blob: 03a761863f656096c8b02036e4393c5b5d8010cc [file] [log] [blame]
Damiend99b0522013-12-21 18:17:45 +00001#include <stdlib.h>
2#include <stdint.h>
3#include <stdarg.h>
4#include <string.h>
5#include <assert.h>
6
7#include "nlr.h"
8#include "misc.h"
9#include "mpconfig.h"
Damien Georgeeb7bfcb2014-01-04 15:57:35 +000010#include "mpqstr.h"
Damiend99b0522013-12-21 18:17:45 +000011#include "obj.h"
12#include "runtime0.h"
13#include "runtime.h"
14
15typedef struct _mp_obj_str_t {
16 mp_obj_base_t base;
17 qstr qstr;
18} mp_obj_str_t;
19
20void str_print(void (*print)(void *env, const char *fmt, ...), void *env, mp_obj_t self_in) {
21 mp_obj_str_t *self = self_in;
22 // TODO need to escape chars etc
23 print(env, "'%s'", qstr_str(self->qstr));
24}
25
26mp_obj_t str_binary_op(int op, mp_obj_t lhs_in, mp_obj_t rhs_in) {
27 mp_obj_str_t *lhs = lhs_in;
28 const char *lhs_str = qstr_str(lhs->qstr);
29 switch (op) {
30 case RT_BINARY_OP_SUBSCR:
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020031 // TODO: need predicate to check for int-like type (bools are such for example)
32 // ["no", "yes"][1 == 2] is common idiom
33 if (MP_OBJ_IS_SMALL_INT(rhs_in)) {
34 // TODO: This implements byte string access for single index so far
Paul Sokolovskyf8b9d3c2014-01-04 01:38:26 +020035 // TODO: Handle negative indexes.
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020036 return mp_obj_new_int(lhs_str[mp_obj_get_int(rhs_in)]);
Paul Sokolovskye606cb62014-01-04 01:34:23 +020037#if MICROPY_ENABLE_SLICE
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020038 } else if (MP_OBJ_IS_TYPE(rhs_in, &slice_type)) {
Damien Georgec8d13842014-01-04 01:06:10 +000039 machine_int_t start, stop, step;
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020040 mp_obj_slice_get(rhs_in, &start, &stop, &step);
41 assert(step == 1);
Paul Sokolovskydecad082014-01-03 23:36:56 +020042 int len = strlen(lhs_str);
43 if (start < 0) {
44 start = len + start;
Paul Sokolovsky6ee1e382014-01-04 03:47:34 +020045 if (start < 0) {
46 start = 0;
47 }
48 } else if (start > len) {
49 start = len;
Paul Sokolovskydecad082014-01-03 23:36:56 +020050 }
51 if (stop <= 0) {
52 stop = len + stop;
Paul Sokolovsky6ee1e382014-01-04 03:47:34 +020053 // CPython returns empty string in such case
54 if (stop < 0) {
55 stop = start;
56 }
57 } else if (stop > len) {
58 stop = len;
Paul Sokolovskydecad082014-01-03 23:36:56 +020059 }
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020060 return mp_obj_new_str(qstr_from_strn_copy(lhs_str + start, stop - start));
Paul Sokolovskye606cb62014-01-04 01:34:23 +020061#endif
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020062 } else {
Paul Sokolovskyf8b9d3c2014-01-04 01:38:26 +020063 // Message doesn't match CPython, but we don't have so much bytes as they
64 // to spend them on verbose wording
Damien Georgeeb7bfcb2014-01-04 15:57:35 +000065 nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "index must be int"));
Paul Sokolovsky31ba60f2014-01-03 02:51:16 +020066 }
Damiend99b0522013-12-21 18:17:45 +000067
68 case RT_BINARY_OP_ADD:
69 case RT_BINARY_OP_INPLACE_ADD:
70 if (MP_OBJ_IS_TYPE(rhs_in, &str_type)) {
71 // add 2 strings
72 const char *rhs_str = qstr_str(((mp_obj_str_t*)rhs_in)->qstr);
Damien Georgefe8fb912014-01-02 16:36:09 +000073 size_t lhs_len = strlen(lhs_str);
74 size_t rhs_len = strlen(rhs_str);
75 int alloc_len = lhs_len + rhs_len + 1;
Damien732407f2013-12-29 19:33:23 +000076 char *val = m_new(char, alloc_len);
Damien Georgefe8fb912014-01-02 16:36:09 +000077 memcpy(val, lhs_str, lhs_len);
78 memcpy(val + lhs_len, rhs_str, rhs_len);
79 val[lhs_len + rhs_len] = '\0';
Damien732407f2013-12-29 19:33:23 +000080 return mp_obj_new_str(qstr_from_str_take(val, alloc_len));
Damiend99b0522013-12-21 18:17:45 +000081 }
82 break;
83 }
84
85 return MP_OBJ_NULL; // op not supported
86}
87
88mp_obj_t str_join(mp_obj_t self_in, mp_obj_t arg) {
89 assert(MP_OBJ_IS_TYPE(self_in, &str_type));
90 mp_obj_str_t *self = self_in;
Damiend99b0522013-12-21 18:17:45 +000091
Damien Georgefe8fb912014-01-02 16:36:09 +000092 // get separation string
93 const char *sep_str = qstr_str(self->qstr);
94 size_t sep_len = strlen(sep_str);
95
96 // process args
Damiend99b0522013-12-21 18:17:45 +000097 uint seq_len;
98 mp_obj_t *seq_items;
99 if (MP_OBJ_IS_TYPE(arg, &tuple_type)) {
100 mp_obj_tuple_get(arg, &seq_len, &seq_items);
101 } else if (MP_OBJ_IS_TYPE(arg, &list_type)) {
102 mp_obj_list_get(arg, &seq_len, &seq_items);
103 } else {
104 goto bad_arg;
105 }
Damien Georgefe8fb912014-01-02 16:36:09 +0000106
107 // count required length
108 int required_len = 0;
Damiend99b0522013-12-21 18:17:45 +0000109 for (int i = 0; i < seq_len; i++) {
110 if (!MP_OBJ_IS_TYPE(seq_items[i], &str_type)) {
111 goto bad_arg;
112 }
Damien Georgefe8fb912014-01-02 16:36:09 +0000113 if (i > 0) {
114 required_len += sep_len;
115 }
Damiend99b0522013-12-21 18:17:45 +0000116 required_len += strlen(qstr_str(mp_obj_str_get(seq_items[i])));
117 }
118
119 // make joined string
120 char *joined_str = m_new(char, required_len + 1);
Damien Georgefe8fb912014-01-02 16:36:09 +0000121 char *s_dest = joined_str;
Damiend99b0522013-12-21 18:17:45 +0000122 for (int i = 0; i < seq_len; i++) {
Damiend99b0522013-12-21 18:17:45 +0000123 if (i > 0) {
Damien Georgefe8fb912014-01-02 16:36:09 +0000124 memcpy(s_dest, sep_str, sep_len);
125 s_dest += sep_len;
Damiend99b0522013-12-21 18:17:45 +0000126 }
Damien Georgefe8fb912014-01-02 16:36:09 +0000127 const char *s2 = qstr_str(mp_obj_str_get(seq_items[i]));
128 size_t s2_len = strlen(s2);
129 memcpy(s_dest, s2, s2_len);
130 s_dest += s2_len;
Damiend99b0522013-12-21 18:17:45 +0000131 }
Damien Georgefe8fb912014-01-02 16:36:09 +0000132 *s_dest = '\0';
133
134 // return joined string
Damien732407f2013-12-29 19:33:23 +0000135 return mp_obj_new_str(qstr_from_str_take(joined_str, required_len + 1));
Damiend99b0522013-12-21 18:17:45 +0000136
137bad_arg:
Damien Georgeeb7bfcb2014-01-04 15:57:35 +0000138 nlr_jump(mp_obj_new_exception_msg(MP_QSTR_TypeError, "?str.join expecting a list of str's"));
Damiend99b0522013-12-21 18:17:45 +0000139}
140
141void vstr_printf_wrapper(void *env, const char *fmt, ...) {
142 va_list args;
143 va_start(args, fmt);
144 vstr_vprintf(env, fmt, args);
145 va_end(args);
146}
147
148mp_obj_t str_format(int n_args, const mp_obj_t *args) {
149 assert(MP_OBJ_IS_TYPE(args[0], &str_type));
150 mp_obj_str_t *self = args[0];
151
152 const char *str = qstr_str(self->qstr);
153 int arg_i = 1;
154 vstr_t *vstr = vstr_new();
155 for (; *str; str++) {
156 if (*str == '{') {
157 str++;
158 if (*str == '{') {
159 vstr_add_char(vstr, '{');
160 } else if (*str == '}') {
161 if (arg_i >= n_args) {
Damien Georgeeb7bfcb2014-01-04 15:57:35 +0000162 nlr_jump(mp_obj_new_exception_msg(MP_QSTR_IndexError, "tuple index out of range"));
Damiend99b0522013-12-21 18:17:45 +0000163 }
164 mp_obj_print_helper(vstr_printf_wrapper, vstr, args[arg_i]);
165 arg_i++;
166 }
167 } else {
168 vstr_add_char(vstr, *str);
169 }
170 }
171
Damien732407f2013-12-29 19:33:23 +0000172 return mp_obj_new_str(qstr_from_str_take(vstr->buf, vstr->alloc));
Damiend99b0522013-12-21 18:17:45 +0000173}
174
175static MP_DEFINE_CONST_FUN_OBJ_2(str_join_obj, str_join);
176static MP_DEFINE_CONST_FUN_OBJ_VAR(str_format_obj, 1, str_format);
177
178const mp_obj_type_t str_type = {
179 { &mp_const_type },
180 "str",
181 str_print, // print
182 NULL, // call_n
183 NULL, // unary_op
184 str_binary_op, // binary_op
185 NULL, // getiter
186 NULL, // iternext
187 { // method list
188 { "join", &str_join_obj },
189 { "format", &str_format_obj },
190 { NULL, NULL }, // end-of-list sentinel
191 },
192};
193
194mp_obj_t mp_obj_new_str(qstr qstr) {
195 mp_obj_str_t *o = m_new_obj(mp_obj_str_t);
196 o->base.type = &str_type;
197 o->qstr = qstr;
198 return o;
199}
200
201qstr mp_obj_str_get(mp_obj_t self_in) {
202 assert(MP_OBJ_IS_TYPE(self_in, &str_type));
203 mp_obj_str_t *self = self_in;
204 return self->qstr;
205}