learn arxiv tex files

* Created a new option remote or -r. Example: /learn -r arxiv <arxiv-id>
* Approach: downloads the tar file for the entire paper into downloads_temp. Then, unzips and collects all .tex files in the tar file and concatenates them. Different authors use various approaches. Some have the entire paper in one tex file, whereas others may have separate tex files for each section, so we need to collect all the tex file into a single file and then hand off to the splitter, embedder. After completion, remove the temp directory. Return a properly formatted error if package arxiv needs to be installed.
* Handle two types of errors: (i) package arxiv not installed. (ii) User enters a wrong paper id.

packages/jupyter-ai/jupyter_ai/chat_handlers/learn.py

@@ -4,7 +4,7 @@
 from typing import Any, Coroutine, List, Optional, Tuple
 
 from dask.distributed import Client as DaskClient
-from jupyter_ai.document_loaders.directory import get_embeddings, split
+from jupyter_ai.document_loaders.directory import get_embeddings, split, arxiv_to_text
 from jupyter_ai.document_loaders.splitter import ExtensionSplitter, NotebookSplitter
 from jupyter_ai.models import (
     DEFAULT_CHUNK_OVERLAP,
@@ -44,6 +44,9 @@ def __init__(self, *args, **kwargs):
         self.parser.add_argument("-v", "--verbose", action="store_true")
         self.parser.add_argument("-d", "--delete", action="store_true")
         self.parser.add_argument("-l", "--list", action="store_true")
+        self.parser.add_argument(
+            "-r", "--remote", action="store" , default=None, type=str
+        )
         self.parser.add_argument(
             "-c", "--chunk-size", action="store", default=DEFAULT_CHUNK_SIZE, type=int
         )
@@ -107,6 +110,17 @@ async def process_message(self, message: HumanChatMessage):
             self.reply(self._build_list_response())
             return
 
+        if args.remote:
+            remote_type = args.remote.lower()
+            if remote_type=="arxiv":
+                try:
+                    id = args.path[0]
+                    args.path = [arxiv_to_text(id)] # call the function in `directory.py``
+                    self.reply(f"Processing arxiv file id {id}, saved in {args.path[0]}.", message)
+                except Exception as e:
+                    self.reply(f"""The arXiv file could not be processed. Check the paper ID ({id}). Or, verify that the `arxiv` package is installed.""")
+                    return
+
         # Make sure the path exists.
         if not len(args.path) == 1:
             self.reply(f"{self.parser.format_usage()}", message)

packages/jupyter-ai/jupyter_ai/document_loaders/directory.py

@@ -5,10 +5,42 @@
 from typing import List
 
 import dask
-from langchain.document_loaders import PyPDFLoader
+from langchain.document_loaders import PyPDFLoader, ArxivLoader
 from langchain.schema import Document
 from langchain.text_splitter import TextSplitter
 
+import tarfile
+import shutil 
+
+# Download a single tar file from arXiv and store in a temp folder for RAG, then run learn on it. 
+try:
+    import arxiv
+except Exception as e:
+    print("Missing package: arxiv")
+
+def arxiv_to_text(id): # id is numbers after "arXiv" in arXiv:xxxx.xxxxx
+    # Get the paper from arxiv
+    outfile = id + ".tex"
+    temp_dir = "downloads_temp"
+    if not os.path.isdir(temp_dir):
+        os.mkdir(temp_dir) 
+    client = arxiv.Client()
+    paper = next(arxiv.Client().results(arxiv.Search(id_list=[id])))
+    paper.download_source(dirpath=temp_dir, filename="downloaded-paper.tar.gz")     
+    # Extract downloaded tar file
+    tar = tarfile.open(temp_dir+"/downloaded-paper.tar.gz")
+    tar.extractall(temp_dir)
+    tar.close()
+    tex_list = os.listdir(temp_dir)
+    tex_list = [j for j in tex_list if j.lower().endswith('.tex')]
+    with open(outfile,'wb') as wfd:
+        for f in tex_list:
+            with open(temp_dir+"/"+f,'rb') as fd:
+                shutil.copyfileobj(fd, wfd)       
+
+    outfile_path = os.path.realpath(outfile)
+    shutil.rmtree(temp_dir) # Delete the temp folder but not the downloaded latex files
+    return outfile_path
 
 # Uses pypdf which is used by PyPDFLoader from langchain
 def pdf_to_text(path):
@@ -50,6 +82,7 @@ def path_to_doc(path):
     ".txt",
     ".html",
     ".pdf",
+    ".tex",
 }
 
 

packages/jupyter-ai/pyproject.toml

@@ -54,7 +54,7 @@ test = [
 
 dev = ["jupyter_ai_magics[dev]"]
 
-all = ["jupyter_ai_magics[all]", "pypdf"]
+all = ["jupyter_ai_magics[all]", "pypdf", "arxiv"]
 
 [tool.hatch.version]
 source = "nodejs"