not query:
-        logger.warning("Empty query string provided.")
-        return []
-
-    # Use 'as_oq' in the cache key too, so different optional terms don't conflict
-    cache_key = f"{query}_{number_of_results}_{offset}_{as_oq or ''}"
-    cached_response = retrieve_output("search_google_serp", cache_key)
-    if cached_response is not None:
-        logger.info("Cache hit for search_google.")
-        return cached_response
-
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    url = "https://serpapi.com/search"
-
-    page_size = 100
-    all_results: List[Dict[str, Any]] = []
-    start_index = offset
-
-    logger.debug(f"Requesting up to {number_of_results} results for '{query}' starting at offset {offset}.")
-
-    async with aiohttp.ClientSession() as session:
-        while len(all_results) < number_of_results:
-            to_fetch = min(page_size, number_of_results - len(all_results))
-            params = {
-                "q": query,
-                "num": to_fetch,
-                "start": start_index,
-                "api_key": SERPAPI_KEY,
-                "location": "United States"
-            }
-
-            # If we have optional terms, add them
-            if as_oq:
-                params["as_oq"] = as_oq
-
-            logger.debug(f"SERP API GET request with params: {params}")
-
-            try:
-                async with session.get(url, params=params) as response:
-                    logger.debug(f"Received response status: {response.status}")
-                    if response.status != 200:
-                        try:
-                            error_content = await response.json()
-                        except Exception:
-                            error_content = await response.text()
-                        logger.warning(f"Non-200 response from SERP API: {error_content}")
-                        return [json.dumps({"error": error_content})]
-
-                    result = await response.json()
-            except Exception as e:
-                logger.exception("Exception during SERP API request.")
-                return [json.dumps({"error": str(e)})]
-
-            organic_results = result.get('organic_results', [])
-            if not organic_results:
-                logger.debug("No more organic results returned; stopping.")
-                break
-
-            all_results.extend(organic_results)
-            start_index += to_fetch
-
-            if len(all_results) >= number_of_results:
-                break
-
-    all_results = all_results[:number_of_results]
-    logger.info(f"Found {len(all_results)} results for query '{query}'.")
-
-    serialized_results = [json.dumps(item) for item in all_results]
-    cache_output("search_google_serp", cache_key, serialized_results)
-    return serialized_results
-
-
-@assistant_tool
-async def search_google_maps(
-    query: str,
-    number_of_results: int = 3,
-    tool_config: Optional[List[Dict]] = None
-) -> List[str]:
-    """
-    Search Google Maps using SERP API and return the results as an array of serialized JSON strings.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The number of results to return.
-    """
-    logger.info("Entering search_google_maps")
-    if not query:
-        logger.warning("Empty query string provided for search_google_maps.")
-        return []
-
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    params = {
-        "q": query,
-        "num": number_of_results,
-        "api_key": SERPAPI_KEY,
-        "engine": "google_maps"
-    }
-    url = "https://serpapi.com/search"
-
-    logger.debug(f"Searching Google Maps with params: {params}")
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(url, params=params) as response:
-                logger.debug(f"Received status: {response.status}")
-                result = await response.json()
-                if response.status != 200:
-                    logger.warning(f"Non-200 response from SERP API: {result}")
-                    return [json.dumps({"error": result})]
-
-                serialized_results = [json.dumps(item) for item in result.get('local_results', [])]
-                logger.info(f"Returning {len(serialized_results)} map results.")
-                return serialized_results
-    except Exception as e:
-        logger.exception("Exception during search_google_maps request.")
-        return [json.dumps({"error": str(e)})]
-
+from dhisana.utils.web_download_parse_tools import fetch_html_content
 
-@assistant_tool
-async def search_google_news(
-    query: str,
-    number_of_results: int = 3,
-    tool_config: Optional[List[Dict]] = None
-) -> List[str]:
-    """
-    Search Google News using SERP API and return the results as an array of serialized JSON strings.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The number of results to return.
-    """
-    logger.info("Entering search_google_news")
-    if not query:
-        logger.warning("Empty query string provided for search_google_news.")
-        return []
 
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    params = {
-        "q": query,
-        "num": number_of_results,
-        "api_key": SERPAPI_KEY,
-        "engine": "google_news"
-    }
-    url = "https://serpapi.com/search"
+class LeadSearchResult(BaseModel):
+    first_name: str = ""
+    last_name: str = ""
+    full_name: str = ""
+    job_title: str = ""
+    linkedin_follower_count: int = 0
+    lead_location: str = ""
+    summary_about_lead: str = ""
+    user_linkedin_url: str = ""
 
-    logger.debug(f"Searching Google News with params: {params}")
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(url, params=params) as response:
-                logger.debug(f"Received status: {response.status}")
-                result = await response.json()
-                if response.status != 200:
-                    logger.warning(f"Non-200 response from SERP API: {result}")
-                    return [json.dumps({"error": result})]
-
-                serialized_results = [json.dumps(item) for item in result.get('news_results', [])]
-                logger.info(f"Returning {len(serialized_results)} news results.")
-                return serialized_results
-    except Exception as e:
-        logger.exception("Exception during search_google_news request.")
-        return [json.dumps({"error": str(e)})]
 
+class LinkedinCandidateChoice(BaseModel):
+    chosen_link: str = ""
+    confidence: float = 0.0
+    reasoning: str = ""
 
-@assistant_tool
-async def search_job_postings(
-    query: str,
-    number_of_results: int,
-    tool_config: Optional[List[Dict]] = None
-) -> List[str]:
-    """
-    Search for job postings using SERP API and return the results as an array of serialized JSON strings.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The number of results to return.
-    """
-    logger.info("Entering search_job_postings")
-    if not query:
-        logger.warning("Empty query string provided for search_job_postings.")
-        return []
 
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    params = {
-        "q": query,
-        "num": number_of_results,
-        "api_key": SERPAPI_KEY,
-        "engine": "google_jobs"
-    }
-    url = "https://serpapi.com/search"
+async def get_structured_output(text: str, tool_config: Optional[List[Dict]] = None) -> LeadSearchResult:
+    """Parse text snippet into ``LeadSearchResult`` using OpenAI."""
 
-    logger.debug(f"Searching Google Jobs with params: {params}")
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(url, params=params) as response:
-                logger.debug(f"Received status: {response.status}")
-                result = await response.json()
-                if response.status != 200:
-                    logger.warning(f"Non-200 response from SERP API: {result}")
-                    return [json.dumps({"error": result})]
-
-                serialized_results = [json.dumps(item) for item in result.get('jobs_results', [])]
-                logger.info(f"Returning {len(serialized_results)} job posting results.")
-                return serialized_results
-    except Exception as e:
-        logger.exception("Exception during search_job_postings request.")
-        return [json.dumps({"error": str(e)})]
+    prompt = (
+        "Extract lead details from the text below.\n"
+        "If follower counts are mentioned, convert values like '1.5k+ followers' to an integer (e.g. 1500).\n"
+        f"Return JSON matching this schema:\n{json.dumps(LeadSearchResult.model_json_schema(), indent=2)}\n\n"
+        f"Text:\n{text}"
+    )
+    result, status = await get_structured_output_internal(
+        prompt, LeadSearchResult, model = "gpt-5.1-chat", tool_config=tool_config
+    )
+    if status != "SUCCESS" or result is None:
+        return LeadSearchResult()
+    return result
 
 
 @assistant_tool
-async def search_google_images(
-    query: str,
-    number_of_results: int,
-    tool_config: Optional[List[Dict]] = None
-) -> List[str]:
-    """
-    Search Google Images using SERP API and return the results as an array of serialized JSON strings.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The number of results to return.
-    """
-    logger.info("Entering search_google_images")
-    if not query:
-        logger.warning("Empty query string provided for search_google_images.")
-        return []
-
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    params = {
-        "q": query,
-        "num": number_of_results,
-        "api_key": SERPAPI_KEY,
-        "engine": "google_images"
-    }
-    url = "https://serpapi.com/search"
-
-    logger.debug(f"Searching Google Images with params: {params}")
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(url, params=params) as response:
-                logger.debug(f"Received status: {response.status}")
-                result = await response.json()
-                if response.status != 200:
-                    logger.warning(f"Non-200 response from SERP API: {result}")
-                    return [json.dumps({"error": result})]
-
-                serialized_results = [json.dumps(item) for item in result.get('images_results', [])]
-                logger.info(f"Returning {len(serialized_results)} image results.")
-                return serialized_results
-    except Exception as e:
-        logger.exception("Exception during search_google_images request.")
-        return [json.dumps({"error": str(e)})]
-
+async def find_user_linkedin_url_with_serper(
+    user_linkedin_url: str,
+    tool_config: Optional[List[Dict]] = None,
+) -> Optional[Dict]:
+    """Search Google via Serper.dev for ``user_linkedin_url`` and parse lead details."""
 
-@assistant_tool
-async def search_google_videos(
-    query: str,
-    number_of_results: int,
-    tool_config: Optional[List[Dict]] = None
-) -> List[str]:
-    """
-    Search Google Videos using SERP API and return the results as an array of serialized JSON strings.
-    
-    Parameters:
-    - query (str): The search query.
-    - number_of_results (int): The number of results to return.
-    """
-    logger.info("Entering search_google_videos")
-    if not query:
-        logger.warning("Empty query string provided for search_google_videos.")
-        return []
+    if not user_linkedin_url:
+        return None
 
-    SERPAPI_KEY = get_serp_api_access_token(tool_config)
-    params = {
-        "q": query,
-        "num": number_of_results,
-        "api_key": SERPAPI_KEY,
-        "engine": "google_videos"
-    }
-    url = "https://serpapi.com/search"
+    normalized_input = extract_user_linkedin_page(user_linkedin_url)
+    results = await search_google_serper(user_linkedin_url, 10, tool_config=tool_config)
+    for item_json in results:
+        try:
+            item = json.loads(item_json)
+        except Exception:
+            continue
+        link = item.get("link", "")
+        if not link:
+            continue
+        if extract_user_linkedin_page(link) == normalized_input:
+            text = " ".join(
+                [item.get("title", ""), item.get("subtitle", ""), item.get("snippet", "")] 
+            ).strip()
+            structured = await get_structured_output(text, tool_config=tool_config)
+            structured.user_linkedin_url = normalized_input
+            return json.loads(structured.model_dump_json())
+    return None
+
+
+async def pick_best_linkedin_candidate_with_llm(
+    email: str,
+    user_name: str,
+    user_title: str,
+    user_location: str,
+    user_company: str,
+    candidates: List[Dict],
+    tool_config: Optional[List[Dict]] = None,
+) -> Optional[LinkedinCandidateChoice]:
+    """Ask the LLM to assess candidate LinkedIn URLs and pick the best match."""
+
+    if not candidates:
+        return None
+
+    candidates_sorted = candidates[-3:]
+    candidate_lines = []
+    for idx, candidate in enumerate(candidates_sorted, start=1):
+        candidate_lines.append(
+            "\n".join(
+                [
+                    f"Candidate {idx}:",
+                    f"  Link: {candidate.get('link', '')}",
+                    f"  Title: {candidate.get('title', '')}",
+                    f"  Snippet: {candidate.get('snippet', '')}",
+                    f"  Subtitle: {candidate.get('subtitle', '')}",
+                    f"  Query: {candidate.get('query', '')}",
+                ]
+            )
+        )
 
-    logger.debug(f"Searching Google Videos with params: {params}")
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(url, params=params) as response:
-                logger.debug(f"Received status: {response.status}")
-                result = await response.json()
-                if response.status != 200:
-                    logger.warning(f"Non-200 response from SERP API: {result}")
-                    return [json.dumps({"error": result})]
-
-                serialized_results = [json.dumps(item) for item in result.get('video_results', [])]
-                logger.info(f"Returning {len(serialized_results)} video results.")
-                return serialized_results
-    except Exception as e:
-        logger.exception("Exception during search_google_videos request.")
-        return [json.dumps({"error": str(e)})]
+    prompt = (
+        "You are validating LinkedIn profile matches for a lead enrichment workflow.\n"
+        "Given the lead context and candidate search results, pick the most likely LinkedIn profile.\n"
+        "If no candidate seems appropriate, return an empty link and confidence 0.\n"
+        "Consider whether the email, name, company, title, or location aligns with the candidate.\n"
+        "Lead context:\n"
+        f"- Email: {email or 'unknown'}\n"
+        f"- Name: {user_name or 'unknown'}\n"
+        f"- Title: {user_title or 'unknown'}\n"
+        f"- Company: {user_company or 'unknown'}\n"
+        f"- Location: {user_location or 'unknown'}\n\n"
+        "Candidates:\n"
+        f"{chr(10).join(candidate_lines)}\n\n"
+        "Return JSON with fields: chosen_link (string), confidence (0-1 float), reasoning (short string)."
+    )
+
+    result, status = await get_structured_output_internal(
+        prompt,
+        LinkedinCandidateChoice,
+        model="gpt-5.1-chat",
+        tool_config=tool_config,
+    )
+
+    if status != "SUCCESS" or result is None:
+        return None
+
+    return result
 
 
 @assistant_tool
@@ -388,14 +154,7 @@ async def get_company_domain_from_google_search(
     tool_config: Optional[List[Dict]] = None
 ) -> str:
     """
-    Tries to find the company domain from the company name using Google search.
-
-    Args:
-        company_name (str): The name of the company to search for.
-        location (str, optional): A location to include in the query.
-
-    Returns:
-        str: The domain of the company's official website if found, otherwise an empty string.
+    Tries to find the company domain from the company name using Google (SerpAPI or Serper.dev).
     """
     logger.info("Entering get_company_domain_from_google_search")
 
@@ -404,22 +163,21 @@ async def get_company_domain_from_google_search(
         logger.debug("Invalid or excluded company_name provided.")
         return ""
 
-    exclude_company_names = ["linkedin", "wikipedia", "facebook", "instagram", "twitter", "youtube", "netflix", "zoominfo", "reditt"]
     query = f"\"{company_name}\" official website"
     if location:
         query = f"\"{company_name}\" official website, {location}"
 
     try:
         logger.debug(f"Performing search with query: {query}")
-        result = await search_google(query, 1, tool_config=tool_config)
+        result = await search_google_with_tools(query, 1, tool_config=tool_config)
         if not isinstance(result, list) or len(result) == 0:
             logger.debug("No results for first attempt, retrying with fallback query.")
             query = f"{company_name} official website"
-            result = await search_google(query, 1, tool_config=tool_config)
+            result = await search_google_with_tools(query, 1, tool_config=tool_config)
             if not isinstance(result, list) or len(result) == 0:
                 logger.debug("No results from fallback query either.")
                 return ''
-    except Exception as e:
+    except Exception:
         logger.exception("Exception during get_company_domain_from_google_search.")
         return ''
 
@@ -471,16 +229,6 @@ async def get_signal_strength(
     """
     Find how strong a match for the keywords in search is by checking
     how many search results contain all desired keywords in the snippet.
-
-    Args:
-        domain_to_search (str): The domain to search inside.
-        keywords (List[str]): The keywords to search for.
-        in_title (List[str]): Keywords that must appear in the title.
-        not_in_title (List[str]): Keywords that must not appear in the title.
-        negative_keywords (List[str]): Keywords to exclude from results.
-
-    Returns:
-        int: A strength score on a scale of 0 to 5.
     """
     logger.info("Entering get_signal_strength")
 
@@ -507,8 +255,8 @@ async def get_signal_strength(
 
     logger.debug(f"Performing get_signal_strength search with query: {final_query}")
     try:
-        results = await search_google(final_query, 5, tool_config=tool_config)
-    except Exception as e:
+        results = await search_google_with_tools(final_query, 5, tool_config=tool_config)
+    except Exception:
         logger.exception("Exception occurred while searching for signal strength.")
         return 0
 
@@ -517,9 +265,9 @@ async def get_signal_strength(
         return 0
 
     score = 0
-    for result in results:
+    for result_item in results:
         try:
-            result_json = json.loads(result)
+            result_json = json.loads(result_item)
             snippet_text = result_json.get('snippet', '').lower()
             if all(kw.lower() in snippet_text for kw in keywords):
                 logger.debug(f"Found match in snippet: {snippet_text[:60]}...")
@@ -543,8 +291,8 @@ def extract_user_linkedin_page(url: str) -> str:
     if not url:
         return ""
 
-    normalized_url = re.sub(r"(https?://)?([\w\-]+\.)?linkedin\.com", "https://www.linkedin.com", url)
-    match = re.match(r"https://www.linkedin.com/in/([\w\-]+)", normalized_url)
+    normalized_url = re.sub(r"^(https?://)?([\w\-]+\.)?linkedin\.com", "https://www.linkedin.com", url)
+    match = re.match(r"https://www\.linkedin\.com/in/([^/?#]+)", normalized_url)
     if match:
         page = f"https://www.linkedin.com/in/{match.group(1)}"
         logger.debug(f"Extracted user LinkedIn page: {page}")
@@ -560,21 +308,12 @@ async def find_user_linkedin_url_google(
     user_title: str,
     user_location: str,
     user_company: str,
+    user_company_domain: str = "",
     use_strict_check: bool = True,
     tool_config: Optional[List[Dict]] = None
 ) -> str:
     """
     Find the LinkedIn URL for a user based on their name, title, location, and company.
-
-    Args:
-        user_name (str): The name of the user.
-        user_title (str): The title of the user.
-        user_location (str): The location of the user.
-        user_company (str): The company of the user.
-        use_strict_check (bool): Whether to use a strict single query or a series of relaxed queries.
-
-    Returns:
-        str: The LinkedIn URL if found, otherwise an empty string.
     """
     logger.info("Entering find_user_linkedin_url_google")
 
@@ -584,7 +323,7 @@ async def find_user_linkedin_url_google(
 
     if use_strict_check:
         queries = [
-            f'site:linkedin.com/in "{user_name}" "{user_location}" "{user_title}" "{user_company}" intitle:"{user_name}" -intitle:"profiles" '
+            f'site:linkedin.com/in ("{user_name}")  ({user_company} | {user_company_domain}) ( {user_title} | ) intitle:"{user_name}" -intitle:"profiles" '
         ]
     else:
         queries = [
@@ -594,14 +333,14 @@ async def find_user_linkedin_url_google(
             f'site:linkedin.com/in "{user_name}" intitle:"{user_name}"'
         ]
 
-    async with aiohttp.ClientSession() as session:  # Not strictly necessary here, but kept for parallel structure
+    async with aiohttp.ClientSession() as session:
         for query in queries:
             if not query.strip():
                 continue
             logger.debug(f"Searching with query: {query}")
             try:
-                results = await search_google(query.strip(), 1, tool_config=tool_config)
-            except Exception as e:
+                results = await search_google_with_tools(query.strip(), 1, tool_config=tool_config)
+            except Exception:
                 logger.exception("Error searching for LinkedIn user URL.")
                 continue
 
@@ -630,6 +369,221 @@ async def find_user_linkedin_url_google(
     return ""
 
 
+@assistant_tool
+async def find_user_linkedin_url_by_email_google(
+    email: str,
+    user_name: str = "",
+    user_title: str = "",
+    user_location: str = "",
+    user_company: str = "",
+    tool_config: Optional[List[Dict]] = None,
+) -> Optional[Dict[str, Any]]:
+    """
+    Find the LinkedIn URL for a user based primarily on their email address.
+
+    Additional profile hints (name, title, location, company) improve query precision
+    when supplied. Returns a dict with the best LinkedIn URL, LLM confidence score,
+    and short reasoning when a match clears the confidence threshold; otherwise ``None``.
+    """
+    logger.info("Entering find_user_linkedin_url_by_email_google")
+
+    if not email:
+        logger.warning("No email provided.")
+        return None
+
+    normalized_email = email.strip().lower()
+    email_local_part = normalized_email.split("@")[0] if "@" in normalized_email else normalized_email
+    email_local_humanized = re.sub(r"[._-]+", " ", email_local_part).strip()
+
+    queries: List[str] = []
+
+    def add_query(query: str) -> None:
+        query = query.strip()
+        if query and query not in queries:
+            queries.append(query)
+
+    def add_query_parts(*parts: str) -> None:
+        tokens = [part.strip() for part in parts if part and part.strip()]
+        if not tokens:
+            return
+        add_query(" ".join(tokens))
+
+    enriched_terms = []
+    if user_name:
+        enriched_terms.append(f'"{user_name}"')
+    if user_company:
+        enriched_terms.append(f'"{user_company}"')
+    if user_title:
+        enriched_terms.append(f'"{user_title}"')
+    if user_location:
+        enriched_terms.append(f'"{user_location}"')
+    base_hint = " ".join(enriched_terms)
+
+    # Prioritise the direct email search variants before broader fallbacks.
+    add_query_parts(normalized_email, "linkedin.com/in", base_hint)
+    add_query_parts(normalized_email, "linkedin.com", base_hint)
+    add_query_parts(normalized_email, "linkedin", base_hint)
+    add_query_parts(normalized_email, base_hint)
+    add_query(f'"{normalized_email}" "linkedin.com/in" {base_hint}')
+    add_query(f'"{normalized_email}" "linkedin.com" {base_hint}')
+    add_query(f'"{normalized_email}" linkedin {base_hint}')
+
+    if email_local_part and email_local_part != normalized_email:
+        add_query_parts(email_local_part, "linkedin.com/in", base_hint)
+        add_query_parts(email_local_part, "linkedin.com", base_hint)
+        add_query_parts(email_local_part, "linkedin", base_hint)
+        add_query(f'"{email_local_part}" "linkedin.com/in" {base_hint}')
+        add_query(f'"{email_local_part}" "linkedin.com" {base_hint}')
+
+    if email_local_humanized and email_local_humanized not in {email_local_part, normalized_email}:
+        add_query_parts(email_local_humanized, "linkedin", base_hint)
+        add_query(f'"{email_local_humanized}" linkedin {base_hint}')
+
+    if normalized_email:
+        add_query(f'site:linkedin.com/in "{normalized_email}" {base_hint}')
+
+    if email_local_part:
+        add_query(f'site:linkedin.com/in "{email_local_part}" {base_hint}')
+
+    if email_local_humanized and email_local_humanized != email_local_part:
+        add_query(f'site:linkedin.com/in "{email_local_humanized}" {base_hint}')
+
+    if base_hint:
+        lookup_hint = user_name or email_local_humanized or email_local_part or normalized_email
+        add_query(
+            f'site:linkedin.com/in "{normalized_email}" {base_hint} '
+            f'intitle:"{lookup_hint}" -intitle:"profiles"'
+        )
+        if email_local_humanized:
+            add_query(
+                f'site:linkedin.com/in "{email_local_humanized}" {base_hint} '
+                f'intitle:"{lookup_hint}" -intitle:"profiles"'
+            )
+
+    candidate_records: List[Dict[str, str]] = []
+    seen_links: Set[str] = set()
+    best_llm_choice: Optional[LinkedinCandidateChoice] = None
+    best_llm_link: str = ""
+    HIGH_CONFIDENCE_THRESHOLD = 0.8
+    MIN_CONFIDENCE_THRESHOLD = 0.75
+
+    async def evaluate_with_llm() -> Optional[LinkedinCandidateChoice]:
+        nonlocal best_llm_choice, best_llm_link
+
+        llm_choice = await pick_best_linkedin_candidate_with_llm(
+            email=email,
+            user_name=user_name,
+            user_title=user_title,
+            user_location=user_location,
+            user_company=user_company,
+            candidates=candidate_records,
+            tool_config=tool_config,
+        )
+
+        if not llm_choice or not llm_choice.chosen_link:
+            return None
+
+        chosen_link = extract_user_linkedin_page(llm_choice.chosen_link)
+        if not chosen_link:
+            return None
+
+        llm_choice.chosen_link = chosen_link
+
+        if best_llm_choice is None or llm_choice.confidence > best_llm_choice.confidence:
+            best_llm_choice = llm_choice
+            best_llm_link = chosen_link
+            logger.debug(
+                "LLM updated best candidate: %s (confidence %.2f) reason: %s",
+                chosen_link,
+                llm_choice.confidence,
+                llm_choice.reasoning,
+            )
+
+        if llm_choice.confidence >= HIGH_CONFIDENCE_THRESHOLD:
+            logger.info(
+                "Returning LinkedIn user page by email via LLM scoring: %s (confidence %.2f)",
+                chosen_link,
+                llm_choice.confidence,
+            )
+            return llm_choice
+
+        return None
+
+    async with aiohttp.ClientSession() as session:
+        for query in queries:
+            query = query.strip()
+            if not query:
+                continue
+            logger.debug(f"Searching with query: {query}")
+
+            try:
+                results = await search_google_with_tools(query, 5, tool_config=tool_config)
+            except Exception:
+                logger.exception("Error searching for LinkedIn user URL by email.")
+                continue
+
+            if not isinstance(results, list) or len(results) == 0:
+                logger.debug("No results for this query, moving to next.")
+                continue
+
+            for result_item in results:
+                try:
+                    result_json = json.loads(result_item)
+                except (json.JSONDecodeError, IndexError):
+                    logger.debug("Failed to parse JSON from the search result.")
+                    continue
+
+                link = result_json.get('link', '')
+                if not link:
+                    continue
+
+                parsed_url = urlparse(link)
+                if 'linkedin.com/in' in (parsed_url.netloc + parsed_url.path):
+                    link = extract_user_linkedin_page(link)
+                    if not link or link in seen_links:
+                        continue
+
+                    title = result_json.get('title', '')
+                    snippet = result_json.get('snippet', '')
+                    subtitle = result_json.get('subtitle', '')
+
+                    candidate_records.append(
+                        {
+                            "link": link,
+                            "title": title,
+                            "snippet": snippet,
+                            "subtitle": subtitle,
+                            "query": query,
+                        }
+                    )
+                    if len(candidate_records) > 6:
+                        candidate_records.pop(0)
+                    seen_links.add(link)
+
+                    high_conf_choice = await evaluate_with_llm()
+                    if high_conf_choice:
+                        return {
+                            "linkedin_url": high_conf_choice.chosen_link,
+                            "confidence": high_conf_choice.confidence,
+                            "reasoning": high_conf_choice.reasoning,
+                        }
+
+    if best_llm_choice and best_llm_link and best_llm_choice.confidence >= MIN_CONFIDENCE_THRESHOLD:
+        logger.info(
+            "Returning LinkedIn user page by email via LLM scoring (best overall): %s (confidence %.2f)",
+            best_llm_link,
+            best_llm_choice.confidence,
+        )
+        return {
+            "linkedin_url": best_llm_link,
+            "confidence": best_llm_choice.confidence,
+            "reasoning": best_llm_choice.reasoning,
+        }
+
+    logger.info("No matching LinkedIn user page found using email queries.")
+    return None
+
+
 @assistant_tool
 async def find_user_linkedin_url_by_job_title_google(
     user_title: str,
@@ -639,14 +593,6 @@ async def find_user_linkedin_url_by_job_title_google(
 ) -> str:
     """
     Find the LinkedIn URL for a user based on their job_title, location, and company.
-
-    Args:
-        user_title (str): The title of the user.
-        user_location (str): The location of the user.
-        user_company (str): The company of the user.
-
-    Returns:
-        str: The LinkedIn URL if found, otherwise an empty string.
     """
     logger.info("Entering find_user_linkedin_url_by_job_title_google")
 
@@ -654,15 +600,15 @@ async def find_user_linkedin_url_by_job_title_google(
         f'site:linkedin.com/in "{user_company}" AND "{user_title}" -intitle:"profiles" ',
     ]
 
-    async with aiohttp.ClientSession() as session:  
+    async with aiohttp.ClientSession() as session:
         for query in queries:
             if not query.strip():
                 continue
             logger.debug(f"Searching with query: {query}")
 
             try:
-                results = await search_google(query.strip(), 1, tool_config=tool_config)
-            except Exception as e:
+                results = await search_google_with_tools(query.strip(), 1, tool_config=tool_config)
+            except Exception:
                 logger.exception("Error searching for LinkedIn URL by job title.")
                 continue
 
@@ -699,14 +645,6 @@ async def find_user_linkedin_url_by_google_search(
 ) -> List[str]:
     """
     Find LinkedIn user URLs based on provided Google search queries.
-    
-    Args:
-        queries (List[str]): A list of Google search queries.
-        number_of_results (int): Number of results to return from each query (default is 5).
-        tool_config (Optional[List[Dict]]): Optional configuration for the SERP API.
-
-    Returns:
-        List[str]: A list of matching LinkedIn user URLs found, or an empty list if none.
     """
     logger.info("Entering find_user_linkedin_url_by_google_search")
     found_urls = []
@@ -717,8 +655,8 @@ async def find_user_linkedin_url_by_google_search(
         logger.debug(f"Searching with query: {query}")
 
         try:
-            results = await search_google(query.strip(), number_of_results, tool_config=tool_config)
-        except Exception as e:
+            results = await search_google_with_tools(query.strip(), number_of_results, tool_config=tool_config)
+        except Exception:
             logger.exception("Error searching for LinkedIn URL using Google search.")
             continue
 
@@ -772,19 +710,12 @@ def extract_company_page(url: str) -> str:
 async def find_organization_linkedin_url_with_google_search(
     company_name: str,
     company_location: Optional[str] = None,
+    company_domain: Optional[str] = None,
     use_strict_check: bool = True,
     tool_config: Optional[List[Dict]] = None,
 ) -> str:
     """
     Find the LinkedIn URL for a company based on its name and optional location using Google search.
-
-    Args:
-        company_name (str): The name of the company.
-        company_location (str, optional): The location of the company.
-        use_strict_check (bool): Whether to use stricter or multiple queries.
-
-    Returns:
-        str: The LinkedIn URL if found, otherwise an empty string.
     """
     logger.info("Entering find_organization_linkedin_url_with_google_search")
 
@@ -793,7 +724,7 @@ async def find_organization_linkedin_url_with_google_search(
         return ""
 
     if use_strict_check:
-        queries = [f'site:linkedin.com/company "{company_name}" {company_location} -intitle:"jobs" ']
+        queries = [f'site:linkedin.com/company "{company_name}" {company_domain} ']
     else:
         if company_location:
             queries = [
@@ -814,8 +745,8 @@ async def find_organization_linkedin_url_with_google_search(
 
             logger.debug(f"Searching with query: {query}")
             try:
-                results = await search_google(query.strip(), 1, tool_config=tool_config)
-            except Exception as e:
+                results = await search_google_with_tools(query.strip(), 1, tool_config=tool_config)
+            except Exception:
                 logger.exception("Error searching for organization LinkedIn URL.")
                 continue
 
@@ -868,7 +799,7 @@ async def get_external_links(url: str) -> List[str]:
                 else:
                     logger.warning(f"Non-200 status ({response.status}) while fetching external links.")
                     return []
-    except Exception as e:
+    except Exception:
         logger.exception("Exception occurred while fetching external links.")
         return []
 
@@ -880,7 +811,7 @@ async def get_resolved_linkedin_links(url: str) -> List[str]:
     logger.debug(f"Entering get_resolved_linkedin_links for URL: {url}")
     try:
         content = await fetch_html_content(url)
-    except Exception as e:
+    except Exception:
         logger.exception("Exception occurred while fetching HTML content.")
         return []
 
@@ -904,7 +835,7 @@ async def get_company_website_from_linkedin_url(linkedin_url: str) -> str:
 
     try:
         links = await get_external_links(linkedin_url)
-    except Exception as e:
+    except Exception:
         logger.exception("Exception occurred while getting external links for LinkedIn URL.")
         return ""