Paper Search MCP

Instructions

Implementation Reference

• paper_search_mcp/server.py:422-432 (handler)

  The MCP tool handler `search_google_scholar` which calls the `async_search` helper.

  async def search_google_scholar(query: str, max_results: int = 10) -> List[Dict]:
      """Search academic papers from Google Scholar.
  
      Args:
          query: Search query string (e.g., 'machine learning').
          max_results: Maximum number of papers to return (default: 10).
      Returns:
          List of paper metadata in dictionary format.
      """
      papers = await async_search(google_scholar_searcher, query, max_results)
      return papers if papers else []

• paper_search_mcp/academic_platforms/google_scholar.py:112-150 (handler)

  The implementation of the search logic for Google Scholar within the `GoogleScholarSearcher` class.

  def search(self, query: str, max_results: int = 10) -> List[Paper]:
      """
      Search Google Scholar with custom parameters
      """
      papers = []
      start = 0
      results_per_page = min(10, max_results)
  
      while len(papers) < max_results:
          try:
              # Construct search parameters
              params = {
                  'q': query,
                  'start': start,
                  'hl': 'en',
                  'as_sdt': '0,5'  # Include articles and citations
              }
  
              response = None
              for attempt in range(self.max_retries):
                  self._rotate_user_agent()
                  time.sleep(random.uniform(1.0, 2.5))
  
                  response = self.session.get(self.SCHOLAR_URL, params=params, timeout=30)
                  if response.status_code == 200:
                      break
  
                  if response.status_code in (403, 429, 503):
                      wait_time = self.retry_delay * (2 ** attempt)
                      wait_time += random.uniform(0, 0.5)
                      logger.warning(
                          "Google Scholar returned %s (attempt %s/%s). Backing off %.1fs",
                          response.status_code,
                          attempt + 1,
                          self.max_retries,
                          wait_time,
                      )
                      time.sleep(wait_time)
                      continue