mypy config update (#237)

* Ran pre-commit autoupdate for newer mypy and added mypy config

* Removed all prior type ignore comments

* Removed redundant cast

* Ran mypy_clean_slate to add type ignores

.pre-commit-config.yaml

@@ -20,11 +20,11 @@ repos:
       - id: black
         language_version: python3
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: "v1.3.0"
+    rev: v1.8.0
     hooks:
       - id: mypy
         args: [--pretty, --ignore-missing-imports]
-        additional_dependencies: [types-requests]
+        additional_dependencies: [types-requests, types-setuptools]
   - repo: https://github.com/PyCQA/isort
     rev: "5.12.0"
     hooks:

dev-requirements.txt

@@ -5,6 +5,7 @@ pyzotero
 python-dotenv
 pymupdf
 build
+types-setuptools
 types-requests
 langchain_openai
 langchain_community

paperqa/contrib/zotero.py

@@ -128,7 +128,7 @@ def get_pdf(self, item: dict) -> Union[Path, None]:
         if pdf_key is None:
             return None
 
-        pdf_path: Path = Path(self.storage / (pdf_key + ".pdf"))  # type: ignore
+        pdf_path: Path = Path(self.storage / (pdf_key + ".pdf"))  # type: ignore[operator]
 
         if not pdf_path.exists():
             pdf_path.parent.mkdir(parents=True, exist_ok=True)

paperqa/docs.py

@@ -422,7 +422,7 @@ async def aadd(
             # get first name and year from citation
             match = re.search(r"([A-Z][a-z]+)", citation)
             if match is not None:
-                author = match.group(1)  # type: ignore
+                author = match.group(1)
             else:
                 # panicking - no word??
                 raise ValueError(
@@ -433,7 +433,7 @@ async def aadd(
             year = ""
             match = re.search(r"(\d{4})", citation)
             if match is not None:
-                year = match.group(1)  # type: ignore
+                year = match.group(1)
             docname = f"{author}{year}"
         docname = self._get_unique_name(docname)
         doc = Doc(docname=docname, citation=citation, dockey=dockey)
@@ -619,7 +619,7 @@ async def aget_evidence(
     ) -> Answer:
         if len(self.docs) == 0 and self.docs_index is None:
             # do we have no docs?
-            return answer
+            return answer  # type: ignore[unreachable]
         self._build_texts_index(keys=answer.dockey_filter)
         _k = k
         if answer.dockey_filter is not None:
@@ -663,13 +663,13 @@ async def process(match):
                 score = 5
             else:
                 if self.prompts.json_summary:
-                    summary_chain = self.summary_llm_model.make_chain(
+                    summary_chain = self.summary_llm_model.make_chain(  # type: ignore[union-attr]
                         client=self._client,
                         prompt=self.prompts.summary_json,
                         system_prompt=self.prompts.summary_json_system,
                     )
                 else:
-                    summary_chain = self.summary_llm_model.make_chain(
+                    summary_chain = self.summary_llm_model.make_chain(  # type: ignore[union-attr]
                         client=self._client,
                         prompt=self.prompts.summary,
                         system_prompt=self.prompts.system,

paperqa/llms.py

@@ -205,9 +205,9 @@ async def execute(
                         f for f in callbacks if not is_coroutine_callable(f)
                     ]
                     async_callbacks = [f for f in callbacks if is_coroutine_callable(f)]
-                    completion = self.achat_iter(client, messages)  # type: ignore
+                    completion = self.achat_iter(client, messages)
                     text_result = []
-                    async for chunk in completion:  # type: ignore
+                    async for chunk in completion:  # type: ignore[attr-defined]
                         if chunk:
                             if result.seconds_to_first_token == 0:
                                 result.seconds_to_first_token = (
@@ -250,12 +250,12 @@ async def execute(
                     ]
                     async_callbacks = [f for f in callbacks if is_coroutine_callable(f)]
 
-                    completion = self.acomplete_iter(  # type: ignore
+                    completion = self.acomplete_iter(
                         client,
                         formatted_prompt,
                     )
                     text_result = []
-                    async for chunk in completion:  # type: ignore
+                    async for chunk in completion:  # type: ignore[attr-defined]
                         if chunk:
                             if result.seconds_to_first_token == 0:
                                 result.seconds_to_first_token = (
@@ -289,7 +289,7 @@ def _check_client(self, client: Any) -> AsyncOpenAI:
             raise ValueError(
                 f"Your client is not a required AsyncOpenAI client. It is a {type(client)}"
             )
-        return cast(AsyncOpenAI, client)
+        return client
 
     @model_validator(mode="after")
     @classmethod
@@ -323,14 +323,14 @@ async def acomplete_iter(self, client: Any, prompt: str) -> Any:
     async def achat(self, client: Any, messages: list[dict[str, str]]) -> str:
         aclient = self._check_client(client)
         completion = await aclient.chat.completions.create(
-            messages=messages, **process_llm_config(self.config)  # type: ignore
+            messages=messages, **process_llm_config(self.config)
         )
         return completion.choices[0].message.content or ""
 
     async def achat_iter(self, client: Any, messages: list[dict[str, str]]) -> Any:
         aclient = self._check_client(client)
         completion = await aclient.chat.completions.create(
-            messages=messages, **process_llm_config(self.config), stream=True  # type: ignore
+            messages=messages, **process_llm_config(self.config), stream=True
         )
         async for chunk in cast(AsyncGenerator, completion):
             yield chunk.choices[0].delta.content
@@ -635,13 +635,13 @@ def add_texts_and_embeddings(self, texts: Sequence[Embeddable]) -> None:
         else:
             raise ValueError("Only embeddings of type Text are supported")
         if self._store is None:
-            self._store = self._store_builder(  # type: ignore
+            self._store = self._store_builder(
                 vec_store_text_and_embeddings,
                 texts,
             )
             if self._store is None or not hasattr(self._store, "add_embeddings"):
                 raise ValueError("store_builder did not return a valid vectorstore")
-        self._store.add_embeddings(  # type: ignore
+        self._store.add_embeddings(
             vec_store_text_and_embeddings,
             metadatas=texts,
         )

paperqa/types.py

@@ -192,7 +192,7 @@ def remove_computed(cls, data: Any) -> Any:
             data.pop("used_contexts", None)
         return data
 
-    @computed_field  # type: ignore
+    @computed_field  # type: ignore[misc]
     @property
     def used_contexts(self) -> set[str]:
         """Return the used contexts."""

pyproject.toml

@@ -0,0 +1,49 @@
+[tool.mypy]
+# Type-checks the interior of functions without type annotations.
+check_untyped_defs = true
+# Allows enabling one or multiple error codes globally. Note: This option will
+# override disabled error codes from the disable_error_code option.
+enable_error_code = [
+    "ignore-without-code",
+    "mutable-override",
+    "redundant-cast",
+    "redundant-expr",
+    "redundant-self",
+    "truthy-bool",
+    "truthy-iterable",
+    "unreachable",
+    "unused-awaitable",
+    "unused-ignore",
+]
+# Shows a short summary line after error messages.
+error_summary = false
+# Use visually nicer output in error messages: use soft word wrap, show source
+# code snippets, and show error location markers.
+pretty = true
+# Shows column numbers in error messages.
+show_column_numbers = true
+# Shows error codes in error messages.
+# SEE: https://mypy.readthedocs.io/en/stable/error_codes.html#error-codes
+show_error_codes = true
+# Prefixes each error with the relevant context.
+show_error_context = true
+# Warns about casting an expression to its inferred type.
+warn_redundant_casts = true
+# Shows a warning when encountering any code inferred to be unreachable or
+# redundant after performing type analysis.
+warn_unreachable = true
+# Warns about per-module sections in the config file that do not match any
+# files processed when invoking mypy.
+warn_unused_configs = true
+# Warns about unneeded `# type: ignore` comments.
+warn_unused_ignores = true
+
+[[tool.mypy.overrides]]
+# Suppresses error messages about imports that cannot be resolved.
+ignore_missing_imports = true
+# Per-module configuration options
+module = [
+    "fitz",
+    "pyzotero",  # SEE: https://github.com/urschrei/pyzotero/issues/110
+    "sentence_transformers",  # SEE: https://github.com/UKPLab/sentence-transformers/issues/1723
+]