Skip to content

Commit fae2363

Browse files
timfeltimfel
committed
[GR-13554] Fixes for Cythonizing a hello_world.pyx
PullRequest: graalpython/384
2 parents c88ed96 + 7cabaf5 commit fae2363

File tree

8 files changed

+116
-41
lines changed

8 files changed

+116
-41
lines changed

‎graalpython/com.oracle.graal.python.test/src/com/oracle/graal/python/test/datatype/StringTests.java

Lines changed: 1 addition & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
/*
2-
* Copyright (c) 2017, 2018, Oracle and/or its affiliates.
2+
* Copyright (c) 2017, 2019, Oracle and/or its affiliates.
33
* Copyright (c) 2013, Regents of the University of California
44
*
55
* All rights reserved.
@@ -88,18 +88,6 @@ public void stringRSplitNoLimit() {
8888
assertPrints("['a', 'b', 'c']\n", source);
8989
}
9090

91-
@Test
92-
public void stringRSplitLimit() {
93-
String source = "print(' a b c '.rsplit(None, 1))\n";
94-
assertPrints("['a b', 'c']\n", source);
95-
}
96-
97-
@Test
98-
public void stringRSplitSep() {
99-
String source = "print('----a---b--c-'.rsplit('-'))\n";
100-
assertPrints("['', '', '', '', 'a', '', '', 'b', '', 'c']\n", source);
101-
}
102-
10391
@Test
10492
public void stringRSplitSepLimit() {
10593
String source = "print('----a---b--c-'.rsplit('-', 5))\n";

‎graalpython/com.oracle.graal.python.test/src/tests/test_range.py

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
# Copyright (c) 2018, Oracle and/or its affiliates. All rights reserved.
1+
# Copyright (c) 2018, 2019, Oracle and/or its affiliates. All rights reserved.
22
# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
33
#
44
# The Universal Permissive License (UPL), Version 1.0
@@ -46,3 +46,5 @@ def test_eq():
4646

4747
# empty ranges are always equal
4848
assert range(10, 20, -30) == range(20, 30, -40)
49+
50+
assert range(True) == range(1)

‎graalpython/com.oracle.graal.python.test/src/tests/test_string.py

Lines changed: 75 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -911,10 +911,10 @@ def test_startswith(self):
911911
self.checkequal(True, 'hello', 'startswith', 'o', -1)
912912
self.checkequal(True, 'hello', 'startswith', '', -3, -3)
913913
self.checkequal(False, 'hello', 'startswith', 'lo', -9)
914-
914+
915915
self.checkraises(TypeError, 'hello', 'startswith')
916916
#self.checkraises(TypeError, 'hello', 'startswith', 42)
917-
917+
918918
# test tuple arguments
919919
self.checkequal(True, 'hello', 'startswith', ('he', 'ha'))
920920
self.checkequal(False, 'hello', 'startswith', ('lo', 'llo'))
@@ -927,11 +927,82 @@ def test_startswith(self):
927927
self.checkequal(True, 'hello', 'startswith', ('lo', 'he'), 0, -1)
928928
self.checkequal(False, 'hello', 'startswith', ('he', 'hel'), 0, 1)
929929
self.checkequal(True, 'hello', 'startswith', ('he', 'hel'), 0, 2)
930-
930+
931931
self.checkraises(TypeError, 'hello', 'startswith', (42,))
932932
self.checkequal(True, 'hello', 'startswith', ('he', 42))
933933
self.checkraises(TypeError, 'hello', 'startswith', ('ne', 42,))
934934

935+
def test_rsplit(self):
936+
# by a char
937+
self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|')
938+
self.checkequal(['a|b|c', 'd'], 'a|b|c|d', 'rsplit', '|', 1)
939+
self.checkequal(['a|b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 2)
940+
self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 3)
941+
self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 4)
942+
self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|',
943+
sys.maxsize-100)
944+
self.checkequal(['a|b|c|d'], 'a|b|c|d', 'rsplit', '|', 0)
945+
self.checkequal(['a||b||c', '', 'd'], 'a||b||c||d', 'rsplit', '|', 2)
946+
self.checkequal(['abcd'], 'abcd', 'rsplit', '|')
947+
self.checkequal([''], '', 'rsplit', '|')
948+
self.checkequal(['', ' begincase'], '| begincase', 'rsplit', '|')
949+
self.checkequal(['endcase ', ''], 'endcase |', 'rsplit', '|')
950+
self.checkequal(['', 'bothcase', ''], '|bothcase|', 'rsplit', '|')
951+
952+
self.checkequal(['a\x00\x00b', 'c', 'd'], 'a\x00\x00b\x00c\x00d', 'rsplit', '\x00', 2)
953+
954+
self.checkequal(['a']*20, ('a|'*20)[:-1], 'rsplit', '|')
955+
self.checkequal(['a|a|a|a|a']+['a']*15,
956+
('a|'*20)[:-1], 'rsplit', '|', 15)
957+
958+
# by string
959+
self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//')
960+
self.checkequal(['a//b//c', 'd'], 'a//b//c//d', 'rsplit', '//', 1)
961+
self.checkequal(['a//b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 2)
962+
self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 3)
963+
self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 4)
964+
self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//',
965+
sys.maxsize-5)
966+
self.checkequal(['a//b//c//d'], 'a//b//c//d', 'rsplit', '//', 0)
967+
self.checkequal(['a////b////c', '', 'd'], 'a////b////c////d', 'rsplit', '//', 2)
968+
self.checkequal(['', ' begincase'], 'test begincase', 'rsplit', 'test')
969+
self.checkequal(['endcase ', ''], 'endcase test', 'rsplit', 'test')
970+
self.checkequal(['', ' bothcase ', ''], 'test bothcase test',
971+
'rsplit', 'test')
972+
self.checkequal(['ab', 'c'], 'abbbc', 'rsplit', 'bb')
973+
self.checkequal(['', ''], 'aaa', 'rsplit', 'aaa')
974+
self.checkequal(['aaa'], 'aaa', 'rsplit', 'aaa', 0)
975+
self.checkequal(['ab', 'ab'], 'abbaab', 'rsplit', 'ba')
976+
self.checkequal(['aaaa'], 'aaaa', 'rsplit', 'aab')
977+
self.checkequal([''], '', 'rsplit', 'aaa')
978+
self.checkequal(['aa'], 'aa', 'rsplit', 'aaa')
979+
self.checkequal(['bbob', 'A'], 'bbobbbobbA', 'rsplit', 'bbobb')
980+
self.checkequal(['', 'B', 'A'], 'bbobbBbbobbA', 'rsplit', 'bbobb')
981+
982+
self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'rsplit', 'BLAH')
983+
self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'rsplit', 'BLAH', 19)
984+
self.checkequal(['aBLAHa'] + ['a']*18, ('aBLAH'*20)[:-4],
985+
'rsplit', 'BLAH', 18)
986+
987+
# with keyword args
988+
self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', sep='|')
989+
self.checkequal(['a|b|c', 'd'],
990+
'a|b|c|d', 'rsplit', '|', maxsplit=1)
991+
self.checkequal(['a|b|c', 'd'],
992+
'a|b|c|d', 'rsplit', sep='|', maxsplit=1)
993+
self.checkequal(['a|b|c', 'd'],
994+
'a|b|c|d', 'rsplit', maxsplit=1, sep='|')
995+
self.checkequal(['a b c', 'd'],
996+
'a b c d', 'rsplit', maxsplit=1)
997+
998+
# argument type
999+
self.checkraises(TypeError, 'hello', 'rsplit', 42, 42, 42)
1000+
1001+
# null case
1002+
self.checkraises(ValueError, 'hello', 'rsplit', '')
1003+
self.checkraises(ValueError, 'hello', 'rsplit', '', 0)
1004+
1005+
9351006
def test_same_id():
9361007
empty_ids = set([id(str()) for i in range(100)])
9371008
assert len(empty_ids) == 1
@@ -957,7 +1028,7 @@ def test_translate_from_byte_table():
9571028
assert "ahoj".translate(table) == "AHOJ"
9581029
assert "ahoj".translate(bytearray(table)) == "AHOJ"
9591030
assert "ahoj".translate(memoryview(table)) == "AHOJ"
960-
1031+
9611032
def test_tranlslate_from_short_table():
9621033
table = b'\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f !"#$%&\'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`ABCDEFGH'
9631034
assert "ahoj".translate(table) == "AHoj"
@@ -983,4 +1054,3 @@ def test_literals():
9831054
assert "hello\[world\]"[6] == "["
9841055
assert "hello\[world\]"[12] == "\\"
9851056
assert "hello\[world\]"[13] == "]"
986-

‎graalpython/com.oracle.graal.python.test/src/tests/test_unicodedata.py

Lines changed: 10 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
# Copyright (c) 2018, Oracle and/or its affiliates. All rights reserved.
1+
# Copyright (c) 2018, 2019, Oracle and/or its affiliates. All rights reserved.
22
# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
33
#
44
# The Universal Permissive License (UPL), Version 1.0
@@ -52,3 +52,12 @@ def test_normalize():
5252
assert_raises(TypeError, unicodedata.normalize)
5353
assert_raises(ValueError, unicodedata.normalize, 'unknown', 'xx')
5454
assert unicodedata.normalize('NFKC', '') == ''
55+
56+
57+
def test_category():
58+
import unicodedata
59+
assert unicodedata.category('\uFFFE') == 'Cn'
60+
assert unicodedata.category('a') == 'Ll'
61+
assert unicodedata.category('A') == 'Lu'
62+
assert_raises(TypeError, unicodedata.category)
63+
assert_raises(TypeError, unicodedata.category, 'xx')

‎graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/modules/BuiltinConstructors.java

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1401,6 +1401,7 @@ private Object callNativeGenericNewNode(PythonNativeClass self, Object[] varargs
14011401
@Builtin(name = RANGE, minNumOfPositionalArgs = 2, maxNumOfPositionalArgs = 4, constructsClass = PythonBuiltinClassType.PRange)
14021402
@GenerateNodeFactory
14031403
@SuppressWarnings("unused")
1404+
@TypeSystemReference(PythonArithmeticTypes.class)
14041405
public abstract static class RangeNode extends PythonBuiltinNode {
14051406

14061407
@Specialization(guards = "caseStop(start,step)")

‎graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/objects/str/StringBuiltins.java

Lines changed: 14 additions & 19 deletions
Original file line numberDiff line numberDiff line change
@@ -1174,7 +1174,7 @@ private PList splitfields(String s, int maxsplit) {
11741174
}
11751175

11761176
// str.split
1177-
@Builtin(name = "rsplit", maxNumOfPositionalArgs = 3)
1177+
@Builtin(name = "rsplit", fixedNumOfPositionalArgs = 1, keywordArguments = {"sep", "maxsplit"})
11781178
@GenerateNodeFactory
11791179
@TypeSystemReference(PythonArithmeticTypes.class)
11801180
public abstract static class RSplitNode extends SplitBaseNode {
@@ -1185,41 +1185,35 @@ public PList doSplit(String self, PNone sep, PNone maxsplit) {
11851185
return rsplitfields(self, -1);
11861186
}
11871187

1188-
@SuppressWarnings("unused")
1189-
@TruffleBoundary
11901188
@Specialization
1191-
public PList doSplit(String self, String sep, PNone maxsplit) {
1192-
PList list = factory().createList();
1193-
String[] strs = self.split(Pattern.quote(sep));
1194-
for (String s : strs) {
1195-
getAppendNode().execute(list, s);
1196-
}
1197-
return list;
1189+
public PList doSplit(String self, String sep, @SuppressWarnings("unused") PNone maxsplit) {
1190+
return doSplit(self, sep, Integer.MAX_VALUE);
11981191
}
11991192

12001193
@Specialization
12011194
public PList doSplit(String self, String sep, int maxsplit) {
1195+
if (sep.length() == 0) {
1196+
throw raise(ValueError, "empty separator");
1197+
}
12021198
PList list = factory().createList();
12031199
int splits = 0;
12041200
int end = self.length();
12051201
String remainder = self;
1202+
int sepLength = sep.length();
12061203
while (splits < maxsplit) {
12071204
int idx = remainder.lastIndexOf(sep);
12081205

12091206
if (idx < 0) {
12101207
break;
12111208
}
12121209

1213-
getAppendNode().execute(list, self.substring(idx + 1, end));
1210+
getAppendNode().execute(list, self.substring(idx + sepLength, end));
12141211
end = idx;
12151212
splits++;
12161213
remainder = remainder.substring(0, end);
12171214
}
12181215

1219-
if (!remainder.isEmpty()) {
1220-
getAppendNode().execute(list, remainder);
1221-
}
1222-
1216+
getAppendNode().execute(list, remainder);
12231217
getReverseNode().execute(list);
12241218
return list;
12251219
}
@@ -1276,18 +1270,19 @@ private PList rsplitfields(String s, int maxsplit) {
12761270
// The next segment runs up to the next next whitespace or end
12771271
for (index = end; index >= 0; index--) {
12781272
if (isWhitespace(s.codePointAt(index))) {
1279-
// Break leaving index pointing at whitespace
1273+
// Break leaving index pointing after the found whitespace
1274+
index++;
12801275
break;
12811276
}
12821277
}
12831278
}
12841279

12851280
// Make a piece from start up to index
1286-
getAppendNode().execute(list, s.substring(index + 1, end + 1));
1281+
getAppendNode().execute(list, s.substring(index, end + 1));
12871282
splits++;
12881283

1289-
// Start next segment search at that point
1290-
end = index;
1284+
// Start next segment search at the whitespace
1285+
end = index - 1;
12911286
}
12921287

12931288
getReverseNode().execute(list);

‎graalpython/lib-graalpython/modules/ginstall.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -60,7 +60,7 @@ def six(*args):
6060
install_from_pypi("six==1.12.0", args)
6161

6262
def Cython(*args):
63-
install_from_pypi("Cython==0.29.2", ['--no-cython-compile'] + args)
63+
install_from_pypi("Cython==0.29.2", ('--no-cython-compile',) + args)
6464

6565
def setuptools(*args):
6666
install_from_pypi("setuptools==40.6.3", args)

‎graalpython/lib-graalpython/unicodedata.py

Lines changed: 11 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
# Copyright (c) 2018, Oracle and/or its affiliates. All rights reserved.
1+
# Copyright (c) 2018, 2019, Oracle and/or its affiliates. All rights reserved.
22
# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
33
#
44
# The Universal Permissive License (UPL), Version 1.0
@@ -41,6 +41,7 @@
4141

4242
c_unicodedata_module = None
4343

44+
4445
@__builtin__
4546
def __get_c_unicodedata():
4647
if c_unicodedata_module is None:
@@ -52,3 +53,12 @@ def __get_c_unicodedata():
5253

5354
east_asian_width = lambda arg: __get_c_unicodedata().east_asian_width(arg)
5455

56+
57+
@__builtin__
58+
def lookup(name):
59+
return __get_c_unicodedata().lookup(name)
60+
61+
62+
@__builtin__
63+
def category(char):
64+
return __get_c_unicodedata().category(char)

0 commit comments

Comments
 (0)