Skip to content

Commit 37fd030

Browse files
bschoenmaeckersbschoenmaeckers
committed
Add helper
1 parent 8a845b8 commit 37fd030

1 file changed

Lines changed: 25 additions & 68 deletions

File tree

‎crates/capi/src/unicodeobject.rs‎

Lines changed: 25 additions & 68 deletions
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@ use core::ffi::{CStr, c_char, c_int};
55
use core::ptr::NonNull;
66
use core::slice;
77
use core::str;
8-
use rustpython_vm::builtins::{PyStr, PyStrRef};
8+
use rustpython_vm::builtins::{PyBytesRef, PyStr, PyStrRef, PyUtf8StrRef};
99
use rustpython_vm::common::wtf8::{CodePoint, Wtf8Buf};
1010
use rustpython_vm::convert::ToPyObject;
1111
use rustpython_vm::{AsObject, PyObjectRef, PyResult, VirtualMachine};
@@ -93,78 +93,50 @@ pub unsafe extern "C" fn PyUnicode_AsUTF8AndSize(
9393
})
9494
}
9595

96+
fn encode_unicode(
97+
vm: &VirtualMachine,
98+
unicode: *mut PyObject,
99+
encoding: &str,
100+
errors: Option<PyUtf8StrRef>,
101+
) -> PyResult<PyBytesRef> {
102+
let unicode = unsafe { &*unicode }
103+
.try_downcast_ref::<PyStr>(vm)?
104+
.to_owned();
105+
vm.state
106+
.codec_registry
107+
.encode_text(unicode, encoding, errors, vm)
108+
}
109+
96110
#[unsafe(no_mangle)]
97111
pub unsafe extern "C" fn PyUnicode_AsASCIIString(unicode: *mut PyObject) -> *mut PyObject {
98-
with_vm(|vm| {
99-
let unicode = unsafe { &*unicode }
100-
.try_downcast_ref::<PyStr>(vm)?
101-
.to_owned();
102-
vm.state
103-
.codec_registry
104-
.encode_text(unicode, "ascii", None, vm)
105-
})
112+
with_vm(|vm| encode_unicode(vm, unicode, "ascii", None))
106113
}
107114

108115
#[unsafe(no_mangle)]
109116
pub unsafe extern "C" fn PyUnicode_AsLatin1String(unicode: *mut PyObject) -> *mut PyObject {
110-
with_vm(|vm| {
111-
let unicode = unsafe { &*unicode }
112-
.try_downcast_ref::<PyStr>(vm)?
113-
.to_owned();
114-
vm.state
115-
.codec_registry
116-
.encode_text(unicode, "latin-1", None, vm)
117-
})
117+
with_vm(|vm| encode_unicode(vm, unicode, "latin-1", None))
118118
}
119119

120120
#[unsafe(no_mangle)]
121121
pub unsafe extern "C" fn PyUnicode_AsRawUnicodeEscapeString(
122122
unicode: *mut PyObject,
123123
) -> *mut PyObject {
124-
with_vm(|vm| {
125-
let unicode = unsafe { &*unicode }
126-
.try_downcast_ref::<PyStr>(vm)?
127-
.to_owned();
128-
vm.state
129-
.codec_registry
130-
.encode_text(unicode, "raw-unicode-escape", None, vm)
131-
})
124+
with_vm(|vm| encode_unicode(vm, unicode, "raw-unicode-escape", None))
132125
}
133126

134127
#[unsafe(no_mangle)]
135128
pub unsafe extern "C" fn PyUnicode_AsUTF16String(unicode: *mut PyObject) -> *mut PyObject {
136-
with_vm(|vm| {
137-
let unicode = unsafe { &*unicode }
138-
.try_downcast_ref::<PyStr>(vm)?
139-
.to_owned();
140-
vm.state
141-
.codec_registry
142-
.encode_text(unicode, "utf-16", None, vm)
143-
})
129+
with_vm(|vm| encode_unicode(vm, unicode, "utf-16", None))
144130
}
145131

146132
#[unsafe(no_mangle)]
147133
pub unsafe extern "C" fn PyUnicode_AsUTF32String(unicode: *mut PyObject) -> *mut PyObject {
148-
with_vm(|vm| {
149-
let unicode = unsafe { &*unicode }
150-
.try_downcast_ref::<PyStr>(vm)?
151-
.to_owned();
152-
vm.state
153-
.codec_registry
154-
.encode_text(unicode, "utf-32", None, vm)
155-
})
134+
with_vm(|vm| encode_unicode(vm, unicode, "utf-32", None))
156135
}
157136

158137
#[unsafe(no_mangle)]
159138
pub unsafe extern "C" fn PyUnicode_AsUnicodeEscapeString(unicode: *mut PyObject) -> *mut PyObject {
160-
with_vm(|vm| {
161-
let unicode = unsafe { &*unicode }
162-
.try_downcast_ref::<PyStr>(vm)?
163-
.to_owned();
164-
vm.state
165-
.codec_registry
166-
.encode_text(unicode, "unicode-escape", None, vm)
167-
})
139+
with_vm(|vm| encode_unicode(vm, unicode, "unicode-escape", None))
168140
}
169141

170142
#[unsafe(no_mangle)]
@@ -174,28 +146,16 @@ pub unsafe extern "C" fn PyUnicode_AsEncodedString(
174146
errors: *const c_char,
175147
) -> *mut PyObject {
176148
with_vm(|vm| {
177-
let unicode = unsafe { &*unicode }
178-
.try_downcast_ref::<PyStr>(vm)?
179-
.to_owned();
180149
let encoding = unsafe { encoding.try_as_str_opt(vm) }?.unwrap_or("utf-8");
181150
let errors =
182151
unsafe { errors.try_as_str_opt(vm) }?.map(|errors| vm.ctx.new_utf8_str(errors));
183-
vm.state
184-
.codec_registry
185-
.encode_text(unicode, encoding, errors, vm)
152+
encode_unicode(vm, unicode, encoding, errors)
186153
})
187154
}
188155

189156
#[unsafe(no_mangle)]
190157
pub unsafe extern "C" fn PyUnicode_AsUTF8String(unicode: *mut PyObject) -> *mut PyObject {
191-
with_vm(|vm| {
192-
let unicode = unsafe { &*unicode }
193-
.try_downcast_ref::<PyStr>(vm)?
194-
.to_owned();
195-
vm.state
196-
.codec_registry
197-
.encode_text(unicode, "utf-8", None, vm)
198-
})
158+
with_vm(|vm| encode_unicode(vm, unicode, "utf-8", None))
199159
}
200160

201161
#[unsafe(no_mangle)]
@@ -414,14 +374,11 @@ pub(crate) fn decode_fsdefault_and_size(
414374
#[unsafe(no_mangle)]
415375
pub unsafe extern "C" fn PyUnicode_EncodeFSDefault(unicode: *mut PyObject) -> *mut PyObject {
416376
with_vm(|vm| {
417-
let unicode = unsafe { &*unicode }
418-
.try_downcast_ref::<PyStr>(vm)?
419-
.to_owned();
420-
vm.state.codec_registry.encode_text(
377+
encode_unicode(
378+
vm,
421379
unicode,
422380
vm.fs_encoding().as_str(),
423381
Some(vm.fs_encode_errors().to_owned()),
424-
vm,
425382
)
426383
})
427384
}

0 commit comments

Comments
 (0)