symbolicai 1.0.0__py3-none-any.whl → 1.1.1__py3-none-any.whl

symai/extended/api_builder.py

@@ -64,7 +64,7 @@ res = run(value) # [MANAGED] must contain this line, do not change
 
 class APIBuilderPreProcessor(PreProcessor):
     def __call__(self, argument):
-        return f'$> {argument.args[0]!s} =>'
+        return f"$> {argument.args[0]!s} =>"
 
 
 class APIBuilder(Expression):
@@ -77,9 +77,12 @@ class APIBuilder(Expression):
         self.sym_return_type = APIBuilder
 
     def forward(self, sym: Symbol, **kwargs) -> Symbol:
-        @core.zero_shot(prompt="Build the API call code:\n",
-                   pre_processors=[APIBuilderPreProcessor()],
-                   post_processors=[CodeExtractPostProcessor()], **kwargs)
+        @core.zero_shot(
+            prompt="Build the API call code:\n",
+            pre_processors=[APIBuilderPreProcessor()],
+            post_processors=[CodeExtractPostProcessor()],
+            **kwargs,
+        )
         def _func(_, text) -> str:
             pass
 
@@ -96,18 +99,20 @@ class StackTraceRetryExecutor(Expression):
     def forward(self, code: Symbol, request: Symbol, **kwargs) -> Symbol:
         code = str(code)
         # Set value that gets passed on to the 'run' function in the generated code
-        value = request.value # do not remove this line
+        value = request.value  # do not remove this line
         # Create the 'run' function
         self._runnable = self.executor(code, locals=locals().copy(), globals=globals().copy())
-        result = self._runnable['locals']['run'](value)
+        result = self._runnable["locals"]["run"](value)
         retry = 0
         # Retry if there is a 'Traceback' in the result
-        while 'Traceback' in result and retry <= self.max_retries:
-            self._runnable = self.executor(code, payload=result, locals=locals().copy(), globals=globals().copy(), **kwargs)
-            result = self._runnable['locals']['run'](value)
+        while "Traceback" in result and retry <= self.max_retries:
+            self._runnable = self.executor(
+                code, payload=result, locals=locals().copy(), globals=globals().copy(), **kwargs
+            )
+            result = self._runnable["locals"]["run"](value)
             retry += 1
-        if 'locals_res' in self._runnable:
-            result = self._runnable['locals_res']
+        if "locals_res" in self._runnable:
+            result = self._runnable["locals_res"]
         return result
 
 
@@ -129,14 +134,14 @@ class APIExecutor(Expression):
     def forward(self, request: Symbol, **_kwargs) -> Symbol:
         self._request = self._to_symbol(request)
         if self._verbose:
-            UserMessage(f'[REQUEST] {self._request}')
+            UserMessage(f"[REQUEST] {self._request}")
         # Generate the code to implement the API call
-        self._code    = self.builder(self._request)
+        self._code = self.builder(self._request)
         if self._verbose:
-            UserMessage(f'[GENERATED_CODE] {self._code}')
+            UserMessage(f"[GENERATED_CODE] {self._code}")
         # Execute the code to define the 'run' function
-        self._result  = self.executor(self._code, request=self._request)
+        self._result = self.executor(self._code, request=self._request)
         if self._verbose:
-            UserMessage(f'[RESULT]: {self._result}')
-        self._value   = self._result
+            UserMessage(f"[RESULT]: {self._result}")
+        self._value = self._result
         return self

symai/extended/arxiv_pdf_parser.py

@@ -12,7 +12,9 @@ from .file_merger import FileMerger
 
 
 class ArxivPdfParser(Expression):
-    def __init__(self, url_pattern: str = r'https://arxiv.org/(?:pdf|abs)/(\d+.\d+)(?:\.pdf)?', **kwargs):
+    def __init__(
+        self, url_pattern: str = r"https://arxiv.org/(?:pdf|abs)/(\d+.\d+)(?:\.pdf)?", **kwargs
+    ):
         super().__init__(**kwargs)
         self.url_pattern = url_pattern
         self.merger = FileMerger()
@@ -22,7 +24,11 @@ class ArxivPdfParser(Expression):
         urls = re.findall(self.url_pattern, str(data))
 
         # Convert all urls to pdf urls
-        pdf_urls = ["https://arxiv.org/pdf/" + (f"{url.split('/')[-1]}.pdf" if 'pdf' not in url else {url.split('/')[-1]}) for url in urls]
+        pdf_urls = [
+            "https://arxiv.org/pdf/"
+            + (f"{url.split('/')[-1]}.pdf" if "pdf" not in url else {url.split("/")[-1]})
+            for url in urls
+        ]
 
         # Create temporary folder in the home directory
         output_path = HOME_PATH / "temp" / "downloads"
@@ -31,7 +37,9 @@ class ArxivPdfParser(Expression):
         pdf_files = []
         with ThreadPoolExecutor() as executor:
             # Download all pdfs in parallel
-            future_to_url = {executor.submit(self.download_pdf, url, output_path): url for url in pdf_urls}
+            future_to_url = {
+                executor.submit(self.download_pdf, url, output_path): url for url in pdf_urls
+            }
             for future in as_completed(future_to_url):
                 url = future_to_url[future]
                 try:
@@ -56,7 +64,7 @@ class ArxivPdfParser(Expression):
     def download_pdf(self, url, output_path):
         # Download pdfs
         response = requests.get(url)
-        file_path = Path(output_path) / f'{url.split("/")[-1]}'
-        with file_path.open('wb') as f:
+        file_path = Path(output_path) / f"{url.split('/')[-1]}"
+        with file_path.open("wb") as f:
             f.write(response.content)
         return str(file_path)

symai/extended/bibtex_parser.py

@@ -68,7 +68,7 @@ Multimodal Few-Shot Learning with Frozen Language Models Maria Tsimpoukelli
 
 class BibTexPreProcessor(PreProcessor):
     def __call__(self, argument):
-        return f'>>>\n{argument.args[0]!s}\n\n<<<\n'
+        return f">>>\n{argument.args[0]!s}\n\n<<<\n"
 
 
 class BibTexParser(Expression):
@@ -81,9 +81,13 @@ class BibTexParser(Expression):
         self.sym_return_type = BibTexParser
 
     def forward(self, sym: Symbol, **kwargs) -> Symbol:
-        @core.zero_shot(prompt="Create bibtex entries:\n",
-                   pre_processors=[BibTexPreProcessor()],
-                   post_processors=[CodeExtractPostProcessor()], **kwargs)
+        @core.zero_shot(
+            prompt="Create bibtex entries:\n",
+            pre_processors=[BibTexPreProcessor()],
+            post_processors=[CodeExtractPostProcessor()],
+            **kwargs,
+        )
         def _func(_, text) -> str:
             pass
+
         return _func(self, sym)

symai/extended/conversation.py

@@ -19,22 +19,25 @@ _DEFAULT_TEXT_CONTAINER_FORMATTER = TextContainerFormatter(text_split=4)
 class CodeFormatter:
     def __call__(self, value: str, *_args: Any, **_kwds: Any) -> Any:
         # extract code from chat conversations or ```<language>\n{code}\n``` blocks
-        return Symbol(value).extract('Only extract code without ``` block markers or chat conversations')
+        return Symbol(value).extract(
+            "Only extract code without ``` block markers or chat conversations"
+        )
 
 
 class Conversation(SlidingWindowStringConcatMemory):
     def __init__(
-            self,
-            init: str | None = None,
-            file_link: list[str] | None = None,
-            url_link: list[str] | None = None,
-            index_name: str | None = None,
-            auto_print: bool = True,
-            truncation_percentage: float = 0.8,
-            truncation_type: str = 'head',
-            with_metadata: bool = False,
-            *args, **kwargs
-        ):
+        self,
+        init: str | None = None,
+        file_link: list[str] | None = None,
+        url_link: list[str] | None = None,
+        index_name: str | None = None,
+        auto_print: bool = True,
+        truncation_percentage: float = 0.8,
+        truncation_type: str = "head",
+        with_metadata: bool = False,
+        *args,
+        **kwargs,
+    ):
         super().__init__(*args, **kwargs)
         self.truncation_percentage = truncation_percentage
         self.truncation_type = truncation_type
@@ -48,9 +51,9 @@ class Conversation(SlidingWindowStringConcatMemory):
         self.index_name = index_name
         self.seo_opt = SEOQueryOptimizer()
         self.reader = FileReader(with_metadata=with_metadata)
-        self.scraper = Interface('naive_webscraping')
-        self.user_tag = 'USER::'
-        self.bot_tag = 'ASSISTANT::'
+        self.scraper = Interface("naive_scrape")
+        self.user_tag = "USER::"
+        self.bot_tag = "ASSISTANT::"
 
         if init is not None:
             self.store_system_message(init, *args, **kwargs)
@@ -63,14 +66,16 @@ class Conversation(SlidingWindowStringConcatMemory):
         self.indexer = None
         self.index = None
         if index_name is not None:
-            UserMessage("Index not supported for conversation class.", raise_with=NotImplementedError)
+            UserMessage(
+                "Index not supported for conversation class.", raise_with=NotImplementedError
+            )
 
     def __getstate__(self):
         state = super().__getstate__().copy()
-        state.pop('seo_opt', None)
-        state.pop('indexer', None)
-        state.pop('index', None)
-        state.pop('reader', None)
+        state.pop("seo_opt", None)
+        state.pop("indexer", None)
+        state.pop("index", None)
+        state.pop("reader", None)
         return state
 
     def __setstate__(self, state):
@@ -78,7 +83,9 @@ class Conversation(SlidingWindowStringConcatMemory):
         self.seo_opt = SEOQueryOptimizer()
         self.reader = FileReader()
         if self.index_name is not None:
-            UserMessage("Index not supported for conversation class.", raise_with=NotImplementedError)
+            UserMessage(
+                "Index not supported for conversation class.", raise_with=NotImplementedError
+            )
 
     def store_system_message(self, message: str, *_args, **_kwargs):
         val = f"[SYSTEM_INSTRUCTION::]: <<<\n{message!s}\n>>>\n"
@@ -100,7 +107,7 @@ class Conversation(SlidingWindowStringConcatMemory):
         path_obj = Path(file_path)
         path_obj.parent.mkdir(parents=True, exist_ok=True)
         # Save the conversation object as a pickle file
-        with path_obj.open('wb') as handle:
+        with path_obj.open("wb") as handle:
             pickle.dump(conversation, handle, protocol=pickle.HIGHEST_PROTOCOL)
 
     def load_conversation_state(self, path: str) -> "Conversation":
@@ -110,7 +117,7 @@ class Conversation(SlidingWindowStringConcatMemory):
             if path_obj.stat().st_size <= 0:
                 UserMessage("File is empty.", raise_with=Exception)
             # Load the conversation object from a pickle file
-            with path_obj.open('rb') as handle:
+            with path_obj.open("rb") as handle:
                 conversation_state = pickle.load(handle)
         else:
             UserMessage("File does not exist or is empty.", raise_with=Exception)
@@ -123,14 +130,16 @@ class Conversation(SlidingWindowStringConcatMemory):
         self._memory = conversation_state._memory
         self.truncation_percentage = conversation_state.truncation_percentage
         self.truncation_type = conversation_state.truncation_type
-        self.auto_print  = conversation_state.auto_print
+        self.auto_print = conversation_state.auto_print
         self.file_link = conversation_state.file_link
         self.url_link = conversation_state.url_link
-        self.index_name  = conversation_state.index_name
+        self.index_name = conversation_state.index_name
         self.seo_opt = SEOQueryOptimizer()
         self.reader = FileReader()
         if self.index_name is not None:
-            UserMessage("Index not supported for conversation class.", raise_with=NotImplementedError)
+            UserMessage(
+                "Index not supported for conversation class.", raise_with=NotImplementedError
+            )
         return self
 
     def commit(self, target_file: str | None = None, formatter: Callable | None = None):
@@ -143,20 +152,22 @@ class Conversation(SlidingWindowStringConcatMemory):
             elif isinstance(file_link, list) and len(file_link) == 1:
                 file_link = file_link[0]
             else:
-                file_link = None # cannot commit to multiple files
-                UserMessage('Cannot commit to multiple files.', raise_with=Exception)
+                file_link = None  # cannot commit to multiple files
+                UserMessage("Cannot commit to multiple files.", raise_with=Exception)
         if file_link:
             # if file extension is .py, then format code
             format_ = formatter
-            formatter = CodeFormatter() if format_ is None and file_link.endswith('.py') else formatter
+            formatter = (
+                CodeFormatter() if format_ is None and file_link.endswith(".py") else formatter
+            )
             val = self.value
             if formatter:
                 val = formatter(val)
             # if file does not exist, create it
-            with Path(file_link).open('w') as file:
+            with Path(file_link).open("w") as file:
                 file.write(str(val))
         else:
-            UserMessage('File link is not set or a set of files.', raise_with=Exception)
+            UserMessage("File link is not set or a set of files.", raise_with=Exception)
 
     def save(self, path: str, replace: bool = False) -> Symbol:
         return Symbol(self._memory).save(path, replace=replace)
@@ -174,7 +185,7 @@ class Conversation(SlidingWindowStringConcatMemory):
         res = self.recall(query, *args, payload=payload, **kwargs)
 
         # if user is requesting to preview the response, then return only the preview result
-        if kwargs.get('preview'):
+        if kwargs.get("preview"):
             if self.auto_print:
                 UserMessage(str(res), style="text")
             return res
@@ -190,12 +201,14 @@ class Conversation(SlidingWindowStringConcatMemory):
         return res
 
     def _apply_truncation_overrides(self, kwargs: dict[str, Any]) -> dict[str, Any]:
-        dynamic_truncation_percentage = kwargs.get('truncation_percentage', self.truncation_percentage)
-        dynamic_truncation_type = kwargs.get('truncation_type', self.truncation_type)
+        dynamic_truncation_percentage = kwargs.get(
+            "truncation_percentage", self.truncation_percentage
+        )
+        dynamic_truncation_type = kwargs.get("truncation_type", self.truncation_type)
         return {
             **kwargs,
-            'truncation_percentage': dynamic_truncation_percentage,
-            'truncation_type': dynamic_truncation_type,
+            "truncation_percentage": dynamic_truncation_percentage,
+            "truncation_type": dynamic_truncation_type,
         }
 
     def _retrieve_index_memory(self, query: Symbol, args: tuple[Any, ...], kwargs: dict[str, Any]):
@@ -205,7 +218,7 @@ class Conversation(SlidingWindowStringConcatMemory):
         memory_split = self._memory.split(self.marker)
         memory_shards = []
         for shard in memory_split:
-            if shard.strip() == '':
+            if shard.strip() == "":
                 continue
             memory_shards.append(shard)
 
@@ -213,33 +226,33 @@ class Conversation(SlidingWindowStringConcatMemory):
         if length_memory_shards > 5:
             memory_shards = memory_shards[:2] + memory_shards[-3:]
         elif length_memory_shards > 3:
-            retained = memory_shards[-(length_memory_shards - 2):]
+            retained = memory_shards[-(length_memory_shards - 2) :]
             memory_shards = memory_shards[:2] + retained
 
-        search_query = query | '\n' | '\n'.join(memory_shards)
-        if kwargs.get('use_seo_opt'):
-            search_query = self.seo_opt('[Query]:' | search_query)
+        search_query = query | "\n" | "\n".join(memory_shards)
+        if kwargs.get("use_seo_opt"):
+            search_query = self.seo_opt("[Query]:" | search_query)
         memory = self.index(search_query, *args, **kwargs)
 
-        if 'raw_result' in kwargs:
+        if "raw_result" in kwargs:
             UserMessage(str(memory), style="text")
         return memory
 
     def _build_payload(self, kwargs: dict[str, Any], memory) -> str:
-        payload = ''
-        if 'payload' in kwargs:
+        payload = ""
+        if "payload" in kwargs:
             payload = f"[Conversation Payload]:\n{kwargs.pop('payload')}\n"
 
-        index_memory = ''
+        index_memory = ""
         if memory:
-            index_memory = f'[Index Retrieval]:\n{str(memory)[:1500]}\n'
-        return f'{index_memory}{payload}'
+            index_memory = f"[Index Retrieval]:\n{str(memory)[:1500]}\n"
+        return f"{index_memory}{payload}"
 
     def _append_interaction_to_memory(self, query: Symbol, res: Symbol) -> None:
         prompt = self.build_tag(self.user_tag, query)
         self.store(prompt)
 
-        self._value = res.value # save last response
+        self._value = res.value  # save last response
         val = self.build_tag(self.bot_tag, res)
         self.store(val)
 
@@ -276,27 +289,33 @@ Responses should be:
 - Referenced to source when applicable
 """
 
+
 @deprecated("Use `Conversation` instead for now. This will be removed/fixed in the future.")
 class RetrievalAugmentedConversation(Conversation):
     def __init__(
-            self,
-            folder_path: str | None = None,
-            *,
-            index_name: str | None = None,
-            max_depth: int | None = 0,
-            auto_print: bool = True,
-            top_k: int = 5,
-            formatter: Callable = _DEFAULT_TEXT_CONTAINER_FORMATTER,
-            overwrite: bool = False,
-            truncation_percentage: float = 0.8,
-            truncation_type: str = 'head',
-            with_metadata: bool = False,
-            raw_result: bool | None = False,
-            new_dim: int | None = None,
-            **kwargs
-        ):
-
-        super().__init__(auto_print=auto_print, truncation_percentage=truncation_percentage, truncation_type=truncation_type, with_metadata=with_metadata, **kwargs)
+        self,
+        folder_path: str | None = None,
+        *,
+        index_name: str | None = None,
+        max_depth: int | None = 0,
+        auto_print: bool = True,
+        top_k: int = 5,
+        formatter: Callable = _DEFAULT_TEXT_CONTAINER_FORMATTER,
+        overwrite: bool = False,
+        truncation_percentage: float = 0.8,
+        truncation_type: str = "head",
+        with_metadata: bool = False,
+        raw_result: bool | None = False,
+        new_dim: int | None = None,
+        **kwargs,
+    ):
+        super().__init__(
+            auto_print=auto_print,
+            truncation_percentage=truncation_percentage,
+            truncation_type=truncation_type,
+            with_metadata=with_metadata,
+            **kwargs,
+        )
 
         self.retriever = DocumentRetriever(
             source=folder_path,
@@ -308,7 +327,7 @@ class RetrievalAugmentedConversation(Conversation):
             with_metadata=with_metadata,
             raw_result=raw_result,
             new_dim=new_dim,
-            **kwargs
+            **kwargs,
         )
 
         self.index = self.retriever.index
@@ -333,14 +352,14 @@ class RetrievalAugmentedConversation(Conversation):
 
         memory = self.index(query, *args, **kwargs)
 
-        if 'raw_result' in kwargs:
+        if "raw_result" in kwargs:
             UserMessage(str(memory), style="text")
             return memory
 
         prompt = self.build_tag(self.user_tag, query)
         self.store(prompt)
 
-        payload = f'[Index Retrieval]:\n{str(memory)[:1500]}\n'
+        payload = f"[Index Retrieval]:\n{str(memory)[:1500]}\n"
 
         res = self.recall(query, *args, payload=payload, **kwargs)
 

symai/extended/document.py

@@ -15,22 +15,24 @@ _DEFAULT_PARAGRAPH_FORMATTER = ParagraphFormatter()
 
 class DocumentRetriever(Expression):
     def __init__(
-            self,
-            source: str | None = None,
-            *,
-            index_name: str = Indexer.DEFAULT,
-            top_k: int = 5,
-            max_depth: int = 1,
-            formatter: Callable = _DEFAULT_PARAGRAPH_FORMATTER,
-            overwrite: bool = False,
-            with_metadata: bool = False,
-            raw_result: bool | None = False,
-            new_dim: int | None = None,
-            **kwargs
-        ):
+        self,
+        source: str | None = None,
+        *,
+        index_name: str = Indexer.DEFAULT,
+        top_k: int = 5,
+        max_depth: int = 1,
+        formatter: Callable = _DEFAULT_PARAGRAPH_FORMATTER,
+        overwrite: bool = False,
+        with_metadata: bool = False,
+        raw_result: bool | None = False,
+        new_dim: int | None = None,
+        **kwargs,
+    ):
         super().__init__(**kwargs)
-        self.indexer = Indexer(index_name=index_name, top_k=top_k, formatter=formatter, auto_add=False, new_dim=new_dim)
-        self.reader  = FileReader(with_metadata=with_metadata)
+        self.indexer = Indexer(
+            index_name=index_name, top_k=top_k, formatter=formatter, auto_add=False, new_dim=new_dim
+        )
+        self.reader = FileReader(with_metadata=with_metadata)
         self.new_dim = new_dim
 
         if overwrite:
@@ -39,21 +41,23 @@ class DocumentRetriever(Expression):
         # we insert the text into the index if (1) index does not exist and (2) there's a specific source
         if source is not None and not self.indexer.exists():
             self.indexer.register()
-            text = self.parse_source(source, with_metadata=with_metadata, max_depth=max_depth, **kwargs)
+            text = self.parse_source(
+                source, with_metadata=with_metadata, max_depth=max_depth, **kwargs
+            )
             self.index = self.indexer(data=text, raw_result=raw_result, **kwargs)
         else:
             # we don't insert the text at initialization since the index already exists and there's no specific source
             self.index = self.indexer(raw_result=raw_result, **kwargs)
 
     def forward(
-            self,
-            query: Symbol,
-            raw_result: bool | None = False,
-        ) -> Symbol:
+        self,
+        query: Symbol,
+        raw_result: bool | None = False,
+    ) -> Symbol:
         return self.index(
-                query,
-                raw_result=raw_result,
-                )
+            query,
+            raw_result=raw_result,
+        )
 
     def insert(self, source: str | Path, **kwargs):
         # dynamically insert data into the index given a session
@@ -61,12 +65,19 @@ class DocumentRetriever(Expression):
         #  - a string (e.g. something that the user wants to insert)
         #  - a file path (e.g. a new file that the user wants to insert)
         #  - a directory path (e.g. a new directory that the user wants to insert)
-        text = self.parse_source(source, with_metadata=kwargs.get('with_metadata', False), max_depth=kwargs.get('max_depth', 1), **kwargs)
-        #NOTE: Do we need `new_dim` here?
+        text = self.parse_source(
+            source,
+            with_metadata=kwargs.get("with_metadata", False),
+            max_depth=kwargs.get("max_depth", 1),
+            **kwargs,
+        )
+        # NOTE: Do we need `new_dim` here?
         self.add(text, index_name=self.indexer.index_name, **kwargs)
         self.config(None, save=True, index_name=self.indexer.index_name, **kwargs)
 
-    def parse_source(self, source: str, with_metadata: bool, max_depth: int, **kwargs) -> list[Union[str, 'TextContainer']]:
+    def parse_source(
+        self, source: str, with_metadata: bool, max_depth: int, **kwargs
+    ) -> list[Union[str, "TextContainer"]]:
         maybe_path = Path(source)
         if isinstance(source, str) and not (maybe_path.is_file() or maybe_path.is_dir()):
             return Symbol(source).zip(new_dim=self.new_dim)
@@ -75,5 +86,7 @@ class DocumentRetriever(Expression):
             return self.reader(files, with_metadata=with_metadata, **kwargs)
         if maybe_path.is_file():
             return self.reader(source, with_metadata=with_metadata, **kwargs)
-        UserMessage(f"Invalid source: {source}; must be a file, directory, or string", raise_with=ValueError)
+        UserMessage(
+            f"Invalid source: {source}; must be a file, directory, or string", raise_with=ValueError
+        )
         return []

symai/extended/file_merger.py

@@ -12,12 +12,48 @@ class FileMerger(Expression):
     Class to merge contents of multiple files into one, specified by their file endings and root path.
     Files specified in the exclude list will not be included.
     """
-    def __init__(self, file_endings: list[str] | None = None,
-                       file_excludes: list[str] | None = None, **kwargs):
+
+    def __init__(
+        self,
+        file_endings: list[str] | None = None,
+        file_excludes: list[str] | None = None,
+        **kwargs,
+    ):
         if file_excludes is None:
-            file_excludes = ['__init__.py', '__pycache__', 'LICENSE', 'requirements.txt', 'environment.yaml', '.git']
+            file_excludes = [
+                "__init__.py",
+                "__pycache__",
+                "LICENSE",
+                "requirements.txt",
+                "environment.yaml",
+                ".git",
+            ]
         if file_endings is None:
-            file_endings = ['.py', '.md', '.txt', '.sh', '.pdf', '.json', '.yaml', '.java', '.cpp', '.hpp', '.c', '.h', '.js', '.css', '.html', '.xml', '.csv', '.tsv', '.yml', '.rst', '.ipynb', '.tex', '.bib']
+            file_endings = [
+                ".py",
+                ".md",
+                ".txt",
+                ".sh",
+                ".pdf",
+                ".json",
+                ".yaml",
+                ".java",
+                ".cpp",
+                ".hpp",
+                ".c",
+                ".h",
+                ".js",
+                ".css",
+                ".html",
+                ".xml",
+                ".csv",
+                ".tsv",
+                ".yml",
+                ".rst",
+                ".ipynb",
+                ".tex",
+                ".bib",
+            ]
         super().__init__(**kwargs)
         self.file_endings = file_endings
         self.file_excludes = file_excludes
@@ -56,9 +92,11 @@ class FileMerger(Expression):
                     file_path_escaped = file_path_str.replace(" ", "\\ ")
 
                     # Append start and end markers for each file
-                    file_content = f"# ----[FILE_START]<PART1/1>{file_path_escaped}[FILE_CONTENT]:\n" + \
-                                   file_content + \
-                                   f"\n# ----[FILE_END]{file_path_escaped}\n"
+                    file_content = (
+                        f"# ----[FILE_START]<PART1/1>{file_path_escaped}[FILE_CONTENT]:\n"
+                        + file_content
+                        + f"\n# ----[FILE_END]{file_path_escaped}\n"
+                    )
 
                     # Merge the file contents
                     merged_file += file_content