Add resolve files

src/datachain/lib/dc.py

@@ -1689,3 +1689,19 @@
     def offset(self, offset: int) -> "Self":
         """Return the results starting with the offset row."""
         return super().offset(offset)
+
+    def resolve_files(self, signal: str = "file") -> "Self":
+        """Check if the file object is valid and update its is_valid field."""
+
+        def check_file(file: File) -> File:
+            if not isinstance(file, File):
+                raise TypeError(f"Signal '{signal}' is not a File object")
+
+            try:
+                with file.open():
+                    file.is_valid = True
+            except OSError:
+                file.is_valid = False
+            return file
+
+        return self.map(**{signal: check_file})

src/datachain/lib/file.py

@@ -119,6 +119,7 @@
     last_modified: datetime = Field(default=TIME_ZERO)
     location: Optional[Union[dict, list[dict]]] = Field(default=None)
     vtype: str = Field(default="")
+    is_valid: Optional[bool] = Field(default=None)
 
     _datachain_column_types: ClassVar[dict[str, Any]] = {
         "source": String,
@@ -191,6 +192,9 @@
     @contextmanager
     def open(self, mode: Literal["rb", "r"] = "rb"):
         """Open the file and return a file object."""
+        if self.is_valid is False:
+            raise FileNotFoundError(f"File {self.path} is not valid")
+
         if self.location:
             with VFileRegistry.resolve(self, self.location) as f:  # type: ignore[arg-type]
                 yield f