Implemented int(str) in UNIX

9 files changed, 248 insertions, 4 deletions

diff --git a/py/obj.c b/py/obj.c
index 81b5c69f7a..73ffe5281f 100644
--- a/py/obj.c
+++ b/py/obj.c
@@ -13,6 +13,7 @@
 #include "runtime0.h"
 #include "runtime.h"
 #include "map.h"
+#include "strtonum.h"
 
 mp_obj_t mp_obj_get_type(mp_obj_t o_in) {
     if (MP_OBJ_IS_SMALL_INT(o_in)) {
@@ -136,6 +137,12 @@ bool mp_obj_less(mp_obj_t o1, mp_obj_t o2) {
 }
 
 machine_int_t mp_obj_get_int(mp_obj_t arg) {
+    return mp_obj_get_int_base(arg, 0);
+}
+
+machine_int_t mp_obj_get_int_base(mp_obj_t arg, mp_obj_t base_arg) {
+    const char *value;
+    int base;
     if (arg == mp_const_false) {
         return 0;
     } else if (arg == mp_const_true) {
@@ -147,6 +154,17 @@ machine_int_t mp_obj_get_int(mp_obj_t arg) {
         // TODO work out if this should be floor, ceil or trunc
         return (machine_int_t)mp_obj_float_get(arg);
 #endif
+    } else if (MP_OBJ_IS_TYPE(arg, &str_type)) {
+        if (base_arg == 0) {
+            value = qstr_str(mp_obj_str_get(arg));
+            return (machine_int_t)strtonum(value, 0);
+        } else if (MP_OBJ_IS_TYPE(base_arg, &int_type)) {
+            base = MP_OBJ_SMALL_INT_VALUE(base_arg);
+            value = qstr_str(mp_obj_str_get(arg));
+            return (machine_int_t)strtonum(value, base);
+        } else {
+            nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "an integer is required"));
+        }
     } else {
         nlr_jump(mp_obj_new_exception_msg_varg(MP_QSTR_TypeError, "can't convert %s to int", mp_obj_get_type_str(arg)));
     }
diff --git a/py/obj.h b/py/obj.h
index b92f1e2a7e..381fcc1f25 100644
--- a/py/obj.h
+++ b/py/obj.h
@@ -235,6 +235,7 @@ bool mp_obj_equal(mp_obj_t o1, mp_obj_t o2);
 bool mp_obj_less(mp_obj_t o1, mp_obj_t o2);
 
 machine_int_t mp_obj_get_int(mp_obj_t arg);
+machine_int_t mp_obj_get_int_base(mp_obj_t arg, mp_obj_t base_arg);
 #if MICROPY_ENABLE_FLOAT
 mp_float_t mp_obj_get_float(mp_obj_t self_in);
 void mp_obj_get_complex(mp_obj_t self_in, mp_float_t *real, mp_float_t *imag);
diff --git a/py/objint.c b/py/objint.c
index 9cd5ebae29..58fc37b3bb 100644
--- a/py/objint.c
+++ b/py/objint.c
@@ -22,8 +22,9 @@ static mp_obj_t int_make_new(mp_obj_t type_in, int n_args, const mp_obj_t *args)
             // TODO allow string as arg and parse it
             return MP_OBJ_NEW_SMALL_INT(mp_obj_get_int(args[0]));
 
-        //case 2:
-            // TODO, parse with given base
+        case 2:
+            // TODO make args[0] and args[1] correct
+            return MP_OBJ_NEW_SMALL_INT(mp_obj_get_int_base(args[0], args[1]));
 
         default:
             nlr_jump(mp_obj_new_exception_msg_1_arg(MP_QSTR_TypeError, "int takes at most 2 arguments, %d given", (void*)(machine_int_t)n_args));
diff --git a/py/py.mk b/py/py.mk
index a44a8bad02..c65fb5e65a 100644
--- a/py/py.mk
+++ b/py/py.mk
@@ -70,6 +70,7 @@ PY_O_BASENAME = \
 	emitinlinethumb.o \
 	runtime.o \
 	map.o \
+	strtonum.o \
 	obj.o \
 	objbool.o \
 	objboundmeth.o \
diff --git a/py/strtonum.c b/py/strtonum.c
new file mode 100644
index 0000000000..7cc45bcba4
--- /dev/null
+++ b/py/strtonum.c
@@ -0,0 +1,168 @@
+#if defined(UNIX)
+
+#include <ctype.h>
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "misc.h"
+#include "mpconfig.h"
+#include "mpqstr.h"
+#include "nlr.h"
+#include "obj.h"
+
+
+long strtonum(const char *restrict s, int base) {
+    int c, neg = 0;
+    const char *p = s;
+    char *num;
+    long found;
+
+    // check radix base
+    if ((base != 0 && base < 2) || base > 36) {
+        printf("ValueError: int() arg 2 must be >=2 and <= 36\n");
+        return 0;
+    }
+    // skip surrounded whitespace
+    while (isspace((c = *(p++))));
+    if (c == 0) {
+        goto value_error;
+    }
+    // preced sign
+    if (c == '+' || c == '-') {
+        neg = - (c == '-');
+        c = *(p++);
+    }
+
+    // find real radix base, and strip preced '0x', '0o' and '0b'
+    if ((base == 0 || base == 16) && c == '0') {
+        c = *(p++);
+        if ((c | 32) == 'x') {
+            base = 16;
+        } else if (base == 0 && (c | 32) == 'o') {
+            base = 8;
+        } else if (base == 0 && (c | 32) == 'b') {
+            base = 2;
+        } else {
+            base = 10;
+            p -= 2;
+        }
+    } else if (base == 8 && c == '0') {
+        if ((c | 32) == 'o') {
+            base = 8;
+        } else {
+            goto value_error;
+        }
+    } else if (base == 2 && c == '0') {
+        if ((c | 32) == 'b') {
+            base = 2;
+        } else {
+            goto value_error;
+        }
+    } else {
+        if (base == 0) base = 10;
+        p--;
+    }
+
+    found = strtol(p, &num, base);
+    if (errno) {
+        goto value_error;
+    } else if (found && *(num) == 0) {
+        goto done;
+    } else if (found || num != p) {
+        goto check_tail_space;
+    } else {
+        goto value_error;
+    }
+
+check_tail_space:
+    if (*(num) != 0) {
+        while (isspace((c = *(num++))));
+        if (c != 0) {
+            goto value_error;
+        }
+    }
+
+done:
+    return (found ^ neg) - neg;
+
+value_error:
+    nlr_jump(mp_obj_new_exception_msg_2_args(MP_QSTR_ValueError, "invalid literal for int() with base %d: '%s'", (void*)(machine_uint_t)base, s));
+    //printf("invalid literal for int() with base %d: '%s'\n", base, s);
+    //return 0;
+}
+
+/*
+int main(void) {
+    printf("  0 %ld\n", strtonum("0", 0));
+    printf("  1 %ld\n", strtonum("1", 0));
+    printf("  1 %ld\n", strtonum("+1", 0));
+    printf(" -1 %ld\n", strtonum("-1", 0));
+    printf("  0 %ld\n", strtonum("+0", 0));
+    printf("  0 %ld\n", strtonum("-0", 0));
+    printf("  1 %ld\n", strtonum("01", 0));
+    printf("  9 %ld\n", strtonum("9", 0));
+    printf(" 10 %ld\n", strtonum("10", 0));
+    printf(" 10 %ld\n", strtonum("+10", 0));
+    printf("-10 %ld\n", strtonum("-10", 0));
+    printf(" 12 %ld\n", strtonum("12", 0));
+    printf("-12 %ld\n", strtonum("-12", 0));
+    printf(" 99 %ld\n", strtonum("99", 0));
+    printf("100 %ld\n", strtonum("100", 0));
+    printf("314 %ld\n", strtonum("314", 0));
+    printf("314 %ld\n", strtonum(" 314", 0));
+    printf("314 %ld\n", strtonum("314 ", 0));
+    printf("314 %ld\n", strtonum("  \t\t 314 \t\t  ", 0));
+    printf("  1 %ld\n", strtonum(" 1 ", 0));
+    printf(" -3 %ld\n", strtonum(" -3 ", 0));
+    printf(" 83 %ld\n", strtonum("0o123", 0));
+    printf(" 18 %ld\n", strtonum("0x12", 0));
+    printf(" 10 %ld\n", strtonum("0o12", 0));
+    printf(" 10 %ld\n", strtonum("0o12", 0));
+    printf(" 10 %ld\n", strtonum("  \t 0o12", 0));
+    printf(" 10 %ld\n", strtonum("0o12   \t  ", 0));
+    printf(" 10 %ld\n", strtonum("0O12", 0));
+    printf("  5 %ld\n", strtonum("0b101", 0));
+    printf("  5 %ld\n", strtonum("0b101", 0));
+
+    printf("  0 %ld\n", strtonum("0", 10));
+    printf("  1 %ld\n", strtonum("1", 10));
+    printf("  1 %ld\n", strtonum(" \t 1 \t ", 10));
+    printf(" 11 %ld\n", strtonum("11", 10));
+    printf(" 17 %ld\n", strtonum("11", 16));
+    printf("  9 %ld\n", strtonum("11", 8));
+    printf("  3 %ld\n", strtonum("11", 2));
+    printf(" 37 %ld\n", strtonum("11", 36));
+    printf("4294967296 %ld\n", strtonum("4294967296", 0));
+
+    printf("%ld\n", strtonum("x", 0));
+    printf("%ld\n", strtonum("1x", 0));
+    printf("%ld\n", strtonum("  1x", 0));
+    printf("%ld\n", strtonum("  1\02  ", 0));
+    printf("%ld\n", strtonum("", 0));
+    printf("%ld\n", strtonum(" ", 0));
+    printf("%ld\n", strtonum(" \t\t  ", 0));
+    printf("%ld\n", strtonum("0x", 0));
+    printf("%ld\n", strtonum("0x", 16));
+    printf("%ld\n", strtonum("0o", 0));
+    printf("%ld\n", strtonum("0o", 8));
+    printf("%ld\n", strtonum("0b", 0));
+    printf("%ld\n", strtonum("0b", 2));
+    printf("%ld\n", strtonum("0b2", 2));
+    printf("%ld\n", strtonum("0o8", 8));
+    printf("%ld\n", strtonum("0xg", 16));
+
+    printf("%ld\n", strtonum("1 1", 0));
+    printf("%ld\n", strtonum("18446744073709551616", 0));
+}
+*/
+
+#else /* defined(UNIX) */
+
+long strtonum(const char *restrict s, int base) {
+    // TODO port strtol to stm
+    return 0;
+}
+
+#endif /* defined(UNIX) */
diff --git a/py/strtonum.h b/py/strtonum.h
new file mode 100644
index 0000000000..03565cca14
--- /dev/null
+++ b/py/strtonum.h
@@ -0,0 +1 @@
+long strtonum(const char *restrict s, int base);
diff --git a/tests/basics/tests/int1.py b/tests/basics/tests/int1.py
new file mode 100644
index 0000000000..d766216d44
--- /dev/null
+++ b/tests/basics/tests/int1.py
@@ -0,0 +1,54 @@
+print(int(0))
+print(int(1))
+print(int(+1))
+print(int(-1))
+
+print(int('0'))
+print(int('+0'))
+print(int('-0'))
+print(int('1'))
+print(int('+1'))
+print(int('-1'))
+print(int('9'))
+print(int('10'))
+print(int('+10'))
+print(int('-10'))
+print(int('99'))
+print(int('100'))
+print(int('314'))
+print(int(' 314'))
+print(int('314 '))
+print(int('  \t\t  314  \t\t  '))
+print(int('  1  '))
+print(int(' -3 '))
+
+print(int('10', 16))
+print(int('0o123', 0))
+print(int('0x123', 16))
+print(int('0X123', 16))
+print(int('0O123', 8))
+print(int('0B100', 2))
+
+
+def test(value, base):
+    try:
+        print(int(value, base))
+    except ValueError:
+        print('ValueError')
+
+
+test('  1x', 0)
+test('  1\02  ', 0)
+test('', 0)
+test(' ', 0)
+test('  \t\t  ', 0)
+test("\u0200", 0)
+test('0x', 16)
+test('0x', 0)
+test('0o', 8)
+test('0o', 0)
+test('0b', 2)
+test('0b', 0)
+test('0b2', 2)
+test('0o8', 8)
+test('0xg', 16)
diff --git a/unix-cpy/Makefile b/unix-cpy/Makefile
index 90e1f733d5..d5397dac9a 100644
--- a/unix-cpy/Makefile
+++ b/unix-cpy/Makefile
@@ -11,7 +11,7 @@ ECHO = @echo
 
 # compiler settings
 CC = gcc
-CFLAGS = -I. -I$(PY_SRC) -Wall -Werror -ansi -std=gnu99 -Os #-DNDEBUG
+CFLAGS = -I. -I$(PY_SRC) -Wall -Werror -ansi -std=gnu99 -Os -DUNIX #-DNDEBUG
 LDFLAGS = -lm
 
 # source files
diff --git a/unix/Makefile b/unix/Makefile
index e9bae4768e..9af86b9855 100644
--- a/unix/Makefile
+++ b/unix/Makefile
@@ -11,7 +11,7 @@ ECHO = @echo
 
 # compiler settings
 CC = gcc
-CFLAGS = -I. -I$(PY_SRC) -Wall -Werror -ansi -std=gnu99 -Os #-DNDEBUG
+CFLAGS = -I. -I$(PY_SRC) -Wall -Werror -ansi -std=gnu99 -Os -DUNIX #-DNDEBUG
 LDFLAGS = -lm
 
 # source files