feat: worked out how to handle files between C and Python

Author: ntamas

poetry.lock

@@ -15,7 +15,7 @@ pytest = "^7.2.1"
 pytest-cov = "^4.0.0"
 richbench = "^1.0.3"
 igraph = "^0.10.6"
-stimulus = { git = "https://github.com/igraph/stimulus.git", tag = "0.14.1" }
+stimulus = { git = "https://github.com/igraph/stimulus.git", tag = "0.15.0" }
 
 [tool.poetry.group.doc.dependencies]
 mkdocs-material = "^9.1.21"

pyproject.toml

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+from contextlib import ExitStack
 from ctypes import c_char_p, c_int
 from typing import Any, Iterable, Optional, TYPE_CHECKING
 
@@ -10,6 +11,7 @@ from .types import (
     BoolArray,
     EdgeLike,
     EdgeSelector,
+    FileLike,
     IntArray,
     MatrixLike,
     MatrixIntLike,

src/codegen/internal_functions.py.in

@@ -1,12 +1,14 @@
 # fmt: off
 
-from ctypes import cdll, c_char_p, c_double, c_int, c_size_t, c_void_p, POINTER
+from ctypes import cdll, c_char_p, c_double, c_int, c_size_t, c_void_p, CDLL, POINTER
 from ctypes.util import find_library
+from platform import system
 from typing import Any
 
 from .errors import handle_igraph_error_t
 from .types import (
     FILE,
+    FilePtr,
     igraph_attribute_combination_t,
     igraph_attribute_table_t,
     igraph_bool_t,
@@ -59,8 +61,35 @@ def _load_igraph_c_library():
     return lib
 
 
+def _load_libc():
+    """Imports the C standard library using `ctypes`."""
+    if system() == "Windows":
+        return CDLL("msvcrt.dll", use_errno=True)
+    elif system() == "Darwin":
+        return CDLL("libc.dylib", use_errno=True)
+    elif system() == "Linux":
+        return CDLL("libc.so.6", use_errno=True)
+    else:
+        raise RuntimeError("Cannot import C standard library on this platform")
+
+
+_libc: Any = _load_libc()
 _lib: Any = _load_igraph_c_library()
 
+# Standard libc functions
+
+fclose = _libc.fclose
+fclose.restype = int
+fclose.argtypes = [POINTER(FILE)]
+
+fflush = _libc.fflush
+fflush.restype = int
+fflush.argtypes = [POINTER(FILE)]
+
+fdopen = _libc.fdopen
+fdopen.restype = POINTER(FILE)
+fdopen.argtypes = [c_int, c_char_p]
+
 # Vector type
 
 igraph_vector_init = _lib.igraph_vector_init

src/codegen/internal_lib.py.in

@@ -233,6 +233,20 @@ BIPARTITE_TYPES:
     INOUT: "%C% = iterable_to_igraph_vector_bool_t(%I%) if %I% is not None else None"
     OUT: "%C% = _VectorBool.create(0)"
 
+INFILE:
+  PY_TYPE: FileLike
+  PY_RETURN_TYPE: c_int
+  INCONV:
+    IN: '%C% = %S%.enter_context(any_to_file_ptr(%I%, "r"))'
+  FLAGS: stack
+
+OUTFILE:
+  PY_TYPE: FileLike
+  PY_RETURN_TYPE: c_int
+  INCONV:
+    IN: '%C% = %S%.enter_context(any_to_file_ptr(%I%, "w"))'
+  FLAGS: stack
+
 ADJACENCY_MODE:
   PY_TYPE: AdjacencyMode
   INCONV:

src/codegen/types.yaml

@@ -4,11 +4,15 @@
 
 import numpy as np
 
-from ctypes import addressof, memmove, POINTER
-from typing import Any, Iterable, Optional, Sequence, TYPE_CHECKING
+from contextlib import contextmanager
+from ctypes import addressof, get_errno, memmove, POINTER
+from os import strerror
+from typing import Any, IO, Iterable, Iterator, Optional, Sequence, TYPE_CHECKING
 
 from .enums import MatrixStorage
 from .lib import (
+    fdopen,
+    fflush,
     igraph_es_all,
     igraph_es_none,
     igraph_es_vector_copy,
@@ -60,6 +64,7 @@
     BoolArray,
     EdgeLike,
     EdgeSelector,
+    FilePtr,
     IntArray,
     MatrixLike,
     MatrixIntLike,
@@ -86,6 +91,7 @@
 
 
 __all__ = (
+    "any_to_file_ptr",
     "any_to_igraph_bool_t",
     "bytes_to_str",
     "edgelike_to_igraph_integer_t",
@@ -137,6 +143,67 @@
 )
 
 
+@contextmanager
+def any_to_file_ptr(obj: Any, mode: str) -> Iterator[Optional[FilePtr]]:
+    """Converts an arbitrary Python object to an open file pointer in the C
+    layer, using the following rules:
+
+    - ``None`` is returned as is.
+
+    - Integers are treated as file handles and a low-level ``fdopen()`` call
+      from the C standard library will be used to convert them into a ``FILE*``
+      pointer.
+
+    - File-like objects with a ``fileno()`` method will be converted into a
+      file handle and then they will be treated the same way as integers above.
+
+    - Anything else is forwarded to ``open()`` to convert them into a file-like
+      object. They will then be treated as any other file-like object. The
+      created object will be _closed_ automatically when the context manager
+      exits.
+    """
+    if obj is None:
+        yield None
+        return
+
+    handle: int
+    fp: IO[Any] | None = None
+    file_ptr: Optional[FilePtr] = None
+
+    if isinstance(obj, int):
+        handle = obj
+    elif hasattr(obj, "fileno") and callable(obj.fileno):
+        # Flush pending writes first to ensure that they do not get mixed up
+        # with the ones performed by igraph's C core
+        if hasattr(obj, "flush") and callable(obj.flush):
+            obj.flush()
+        handle = obj.fileno()
+    else:
+        fp = open(obj, mode)
+        try:
+            if hasattr(fp, "fileno") and callable(fp.fileno):
+                handle = fp.fileno()
+            else:
+                raise TypeError("open() returned an object without a file handle")
+        except Exception:
+            fp.close()
+            fp = None
+            raise
+
+    try:
+        file_ptr = fdopen(
+            handle, mode.encode("ascii") if isinstance(mode, str) else mode
+        )
+        if not file_ptr:
+            errno = get_errno()
+            raise OSError(errno, strerror(errno))
+        yield file_ptr
+        fflush(file_ptr)
+    finally:
+        if fp is not None:
+            fp.close()  # takes care of fclose() on file_ptr
+
+
 def any_to_igraph_bool_t(obj: Any) -> igraph_bool_t:
     """Converts an arbitrary Python object to an igraph boolean by taking its
     truth value.