academia-mcp 1.4.2__py3-none-any.whl → 1.5.0__py3-none-any.whl

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/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/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/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/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/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/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-1.4.2.dist-info → academia_mcp-1.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: academia-mcp
-Version: 1.4.2
+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-1.4.2.dist-info → academia_mcp-1.5.0.dist-info}/RECORD

@@ -1,27 +1,29 @@
 academia_mcp/__init__.py,sha256=2Ru2I5u4cE7DrkkAsibDUEF1K6sYtqppb9VyFrRoQKI,94
 academia_mcp/__main__.py,sha256=rcmsOtJd3SA82exjrcGBuxuptcoxF8AXI7jNjiVq2BY,59
 academia_mcp/files.py,sha256=tvt3OPr5q6pAPCZ0XvRHHL9ZWuTXINRZvqjeRFmx5YE,815
-academia_mcp/llm.py,sha256=o84FQNSbjjVSk9DlvFXWsUDiz5IOaavYU6kOqnPEG7E,1071
+academia_mcp/llm.py,sha256=jh-_H3_gNyRsvpFqFx-yWVhGznnXxehFP79inUy4vVQ,995
+academia_mcp/pdf.py,sha256=9PlXzHGhb6ay3ldbTdxCcTWvH4TkET3bnb64mgoh9i0,1273
 academia_mcp/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 academia_mcp/server.py,sha256=FRrPAacAPs1IZ3LRKDFZi7copAqzy_aPGYd2RLsA01U,2974
 academia_mcp/utils.py,sha256=P9U3RjYzcztE0KxXvJSy5wSBaUg2CM9tpByljYrsrl4,4607
 academia_mcp/latex_templates/agents4science_2025/agents4science_2025.sty,sha256=hGcEPCYBJS4vdhWvN_yEaJC4GvT_yDroI94CfY2Oguk,12268
-academia_mcp/latex_templates/agents4science_2025/agents4science_2025.tex,sha256=Nu_nL-3WrnIrPZ03qswhKpKUOXm0Dr8gFmUfASkLMBc,33800
+academia_mcp/latex_templates/agents4science_2025/agents4science_2025.tex,sha256=VxuE1Va7_QtZ87P1iUm8-JOW1e_9tdbSSvhH_hz7Ras,33801
 academia_mcp/tools/__init__.py,sha256=u_6HkChV2P46zXxGp92s1cTSjkkd42udhCM3BFDYQ_c,1137
 academia_mcp/tools/anthology_search.py,sha256=rhFpJZqGLABgr0raDuH0CARBiAJNJtEI4dlMrKNHfDQ,7669
-academia_mcp/tools/arxiv_download.py,sha256=soa9nPDHV1-ZgTLtqh-Fj69WkRsrb1hRKouWDa2ePng,11269
+academia_mcp/tools/arxiv_download.py,sha256=gBY0_Kz0yGtVkLMwn6GrAyfBjovZVgcSMuyy67p65Cw,10474
 academia_mcp/tools/arxiv_search.py,sha256=pzM18qrF3QL03A53w003kE7hQi3s3QKtjgw0m7K88UY,8355
-academia_mcp/tools/bitflip.py,sha256=u0hSOPWbnCDu2EbA_RkueX496SvTKz9QhZcXugshSfI,10949
-academia_mcp/tools/document_qa.py,sha256=04pJpiYCg27EFiZhfmTaMjeobu8SMT0Dls7OAFDoH00,2392
+academia_mcp/tools/bitflip.py,sha256=Lu2UASqabiMQ-F-s1BTDWMNuIceys9NIOM9M1bpJczk,11118
+academia_mcp/tools/document_qa.py,sha256=t9mygYQ7AFIAPiha1nZ-y043luQlkTCBdWb_SDnzEsE,2444
 academia_mcp/tools/hf_datasets_search.py,sha256=KiBkqT4rXjEN4oc1AWZOPnqN_Go90TQogY5-DUm3LQo,2854
 academia_mcp/tools/latex.py,sha256=bf8VZUgCByzBAMTZCeqrRrmakotext3d3DbtkiOTh1k,5892
 academia_mcp/tools/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+academia_mcp/tools/review.py,sha256=YR72fl8pdzPur8TfdQeBZHfjpKjGTl-bngPjhYafsRI,1409
 academia_mcp/tools/s2_citations.py,sha256=dqrBp76RrX1zH2XzcMAoWBbvbtyhxLeF-xnqOKD_JiM,4852
 academia_mcp/tools/visit_webpage.py,sha256=OZdqDkVPIbANyFw5o5jIjU5Rr_dolxrGDs63Ud-GmRM,1966
 academia_mcp/tools/web_search.py,sha256=mobKm4iqKppn8pduZYMzWRo1MQBjkAqmMtrFLI5XY2Y,6296
-academia_mcp-1.4.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-academia_mcp-1.4.2.dist-info/METADATA,sha256=aGkaWNjsNm7aB6o-mwEEvekh8ybOOuE5vuuIJW2rkog,3172
-academia_mcp-1.4.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-academia_mcp-1.4.2.dist-info/entry_points.txt,sha256=gxkiKJ74w2FwJpSECpjA3XtCfI5ZfrM6N8cqnwsq4yY,51
-academia_mcp-1.4.2.dist-info/top_level.txt,sha256=CzGpRFsRRJRqWEb1e3SUlcfGqRzOxevZGaJWrtGF8W0,13
-academia_mcp-1.4.2.dist-info/RECORD,,
+academia_mcp-1.5.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+academia_mcp-1.5.0.dist-info/METADATA,sha256=caqX8xdo4NeS5eUtSWkXYr5Y6kvXJISb4HhUsjZvhRo,3233
+academia_mcp-1.5.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+academia_mcp-1.5.0.dist-info/entry_points.txt,sha256=gxkiKJ74w2FwJpSECpjA3XtCfI5ZfrM6N8cqnwsq4yY,51
+academia_mcp-1.5.0.dist-info/top_level.txt,sha256=CzGpRFsRRJRqWEb1e3SUlcfGqRzOxevZGaJWrtGF8W0,13
+academia_mcp-1.5.0.dist-info/RECORD,,