compile.c 115 KB
Newer Older
Damien's avatar
Damien committed
1
2
3
4
5
6
7
8
#include <unistd.h>
#include <stdlib.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>

#include "misc.h"
9
#include "mpyconfig.h"
Damien's avatar
Damien committed
10
11
12
13
14
15
16
17
18
#include "lexer.h"
#include "parse.h"
#include "scope.h"
#include "compile.h"
#include "runtime.h"
#include "emit.h"

// TODO need to mangle __attr names

19
20
#define MICROPY_EMIT_NATIVE (MICROPY_EMIT_X64 || MICROPY_EMIT_THUMB)

Damien's avatar
Damien committed
21
22
23
24
25
26
27
28
typedef enum {
    PN_none = 0,
#define DEF_RULE(rule, comp, kind, arg...) PN_##rule,
#include "grammar.h"
#undef DEF_RULE
    PN_maximum_number_of,
} pn_kind_t;

29
#define EMIT(fun, arg...) (comp->emit_method_table->fun(comp->emit, ##arg))
30
#define EMIT_INLINE_ASM(fun, arg...) (comp->emit_inline_asm_method_table->fun(comp->emit_inline_asm, ##arg))
Damien's avatar
Damien committed
31

32
33
34
#define EMIT_OPT_NONE           (0)
#define EMIT_OPT_BYTE_CODE      (1)
#define EMIT_OPT_NATIVE_PYTHON  (2)
Damien's avatar
Damien committed
35
36
#define EMIT_OPT_VIPER          (3)
#define EMIT_OPT_ASM_THUMB      (4)
37

Damien's avatar
Damien committed
38
39
40
41
42
43
44
45
typedef struct _compiler_t {
    qstr qstr___class__;
    qstr qstr___locals__;
    qstr qstr___name__;
    qstr qstr___module__;
    qstr qstr___qualname__;
    qstr qstr___doc__;
    qstr qstr_assertion_error;
46
    qstr qstr_micropython;
Damien's avatar
Damien committed
47
    qstr qstr_byte_code;
48
    qstr qstr_native;
Damien's avatar
Damien committed
49
    qstr qstr_viper;
50
    qstr qstr_asm_thumb;
51
    qstr qstr_range;
Damien's avatar
Damien committed
52

Damien's avatar
Damien committed
53
    bool is_repl;
Damien's avatar
Damien committed
54
    pass_kind_t pass;
Damien's avatar
Damien committed
55
    bool had_error; // try to keep compiler clean from nlr
Damien's avatar
Damien committed
56

57
58
    int next_label;

Damien's avatar
Damien committed
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
    int break_label;
    int continue_label;
    int except_nest_level;

    int n_arg_keyword;
    bool have_star_arg;
    bool have_dbl_star_arg;
    bool have_bare_star;
    int param_pass;
    int param_pass_num_dict_params;
    int param_pass_num_default_params;

    scope_t *scope_head;
    scope_t *scope_cur;

74
75
    emit_t *emit;                                   // current emitter
    const emit_method_table_t *emit_method_table;   // current emit method table
76
77
78

    emit_inline_asm_t *emit_inline_asm;                                   // current emitter for inline asm
    const emit_inline_asm_method_table_t *emit_inline_asm_method_table;   // current emit method table for inline asm
Damien's avatar
Damien committed
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
} compiler_t;

py_parse_node_t fold_constants(py_parse_node_t pn) {
    if (PY_PARSE_NODE_IS_STRUCT(pn)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns);

        // fold arguments first
        for (int i = 0; i < n; i++) {
            pns->nodes[i] = fold_constants(pns->nodes[i]);
        }

        switch (PY_PARSE_NODE_STRUCT_KIND(pns)) {
            case PN_shift_expr:
                if (n == 3 && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
                    int arg0 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
                    int arg1 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
                    if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_DBL_LESS)) {
97
#if MICROPY_EMIT_CPYTHON
98
99
100
                        // can overflow; enabled only to compare with CPython
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg0 << arg1);
#endif
Damien's avatar
Damien committed
101
102
103
104
105
106
107
108
109
110
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_DBL_MORE)) {
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg0 >> arg1);
                    } else {
                        // shouldn't happen
                        assert(0);
                    }
                }
                break;

            case PN_arith_expr:
111
                // overflow checking here relies on SMALL_INT being strictly smaller than machine_int_t
Damien's avatar
Damien committed
112
                if (n == 3 && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
113
114
115
                    machine_int_t arg0 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
                    machine_int_t arg1 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
                    machine_int_t res;
Damien's avatar
Damien committed
116
                    if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_PLUS)) {
117
                        res = arg0 + arg1;
Damien's avatar
Damien committed
118
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_MINUS)) {
119
                        res = arg0 - arg1;
Damien's avatar
Damien committed
120
121
122
                    } else {
                        // shouldn't happen
                        assert(0);
123
124
125
126
                        res = 0;
                    }
                    if (PY_FIT_SMALL_INT(res)) {
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, res);
Damien's avatar
Damien committed
127
128
129
130
131
132
133
134
135
                    }
                }
                break;

            case PN_term:
                if (n == 3 && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
                    int arg0 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
                    int arg1 = PY_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
                    if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_STAR)) {
136
#if MICROPY_EMIT_CPYTHON
137
                        // can overflow; enabled only to compare with CPython
Damien's avatar
Damien committed
138
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg0 * arg1);
139
#endif
Damien's avatar
Damien committed
140
141
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_SLASH)) {
                        ; // pass
142
143
144
145
146
147
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_PERCENT)) {
                        // XXX implement this properly as Python's % operator acts differently to C's
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg0 % arg1);
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], PY_TOKEN_OP_DBL_SLASH)) {
                        // XXX implement this properly as Python's // operator acts differently to C's
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg0 / arg1);
Damien's avatar
Damien committed
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
                    } else {
                        // shouldn't happen
                        assert(0);
                    }
                }
                break;

            case PN_factor_2:
                if (PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[1])) {
                    machine_int_t arg = PY_PARSE_NODE_LEAF_ARG(pns->nodes[1]);
                    if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[0], PY_TOKEN_OP_PLUS)) {
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, arg);
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[0], PY_TOKEN_OP_MINUS)) {
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, -arg);
                    } else if (PY_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[0], PY_TOKEN_OP_TILDE)) {
                        pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, ~arg);
                    } else {
                        // shouldn't happen
                        assert(0);
                    }
                }
                break;

171
#if MICROPY_EMIT_CPYTHON
Damien's avatar
Damien committed
172
            case PN_power:
173
                // can overflow; enabled only to compare with CPython
Damien's avatar
Damien committed
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
                if (PY_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && PY_PARSE_NODE_IS_NULL(pns->nodes[1]) && !PY_PARSE_NODE_IS_NULL(pns->nodes[2])) {
                    py_parse_node_struct_t* pns2 = (py_parse_node_struct_t*)pns->nodes[2];
                    if (PY_PARSE_NODE_IS_SMALL_INT(pns2->nodes[0])) {
                        int power = PY_PARSE_NODE_LEAF_ARG(pns2->nodes[0]);
                        if (power >= 0) {
                            int ans = 1;
                            int base = PY_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
                            for (; power > 0; power--) {
                                ans *= base;
                            }
                            pn = py_parse_node_new_leaf(PY_PARSE_NODE_SMALL_INT, ans);
                        }
                    }
                }
                break;
189
#endif
Damien's avatar
Damien committed
190
191
192
193
194
195
196
197
        }
    }

    return pn;
}

void compile_node(compiler_t *comp, py_parse_node_t pn);

198
199
200
201
static int comp_next_label(compiler_t *comp) {
    return comp->next_label++;
}

202
static scope_t *scope_new_and_link(compiler_t *comp, scope_kind_t kind, py_parse_node_t pn, uint emit_options) {
Damien's avatar
Damien committed
203
    scope_t *scope = scope_new(kind, pn, rt_get_unique_code_id(kind == SCOPE_MODULE), emit_options);
Damien's avatar
Damien committed
204
205
206
207
208
209
210
211
212
213
214
215
216
217
    scope->parent = comp->scope_cur;
    scope->next = NULL;
    if (comp->scope_head == NULL) {
        comp->scope_head = scope;
    } else {
        scope_t *s = comp->scope_head;
        while (s->next != NULL) {
            s = s->next;
        }
        s->next = scope;
    }
    return scope;
}

218
static int list_len(py_parse_node_t pn, int pn_kind) {
Damien's avatar
Damien committed
219
220
221
222
223
224
225
226
227
228
229
230
231
232
    if (PY_PARSE_NODE_IS_NULL(pn)) {
        return 0;
    } else if (PY_PARSE_NODE_IS_LEAF(pn)) {
        return 1;
    } else {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        if (PY_PARSE_NODE_STRUCT_KIND(pns) != pn_kind) {
            return 1;
        } else {
            return PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
        }
    }
}

233
static void apply_to_single_or_list(compiler_t *comp, py_parse_node_t pn, int pn_list_kind, void (*f)(compiler_t*, py_parse_node_t)) {
Damien's avatar
Damien committed
234
235
236
237
238
239
240
241
242
243
244
    if (PY_PARSE_NODE_IS_STRUCT(pn) && PY_PARSE_NODE_STRUCT_KIND((py_parse_node_struct_t*)pn) == pn_list_kind) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        int num_nodes = PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
        for (int i = 0; i < num_nodes; i++) {
            f(comp, pns->nodes[i]);
        }
    } else if (!PY_PARSE_NODE_IS_NULL(pn)) {
        f(comp, pn);
    }
}

245
static int list_get(py_parse_node_t *pn, int pn_kind, py_parse_node_t **nodes) {
Damien's avatar
Damien committed
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
    if (PY_PARSE_NODE_IS_NULL(*pn)) {
        *nodes = NULL;
        return 0;
    } else if (PY_PARSE_NODE_IS_LEAF(*pn)) {
        *nodes = pn;
        return 1;
    } else {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)(*pn);
        if (PY_PARSE_NODE_STRUCT_KIND(pns) != pn_kind) {
            *nodes = pn;
            return 1;
        } else {
            *nodes = pns->nodes;
            return PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
        }
    }
}

void compile_do_nothing(compiler_t *comp, py_parse_node_struct_t *pns) {
}

void compile_generic_all_nodes(compiler_t *comp, py_parse_node_struct_t *pns) {
    int num_nodes = PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
    for (int i = 0; i < num_nodes; i++) {
        compile_node(comp, pns->nodes[i]);
    }
}

274
#if MICROPY_EMIT_CPYTHON
275
static bool cpython_c_tuple_is_const(py_parse_node_t pn) {
Damien's avatar
Damien committed
276
277
278
279
280
281
282
283
284
    if (!PY_PARSE_NODE_IS_LEAF(pn)) {
        return false;
    }
    if (PY_PARSE_NODE_IS_ID(pn)) {
        return false;
    }
    return true;
}

285
static void cpython_c_tuple_emit_const(compiler_t *comp, py_parse_node_t pn) {
Damien's avatar
Damien committed
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
    assert(PY_PARSE_NODE_IS_LEAF(pn));
    int arg = PY_PARSE_NODE_LEAF_ARG(pn);
    switch (PY_PARSE_NODE_LEAF_KIND(pn)) {
        case PY_PARSE_NODE_ID: assert(0);
        case PY_PARSE_NODE_SMALL_INT: EMIT(load_const_verbatim_int, arg); break;
        case PY_PARSE_NODE_INTEGER: EMIT(load_const_verbatim_str, qstr_str(arg)); break;
        case PY_PARSE_NODE_DECIMAL: EMIT(load_const_verbatim_str, qstr_str(arg)); break;
        case PY_PARSE_NODE_STRING: EMIT(load_const_verbatim_quoted_str, arg, false); break;
        case PY_PARSE_NODE_BYTES: EMIT(load_const_verbatim_quoted_str, arg, true); break;
        case PY_PARSE_NODE_TOKEN:
            switch (arg) {
                case PY_TOKEN_KW_FALSE: EMIT(load_const_verbatim_str, "False"); break;
                case PY_TOKEN_KW_NONE: EMIT(load_const_verbatim_str, "None"); break;
                case PY_TOKEN_KW_TRUE: EMIT(load_const_verbatim_str, "True"); break;
                default: assert(0);
            }
            break;
        default: assert(0);
    }
}

307
static void cpython_c_tuple(compiler_t *comp, py_parse_node_t pn, py_parse_node_struct_t *pns_list) {
Damien's avatar
Damien committed
308
309
310
311
312
313
314
315
    int n = 0;
    if (pns_list != NULL) {
        n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns_list);
    }
    int total = n;
    bool is_const = true;
    if (!PY_PARSE_NODE_IS_NULL(pn)) {
        total += 1;
316
        if (!cpython_c_tuple_is_const(pn)) {
Damien's avatar
Damien committed
317
318
319
320
            is_const = false;
        }
    }
    for (int i = 0; i < n; i++) {
321
        if (!cpython_c_tuple_is_const(pns_list->nodes[i])) {
Damien's avatar
Damien committed
322
323
324
325
326
327
328
329
330
            is_const = false;
            break;
        }
    }
    if (total > 0 && is_const) {
        bool need_comma = false;
        EMIT(load_const_verbatim_start);
        EMIT(load_const_verbatim_str, "(");
        if (!PY_PARSE_NODE_IS_NULL(pn)) {
331
            cpython_c_tuple_emit_const(comp, pn);
Damien's avatar
Damien committed
332
333
334
335
336
337
            need_comma = true;
        }
        for (int i = 0; i < n; i++) {
            if (need_comma) {
                EMIT(load_const_verbatim_str, ", ");
            }
338
            cpython_c_tuple_emit_const(comp, pns_list->nodes[i]);
Damien's avatar
Damien committed
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
            need_comma = true;
        }
        if (total == 1) {
            EMIT(load_const_verbatim_str, ",)");
        } else {
            EMIT(load_const_verbatim_str, ")");
        }
        EMIT(load_const_verbatim_end);
    } else {
        if (!PY_PARSE_NODE_IS_NULL(pn)) {
            compile_node(comp, pn);
        }
        for (int i = 0; i < n; i++) {
            compile_node(comp, pns_list->nodes[i]);
        }
        EMIT(build_tuple, total);
    }
}
357
358
359
360
#endif

// funnelling all tuple creations through this function is purely so we can optionally agree with CPython
void c_tuple(compiler_t *comp, py_parse_node_t pn, py_parse_node_struct_t *pns_list) {
361
#if MICROPY_EMIT_CPYTHON
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
    cpython_c_tuple(comp, pn, pns_list);
#else
    int total = 0;
    if (!PY_PARSE_NODE_IS_NULL(pn)) {
        compile_node(comp, pn);
        total += 1;
    }
    if (pns_list != NULL) {
        int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns_list);
        for (int i = 0; i < n; i++) {
            compile_node(comp, pns_list->nodes[i]);
        }
        total += n;
    }
    EMIT(build_tuple, total);
#endif
}
Damien's avatar
Damien committed
379
380
381
382
383
384

void compile_generic_tuple(compiler_t *comp, py_parse_node_struct_t *pns) {
    // a simple tuple expression
    c_tuple(comp, PY_PARSE_NODE_NULL, pns);
}

385
static bool node_is_const_false(py_parse_node_t pn) {
Damien's avatar
Damien committed
386
387
388
389
    return PY_PARSE_NODE_IS_TOKEN_KIND(pn, PY_TOKEN_KW_FALSE);
    // untested: || (PY_PARSE_NODE_IS_SMALL_INT(pn) && PY_PARSE_NODE_LEAF_ARG(pn) == 1);
}

390
static bool node_is_const_true(py_parse_node_t pn) {
Damien's avatar
Damien committed
391
392
393
    return PY_PARSE_NODE_IS_TOKEN_KIND(pn, PY_TOKEN_KW_TRUE) || (PY_PARSE_NODE_IS_SMALL_INT(pn) && PY_PARSE_NODE_LEAF_ARG(pn) == 1);
}

394
#if MICROPY_EMIT_CPYTHON
395
396
// the is_nested variable is purely to match with CPython, which doesn't fully optimise not's
static void cpython_c_if_cond(compiler_t *comp, py_parse_node_t pn, bool jump_if, int label, bool is_nested) {
Damien's avatar
Damien committed
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
    if (node_is_const_false(pn)) {
        if (jump_if == false) {
            EMIT(jump, label);
        }
        return;
    } else if (node_is_const_true(pn)) {
        if (jump_if == true) {
            EMIT(jump, label);
        }
        return;
    } else if (PY_PARSE_NODE_IS_STRUCT(pn)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
        if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_or_test) {
            if (jump_if == false) {
412
                int label2 = comp_next_label(comp);
Damien's avatar
Damien committed
413
                for (int i = 0; i < n - 1; i++) {
414
                    cpython_c_if_cond(comp, pns->nodes[i], true, label2, true);
Damien's avatar
Damien committed
415
                }
416
                cpython_c_if_cond(comp, pns->nodes[n - 1], false, label, true);
Damien's avatar
Damien committed
417
418
419
                EMIT(label_assign, label2);
            } else {
                for (int i = 0; i < n; i++) {
420
                    cpython_c_if_cond(comp, pns->nodes[i], true, label, true);
Damien's avatar
Damien committed
421
422
423
424
425
426
                }
            }
            return;
        } else if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_and_test) {
            if (jump_if == false) {
                for (int i = 0; i < n; i++) {
427
                    cpython_c_if_cond(comp, pns->nodes[i], false, label, true);
Damien's avatar
Damien committed
428
429
                }
            } else {
430
                int label2 = comp_next_label(comp);
Damien's avatar
Damien committed
431
                for (int i = 0; i < n - 1; i++) {
432
                    cpython_c_if_cond(comp, pns->nodes[i], false, label2, true);
Damien's avatar
Damien committed
433
                }
434
                cpython_c_if_cond(comp, pns->nodes[n - 1], true, label, true);
Damien's avatar
Damien committed
435
436
437
438
                EMIT(label_assign, label2);
            }
            return;
        } else if (!is_nested && PY_PARSE_NODE_STRUCT_KIND(pns) == PN_not_test_2) {
439
            cpython_c_if_cond(comp, pns->nodes[0], !jump_if, label, true);
Damien's avatar
Damien committed
440
441
442
443
444
445
446
447
448
449
450
451
            return;
        }
    }

    // nothing special, fall back to default compiling for node and jump
    compile_node(comp, pn);
    if (jump_if == false) {
        EMIT(pop_jump_if_false, label);
    } else {
        EMIT(pop_jump_if_true, label);
    }
}
452
#endif
Damien's avatar
Damien committed
453

454
static void c_if_cond(compiler_t *comp, py_parse_node_t pn, bool jump_if, int label) {
455
#if MICROPY_EMIT_CPYTHON
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
    cpython_c_if_cond(comp, pn, jump_if, label, false);
#else
    if (node_is_const_false(pn)) {
        if (jump_if == false) {
            EMIT(jump, label);
        }
        return;
    } else if (node_is_const_true(pn)) {
        if (jump_if == true) {
            EMIT(jump, label);
        }
        return;
    } else if (PY_PARSE_NODE_IS_STRUCT(pn)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns);
        if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_or_test) {
            if (jump_if == false) {
                int label2 = comp_next_label(comp);
                for (int i = 0; i < n - 1; i++) {
                    c_if_cond(comp, pns->nodes[i], true, label2);
                }
                c_if_cond(comp, pns->nodes[n - 1], false, label);
                EMIT(label_assign, label2);
            } else {
                for (int i = 0; i < n; i++) {
                    c_if_cond(comp, pns->nodes[i], true, label);
                }
            }
            return;
        } else if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_and_test) {
            if (jump_if == false) {
                for (int i = 0; i < n; i++) {
                    c_if_cond(comp, pns->nodes[i], false, label);
                }
            } else {
                int label2 = comp_next_label(comp);
                for (int i = 0; i < n - 1; i++) {
                    c_if_cond(comp, pns->nodes[i], false, label2);
                }
                c_if_cond(comp, pns->nodes[n - 1], true, label);
                EMIT(label_assign, label2);
            }
            return;
        } else if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_not_test_2) {
            c_if_cond(comp, pns->nodes[0], !jump_if, label);
            return;
        }
    }

    // nothing special, fall back to default compiling for node and jump
    compile_node(comp, pn);
    if (jump_if == false) {
        EMIT(pop_jump_if_false, label);
    } else {
        EMIT(pop_jump_if_true, label);
    }
#endif
Damien's avatar
Damien committed
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
}

typedef enum { ASSIGN_STORE, ASSIGN_AUG_LOAD, ASSIGN_AUG_STORE } assign_kind_t;
void c_assign(compiler_t *comp, py_parse_node_t pn, assign_kind_t kind);

void c_assign_power(compiler_t *comp, py_parse_node_struct_t *pns, assign_kind_t assign_kind) {
    if (assign_kind != ASSIGN_AUG_STORE) {
        compile_node(comp, pns->nodes[0]);
    }

    if (PY_PARSE_NODE_IS_STRUCT(pns->nodes[1])) {
        py_parse_node_struct_t *pns1 = (py_parse_node_struct_t*)pns->nodes[1];
        if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_power_trailers) {
            int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns1);
            if (assign_kind != ASSIGN_AUG_STORE) {
                for (int i = 0; i < n - 1; i++) {
                    compile_node(comp, pns1->nodes[i]);
                }
            }
            assert(PY_PARSE_NODE_IS_STRUCT(pns1->nodes[n - 1]));
            pns1 = (py_parse_node_struct_t*)pns1->nodes[n - 1];
        }
        if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_paren) {
            printf("SyntaxError: can't assign to function call\n");
            return;
        } else if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_bracket) {
            if (assign_kind == ASSIGN_AUG_STORE) {
                EMIT(rot_three);
                EMIT(store_subscr);
            } else {
                compile_node(comp, pns1->nodes[0]);
                if (assign_kind == ASSIGN_AUG_LOAD) {
                    EMIT(dup_top_two);
                    EMIT(binary_op, RT_BINARY_OP_SUBSCR);
                } else {
                    EMIT(store_subscr);
                }
            }
        } else if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_period) {
            assert(PY_PARSE_NODE_IS_ID(pns1->nodes[0]));
            if (assign_kind == ASSIGN_AUG_LOAD) {
                EMIT(dup_top);
                EMIT(load_attr, PY_PARSE_NODE_LEAF_ARG(pns1->nodes[0]));
            } else {
                if (assign_kind == ASSIGN_AUG_STORE) {
                    EMIT(rot_two);
                }
                EMIT(store_attr, PY_PARSE_NODE_LEAF_ARG(pns1->nodes[0]));
            }
        } else {
            // shouldn't happen
            assert(0);
        }
    } else {
        // shouldn't happen
        assert(0);
    }

    if (!PY_PARSE_NODE_IS_NULL(pns->nodes[2])) {
        // SyntaxError, cannot assign
        assert(0);
    }
}

void c_assign_tuple(compiler_t *comp, int n, py_parse_node_t *nodes) {
    assert(n >= 0);
    int have_star_index = -1;
    for (int i = 0; i < n; i++) {
        if (PY_PARSE_NODE_IS_STRUCT_KIND(nodes[i], PN_star_expr)) {
            if (have_star_index < 0) {
                EMIT(unpack_ex, i, n - i - 1);
                have_star_index = i;
            } else {
                printf("SyntaxError: two starred expressions in assignment\n");
                return;
            }
        }
    }
    if (have_star_index < 0) {
        EMIT(unpack_sequence, n);
    }
    for (int i = 0; i < n; i++) {
        if (i == have_star_index) {
            c_assign(comp, ((py_parse_node_struct_t*)nodes[i])->nodes[0], ASSIGN_STORE);
        } else {
            c_assign(comp, nodes[i], ASSIGN_STORE);
        }
    }
}

// assigns top of stack to pn
void c_assign(compiler_t *comp, py_parse_node_t pn, assign_kind_t assign_kind) {
    tail_recursion:
    if (PY_PARSE_NODE_IS_NULL(pn)) {
        assert(0);
    } else if (PY_PARSE_NODE_IS_LEAF(pn)) {
        if (PY_PARSE_NODE_IS_ID(pn)) {
            int arg = PY_PARSE_NODE_LEAF_ARG(pn);
            switch (assign_kind) {
                case ASSIGN_STORE:
                case ASSIGN_AUG_STORE:
614
                    EMIT(store_id, arg);
Damien's avatar
Damien committed
615
616
                    break;
                case ASSIGN_AUG_LOAD:
617
                    EMIT(load_id, arg);
Damien's avatar
Damien committed
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
                    break;
            }
        } else {
            printf("SyntaxError: can't assign to literal\n");
            return;
        }
    } else {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
        switch (PY_PARSE_NODE_STRUCT_KIND(pns)) {
            case PN_power:
                // lhs is an index or attribute
                c_assign_power(comp, pns, assign_kind);
                break;

            case PN_testlist_star_expr:
            case PN_exprlist:
                // lhs is a tuple
                if (assign_kind != ASSIGN_STORE) {
                    goto bad_aug;
                }
                c_assign_tuple(comp, PY_PARSE_NODE_STRUCT_NUM_NODES(pns), pns->nodes);
                break;

            case PN_atom_paren:
                // lhs is something in parenthesis
                if (PY_PARSE_NODE_IS_NULL(pns->nodes[0])) {
                    // empty tuple
                    printf("SyntaxError: can't assign to ()\n");
                    return;
                } else if (PY_PARSE_NODE_IS_STRUCT_KIND(pns->nodes[0], PN_testlist_comp)) {
                    pns = (py_parse_node_struct_t*)pns->nodes[0];
                    goto testlist_comp;
                } else {
                    // parenthesis around 1 item, is just that item
                    pn = pns->nodes[0];
                    goto tail_recursion;
                }
                break;

            case PN_atom_bracket:
                // lhs is something in brackets
                if (assign_kind != ASSIGN_STORE) {
                    goto bad_aug;
                }
                if (PY_PARSE_NODE_IS_NULL(pns->nodes[0])) {
                    // empty list, assignment allowed
                    c_assign_tuple(comp, 0, NULL);
                } else if (PY_PARSE_NODE_IS_STRUCT_KIND(pns->nodes[0], PN_testlist_comp)) {
                    pns = (py_parse_node_struct_t*)pns->nodes[0];
                    goto testlist_comp;
                } else {
                    // brackets around 1 item
                    c_assign_tuple(comp, 1, &pns->nodes[0]);
                }
                break;

            default:
                printf("unknown assign, %u\n", (uint)PY_PARSE_NODE_STRUCT_KIND(pns));
                assert(0);
        }
        return;

        testlist_comp:
        // lhs is a sequence
        if (PY_PARSE_NODE_IS_STRUCT(pns->nodes[1])) {
            py_parse_node_struct_t *pns2 = (py_parse_node_struct_t*)pns->nodes[1];
            if (PY_PARSE_NODE_STRUCT_KIND(pns2) == PN_testlist_comp_3b) {
                // sequence of one item, with trailing comma
                assert(PY_PARSE_NODE_IS_NULL(pns2->nodes[0]));
                c_assign_tuple(comp, 1, &pns->nodes[0]);
            } else if (PY_PARSE_NODE_STRUCT_KIND(pns2) == PN_testlist_comp_3c) {
                // sequence of many items
                // TODO call c_assign_tuple instead
                int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns2);
                EMIT(unpack_sequence, 1 + n);
                c_assign(comp, pns->nodes[0], ASSIGN_STORE);
                for (int i = 0; i < n; i++) {
                    c_assign(comp, pns2->nodes[i], ASSIGN_STORE);
                }
            } else if (PY_PARSE_NODE_STRUCT_KIND(pns) == PN_comp_for) {
                // TODO not implemented
                assert(0);
            } else {
                // sequence with 2 items
                goto sequence_with_2_items;
            }
        } else {
            // sequence with 2 items
            sequence_with_2_items:
            c_assign_tuple(comp, 2, pns->nodes);
        }
        return;
    }
    return;

    bad_aug:
    printf("SyntaxError: illegal expression for augmented assignment\n");
}

// stuff for lambda and comprehensions and generators
void close_over_variables_etc(compiler_t *comp, scope_t *this_scope, int n_dict_params, int n_default_params) {
    // make closed over variables, if any
    int nfree = 0;
    if (comp->scope_cur->kind != SCOPE_MODULE) {
        for (int i = 0; i < this_scope->id_info_len; i++) {
            id_info_t *id_info = &this_scope->id_info[i];
            if (id_info->kind == ID_INFO_KIND_FREE) {
725
                EMIT(load_closure, id_info->qstr, id_info->local_num);
Damien's avatar
Damien committed
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
                nfree += 1;
            }
        }
    }
    if (nfree > 0) {
        EMIT(build_tuple, nfree);
    }

    // make the function/closure
    if (nfree == 0) {
        EMIT(make_function, this_scope, n_dict_params, n_default_params);
    } else {
        EMIT(make_closure, this_scope, n_dict_params, n_default_params);
    }
}

void compile_funcdef_param(compiler_t *comp, py_parse_node_t pn) {
743
744
    if (PY_PARSE_NODE_IS_STRUCT_KIND(pn, PN_typedargslist_name)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
Damien's avatar
Damien committed
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
        if (!PY_PARSE_NODE_IS_NULL(pns->nodes[2])) {
            // this parameter has a default value
            // in CPython, None (and True, False?) as default parameters are loaded with LOAD_NAME; don't understandy why
            if (comp->have_bare_star) {
                comp->param_pass_num_dict_params += 1;
                if (comp->param_pass == 1) {
                    EMIT(load_const_id, PY_PARSE_NODE_LEAF_ARG(pns->nodes[0]));
                    compile_node(comp, pns->nodes[2]);
                }
            } else {
                comp->param_pass_num_default_params += 1;
                if (comp->param_pass == 2) {
                    compile_node(comp, pns->nodes[2]);
                }
            }
        }
761
762
    } else if (PY_PARSE_NODE_IS_STRUCT_KIND(pn, PN_typedargslist_star)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
Damien's avatar
Damien committed
763
764
765
766
767
768
769
770
771
        if (PY_PARSE_NODE_IS_NULL(pns->nodes[0])) {
            // bare star
            comp->have_bare_star = true;
        }
    }
}

// leaves function object on stack
// returns function name
772
qstr compile_funcdef_helper(compiler_t *comp, py_parse_node_struct_t *pns, uint emit_options) {
Damien's avatar
Damien committed
773
774
    if (comp->pass == PASS_1) {
        // create a new scope for this function
775
        scope_t *s = scope_new_and_link(comp, SCOPE_FUNCTION, (py_parse_node_t)pns, emit_options);
Damien's avatar
Damien committed
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
        // store the function scope so the compiling function can use it at each pass
        pns->nodes[4] = (py_parse_node_t)s;
    }

    // save variables (probably don't need to do this, since we can't have nested definitions..?)
    bool old_have_bare_star = comp->have_bare_star;
    int old_param_pass = comp->param_pass;
    int old_param_pass_num_dict_params = comp->param_pass_num_dict_params;
    int old_param_pass_num_default_params = comp->param_pass_num_default_params;

    // compile default parameters
    comp->have_bare_star = false;
    comp->param_pass = 1; // pass 1 does any default parameters after bare star
    comp->param_pass_num_dict_params = 0;
    comp->param_pass_num_default_params = 0;
    apply_to_single_or_list(comp, pns->nodes[1], PN_typedargslist, compile_funcdef_param);
    comp->have_bare_star = false;
    comp->param_pass = 2; // pass 2 does any default parameters before bare star
    comp->param_pass_num_dict_params = 0;
    comp->param_pass_num_default_params = 0;
    apply_to_single_or_list(comp, pns->nodes[1], PN_typedargslist, compile_funcdef_param);

    // get the scope for this function
    scope_t *fscope = (scope_t*)pns->nodes[4];

    // make the function
    close_over_variables_etc(comp, fscope, comp->param_pass_num_dict_params, comp->param_pass_num_default_params);

    // restore variables
    comp->have_bare_star = old_have_bare_star;
    comp->param_pass = old_param_pass;
    comp->param_pass_num_dict_params = old_param_pass_num_dict_params;
    comp->param_pass_num_default_params = old_param_pass_num_default_params;

    // return its name (the 'f' in "def f(...):")
    return fscope->simple_name;
}

// leaves class object on stack
// returns class name
816
qstr compile_classdef_helper(compiler_t *comp, py_parse_node_struct_t *pns, uint emit_options) {
Damien's avatar
Damien committed
817
818
    if (comp->pass == PASS_1) {
        // create a new scope for this class
819
        scope_t *s = scope_new_and_link(comp, SCOPE_CLASS, (py_parse_node_t)pns, emit_options);
Damien's avatar
Damien committed
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
        // store the class scope so the compiling function can use it at each pass
        pns->nodes[3] = (py_parse_node_t)s;
    }

    EMIT(load_build_class);

    // scope for this class
    scope_t *cscope = (scope_t*)pns->nodes[3];

    // compile the class
    close_over_variables_etc(comp, cscope, 0, 0);

    // get its name
    EMIT(load_const_id, cscope->simple_name);

    // nodes[1] has parent classes, if any
    if (PY_PARSE_NODE_IS_NULL(pns->nodes[1])) {
        // no parent classes
        EMIT(call_function, 2, 0, false, false);
    } else {
        // have a parent class or classes
        // TODO what if we have, eg, *a or **a in the parent list?
        compile_node(comp, pns->nodes[1]);
        EMIT(call_function, 2 + list_len(pns->nodes[1], PN_arglist), 0, false, false);
    }

    // return its name (the 'C' in class C(...):")
    return cscope->simple_name;
}

850
851
852
853
854
855
856
857
858
859
860
861
// returns true if it was a built-in decorator (even if the built-in had an error)
static bool compile_built_in_decorator(compiler_t *comp, int name_len, py_parse_node_t *name_nodes, uint *emit_options) {
    if (PY_PARSE_NODE_LEAF_ARG(name_nodes[0]) != comp->qstr_micropython) {
        return false;
    }

    if (name_len != 2) {
        printf("SyntaxError: invalid micropython decorator\n");
        return true;
    }

    qstr attr = PY_PARSE_NODE_LEAF_ARG(name_nodes[1]);
Damien's avatar
Damien committed
862
863
    if (attr == comp->qstr_byte_code) {
        *emit_options = EMIT_OPT_BYTE_CODE;
864
865
#if MICROPY_EMIT_NATIVE
    } else if (attr == comp->qstr_native) {
866
        *emit_options = EMIT_OPT_NATIVE_PYTHON;
Damien's avatar
Damien committed
867
868
    } else if (attr == comp->qstr_viper) {
        *emit_options = EMIT_OPT_VIPER;
869
#endif
870
#if MICROPY_EMIT_INLINE_THUMB
871
872
    } else if (attr == comp->qstr_asm_thumb) {
        *emit_options = EMIT_OPT_ASM_THUMB;
873
#endif
874
    } else {
875
        printf("SyntaxError: invalid micropython decorator '%s'\n", qstr_str(attr));
876
877
878
879
880
    }

    return true;
}

Damien's avatar
Damien committed
881
882
883
884
885
void compile_decorated(compiler_t *comp, py_parse_node_struct_t *pns) {
    // get the list of decorators
    py_parse_node_t *nodes;
    int n = list_get(&pns->nodes[0], PN_decorators, &nodes);

886
887
888
889
890
    // inherit emit options for this function/class definition
    uint emit_options = comp->scope_cur->emit_options;

    // compile each decorator
    int num_built_in_decorators = 0;
Damien's avatar
Damien committed
891
892
893
    for (int i = 0; i < n; i++) {
        assert(PY_PARSE_NODE_IS_STRUCT_KIND(nodes[i], PN_decorator)); // should be
        py_parse_node_struct_t *pns_decorator = (py_parse_node_struct_t*)nodes[i];
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918

        // nodes[0] contains the decorator function, which is a dotted name
        py_parse_node_t *name_nodes;
        int name_len = list_get(&pns_decorator->nodes[0], PN_dotted_name, &name_nodes);

        // check for built-in decorators
        if (compile_built_in_decorator(comp, name_len, name_nodes, &emit_options)) {
            // this was a built-in
            num_built_in_decorators += 1;

        } else {
            // not a built-in, compile normally

            // compile the decorator function
            compile_node(comp, name_nodes[0]);
            for (int i = 1; i < name_len; i++) {
                assert(PY_PARSE_NODE_IS_ID(name_nodes[i])); // should be
                EMIT(load_attr, PY_PARSE_NODE_LEAF_ARG(name_nodes[i]));
            }

            // nodes[1] contains arguments to the decorator function, if any
            if (!PY_PARSE_NODE_IS_NULL(pns_decorator->nodes[1])) {
                // call the decorator function with the arguments in nodes[1]
                compile_node(comp, pns_decorator->nodes[1]);
            }
Damien's avatar
Damien committed
919
920
921
922
923
924
925
        }
    }

    // compile the body (funcdef or classdef) and get its name
    py_parse_node_struct_t *pns_body = (py_parse_node_struct_t*)pns->nodes[1];
    qstr body_name = 0;
    if (PY_PARSE_NODE_STRUCT_KIND(pns_body) == PN_funcdef) {
926
        body_name = compile_funcdef_helper(comp, pns_body, emit_options);
Damien's avatar
Damien committed
927
    } else if (PY_PARSE_NODE_STRUCT_KIND(pns_body) == PN_classdef) {
928
        body_name = compile_classdef_helper(comp, pns_body, emit_options);
Damien's avatar
Damien committed
929
930
931
932
933
934
    } else {
        // shouldn't happen
        assert(0);
    }

    // call each decorator
935
    for (int i = 0; i < n - num_built_in_decorators; i++) {
Damien's avatar
Damien committed
936
937
938
939
        EMIT(call_function, 1, 0, false, false);
    }

    // store func/class object into name
940
    EMIT(store_id, body_name);
Damien's avatar
Damien committed
941
942
943
}

void compile_funcdef(compiler_t *comp, py_parse_node_struct_t *pns) {
944
    qstr fname = compile_funcdef_helper(comp, pns, comp->scope_cur->emit_options);
Damien's avatar
Damien committed
945
    // store function object into function name
946
    EMIT(store_id, fname);
Damien's avatar
Damien committed
947
948
949
950
}

void c_del_stmt(compiler_t *comp, py_parse_node_t pn) {
    if (PY_PARSE_NODE_IS_ID(pn)) {
951
        EMIT(delete_id, PY_PARSE_NODE_LEAF_ARG(pn));
Damien's avatar
Damien committed
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
    } else if (PY_PARSE_NODE_IS_STRUCT_KIND(pn, PN_power)) {
        py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;

        compile_node(comp, pns->nodes[0]); // base of the power node

        if (PY_PARSE_NODE_IS_STRUCT(pns->nodes[1])) {
            py_parse_node_struct_t *pns1 = (py_parse_node_struct_t*)pns->nodes[1];
            if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_power_trailers) {
                int n = PY_PARSE_NODE_STRUCT_NUM_NODES(pns1);
                for (int i = 0; i < n - 1; i++) {
                    compile_node(comp, pns1->nodes[i]);
                }
                assert(PY_PARSE_NODE_IS_STRUCT(pns1->nodes[n - 1]));
                pns1 = (py_parse_node_struct_t*)pns1->nodes[n - 1];
            }
            if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_paren) {
                // SyntaxError: can't delete a function call
                assert(0);
            } else if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_bracket) {
                compile_node(comp, pns1->nodes[0]);
                EMIT(delete_subscr);
            } else if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_trailer_period) {
                assert(PY_PARSE_NODE_IS_ID(pns1->nodes[0]));
                EMIT(delete_attr, PY_PARSE_NODE_LEAF_ARG(pns1->nodes[0]));
            } else {
                // shouldn't happen
                assert(0);
            }
        } else {
            // shouldn't happen
            assert(0);
        }

        if (!PY_PARSE_NODE_IS_NULL(pns->nodes[2])) {
            // SyntaxError, cannot delete
            assert(0);
        }
    } else if (PY_PARSE_NODE_IS_STRUCT_KIND(pn, PN_atom_paren)) {
        pn = ((py_parse_node_struct_t*)pn)->nodes[0];
        if (PY_PARSE_NODE_IS_STRUCT_KIND(pn, PN_testlist_comp)) {
            py_parse_node_struct_t *pns = (py_parse_node_struct_t*)pn;
            // TODO perhaps factorise testlist_comp code with other uses of PN_testlist_comp

            if (PY_PARSE_NODE_IS_STRUCT(pns->nodes[1])) {
                py_parse_node_struct_t *pns1 = (py_parse_node_struct_t*)pns->nodes[1];
                if (PY_PARSE_NODE_STRUCT_KIND(pns1) == PN_testlist_comp_3b) {
                    // sequence of one item, with trailing comma
                    assert(PY_PARSE_NODE_IS_NULL(pns1->nodes[0]));
                    c_del_stmt(comp, pns->nodes[0]);