academia-mcp 1.4.1__tar.gz → 1.5.0__tar.gz

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: academia-mcp
-Version: 1.4.1
+Version: 1.5.0
 Summary: MCP server that provides different tools to search for scientific publications
 Author-email: Ilya Gusev <phoenixilya@gmail.com>
 Project-URL: Homepage, https://github.com/IlyaGusev/academia_mcp
@@ -27,6 +27,8 @@ Requires-Dist: fire>=0.7.0
 Requires-Dist: openai>=1.97.1
 Requires-Dist: jinja2>=3.1.6
 Requires-Dist: datasets>=4.0.0
+Requires-Dist: pymupdf>=1.26.4
+Requires-Dist: pillow>=11.3.0
 Dynamic: license-file
 
 # Academia MCP

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/latex_templates/agents4science_2025/agents4science_2025.tex

@@ -5,11 +5,11 @@
 % before loading agents4science_2025
 
 % ready for submission
-\usepackage{agents4science_2025}
+% \usepackage{agents4science_2025}
 
 % to compile a preprint version, e.g., for submission to arXiv, add the
 % [preprint] option:
-% \usepackage[preprint]{agents4science_2025}
+\usepackage[preprint]{agents4science_2025}
 
 % to compile a camera-ready version, add the [final] option, e.g.:
 %     \usepackage[final]{agents4science_2025}
@@ -683,4 +683,4 @@ IMPORTANT, please:
 \end{enumerate}
 
 
-\end{document}
+\end{document}

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/llm.py

@@ -14,14 +14,11 @@ class ChatMessage(BaseModel):  # type: ignore
 ChatMessages = List[ChatMessage]
 
 
-async def llm_acall(model_name: str, prompt: str) -> str:
+async def llm_acall(model_name: str, messages: ChatMessages) -> str:
     key = os.getenv("OPENROUTER_API_KEY", "")
     assert key, "Please set OPENROUTER_API_KEY in the environment variables"
     base_url = os.getenv("BASE_URL", "https://openrouter.ai/api/v1")
 
-    messages: ChatMessages = [
-        ChatMessage(role="user", content=prompt),
-    ]
     client = AsyncOpenAI(base_url=base_url, api_key=key)
     response: ChatCompletionMessage = (
         (

academia_mcp-1.5.0/academia_mcp/pdf.py

@@ -0,0 +1,44 @@
+from pathlib import Path
+from typing import List
+
+from pypdf import PdfReader
+from PIL import Image
+import pymupdf  # type: ignore
+
+from academia_mcp.utils import get_with_retries
+
+
+def download_pdf(url: str, output_path: Path) -> None:
+    response = get_with_retries(url)
+    response.raise_for_status()
+    content_type = response.headers.get("content-type")
+    assert content_type
+    assert "application/pdf" in content_type.lower()
+    with open(output_path.resolve(), "wb") as fp:
+        fp.write(response.content)
+
+
+def parse_pdf_file(pdf_path: Path) -> List[str]:
+    # Why not Marker? Because it is too heavy.
+    reader = PdfReader(str(pdf_path.resolve()))
+
+    pages = []
+    for page_number, page in enumerate(reader.pages, start=1):
+        try:
+            text = page.extract_text()
+            if not text:
+                continue
+            prefix = f"## Page {page_number}\n\n"
+            pages.append(prefix + text)
+        except Exception:
+            continue
+    return pages
+
+
+def parse_pdf_file_to_images(pdf_path: Path) -> List[Image.Image]:
+    doc = pymupdf.open(str(pdf_path.resolve()))
+    images = []
+    for page in doc:
+        pil_image: Image.Image = page.get_pixmap().pil_image()
+        images.append(pil_image)
+    return images

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/tools/arxiv_download.py

@@ -12,36 +12,9 @@ from dataclasses import dataclass, field
 import requests
 import bs4
 from markdownify import MarkdownConverter  # type: ignore
-from pypdf import PdfReader
 
 from academia_mcp.utils import get_with_retries
-
-
-def download_pdf(url: str, output_path: Path) -> None:
-    response = get_with_retries(url)
-    response.raise_for_status()
-    content_type = response.headers.get("content-type")
-    assert content_type
-    assert "application/pdf" in content_type.lower()
-    with open(output_path.resolve(), "wb") as fp:
-        fp.write(response.content)
-
-
-def parse_pdf_file(pdf_path: Path) -> List[str]:
-    # Why not Marker? Because it is too heavy.
-    reader = PdfReader(str(pdf_path.resolve()))
-
-    pages = []
-    for page_number, page in enumerate(reader.pages, start=1):
-        try:
-            text = page.extract_text()
-            if not text:
-                continue
-            prefix = f"## Page {page_number}\n\n"
-            pages.append(prefix + text)
-        except Exception:
-            continue
-    return pages
+from academia_mcp.pdf import parse_pdf_file, download_pdf
 
 
 HTML_URL = "https://arxiv.org/html/{paper_id}"

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/tools/bitflip.py

@@ -11,7 +11,7 @@ from datasets import load_dataset  # type: ignore
 
 from academia_mcp.tools.arxiv_download import arxiv_download
 from academia_mcp.utils import extract_json, encode_prompt
-from academia_mcp.llm import llm_acall
+from academia_mcp.llm import llm_acall, ChatMessage
 
 
 class ProposalDataset:
@@ -208,7 +208,9 @@ async def extract_bitflip_info(arxiv_id: str) -> str:
     paper = arxiv_download(arxiv_id)
     abstract = json.loads(paper)["abstract"]
     prompt = encode_prompt(EXTRACT_PROMPT, abstract=abstract)
-    content = await llm_acall(model_name=model_name, prompt=prompt)
+    content = await llm_acall(
+        model_name=model_name, messages=[ChatMessage(role="user", content=prompt)]
+    )
     result = extract_json(content)
     bitflip_info: BitFlipInfo = BitFlipInfo.model_validate(result)
     return str(bitflip_info.model_dump_json())
@@ -240,7 +242,9 @@ async def generate_research_proposal(bit: str, additional_context: str = "") ->
     prompt = encode_prompt(
         IMPROVEMENT_PROMPT, bit=bit, examples=examples, additional_context=additional_context
     )
-    content = await llm_acall(model_name=model_name, prompt=prompt)
+    content = await llm_acall(
+        model_name=model_name, messages=[ChatMessage(role="user", content=prompt)]
+    )
     result = extract_json(content)
     proposal: Proposal = Proposal.model_validate(result)
     proposal.proposal_id = random.randint(0, 1000000)
@@ -276,7 +280,9 @@ async def score_research_proposals(proposals: List[str]) -> str:
     model_name = os.getenv("BITFLIP_MODEL_NAME", "deepseek/deepseek-chat-v3-0324")
     proposals = [Proposal.model_validate_json(proposal) for proposal in proposals]
     prompt = encode_prompt(SCORE_PROMPT, proposals=proposals)
-    content = await llm_acall(model_name=model_name, prompt=prompt)
+    content = await llm_acall(
+        model_name=model_name, messages=[ChatMessage(role="user", content=prompt)]
+    )
     scores = extract_json(content)
     final_scores = [ProposalScores.model_validate(score) for score in scores]
     return json.dumps([s.model_dump() for s in final_scores], ensure_ascii=False)

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/tools/document_qa.py

@@ -70,5 +70,7 @@ async def document_qa(
 
     model_name = os.getenv("DOCUMENT_QA_MODEL_NAME", "deepseek/deepseek-chat-v3-0324")
     prompt = PROMPT.format(question=question, document=document)
-    content = await llm_acall(model_name=model_name, prompt=prompt)
+    content = await llm_acall(
+        model_name=model_name, messages=[ChatMessage(role="user", content=prompt)]
+    )
     return content.strip()

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp/tools/latex.py

@@ -58,7 +58,11 @@ def compile_latex_from_file(
         output_filename: The path to the output pdf file.
         timeout: The timeout for the compilation. 60 seconds by default.
     """
-    with open(input_filename, "r", encoding="utf-8") as file:
+    input_filename_path = Path(input_filename)
+    if not input_filename_path.exists():
+        input_filename_path = Path(get_workspace_dir()) / input_filename
+    assert input_filename_path.exists(), f"Input file {input_filename} does not exist"
+    with open(input_filename_path, "r", encoding="utf-8") as file:
         latex_code = file.read()
     return compile_latex_from_str(latex_code, output_filename, timeout)
 

academia_mcp-1.5.0/academia_mcp/tools/review.py

@@ -0,0 +1,51 @@
+import base64
+from io import BytesIO
+from pathlib import Path
+
+from academia_mcp.pdf import parse_pdf_file_to_images
+from academia_mcp.llm import llm_acall, ChatMessage
+from academia_mcp.files import get_workspace_dir
+
+
+PROMPT = """
+Find problems with the paper formatiing.
+"""
+
+
+async def review_pdf(pdf_filename: str) -> str:
+    """
+    Review a pdf file.
+
+    Args:
+        pdf_path: The path to the pdf file.
+    """
+    pdf_filename_path = Path(pdf_filename)
+    if not pdf_filename_path.exists():
+        pdf_filename_path = Path(get_workspace_dir()) / pdf_filename
+
+    images = parse_pdf_file_to_images(pdf_filename_path)
+    content_parts = []
+    for image in images:
+        buffer_io = BytesIO()
+        image.save(buffer_io, format="PNG")
+        img_bytes = buffer_io.getvalue()
+        image_base64 = base64.b64encode(img_bytes).decode("utf-8")
+        image_content = {
+            "type": "image_url",
+            "image_url": {"url": f"data:image/png;base64,{image_base64}"},
+        }
+        content_parts.append(image_content)
+
+    content_parts.append(
+        {
+            "type": "text",
+            "text": "Please review the paper and provide a summary of its content.",
+        }
+    )
+    llm_response = await llm_acall(
+        model_name="gpt-4o",
+        messages=[
+            ChatMessage(role="user", content=content_parts),
+        ],
+    )
+    return llm_response.strip()

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: academia-mcp
-Version: 1.4.1
+Version: 1.5.0
 Summary: MCP server that provides different tools to search for scientific publications
 Author-email: Ilya Gusev <phoenixilya@gmail.com>
 Project-URL: Homepage, https://github.com/IlyaGusev/academia_mcp
@@ -27,6 +27,8 @@ Requires-Dist: fire>=0.7.0
 Requires-Dist: openai>=1.97.1
 Requires-Dist: jinja2>=3.1.6
 Requires-Dist: datasets>=4.0.0
+Requires-Dist: pymupdf>=1.26.4
+Requires-Dist: pillow>=11.3.0
 Dynamic: license-file
 
 # Academia MCP

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp.egg-info/SOURCES.txt

@@ -5,6 +5,7 @@ academia_mcp/__init__.py
 academia_mcp/__main__.py
 academia_mcp/files.py
 academia_mcp/llm.py
+academia_mcp/pdf.py
 academia_mcp/py.typed
 academia_mcp/server.py
 academia_mcp/utils.py
@@ -25,6 +26,7 @@ academia_mcp/tools/document_qa.py
 academia_mcp/tools/hf_datasets_search.py
 academia_mcp/tools/latex.py
 academia_mcp/tools/py.typed
+academia_mcp/tools/review.py
 academia_mcp/tools/s2_citations.py
 academia_mcp/tools/visit_webpage.py
 academia_mcp/tools/web_search.py
@@ -36,6 +38,7 @@ tests/test_document_qa.py
 tests/test_extract_json.py
 tests/test_hf_dataset_search.py
 tests/test_latex.py
+tests/test_review.py
 tests/test_s2_citations.py
 tests/test_visit_webpage.py
 tests/test_web_search.py

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/academia_mcp.egg-info/requires.txt

@@ -15,3 +15,5 @@ fire>=0.7.0
 openai>=1.97.1
 jinja2>=3.1.6
 datasets>=4.0.0
+pymupdf>=1.26.4
+pillow>=11.3.0

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "academia-mcp"
-version = "1.4.1"
+version = "1.5.0"
 description = "MCP server that provides different tools to search for scientific publications"
 readme = "README.md"
 authors = [
@@ -34,6 +34,8 @@ dependencies = [
     "openai>=1.97.1",
     "jinja2>=3.1.6",
     "datasets>=4.0.0",
+    "pymupdf>=1.26.4",
+    "pillow>=11.3.0",
 ]
 
 [dependency-groups]
@@ -71,3 +73,9 @@ follow_imports = "skip"
 [tool.pytest.ini_options]
 asyncio_mode = "auto"
 asyncio_default_test_loop_scope = "function"
+filterwarnings = [
+    "ignore:builtin type SwigPyPacked has no __module__ attribute:DeprecationWarning",
+    "ignore:builtin type SwigPyObject has no __module__ attribute:DeprecationWarning",
+    "ignore:builtin type swigvarlink has no __module__ attribute:DeprecationWarning",
+    "ignore:The 'warn' method is deprecated, use 'warning' instead:DeprecationWarning:acl_anthology.text.texmath",
+]

{academia_mcp-1.4.1 → academia_mcp-1.5.0}/tests/test_latex.py

@@ -36,6 +36,7 @@ def test_latex_compile_latex_from_file() -> None:
         temp_dir_path = Path(temp_dir)
         tex_filename = "temp.tex"
         tex_file_path = temp_dir_path / tex_filename
+        pdf_filename = "test.pdf"
         tex_file_path.write_text(template["template"], encoding="utf-8")
-        result = compile_latex_from_file(str(tex_file_path), "test.pdf")
+        result = compile_latex_from_file(str(tex_file_path), pdf_filename)
     assert "Compilation successful" in result

academia_mcp-1.5.0/tests/test_review.py

@@ -0,0 +1,20 @@
+import json
+import tempfile
+from pathlib import Path
+
+from academia_mcp.tools.review import review_pdf
+from academia_mcp.tools.latex import compile_latex_from_file, get_latex_template
+
+
+async def test_review_pdf() -> None:
+    template = json.loads(get_latex_template("agents4science_2025"))
+    with tempfile.TemporaryDirectory() as temp_dir:
+        temp_dir_path = Path(temp_dir)
+        tex_filename = "temp.tex"
+        tex_file_path = temp_dir_path / tex_filename
+        pdf_filename = "test.pdf"
+        tex_file_path.write_text(template["template"], encoding="utf-8")
+        result = compile_latex_from_file(str(tex_file_path), pdf_filename)
+        assert "Compilation successful" in result
+        review = await review_pdf(str(pdf_filename))
+        print(review)