source: vendor/python/2.5/Python/compile.c

Last change on this file was 3225, checked in by bird, 18 years ago

Python 2.5
File size: 108.2 KB
Line 
1/*
2 * This file compiles an abstract syntax tree (AST) into Python bytecode.
3 *
4 * The primary entry point is PyAST_Compile(), which returns a
5 * PyCodeObject. The compiler makes several passes to build the code
6 * object:
7 * 1. Checks for future statements. See future.c
8 * 2. Builds a symbol table. See symtable.c.
9 * 3. Generate code for basic blocks. See compiler_mod() in this file.
10 * 4. Assemble the basic blocks into final code. See assemble() in
11 * this file.
12 *
13 * Note that compiler_mod() suggests module, but the module ast type
14 * (mod_ty) has cases for expressions and interactive statements.
15 *
16 * CAUTION: The VISIT_* macros abort the current function when they
17 * encounter a problem. So don't invoke them when there is memory
18 * which needs to be released. Code blocks are OK, as the compiler
19 * structure takes care of releasing those.
20 */
21
22#include "Python.h"
23
24#include "Python-ast.h"
25#include "node.h"
26#include "pyarena.h"
27#include "ast.h"
28#include "code.h"
29#include "compile.h"
30#include "symtable.h"
31#include "opcode.h"
32
33int Py_OptimizeFlag = 0;
34
35/*
36 ISSUES:
37
38 opcode_stack_effect() function should be reviewed since stack depth bugs
39 could be really hard to find later.
40
41 Dead code is being generated (i.e. after unconditional jumps).
42 XXX(nnorwitz): not sure this is still true
43*/
44
45#define DEFAULT_BLOCK_SIZE 16
46#define DEFAULT_BLOCKS 8
47#define DEFAULT_CODE_SIZE 128
48#define DEFAULT_LNOTAB_SIZE 16
49
50struct instr {
51 unsigned i_jabs : 1;
52 unsigned i_jrel : 1;
53 unsigned i_hasarg : 1;
54 unsigned char i_opcode;
55 int i_oparg;
56 struct basicblock_ *i_target; /* target block (if jump instruction) */
57 int i_lineno;
58};
59
60typedef struct basicblock_ {
61 /* Each basicblock in a compilation unit is linked via b_list in the
62 reverse order that the block are allocated. b_list points to the next
63 block, not to be confused with b_next, which is next by control flow. */
64 struct basicblock_ *b_list;
65 /* number of instructions used */
66 int b_iused;
67 /* length of instruction array (b_instr) */
68 int b_ialloc;
69 /* pointer to an array of instructions, initially NULL */
70 struct instr *b_instr;
71 /* If b_next is non-NULL, it is a pointer to the next
72 block reached by normal control flow. */
73 struct basicblock_ *b_next;
74 /* b_seen is used to perform a DFS of basicblocks. */
75 unsigned b_seen : 1;
76 /* b_return is true if a RETURN_VALUE opcode is inserted. */
77 unsigned b_return : 1;
78 /* depth of stack upon entry of block, computed by stackdepth() */
79 int b_startdepth;
80 /* instruction offset for block, computed by assemble_jump_offsets() */
81 int b_offset;
82} basicblock;
83
84/* fblockinfo tracks the current frame block.
85
86A frame block is used to handle loops, try/except, and try/finally.
87It's called a frame block to distinguish it from a basic block in the
88compiler IR.
89*/
90
91enum fblocktype { LOOP, EXCEPT, FINALLY_TRY, FINALLY_END };
92
93struct fblockinfo {
94 enum fblocktype fb_type;
95 basicblock *fb_block;
96};
97
98/* The following items change on entry and exit of code blocks.
99 They must be saved and restored when returning to a block.
100*/
101struct compiler_unit {
102 PySTEntryObject *u_ste;
103
104 PyObject *u_name;
105 /* The following fields are dicts that map objects to
106 the index of them in co_XXX. The index is used as
107 the argument for opcodes that refer to those collections.
108 */
109 PyObject *u_consts; /* all constants */
110 PyObject *u_names; /* all names */
111 PyObject *u_varnames; /* local variables */
112 PyObject *u_cellvars; /* cell variables */
113 PyObject *u_freevars; /* free variables */
114
115 PyObject *u_private; /* for private name mangling */
116
117 int u_argcount; /* number of arguments for block */
118 /* Pointer to the most recently allocated block. By following b_list
119 members, you can reach all early allocated blocks. */
120 basicblock *u_blocks;
121 basicblock *u_curblock; /* pointer to current block */
122 int u_tmpname; /* temporary variables for list comps */
123
124 int u_nfblocks;
125 struct fblockinfo u_fblock[CO_MAXBLOCKS];
126
127 int u_firstlineno; /* the first lineno of the block */
128 int u_lineno; /* the lineno for the current stmt */
129 bool u_lineno_set; /* boolean to indicate whether instr
130 has been generated with current lineno */
131};
132
133/* This struct captures the global state of a compilation.
134
135The u pointer points to the current compilation unit, while units
136for enclosing blocks are stored in c_stack. The u and c_stack are
137managed by compiler_enter_scope() and compiler_exit_scope().
138*/
139
140struct compiler {
141 const char *c_filename;
142 struct symtable *c_st;
143 PyFutureFeatures *c_future; /* pointer to module's __future__ */
144 PyCompilerFlags *c_flags;
145
146 int c_interactive; /* true if in interactive mode */
147 int c_nestlevel;
148
149 struct compiler_unit *u; /* compiler state for current block */
150 PyObject *c_stack; /* Python list holding compiler_unit ptrs */
151 char *c_encoding; /* source encoding (a borrowed reference) */
152 PyArena *c_arena; /* pointer to memory allocation arena */
153};
154
155struct assembler {
156 PyObject *a_bytecode; /* string containing bytecode */
157 int a_offset; /* offset into bytecode */
158 int a_nblocks; /* number of reachable blocks */
159 basicblock **a_postorder; /* list of blocks in dfs postorder */
160 PyObject *a_lnotab; /* string containing lnotab */
161 int a_lnotab_off; /* offset into lnotab */
162 int a_lineno; /* last lineno of emitted instruction */
163 int a_lineno_off; /* bytecode offset of last lineno */
164};
165
166static int compiler_enter_scope(struct compiler *, identifier, void *, int);
167static void compiler_free(struct compiler *);
168static basicblock *compiler_new_block(struct compiler *);
169static int compiler_next_instr(struct compiler *, basicblock *);
170static int compiler_addop(struct compiler *, int);
171static int compiler_addop_o(struct compiler *, int, PyObject *, PyObject *);
172static int compiler_addop_i(struct compiler *, int, int);
173static int compiler_addop_j(struct compiler *, int, basicblock *, int);
174static basicblock *compiler_use_new_block(struct compiler *);
175static int compiler_error(struct compiler *, const char *);
176static int compiler_nameop(struct compiler *, identifier, expr_context_ty);
177
178static PyCodeObject *compiler_mod(struct compiler *, mod_ty);
179static int compiler_visit_stmt(struct compiler *, stmt_ty);
180static int compiler_visit_keyword(struct compiler *, keyword_ty);
181static int compiler_visit_expr(struct compiler *, expr_ty);
182static int compiler_augassign(struct compiler *, stmt_ty);
183static int compiler_visit_slice(struct compiler *, slice_ty,
184 expr_context_ty);
185
186static int compiler_push_fblock(struct compiler *, enum fblocktype,
187 basicblock *);
188static void compiler_pop_fblock(struct compiler *, enum fblocktype,
189 basicblock *);
190
191static int inplace_binop(struct compiler *, operator_ty);
192static int expr_constant(expr_ty e);
193
194static int compiler_with(struct compiler *, stmt_ty);
195
196static PyCodeObject *assemble(struct compiler *, int addNone);
197static PyObject *__doc__;
198
199PyObject *
200_Py_Mangle(PyObject *privateobj, PyObject *ident)
201{
202 /* Name mangling: __private becomes _classname__private.
203 This is independent from how the name is used. */
204 const char *p, *name = PyString_AsString(ident);
205 char *buffer;
206 size_t nlen, plen;
207 if (privateobj == NULL || !PyString_Check(privateobj) ||
208 name == NULL || name[0] != '_' || name[1] != '_') {
209 Py_INCREF(ident);
210 return ident;
211 }
212 p = PyString_AsString(privateobj);
213 nlen = strlen(name);
214 if (name[nlen-1] == '_' && name[nlen-2] == '_') {
215 Py_INCREF(ident);
216 return ident; /* Don't mangle __whatever__ */
217 }
218 /* Strip leading underscores from class name */
219 while (*p == '_')
220 p++;
221 if (*p == '\0') {
222 Py_INCREF(ident);
223 return ident; /* Don't mangle if class is just underscores */
224 }
225 plen = strlen(p);
226 ident = PyString_FromStringAndSize(NULL, 1 + nlen + plen);
227 if (!ident)
228 return 0;
229 /* ident = "_" + p[:plen] + name # i.e. 1+plen+nlen bytes */
230 buffer = PyString_AS_STRING(ident);
231 buffer[0] = '_';
232 strncpy(buffer+1, p, plen);
233 strcpy(buffer+1+plen, name);
234 return ident;
235}
236
237static int
238compiler_init(struct compiler *c)
239{
240 memset(c, 0, sizeof(struct compiler));
241
242 c->c_stack = PyList_New(0);
243 if (!c->c_stack)
244 return 0;
245
246 return 1;
247}
248
249PyCodeObject *
250PyAST_Compile(mod_ty mod, const char *filename, PyCompilerFlags *flags,
251 PyArena *arena)
252{
253 struct compiler c;
254 PyCodeObject *co = NULL;
255 PyCompilerFlags local_flags;
256 int merged;
257
258 if (!__doc__) {
259 __doc__ = PyString_InternFromString("__doc__");
260 if (!__doc__)
261 return NULL;
262 }
263
264 if (!compiler_init(&c))
265 return NULL;
266 c.c_filename = filename;
267 c.c_arena = arena;
268 c.c_future = PyFuture_FromAST(mod, filename);
269 if (c.c_future == NULL)
270 goto finally;
271 if (!flags) {
272 local_flags.cf_flags = 0;
273 flags = &local_flags;
274 }
275 merged = c.c_future->ff_features | flags->cf_flags;
276 c.c_future->ff_features = merged;
277 flags->cf_flags = merged;
278 c.c_flags = flags;
279 c.c_nestlevel = 0;
280
281 c.c_st = PySymtable_Build(mod, filename, c.c_future);
282 if (c.c_st == NULL) {
283 if (!PyErr_Occurred())
284 PyErr_SetString(PyExc_SystemError, "no symtable");
285 goto finally;
286 }
287
288 /* XXX initialize to NULL for now, need to handle */
289 c.c_encoding = NULL;
290
291 co = compiler_mod(&c, mod);
292
293 finally:
294 compiler_free(&c);
295 assert(co || PyErr_Occurred());
296 return co;
297}
298
299PyCodeObject *
300PyNode_Compile(struct _node *n, const char *filename)
301{
302 PyCodeObject *co = NULL;
303 mod_ty mod;
304 PyArena *arena = PyArena_New();
305 if (!arena)
306 return NULL;
307 mod = PyAST_FromNode(n, NULL, filename, arena);
308 if (mod)
309 co = PyAST_Compile(mod, filename, NULL, arena);
310 PyArena_Free(arena);
311 return co;
312}
313
314static void
315compiler_free(struct compiler *c)
316{
317 if (c->c_st)
318 PySymtable_Free(c->c_st);
319 if (c->c_future)
320 PyObject_Free(c->c_future);
321 Py_DECREF(c->c_stack);
322}
323
324static PyObject *
325list2dict(PyObject *list)
326{
327 Py_ssize_t i, n;
328 PyObject *v, *k;
329 PyObject *dict = PyDict_New();
330 if (!dict) return NULL;
331
332 n = PyList_Size(list);
333 for (i = 0; i < n; i++) {
334 v = PyInt_FromLong(i);
335 if (!v) {
336 Py_DECREF(dict);
337 return NULL;
338 }
339 k = PyList_GET_ITEM(list, i);
340 k = PyTuple_Pack(2, k, k->ob_type);
341 if (k == NULL || PyDict_SetItem(dict, k, v) < 0) {
342 Py_XDECREF(k);
343 Py_DECREF(v);
344 Py_DECREF(dict);
345 return NULL;
346 }
347 Py_DECREF(k);
348 Py_DECREF(v);
349 }
350 return dict;
351}
352
353/* Return new dict containing names from src that match scope(s).
354
355src is a symbol table dictionary. If the scope of a name matches
356either scope_type or flag is set, insert it into the new dict. The
357values are integers, starting at offset and increasing by one for
358each key.
359*/
360
361static PyObject *
362dictbytype(PyObject *src, int scope_type, int flag, int offset)
363{
364 Py_ssize_t pos = 0, i = offset, scope;
365 PyObject *k, *v, *dest = PyDict_New();
366
367 assert(offset >= 0);
368 if (dest == NULL)
369 return NULL;
370
371 while (PyDict_Next(src, &pos, &k, &v)) {
372 /* XXX this should probably be a macro in symtable.h */
373 assert(PyInt_Check(v));
374 scope = (PyInt_AS_LONG(v) >> SCOPE_OFF) & SCOPE_MASK;
375
376 if (scope == scope_type || PyInt_AS_LONG(v) & flag) {
377 PyObject *tuple, *item = PyInt_FromLong(i);
378 if (item == NULL) {
379 Py_DECREF(dest);
380 return NULL;
381 }
382 i++;
383 tuple = PyTuple_Pack(2, k, k->ob_type);
384 if (!tuple || PyDict_SetItem(dest, tuple, item) < 0) {
385 Py_DECREF(item);
386 Py_DECREF(dest);
387 Py_XDECREF(tuple);
388 return NULL;
389 }
390 Py_DECREF(item);
391 Py_DECREF(tuple);
392 }
393 }
394 return dest;
395}
396
397/* Begin: Peephole optimizations ----------------------------------------- */
398
399#define GETARG(arr, i) ((int)((arr[i+2]<<8) + arr[i+1]))
400#define UNCONDITIONAL_JUMP(op) (op==JUMP_ABSOLUTE || op==JUMP_FORWARD)
401#define ABSOLUTE_JUMP(op) (op==JUMP_ABSOLUTE || op==CONTINUE_LOOP)
402#define GETJUMPTGT(arr, i) (GETARG(arr,i) + (ABSOLUTE_JUMP(arr[i]) ? 0 : i+3))
403#define SETARG(arr, i, val) arr[i+2] = val>>8; arr[i+1] = val & 255
404#define CODESIZE(op) (HAS_ARG(op) ? 3 : 1)
405#define ISBASICBLOCK(blocks, start, bytes) \
406 (blocks[start]==blocks[start+bytes-1])
407
408/* Replace LOAD_CONST c1. LOAD_CONST c2 ... LOAD_CONST cn BUILD_TUPLE n
409 with LOAD_CONST (c1, c2, ... cn).
410 The consts table must still be in list form so that the
411 new constant (c1, c2, ... cn) can be appended.
412 Called with codestr pointing to the first LOAD_CONST.
413 Bails out with no change if one or more of the LOAD_CONSTs is missing.
414 Also works for BUILD_LIST when followed by an "in" or "not in" test.
415*/
416static int
417tuple_of_constants(unsigned char *codestr, int n, PyObject *consts)
418{
419 PyObject *newconst, *constant;
420 Py_ssize_t i, arg, len_consts;
421
422 /* Pre-conditions */
423 assert(PyList_CheckExact(consts));
424 assert(codestr[n*3] == BUILD_TUPLE || codestr[n*3] == BUILD_LIST);
425 assert(GETARG(codestr, (n*3)) == n);
426 for (i=0 ; i<n ; i++)
427 assert(codestr[i*3] == LOAD_CONST);
428
429 /* Buildup new tuple of constants */
430 newconst = PyTuple_New(n);
431 if (newconst == NULL)
432 return 0;
433 len_consts = PyList_GET_SIZE(consts);
434 for (i=0 ; i<n ; i++) {
435 arg = GETARG(codestr, (i*3));
436 assert(arg < len_consts);
437 constant = PyList_GET_ITEM(consts, arg);
438 Py_INCREF(constant);
439 PyTuple_SET_ITEM(newconst, i, constant);
440 }
441
442 /* Append folded constant onto consts */
443 if (PyList_Append(consts, newconst)) {
444 Py_DECREF(newconst);
445 return 0;
446 }
447 Py_DECREF(newconst);
448
449 /* Write NOPs over old LOAD_CONSTS and
450 add a new LOAD_CONST newconst on top of the BUILD_TUPLE n */
451 memset(codestr, NOP, n*3);
452 codestr[n*3] = LOAD_CONST;
453 SETARG(codestr, (n*3), len_consts);
454 return 1;
455}
456
457/* Replace LOAD_CONST c1. LOAD_CONST c2 BINOP
458 with LOAD_CONST binop(c1,c2)
459 The consts table must still be in list form so that the
460 new constant can be appended.
461 Called with codestr pointing to the first LOAD_CONST.
462 Abandons the transformation if the folding fails (i.e. 1+'a').
463 If the new constant is a sequence, only folds when the size
464 is below a threshold value. That keeps pyc files from
465 becoming large in the presence of code like: (None,)*1000.
466*/
467static int
468fold_binops_on_constants(unsigned char *codestr, PyObject *consts)
469{
470 PyObject *newconst, *v, *w;
471 Py_ssize_t len_consts, size;
472 int opcode;
473
474 /* Pre-conditions */
475 assert(PyList_CheckExact(consts));
476 assert(codestr[0] == LOAD_CONST);
477 assert(codestr[3] == LOAD_CONST);
478
479 /* Create new constant */
480 v = PyList_GET_ITEM(consts, GETARG(codestr, 0));
481 w = PyList_GET_ITEM(consts, GETARG(codestr, 3));
482 opcode = codestr[6];
483 switch (opcode) {
484 case BINARY_POWER:
485 newconst = PyNumber_Power(v, w, Py_None);
486 break;
487 case BINARY_MULTIPLY:
488 newconst = PyNumber_Multiply(v, w);
489 break;
490 case BINARY_DIVIDE:
491 /* Cannot fold this operation statically since
492 the result can depend on the run-time presence
493 of the -Qnew flag */
494 return 0;
495 case BINARY_TRUE_DIVIDE:
496 newconst = PyNumber_TrueDivide(v, w);
497 break;
498 case BINARY_FLOOR_DIVIDE:
499 newconst = PyNumber_FloorDivide(v, w);
500 break;
501 case BINARY_MODULO:
502 newconst = PyNumber_Remainder(v, w);
503 break;
504 case BINARY_ADD:
505 newconst = PyNumber_Add(v, w);
506 break;
507 case BINARY_SUBTRACT:
508 newconst = PyNumber_Subtract(v, w);
509 break;
510 case BINARY_SUBSCR:
511 newconst = PyObject_GetItem(v, w);
512 break;
513 case BINARY_LSHIFT:
514 newconst = PyNumber_Lshift(v, w);
515 break;
516 case BINARY_RSHIFT:
517 newconst = PyNumber_Rshift(v, w);
518 break;
519 case BINARY_AND:
520 newconst = PyNumber_And(v, w);
521 break;
522 case BINARY_XOR:
523 newconst = PyNumber_Xor(v, w);
524 break;
525 case BINARY_OR:
526 newconst = PyNumber_Or(v, w);
527 break;
528 default:
529 /* Called with an unknown opcode */
530 PyErr_Format(PyExc_SystemError,
531 "unexpected binary operation %d on a constant",
532 opcode);
533 return 0;
534 }
535 if (newconst == NULL) {
536 PyErr_Clear();
537 return 0;
538 }
539 size = PyObject_Size(newconst);
540 if (size == -1)
541 PyErr_Clear();
542 else if (size > 20) {
543 Py_DECREF(newconst);
544 return 0;
545 }
546
547 /* Append folded constant into consts table */
548 len_consts = PyList_GET_SIZE(consts);
549 if (PyList_Append(consts, newconst)) {
550 Py_DECREF(newconst);
551 return 0;
552 }
553 Py_DECREF(newconst);
554
555 /* Write NOP NOP NOP NOP LOAD_CONST newconst */
556 memset(codestr, NOP, 4);
557 codestr[4] = LOAD_CONST;
558 SETARG(codestr, 4, len_consts);
559 return 1;
560}
561
562static int
563fold_unaryops_on_constants(unsigned char *codestr, PyObject *consts)
564{
565 PyObject *newconst=NULL, *v;
566 Py_ssize_t len_consts;
567 int opcode;
568
569 /* Pre-conditions */
570 assert(PyList_CheckExact(consts));
571 assert(codestr[0] == LOAD_CONST);
572
573 /* Create new constant */
574 v = PyList_GET_ITEM(consts, GETARG(codestr, 0));
575 opcode = codestr[3];
576 switch (opcode) {
577 case UNARY_NEGATIVE:
578 /* Preserve the sign of -0.0 */
579 if (PyObject_IsTrue(v) == 1)
580 newconst = PyNumber_Negative(v);
581 break;
582 case UNARY_CONVERT:
583 newconst = PyObject_Repr(v);
584 break;
585 case UNARY_INVERT:
586 newconst = PyNumber_Invert(v);
587 break;
588 default:
589 /* Called with an unknown opcode */
590 PyErr_Format(PyExc_SystemError,
591 "unexpected unary operation %d on a constant",
592 opcode);
593 return 0;
594 }
595 if (newconst == NULL) {
596 PyErr_Clear();
597 return 0;
598 }
599
600 /* Append folded constant into consts table */
601 len_consts = PyList_GET_SIZE(consts);
602 if (PyList_Append(consts, newconst)) {
603 Py_DECREF(newconst);
604 return 0;
605 }
606 Py_DECREF(newconst);
607
608 /* Write NOP LOAD_CONST newconst */
609 codestr[0] = NOP;
610 codestr[1] = LOAD_CONST;
611 SETARG(codestr, 1, len_consts);
612 return 1;
613}
614
615static unsigned int *
616markblocks(unsigned char *code, int len)
617{
618 unsigned int *blocks = (unsigned int *)PyMem_Malloc(len*sizeof(int));
619 int i,j, opcode, blockcnt = 0;
620
621 if (blocks == NULL) {
622 PyErr_NoMemory();
623 return NULL;
624 }
625 memset(blocks, 0, len*sizeof(int));
626
627 /* Mark labels in the first pass */
628 for (i=0 ; i<len ; i+=CODESIZE(opcode)) {
629 opcode = code[i];
630 switch (opcode) {
631 case FOR_ITER:
632 case JUMP_FORWARD:
633 case JUMP_IF_FALSE:
634 case JUMP_IF_TRUE:
635 case JUMP_ABSOLUTE:
636 case CONTINUE_LOOP:
637 case SETUP_LOOP:
638 case SETUP_EXCEPT:
639 case SETUP_FINALLY:
640 j = GETJUMPTGT(code, i);
641 blocks[j] = 1;
642 break;
643 }
644 }
645 /* Build block numbers in the second pass */
646 for (i=0 ; i<len ; i++) {
647 blockcnt += blocks[i]; /* increment blockcnt over labels */
648 blocks[i] = blockcnt;
649 }
650 return blocks;
651}
652
653/* Perform basic peephole optimizations to components of a code object.
654 The consts object should still be in list form to allow new constants
655 to be appended.
656
657 To keep the optimizer simple, it bails out (does nothing) for code
658 containing extended arguments or that has a length over 32,700. That
659 allows us to avoid overflow and sign issues. Likewise, it bails when
660 the lineno table has complex encoding for gaps >= 255.
661
662 Optimizations are restricted to simple transformations occuring within a
663 single basic block. All transformations keep the code size the same or
664 smaller. For those that reduce size, the gaps are initially filled with
665 NOPs. Later those NOPs are removed and the jump addresses retargeted in
666 a single pass. Line numbering is adjusted accordingly. */
667
668static PyObject *
669optimize_code(PyObject *code, PyObject* consts, PyObject *names,
670 PyObject *lineno_obj)
671{
672 Py_ssize_t i, j, codelen;
673 int nops, h, adj;
674 int tgt, tgttgt, opcode;
675 unsigned char *codestr = NULL;
676 unsigned char *lineno;
677 int *addrmap = NULL;
678 int new_line, cum_orig_line, last_line, tabsiz;
679 int cumlc=0, lastlc=0; /* Count runs of consecutive LOAD_CONSTs */
680 unsigned int *blocks = NULL;
681 char *name;
682
683 /* Bail out if an exception is set */
684 if (PyErr_Occurred())
685 goto exitUnchanged;
686
687 /* Bypass optimization when the lineno table is too complex */
688 assert(PyString_Check(lineno_obj));
689 lineno = (unsigned char*)PyString_AS_STRING(lineno_obj);
690 tabsiz = PyString_GET_SIZE(lineno_obj);
691 if (memchr(lineno, 255, tabsiz) != NULL)
692 goto exitUnchanged;
693
694 /* Avoid situations where jump retargeting could overflow */
695 assert(PyString_Check(code));
696 codelen = PyString_Size(code);
697 if (codelen > 32700)
698 goto exitUnchanged;
699
700 /* Make a modifiable copy of the code string */
701 codestr = (unsigned char *)PyMem_Malloc(codelen);
702 if (codestr == NULL)
703 goto exitUnchanged;
704 codestr = (unsigned char *)memcpy(codestr,
705 PyString_AS_STRING(code), codelen);
706
707 /* Verify that RETURN_VALUE terminates the codestring. This allows
708 the various transformation patterns to look ahead several
709 instructions without additional checks to make sure they are not
710 looking beyond the end of the code string.
711 */
712 if (codestr[codelen-1] != RETURN_VALUE)
713 goto exitUnchanged;
714
715 /* Mapping to new jump targets after NOPs are removed */
716 addrmap = (int *)PyMem_Malloc(codelen * sizeof(int));
717 if (addrmap == NULL)
718 goto exitUnchanged;
719
720 blocks = markblocks(codestr, codelen);
721 if (blocks == NULL)
722 goto exitUnchanged;
723 assert(PyList_Check(consts));
724
725 for (i=0 ; i<codelen ; i += CODESIZE(codestr[i])) {
726 opcode = codestr[i];
727
728 lastlc = cumlc;
729 cumlc = 0;
730
731 switch (opcode) {
732
733 /* Replace UNARY_NOT JUMP_IF_FALSE POP_TOP with
734 with JUMP_IF_TRUE POP_TOP */
735 case UNARY_NOT:
736 if (codestr[i+1] != JUMP_IF_FALSE ||
737 codestr[i+4] != POP_TOP ||
738 !ISBASICBLOCK(blocks,i,5))
739 continue;
740 tgt = GETJUMPTGT(codestr, (i+1));
741 if (codestr[tgt] != POP_TOP)
742 continue;
743 j = GETARG(codestr, i+1) + 1;
744 codestr[i] = JUMP_IF_TRUE;
745 SETARG(codestr, i, j);
746 codestr[i+3] = POP_TOP;
747 codestr[i+4] = NOP;
748 break;
749
750 /* not a is b --> a is not b
751 not a in b --> a not in b
752 not a is not b --> a is b
753 not a not in b --> a in b
754 */
755 case COMPARE_OP:
756 j = GETARG(codestr, i);
757 if (j < 6 || j > 9 ||
758 codestr[i+3] != UNARY_NOT ||
759 !ISBASICBLOCK(blocks,i,4))
760 continue;
761 SETARG(codestr, i, (j^1));
762 codestr[i+3] = NOP;
763 break;
764
765 /* Replace LOAD_GLOBAL/LOAD_NAME None
766 with LOAD_CONST None */
767 case LOAD_NAME:
768 case LOAD_GLOBAL:
769 j = GETARG(codestr, i);
770 name = PyString_AsString(PyTuple_GET_ITEM(names, j));
771 if (name == NULL || strcmp(name, "None") != 0)
772 continue;
773 for (j=0 ; j < PyList_GET_SIZE(consts) ; j++) {
774 if (PyList_GET_ITEM(consts, j) == Py_None) {
775 codestr[i] = LOAD_CONST;
776 SETARG(codestr, i, j);
777 cumlc = lastlc + 1;
778 break;
779 }
780 }
781 break;
782
783 /* Skip over LOAD_CONST trueconst
784 JUMP_IF_FALSE xx POP_TOP */
785 case LOAD_CONST:
786 cumlc = lastlc + 1;
787 j = GETARG(codestr, i);
788 if (codestr[i+3] != JUMP_IF_FALSE ||
789 codestr[i+6] != POP_TOP ||
790 !ISBASICBLOCK(blocks,i,7) ||
791 !PyObject_IsTrue(PyList_GET_ITEM(consts, j)))
792 continue;
793 memset(codestr+i, NOP, 7);
794 cumlc = 0;
795 break;
796
797 /* Try to fold tuples of constants (includes a case for lists
798 which are only used for "in" and "not in" tests).
799 Skip over BUILD_SEQN 1 UNPACK_SEQN 1.
800 Replace BUILD_SEQN 2 UNPACK_SEQN 2 with ROT2.
801 Replace BUILD_SEQN 3 UNPACK_SEQN 3 with ROT3 ROT2. */
802 case BUILD_TUPLE:
803 case BUILD_LIST:
804 j = GETARG(codestr, i);
805 h = i - 3 * j;
806 if (h >= 0 &&
807 j <= lastlc &&
808 ((opcode == BUILD_TUPLE &&
809 ISBASICBLOCK(blocks, h, 3*(j+1))) ||
810 (opcode == BUILD_LIST &&
811 codestr[i+3]==COMPARE_OP &&
812 ISBASICBLOCK(blocks, h, 3*(j+2)) &&
813 (GETARG(codestr,i+3)==6 ||
814 GETARG(codestr,i+3)==7))) &&
815 tuple_of_constants(&codestr[h], j, consts)) {
816 assert(codestr[i] == LOAD_CONST);
817 cumlc = 1;
818 break;
819 }
820 if (codestr[i+3] != UNPACK_SEQUENCE ||
821 !ISBASICBLOCK(blocks,i,6) ||
822 j != GETARG(codestr, i+3))
823 continue;
824 if (j == 1) {
825 memset(codestr+i, NOP, 6);
826 } else if (j == 2) {
827 codestr[i] = ROT_TWO;
828 memset(codestr+i+1, NOP, 5);
829 } else if (j == 3) {
830 codestr[i] = ROT_THREE;
831 codestr[i+1] = ROT_TWO;
832 memset(codestr+i+2, NOP, 4);
833 }
834 break;
835
836 /* Fold binary ops on constants.
837 LOAD_CONST c1 LOAD_CONST c2 BINOP --> LOAD_CONST binop(c1,c2) */
838 case BINARY_POWER:
839 case BINARY_MULTIPLY:
840 case BINARY_TRUE_DIVIDE:
841 case BINARY_FLOOR_DIVIDE:
842 case BINARY_MODULO:
843 case BINARY_ADD:
844 case BINARY_SUBTRACT:
845 case BINARY_SUBSCR:
846 case BINARY_LSHIFT:
847 case BINARY_RSHIFT:
848 case BINARY_AND:
849 case BINARY_XOR:
850 case BINARY_OR:
851 if (lastlc >= 2 &&
852 ISBASICBLOCK(blocks, i-6, 7) &&
853 fold_binops_on_constants(&codestr[i-6], consts)) {
854 i -= 2;
855 assert(codestr[i] == LOAD_CONST);
856 cumlc = 1;
857 }
858 break;
859
860 /* Fold unary ops on constants.
861 LOAD_CONST c1 UNARY_OP --> LOAD_CONST unary_op(c) */
862 case UNARY_NEGATIVE:
863 case UNARY_CONVERT:
864 case UNARY_INVERT:
865 if (lastlc >= 1 &&
866 ISBASICBLOCK(blocks, i-3, 4) &&
867 fold_unaryops_on_constants(&codestr[i-3], consts)) {
868 i -= 2;
869 assert(codestr[i] == LOAD_CONST);
870 cumlc = 1;
871 }
872 break;
873
874 /* Simplify conditional jump to conditional jump where the
875 result of the first test implies the success of a similar
876 test or the failure of the opposite test.
877 Arises in code like:
878 "if a and b:"
879 "if a or b:"
880 "a and b or c"
881 "(a and b) and c"
882 x:JUMP_IF_FALSE y y:JUMP_IF_FALSE z --> x:JUMP_IF_FALSE z
883 x:JUMP_IF_FALSE y y:JUMP_IF_TRUE z --> x:JUMP_IF_FALSE y+3
884 where y+3 is the instruction following the second test.
885 */
886 case JUMP_IF_FALSE:
887 case JUMP_IF_TRUE:
888 tgt = GETJUMPTGT(codestr, i);
889 j = codestr[tgt];
890 if (j == JUMP_IF_FALSE || j == JUMP_IF_TRUE) {
891 if (j == opcode) {
892 tgttgt = GETJUMPTGT(codestr, tgt) - i - 3;
893 SETARG(codestr, i, tgttgt);
894 } else {
895 tgt -= i;
896 SETARG(codestr, i, tgt);
897 }
898 break;
899 }
900 /* Intentional fallthrough */
901
902 /* Replace jumps to unconditional jumps */
903 case FOR_ITER:
904 case JUMP_FORWARD:
905 case JUMP_ABSOLUTE:
906 case CONTINUE_LOOP:
907 case SETUP_LOOP:
908 case SETUP_EXCEPT:
909 case SETUP_FINALLY:
910 tgt = GETJUMPTGT(codestr, i);
911 if (!UNCONDITIONAL_JUMP(codestr[tgt]))
912 continue;
913 tgttgt = GETJUMPTGT(codestr, tgt);
914 if (opcode == JUMP_FORWARD) /* JMP_ABS can go backwards */
915 opcode = JUMP_ABSOLUTE;
916 if (!ABSOLUTE_JUMP(opcode))
917 tgttgt -= i + 3; /* Calc relative jump addr */
918 if (tgttgt < 0) /* No backward relative jumps */
919 continue;
920 codestr[i] = opcode;
921 SETARG(codestr, i, tgttgt);
922 break;
923
924 case EXTENDED_ARG:
925 goto exitUnchanged;
926
927 /* Replace RETURN LOAD_CONST None RETURN with just RETURN */
928 case RETURN_VALUE:
929 if (i+4 >= codelen ||
930 codestr[i+4] != RETURN_VALUE ||
931 !ISBASICBLOCK(blocks,i,5))
932 continue;
933 memset(codestr+i+1, NOP, 4);
934 break;
935 }
936 }
937
938 /* Fixup linenotab */
939 for (i=0, nops=0 ; i<codelen ; i += CODESIZE(codestr[i])) {
940 addrmap[i] = i - nops;
941 if (codestr[i] == NOP)
942 nops++;
943 }
944 cum_orig_line = 0;
945 last_line = 0;
946 for (i=0 ; i < tabsiz ; i+=2) {
947 cum_orig_line += lineno[i];
948 new_line = addrmap[cum_orig_line];
949 assert (new_line - last_line < 255);
950 lineno[i] =((unsigned char)(new_line - last_line));
951 last_line = new_line;
952 }
953
954 /* Remove NOPs and fixup jump targets */
955 for (i=0, h=0 ; i<codelen ; ) {
956 opcode = codestr[i];
957 switch (opcode) {
958 case NOP:
959 i++;
960 continue;
961
962 case JUMP_ABSOLUTE:
963 case CONTINUE_LOOP:
964 j = addrmap[GETARG(codestr, i)];
965 SETARG(codestr, i, j);
966 break;
967
968 case FOR_ITER:
969 case JUMP_FORWARD:
970 case JUMP_IF_FALSE:
971 case JUMP_IF_TRUE:
972 case SETUP_LOOP:
973 case SETUP_EXCEPT:
974 case SETUP_FINALLY:
975 j = addrmap[GETARG(codestr, i) + i + 3] - addrmap[i] - 3;
976 SETARG(codestr, i, j);
977 break;
978 }
979 adj = CODESIZE(opcode);
980 while (adj--)
981 codestr[h++] = codestr[i++];
982 }
983 assert(h + nops == codelen);
984
985 code = PyString_FromStringAndSize((char *)codestr, h);
986 PyMem_Free(addrmap);
987 PyMem_Free(codestr);
988 PyMem_Free(blocks);
989 return code;
990
991 exitUnchanged:
992 if (blocks != NULL)
993 PyMem_Free(blocks);
994 if (addrmap != NULL)
995 PyMem_Free(addrmap);
996 if (codestr != NULL)
997 PyMem_Free(codestr);
998 Py_INCREF(code);
999 return code;
1000}
1001
1002/* End: Peephole optimizations ----------------------------------------- */
1003
1004/*
1005
1006Leave this debugging code for just a little longer.
1007
1008static void
1009compiler_display_symbols(PyObject *name, PyObject *symbols)
1010{
1011PyObject *key, *value;
1012int flags;
1013Py_ssize_t pos = 0;
1014
1015fprintf(stderr, "block %s\n", PyString_AS_STRING(name));
1016while (PyDict_Next(symbols, &pos, &key, &value)) {
1017flags = PyInt_AsLong(value);
1018fprintf(stderr, "var %s:", PyString_AS_STRING(key));
1019if (flags & DEF_GLOBAL)
1020fprintf(stderr, " declared_global");
1021if (flags & DEF_LOCAL)
1022fprintf(stderr, " local");
1023if (flags & DEF_PARAM)
1024fprintf(stderr, " param");
1025if (flags & DEF_STAR)
1026fprintf(stderr, " stararg");
1027if (flags & DEF_DOUBLESTAR)
1028fprintf(stderr, " starstar");
1029if (flags & DEF_INTUPLE)
1030fprintf(stderr, " tuple");
1031if (flags & DEF_FREE)
1032fprintf(stderr, " free");
1033if (flags & DEF_FREE_GLOBAL)
1034fprintf(stderr, " global");
1035if (flags & DEF_FREE_CLASS)
1036fprintf(stderr, " free/class");
1037if (flags & DEF_IMPORT)
1038fprintf(stderr, " import");
1039fprintf(stderr, "\n");
1040}
1041 fprintf(stderr, "\n");
1042}
1043*/
1044
1045static void
1046compiler_unit_check(struct compiler_unit *u)
1047{
1048 basicblock *block;
1049 for (block = u->u_blocks; block != NULL; block = block->b_list) {
1050 assert(block != (void *)0xcbcbcbcb);
1051 assert(block != (void *)0xfbfbfbfb);
1052 assert(block != (void *)0xdbdbdbdb);
1053 if (block->b_instr != NULL) {
1054 assert(block->b_ialloc > 0);
1055 assert(block->b_iused > 0);
1056 assert(block->b_ialloc >= block->b_iused);
1057 }
1058 else {
1059 assert (block->b_iused == 0);
1060 assert (block->b_ialloc == 0);
1061 }
1062 }
1063}
1064
1065static void
1066compiler_unit_free(struct compiler_unit *u)
1067{
1068 basicblock *b, *next;
1069
1070 compiler_unit_check(u);
1071 b = u->u_blocks;
1072 while (b != NULL) {
1073 if (b->b_instr)
1074 PyObject_Free((void *)b->b_instr);
1075 next = b->b_list;
1076 PyObject_Free((void *)b);
1077 b = next;
1078 }
1079 Py_CLEAR(u->u_ste);
1080 Py_CLEAR(u->u_name);
1081 Py_CLEAR(u->u_consts);
1082 Py_CLEAR(u->u_names);
1083 Py_CLEAR(u->u_varnames);
1084 Py_CLEAR(u->u_freevars);
1085 Py_CLEAR(u->u_cellvars);
1086 Py_CLEAR(u->u_private);
1087 PyObject_Free(u);
1088}
1089
1090static int
1091compiler_enter_scope(struct compiler *c, identifier name, void *key,
1092 int lineno)
1093{
1094 struct compiler_unit *u;
1095
1096 u = (struct compiler_unit *)PyObject_Malloc(sizeof(
1097 struct compiler_unit));
1098 if (!u) {
1099 PyErr_NoMemory();
1100 return 0;
1101 }
1102 memset(u, 0, sizeof(struct compiler_unit));
1103 u->u_argcount = 0;
1104 u->u_ste = PySymtable_Lookup(c->c_st, key);
1105 if (!u->u_ste) {
1106 compiler_unit_free(u);
1107 return 0;
1108 }
1109 Py_INCREF(name);
1110 u->u_name = name;
1111 u->u_varnames = list2dict(u->u_ste->ste_varnames);
1112 u->u_cellvars = dictbytype(u->u_ste->ste_symbols, CELL, 0, 0);
1113 if (!u->u_varnames || !u->u_cellvars) {
1114 compiler_unit_free(u);
1115 return 0;
1116 }
1117
1118 u->u_freevars = dictbytype(u->u_ste->ste_symbols, FREE, DEF_FREE_CLASS,
1119 PyDict_Size(u->u_cellvars));
1120 if (!u->u_freevars) {
1121 compiler_unit_free(u);
1122 return 0;
1123 }
1124
1125 u->u_blocks = NULL;
1126 u->u_tmpname = 0;
1127 u->u_nfblocks = 0;
1128 u->u_firstlineno = lineno;
1129 u->u_lineno = 0;
1130 u->u_lineno_set = false;
1131 u->u_consts = PyDict_New();
1132 if (!u->u_consts) {
1133 compiler_unit_free(u);
1134 return 0;
1135 }
1136 u->u_names = PyDict_New();
1137 if (!u->u_names) {
1138 compiler_unit_free(u);
1139 return 0;
1140 }
1141
1142 u->u_private = NULL;
1143
1144 /* Push the old compiler_unit on the stack. */
1145 if (c->u) {
1146 PyObject *wrapper = PyCObject_FromVoidPtr(c->u, NULL);
1147 if (!wrapper || PyList_Append(c->c_stack, wrapper) < 0) {
1148 Py_XDECREF(wrapper);
1149 compiler_unit_free(u);
1150 return 0;
1151 }
1152 Py_DECREF(wrapper);
1153 u->u_private = c->u->u_private;
1154 Py_XINCREF(u->u_private);
1155 }
1156 c->u = u;
1157
1158 c->c_nestlevel++;
1159 if (compiler_use_new_block(c) == NULL)
1160 return 0;
1161
1162 return 1;
1163}
1164
1165static void
1166compiler_exit_scope(struct compiler *c)
1167{
1168 int n;
1169 PyObject *wrapper;
1170
1171 c->c_nestlevel--;
1172 compiler_unit_free(c->u);
1173 /* Restore c->u to the parent unit. */
1174 n = PyList_GET_SIZE(c->c_stack) - 1;
1175 if (n >= 0) {
1176 wrapper = PyList_GET_ITEM(c->c_stack, n);
1177 c->u = (struct compiler_unit *)PyCObject_AsVoidPtr(wrapper);
1178 /* we are deleting from a list so this really shouldn't fail */
1179 if (PySequence_DelItem(c->c_stack, n) < 0)
1180 Py_FatalError("compiler_exit_scope()");
1181 compiler_unit_check(c->u);
1182 }
1183 else
1184 c->u = NULL;
1185
1186}
1187
1188/* Allocate a new "anonymous" local variable.
1189 Used by list comprehensions and with statements.
1190*/
1191
1192static PyObject *
1193compiler_new_tmpname(struct compiler *c)
1194{
1195 char tmpname[256];
1196 PyOS_snprintf(tmpname, sizeof(tmpname), "_[%d]", ++c->u->u_tmpname);
1197 return PyString_FromString(tmpname);
1198}
1199
1200/* Allocate a new block and return a pointer to it.
1201 Returns NULL on error.
1202*/
1203
1204static basicblock *
1205compiler_new_block(struct compiler *c)
1206{
1207 basicblock *b;
1208 struct compiler_unit *u;
1209
1210 u = c->u;
1211 b = (basicblock *)PyObject_Malloc(sizeof(basicblock));
1212 if (b == NULL) {
1213 PyErr_NoMemory();
1214 return NULL;
1215 }
1216 memset((void *)b, 0, sizeof(basicblock));
1217 /* Extend the singly linked list of blocks with new block. */
1218 b->b_list = u->u_blocks;
1219 u->u_blocks = b;
1220 return b;
1221}
1222
1223static basicblock *
1224compiler_use_new_block(struct compiler *c)
1225{
1226 basicblock *block = compiler_new_block(c);
1227 if (block == NULL)
1228 return NULL;
1229 c->u->u_curblock = block;
1230 return block;
1231}
1232
1233static basicblock *
1234compiler_next_block(struct compiler *c)
1235{
1236 basicblock *block = compiler_new_block(c);
1237 if (block == NULL)
1238 return NULL;
1239 c->u->u_curblock->b_next = block;
1240 c->u->u_curblock = block;
1241 return block;
1242}
1243
1244static basicblock *
1245compiler_use_next_block(struct compiler *c, basicblock *block)
1246{
1247 assert(block != NULL);
1248 c->u->u_curblock->b_next = block;
1249 c->u->u_curblock = block;
1250 return block;
1251}
1252
1253/* Returns the offset of the next instruction in the current block's
1254 b_instr array. Resizes the b_instr as necessary.
1255 Returns -1 on failure.
1256 */
1257
1258static int
1259compiler_next_instr(struct compiler *c, basicblock *b)
1260{
1261 assert(b != NULL);
1262 if (b->b_instr == NULL) {
1263 b->b_instr = (struct instr *)PyObject_Malloc(
1264 sizeof(struct instr) * DEFAULT_BLOCK_SIZE);
1265 if (b->b_instr == NULL) {
1266 PyErr_NoMemory();
1267 return -1;
1268 }
1269 b->b_ialloc = DEFAULT_BLOCK_SIZE;
1270 memset((char *)b->b_instr, 0,
1271 sizeof(struct instr) * DEFAULT_BLOCK_SIZE);
1272 }
1273 else if (b->b_iused == b->b_ialloc) {
1274 struct instr *tmp;
1275 size_t oldsize, newsize;
1276 oldsize = b->b_ialloc * sizeof(struct instr);
1277 newsize = oldsize << 1;
1278 if (newsize == 0) {
1279 PyErr_NoMemory();
1280 return -1;
1281 }
1282 b->b_ialloc <<= 1;
1283 tmp = (struct instr *)PyObject_Realloc(
1284 (void *)b->b_instr, newsize);
1285 if (tmp == NULL) {
1286 PyErr_NoMemory();
1287 return -1;
1288 }
1289 b->b_instr = tmp;
1290 memset((char *)b->b_instr + oldsize, 0, newsize - oldsize);
1291 }
1292 return b->b_iused++;
1293}
1294
1295/* Set the i_lineno member of the instruction at offse off if the
1296 line number for the current expression/statement (?) has not
1297 already been set. If it has been set, the call has no effect.
1298
1299 Every time a new node is b
1300 */
1301
1302static void
1303compiler_set_lineno(struct compiler *c, int off)
1304{
1305 basicblock *b;
1306 if (c->u->u_lineno_set)
1307 return;
1308 c->u->u_lineno_set = true;
1309 b = c->u->u_curblock;
1310 b->b_instr[off].i_lineno = c->u->u_lineno;
1311}
1312
1313static int
1314opcode_stack_effect(int opcode, int oparg)
1315{
1316 switch (opcode) {
1317 case POP_TOP:
1318 return -1;
1319 case ROT_TWO:
1320 case ROT_THREE:
1321 return 0;
1322 case DUP_TOP:
1323 return 1;
1324 case ROT_FOUR:
1325 return 0;
1326
1327 case UNARY_POSITIVE:
1328 case UNARY_NEGATIVE:
1329 case UNARY_NOT:
1330 case UNARY_CONVERT:
1331 case UNARY_INVERT:
1332 return 0;
1333
1334 case LIST_APPEND:
1335 return -2;
1336
1337 case BINARY_POWER:
1338 case BINARY_MULTIPLY:
1339 case BINARY_DIVIDE:
1340 case BINARY_MODULO:
1341 case BINARY_ADD:
1342 case BINARY_SUBTRACT:
1343 case BINARY_SUBSCR:
1344 case BINARY_FLOOR_DIVIDE:
1345 case BINARY_TRUE_DIVIDE:
1346 return -1;
1347 case INPLACE_FLOOR_DIVIDE:
1348 case INPLACE_TRUE_DIVIDE:
1349 return -1;
1350
1351 case SLICE+0:
1352 return 1;
1353 case SLICE+1:
1354 return 0;
1355 case SLICE+2:
1356 return 0;
1357 case SLICE+3:
1358 return -1;
1359
1360 case STORE_SLICE+0:
1361 return -2;
1362 case STORE_SLICE+1:
1363 return -3;
1364 case STORE_SLICE+2:
1365 return -3;
1366 case STORE_SLICE+3:
1367 return -4;
1368
1369 case DELETE_SLICE+0:
1370 return -1;
1371 case DELETE_SLICE+1:
1372 return -2;
1373 case DELETE_SLICE+2:
1374 return -2;
1375 case DELETE_SLICE+3:
1376 return -3;
1377
1378 case INPLACE_ADD:
1379 case INPLACE_SUBTRACT:
1380 case INPLACE_MULTIPLY:
1381 case INPLACE_DIVIDE:
1382 case INPLACE_MODULO:
1383 return -1;
1384 case STORE_SUBSCR:
1385 return -3;
1386 case DELETE_SUBSCR:
1387 return -2;
1388
1389 case BINARY_LSHIFT:
1390 case BINARY_RSHIFT:
1391 case BINARY_AND:
1392 case BINARY_XOR:
1393 case BINARY_OR:
1394 return -1;
1395 case INPLACE_POWER:
1396 return -1;
1397 case GET_ITER:
1398 return 0;
1399
1400 case PRINT_EXPR:
1401 return -1;
1402 case PRINT_ITEM:
1403 return -1;
1404 case PRINT_NEWLINE:
1405 return 0;
1406 case PRINT_ITEM_TO:
1407 return -2;
1408 case PRINT_NEWLINE_TO:
1409 return -1;
1410 case INPLACE_LSHIFT:
1411 case INPLACE_RSHIFT:
1412 case INPLACE_AND:
1413 case INPLACE_XOR:
1414 case INPLACE_OR:
1415 return -1;
1416 case BREAK_LOOP:
1417 return 0;
1418 case WITH_CLEANUP:
1419 return -1; /* XXX Sometimes more */
1420 case LOAD_LOCALS:
1421 return 1;
1422 case RETURN_VALUE:
1423 return -1;
1424 case IMPORT_STAR:
1425 return -1;
1426 case EXEC_STMT:
1427 return -3;
1428 case YIELD_VALUE:
1429 return 0;
1430
1431 case POP_BLOCK:
1432 return 0;
1433 case END_FINALLY:
1434 return -1; /* or -2 or -3 if exception occurred */
1435 case BUILD_CLASS:
1436 return -2;
1437
1438 case STORE_NAME:
1439 return -1;
1440 case DELETE_NAME:
1441 return 0;
1442 case UNPACK_SEQUENCE:
1443 return oparg-1;
1444 case FOR_ITER:
1445 return 1;
1446
1447 case STORE_ATTR:
1448 return -2;
1449 case DELETE_ATTR:
1450 return -1;
1451 case STORE_GLOBAL:
1452 return -1;
1453 case DELETE_GLOBAL:
1454 return 0;
1455 case DUP_TOPX:
1456 return oparg;
1457 case LOAD_CONST:
1458 return 1;
1459 case LOAD_NAME:
1460 return 1;
1461 case BUILD_TUPLE:
1462 case BUILD_LIST:
1463 return 1-oparg;
1464 case BUILD_MAP:
1465 return 1;
1466 case LOAD_ATTR:
1467 return 0;
1468 case COMPARE_OP:
1469 return -1;
1470 case IMPORT_NAME:
1471 return 0;
1472 case IMPORT_FROM:
1473 return 1;
1474
1475 case JUMP_FORWARD:
1476 case JUMP_IF_FALSE:
1477 case JUMP_IF_TRUE:
1478 case JUMP_ABSOLUTE:
1479 return 0;
1480
1481 case LOAD_GLOBAL:
1482 return 1;
1483
1484 case CONTINUE_LOOP:
1485 return 0;
1486 case SETUP_LOOP:
1487 return 0;
1488 case SETUP_EXCEPT:
1489 case SETUP_FINALLY:
1490 return 3; /* actually pushed by an exception */
1491
1492 case LOAD_FAST:
1493 return 1;
1494 case STORE_FAST:
1495 return -1;
1496 case DELETE_FAST:
1497 return 0;
1498
1499 case RAISE_VARARGS:
1500 return -oparg;
1501#define NARGS(o) (((o) % 256) + 2*((o) / 256))
1502 case CALL_FUNCTION:
1503 return -NARGS(oparg);
1504 case CALL_FUNCTION_VAR:
1505 case CALL_FUNCTION_KW:
1506 return -NARGS(oparg)-1;
1507 case CALL_FUNCTION_VAR_KW:
1508 return -NARGS(oparg)-2;
1509#undef NARGS
1510 case MAKE_FUNCTION:
1511 return -oparg;
1512 case BUILD_SLICE:
1513 if (oparg == 3)
1514 return -2;
1515 else
1516 return -1;
1517
1518 case MAKE_CLOSURE:
1519 return -oparg;
1520 case LOAD_CLOSURE:
1521 return 1;
1522 case LOAD_DEREF:
1523 return 1;
1524 case STORE_DEREF:
1525 return -1;
1526 default:
1527 fprintf(stderr, "opcode = %d\n", opcode);
1528 Py_FatalError("opcode_stack_effect()");
1529
1530 }
1531 return 0; /* not reachable */
1532}
1533
1534/* Add an opcode with no argument.
1535 Returns 0 on failure, 1 on success.
1536*/
1537
1538static int
1539compiler_addop(struct compiler *c, int opcode)
1540{
1541 basicblock *b;
1542 struct instr *i;
1543 int off;
1544 off = compiler_next_instr(c, c->u->u_curblock);
1545 if (off < 0)
1546 return 0;
1547 b = c->u->u_curblock;
1548 i = &b->b_instr[off];
1549 i->i_opcode = opcode;
1550 i->i_hasarg = 0;
1551 if (opcode == RETURN_VALUE)
1552 b->b_return = 1;
1553 compiler_set_lineno(c, off);
1554 return 1;
1555}
1556
1557static int
1558compiler_add_o(struct compiler *c, PyObject *dict, PyObject *o)
1559{
1560 PyObject *t, *v;
1561 Py_ssize_t arg;
1562
1563 /* necessary to make sure types aren't coerced (e.g., int and long) */
1564 t = PyTuple_Pack(2, o, o->ob_type);
1565 if (t == NULL)
1566 return -1;
1567
1568 v = PyDict_GetItem(dict, t);
1569 if (!v) {
1570 arg = PyDict_Size(dict);
1571 v = PyInt_FromLong(arg);
1572 if (!v) {
1573 Py_DECREF(t);
1574 return -1;
1575 }
1576 if (PyDict_SetItem(dict, t, v) < 0) {
1577 Py_DECREF(t);
1578 Py_DECREF(v);
1579 return -1;
1580 }
1581 Py_DECREF(v);
1582 }
1583 else
1584 arg = PyInt_AsLong(v);
1585 Py_DECREF(t);
1586 return arg;
1587}
1588
1589static int
1590compiler_addop_o(struct compiler *c, int opcode, PyObject *dict,
1591 PyObject *o)
1592{
1593 int arg = compiler_add_o(c, dict, o);
1594 if (arg < 0)
1595 return 0;
1596 return compiler_addop_i(c, opcode, arg);
1597}
1598
1599static int
1600compiler_addop_name(struct compiler *c, int opcode, PyObject *dict,
1601 PyObject *o)
1602{
1603 int arg;
1604 PyObject *mangled = _Py_Mangle(c->u->u_private, o);
1605 if (!mangled)
1606 return 0;
1607 arg = compiler_add_o(c, dict, mangled);
1608 Py_DECREF(mangled);
1609 if (arg < 0)
1610 return 0;
1611 return compiler_addop_i(c, opcode, arg);
1612}
1613
1614/* Add an opcode with an integer argument.
1615 Returns 0 on failure, 1 on success.
1616*/
1617
1618static int
1619compiler_addop_i(struct compiler *c, int opcode, int oparg)
1620{
1621 struct instr *i;
1622 int off;
1623 off = compiler_next_instr(c, c->u->u_curblock);
1624 if (off < 0)
1625 return 0;
1626 i = &c->u->u_curblock->b_instr[off];
1627 i->i_opcode = opcode;
1628 i->i_oparg = oparg;
1629 i->i_hasarg = 1;
1630 compiler_set_lineno(c, off);
1631 return 1;
1632}
1633
1634static int
1635compiler_addop_j(struct compiler *c, int opcode, basicblock *b, int absolute)
1636{
1637 struct instr *i;
1638 int off;
1639
1640 assert(b != NULL);
1641 off = compiler_next_instr(c, c->u->u_curblock);
1642 if (off < 0)
1643 return 0;
1644 i = &c->u->u_curblock->b_instr[off];
1645 i->i_opcode = opcode;
1646 i->i_target = b;
1647 i->i_hasarg = 1;
1648 if (absolute)
1649 i->i_jabs = 1;
1650 else
1651 i->i_jrel = 1;
1652 compiler_set_lineno(c, off);
1653 return 1;
1654}
1655
1656/* The distinction between NEW_BLOCK and NEXT_BLOCK is subtle. (I'd
1657 like to find better names.) NEW_BLOCK() creates a new block and sets
1658 it as the current block. NEXT_BLOCK() also creates an implicit jump
1659 from the current block to the new block.
1660*/
1661
1662/* XXX The returns inside these macros make it impossible to decref
1663 objects created in the local function.
1664*/
1665
1666
1667#define NEW_BLOCK(C) { \
1668 if (compiler_use_new_block((C)) == NULL) \
1669 return 0; \
1670}
1671
1672#define NEXT_BLOCK(C) { \
1673 if (compiler_next_block((C)) == NULL) \
1674 return 0; \
1675}
1676
1677#define ADDOP(C, OP) { \
1678 if (!compiler_addop((C), (OP))) \
1679 return 0; \
1680}
1681
1682#define ADDOP_IN_SCOPE(C, OP) { \
1683 if (!compiler_addop((C), (OP))) { \
1684 compiler_exit_scope(c); \
1685 return 0; \
1686 } \
1687}
1688
1689#define ADDOP_O(C, OP, O, TYPE) { \
1690 if (!compiler_addop_o((C), (OP), (C)->u->u_ ## TYPE, (O))) \
1691 return 0; \
1692}
1693
1694#define ADDOP_NAME(C, OP, O, TYPE) { \
1695 if (!compiler_addop_name((C), (OP), (C)->u->u_ ## TYPE, (O))) \
1696 return 0; \
1697}
1698
1699#define ADDOP_I(C, OP, O) { \
1700 if (!compiler_addop_i((C), (OP), (O))) \
1701 return 0; \
1702}
1703
1704#define ADDOP_JABS(C, OP, O) { \
1705 if (!compiler_addop_j((C), (OP), (O), 1)) \
1706 return 0; \
1707}
1708
1709#define ADDOP_JREL(C, OP, O) { \
1710 if (!compiler_addop_j((C), (OP), (O), 0)) \
1711 return 0; \
1712}
1713
1714/* VISIT and VISIT_SEQ takes an ASDL type as their second argument. They use
1715 the ASDL name to synthesize the name of the C type and the visit function.
1716*/
1717
1718#define VISIT(C, TYPE, V) {\
1719 if (!compiler_visit_ ## TYPE((C), (V))) \
1720 return 0; \
1721}
1722
1723#define VISIT_IN_SCOPE(C, TYPE, V) {\
1724 if (!compiler_visit_ ## TYPE((C), (V))) { \
1725 compiler_exit_scope(c); \
1726 return 0; \
1727 } \
1728}
1729
1730#define VISIT_SLICE(C, V, CTX) {\
1731 if (!compiler_visit_slice((C), (V), (CTX))) \
1732 return 0; \
1733}
1734
1735#define VISIT_SEQ(C, TYPE, SEQ) { \
1736 int _i; \
1737 asdl_seq *seq = (SEQ); /* avoid variable capture */ \
1738 for (_i = 0; _i < asdl_seq_LEN(seq); _i++) { \
1739 TYPE ## _ty elt = (TYPE ## _ty)asdl_seq_GET(seq, _i); \
1740 if (!compiler_visit_ ## TYPE((C), elt)) \
1741 return 0; \
1742 } \
1743}
1744
1745#define VISIT_SEQ_IN_SCOPE(C, TYPE, SEQ) { \
1746 int _i; \
1747 asdl_seq *seq = (SEQ); /* avoid variable capture */ \
1748 for (_i = 0; _i < asdl_seq_LEN(seq); _i++) { \
1749 TYPE ## _ty elt = (TYPE ## _ty)asdl_seq_GET(seq, _i); \
1750 if (!compiler_visit_ ## TYPE((C), elt)) { \
1751 compiler_exit_scope(c); \
1752 return 0; \
1753 } \
1754 } \
1755}
1756
1757static int
1758compiler_isdocstring(stmt_ty s)
1759{
1760 if (s->kind != Expr_kind)
1761 return 0;
1762 return s->v.Expr.value->kind == Str_kind;
1763}
1764
1765/* Compile a sequence of statements, checking for a docstring. */
1766
1767static int
1768compiler_body(struct compiler *c, asdl_seq *stmts)
1769{
1770 int i = 0;
1771 stmt_ty st;
1772
1773 if (!asdl_seq_LEN(stmts))
1774 return 1;
1775 st = (stmt_ty)asdl_seq_GET(stmts, 0);
1776 if (compiler_isdocstring(st)) {
1777 i = 1;
1778 VISIT(c, expr, st->v.Expr.value);
1779 if (!compiler_nameop(c, __doc__, Store))
1780 return 0;
1781 }
1782 for (; i < asdl_seq_LEN(stmts); i++)
1783 VISIT(c, stmt, (stmt_ty)asdl_seq_GET(stmts, i));
1784 return 1;
1785}
1786
1787static PyCodeObject *
1788compiler_mod(struct compiler *c, mod_ty mod)
1789{
1790 PyCodeObject *co;
1791 int addNone = 1;
1792 static PyObject *module;
1793 if (!module) {
1794 module = PyString_FromString("<module>");
1795 if (!module)
1796 return NULL;
1797 }
1798 /* Use 0 for firstlineno initially, will fixup in assemble(). */
1799 if (!compiler_enter_scope(c, module, mod, 0))
1800 return NULL;
1801 switch (mod->kind) {
1802 case Module_kind:
1803 if (!compiler_body(c, mod->v.Module.body)) {
1804 compiler_exit_scope(c);
1805 return 0;
1806 }
1807 break;
1808 case Interactive_kind:
1809 c->c_interactive = 1;
1810 VISIT_SEQ_IN_SCOPE(c, stmt,
1811 mod->v.Interactive.body);
1812 break;
1813 case Expression_kind:
1814 VISIT_IN_SCOPE(c, expr, mod->v.Expression.body);
1815 addNone = 0;
1816 break;
1817 case Suite_kind:
1818 PyErr_SetString(PyExc_SystemError,
1819 "suite should not be possible");
1820 return 0;
1821 default:
1822 PyErr_Format(PyExc_SystemError,
1823 "module kind %d should not be possible",
1824 mod->kind);
1825 return 0;
1826 }
1827 co = assemble(c, addNone);
1828 compiler_exit_scope(c);
1829 return co;
1830}
1831
1832/* The test for LOCAL must come before the test for FREE in order to
1833 handle classes where name is both local and free. The local var is
1834 a method and the free var is a free var referenced within a method.
1835*/
1836
1837static int
1838get_ref_type(struct compiler *c, PyObject *name)
1839{
1840 int scope = PyST_GetScope(c->u->u_ste, name);
1841 if (scope == 0) {
1842 char buf[350];
1843 PyOS_snprintf(buf, sizeof(buf),
1844 "unknown scope for %.100s in %.100s(%s) in %s\n"
1845 "symbols: %s\nlocals: %s\nglobals: %s\n",
1846 PyString_AS_STRING(name),
1847 PyString_AS_STRING(c->u->u_name),
1848 PyObject_REPR(c->u->u_ste->ste_id),
1849 c->c_filename,
1850 PyObject_REPR(c->u->u_ste->ste_symbols),
1851 PyObject_REPR(c->u->u_varnames),
1852 PyObject_REPR(c->u->u_names)
1853 );
1854 Py_FatalError(buf);
1855 }
1856
1857 return scope;
1858}
1859
1860static int
1861compiler_lookup_arg(PyObject *dict, PyObject *name)
1862{
1863 PyObject *k, *v;
1864 k = PyTuple_Pack(2, name, name->ob_type);
1865 if (k == NULL)
1866 return -1;
1867 v = PyDict_GetItem(dict, k);
1868 Py_DECREF(k);
1869 if (v == NULL)
1870 return -1;
1871 return PyInt_AS_LONG(v);
1872}
1873
1874static int
1875compiler_make_closure(struct compiler *c, PyCodeObject *co, int args)
1876{
1877 int i, free = PyCode_GetNumFree(co);
1878 if (free == 0) {
1879 ADDOP_O(c, LOAD_CONST, (PyObject*)co, consts);
1880 ADDOP_I(c, MAKE_FUNCTION, args);
1881 return 1;
1882 }
1883 for (i = 0; i < free; ++i) {
1884 /* Bypass com_addop_varname because it will generate
1885 LOAD_DEREF but LOAD_CLOSURE is needed.
1886 */
1887 PyObject *name = PyTuple_GET_ITEM(co->co_freevars, i);
1888 int arg, reftype;
1889
1890 /* Special case: If a class contains a method with a
1891 free variable that has the same name as a method,
1892 the name will be considered free *and* local in the
1893 class. It should be handled by the closure, as
1894 well as by the normal name loookup logic.
1895 */
1896 reftype = get_ref_type(c, name);
1897 if (reftype == CELL)
1898 arg = compiler_lookup_arg(c->u->u_cellvars, name);
1899 else /* (reftype == FREE) */
1900 arg = compiler_lookup_arg(c->u->u_freevars, name);
1901 if (arg == -1) {
1902 printf("lookup %s in %s %d %d\n"
1903 "freevars of %s: %s\n",
1904 PyObject_REPR(name),
1905 PyString_AS_STRING(c->u->u_name),
1906 reftype, arg,
1907 PyString_AS_STRING(co->co_name),
1908 PyObject_REPR(co->co_freevars));
1909 Py_FatalError("compiler_make_closure()");
1910 }
1911 ADDOP_I(c, LOAD_CLOSURE, arg);
1912 }
1913 ADDOP_I(c, BUILD_TUPLE, free);
1914 ADDOP_O(c, LOAD_CONST, (PyObject*)co, consts);
1915 ADDOP_I(c, MAKE_CLOSURE, args);
1916 return 1;
1917}
1918
1919static int
1920compiler_decorators(struct compiler *c, asdl_seq* decos)
1921{
1922 int i;
1923
1924 if (!decos)
1925 return 1;
1926
1927 for (i = 0; i < asdl_seq_LEN(decos); i++) {
1928 VISIT(c, expr, (expr_ty)asdl_seq_GET(decos, i));
1929 }
1930 return 1;
1931}
1932
1933static int
1934compiler_arguments(struct compiler *c, arguments_ty args)
1935{
1936 int i;
1937 int n = asdl_seq_LEN(args->args);
1938 /* Correctly handle nested argument lists */
1939 for (i = 0; i < n; i++) {
1940 expr_ty arg = (expr_ty)asdl_seq_GET(args->args, i);
1941 if (arg->kind == Tuple_kind) {
1942 PyObject *id = PyString_FromFormat(".%d", i);
1943 if (id == NULL) {
1944 return 0;
1945 }
1946 if (!compiler_nameop(c, id, Load)) {
1947 Py_DECREF(id);
1948 return 0;
1949 }
1950 Py_DECREF(id);
1951 VISIT(c, expr, arg);
1952 }
1953 }
1954 return 1;
1955}
1956
1957static int
1958compiler_function(struct compiler *c, stmt_ty s)
1959{
1960 PyCodeObject *co;
1961 PyObject *first_const = Py_None;
1962 arguments_ty args = s->v.FunctionDef.args;
1963 asdl_seq* decos = s->v.FunctionDef.decorators;
1964 stmt_ty st;
1965 int i, n, docstring;
1966
1967 assert(s->kind == FunctionDef_kind);
1968
1969 if (!compiler_decorators(c, decos))
1970 return 0;
1971 if (args->defaults)
1972 VISIT_SEQ(c, expr, args->defaults);
1973 if (!compiler_enter_scope(c, s->v.FunctionDef.name, (void *)s,
1974 s->lineno))
1975 return 0;
1976
1977 st = (stmt_ty)asdl_seq_GET(s->v.FunctionDef.body, 0);
1978 docstring = compiler_isdocstring(st);
1979 if (docstring)
1980 first_const = st->v.Expr.value->v.Str.s;
1981 if (compiler_add_o(c, c->u->u_consts, first_const) < 0) {
1982 compiler_exit_scope(c);
1983 return 0;
1984 }
1985
1986 /* unpack nested arguments */
1987 compiler_arguments(c, args);
1988
1989 c->u->u_argcount = asdl_seq_LEN(args->args);
1990 n = asdl_seq_LEN(s->v.FunctionDef.body);
1991 /* if there was a docstring, we need to skip the first statement */
1992 for (i = docstring; i < n; i++) {
1993 st = (stmt_ty)asdl_seq_GET(s->v.FunctionDef.body, i);
1994 VISIT_IN_SCOPE(c, stmt, st);
1995 }
1996 co = assemble(c, 1);
1997 compiler_exit_scope(c);
1998 if (co == NULL)
1999 return 0;
2000
2001 compiler_make_closure(c, co, asdl_seq_LEN(args->defaults));
2002 Py_DECREF(co);
2003
2004 for (i = 0; i < asdl_seq_LEN(decos); i++) {
2005 ADDOP_I(c, CALL_FUNCTION, 1);
2006 }
2007
2008 return compiler_nameop(c, s->v.FunctionDef.name, Store);
2009}
2010
2011static int
2012compiler_class(struct compiler *c, stmt_ty s)
2013{
2014 int n;
2015 PyCodeObject *co;
2016 PyObject *str;
2017 /* push class name on stack, needed by BUILD_CLASS */
2018 ADDOP_O(c, LOAD_CONST, s->v.ClassDef.name, consts);
2019 /* push the tuple of base classes on the stack */
2020 n = asdl_seq_LEN(s->v.ClassDef.bases);
2021 if (n > 0)
2022 VISIT_SEQ(c, expr, s->v.ClassDef.bases);
2023 ADDOP_I(c, BUILD_TUPLE, n);
2024 if (!compiler_enter_scope(c, s->v.ClassDef.name, (void *)s,
2025 s->lineno))
2026 return 0;
2027 c->u->u_private = s->v.ClassDef.name;
2028 Py_INCREF(c->u->u_private);
2029 str = PyString_InternFromString("__name__");
2030 if (!str || !compiler_nameop(c, str, Load)) {
2031 Py_XDECREF(str);
2032 compiler_exit_scope(c);
2033 return 0;
2034 }
2035
2036 Py_DECREF(str);
2037 str = PyString_InternFromString("__module__");
2038 if (!str || !compiler_nameop(c, str, Store)) {
2039 Py_XDECREF(str);
2040 compiler_exit_scope(c);
2041 return 0;
2042 }
2043 Py_DECREF(str);
2044
2045 if (!compiler_body(c, s->v.ClassDef.body)) {
2046 compiler_exit_scope(c);
2047 return 0;
2048 }
2049
2050 ADDOP_IN_SCOPE(c, LOAD_LOCALS);
2051 ADDOP_IN_SCOPE(c, RETURN_VALUE);
2052 co = assemble(c, 1);
2053 compiler_exit_scope(c);
2054 if (co == NULL)
2055 return 0;
2056
2057 compiler_make_closure(c, co, 0);
2058 Py_DECREF(co);
2059
2060 ADDOP_I(c, CALL_FUNCTION, 0);
2061 ADDOP(c, BUILD_CLASS);
2062 if (!compiler_nameop(c, s->v.ClassDef.name, Store))
2063 return 0;
2064 return 1;
2065}
2066
2067static int
2068compiler_ifexp(struct compiler *c, expr_ty e)
2069{
2070 basicblock *end, *next;
2071
2072 assert(e->kind == IfExp_kind);
2073 end = compiler_new_block(c);
2074 if (end == NULL)
2075 return 0;
2076 next = compiler_new_block(c);
2077 if (next == NULL)
2078 return 0;
2079 VISIT(c, expr, e->v.IfExp.test);
2080 ADDOP_JREL(c, JUMP_IF_FALSE, next);
2081 ADDOP(c, POP_TOP);
2082 VISIT(c, expr, e->v.IfExp.body);
2083 ADDOP_JREL(c, JUMP_FORWARD, end);
2084 compiler_use_next_block(c, next);
2085 ADDOP(c, POP_TOP);
2086 VISIT(c, expr, e->v.IfExp.orelse);
2087 compiler_use_next_block(c, end);
2088 return 1;
2089}
2090
2091static int
2092compiler_lambda(struct compiler *c, expr_ty e)
2093{
2094 PyCodeObject *co;
2095 static identifier name;
2096 arguments_ty args = e->v.Lambda.args;
2097 assert(e->kind == Lambda_kind);
2098
2099 if (!name) {
2100 name = PyString_InternFromString("<lambda>");
2101 if (!name)
2102 return 0;
2103 }
2104
2105 if (args->defaults)
2106 VISIT_SEQ(c, expr, args->defaults);
2107 if (!compiler_enter_scope(c, name, (void *)e, e->lineno))
2108 return 0;
2109
2110 /* unpack nested arguments */
2111 compiler_arguments(c, args);
2112
2113 c->u->u_argcount = asdl_seq_LEN(args->args);
2114 VISIT_IN_SCOPE(c, expr, e->v.Lambda.body);
2115 ADDOP_IN_SCOPE(c, RETURN_VALUE);
2116 co = assemble(c, 1);
2117 compiler_exit_scope(c);
2118 if (co == NULL)
2119 return 0;
2120
2121 compiler_make_closure(c, co, asdl_seq_LEN(args->defaults));
2122 Py_DECREF(co);
2123
2124 return 1;
2125}
2126
2127static int
2128compiler_print(struct compiler *c, stmt_ty s)
2129{
2130 int i, n;
2131 bool dest;
2132
2133 assert(s->kind == Print_kind);
2134 n = asdl_seq_LEN(s->v.Print.values);
2135 dest = false;
2136 if (s->v.Print.dest) {
2137 VISIT(c, expr, s->v.Print.dest);
2138 dest = true;
2139 }
2140 for (i = 0; i < n; i++) {
2141 expr_ty e = (expr_ty)asdl_seq_GET(s->v.Print.values, i);
2142 if (dest) {
2143 ADDOP(c, DUP_TOP);
2144 VISIT(c, expr, e);
2145 ADDOP(c, ROT_TWO);
2146 ADDOP(c, PRINT_ITEM_TO);
2147 }
2148 else {
2149 VISIT(c, expr, e);
2150 ADDOP(c, PRINT_ITEM);
2151 }
2152 }
2153 if (s->v.Print.nl) {
2154 if (dest)
2155 ADDOP(c, PRINT_NEWLINE_TO)
2156 else
2157 ADDOP(c, PRINT_NEWLINE)
2158 }
2159 else if (dest)
2160 ADDOP(c, POP_TOP);
2161 return 1;
2162}
2163
2164static int
2165compiler_if(struct compiler *c, stmt_ty s)
2166{
2167 basicblock *end, *next;
2168 int constant;
2169 assert(s->kind == If_kind);
2170 end = compiler_new_block(c);
2171 if (end == NULL)
2172 return 0;
2173 next = compiler_new_block(c);
2174 if (next == NULL)
2175 return 0;
2176
2177 constant = expr_constant(s->v.If.test);
2178 /* constant = 0: "if 0"
2179 * constant = 1: "if 1", "if 2", ...
2180 * constant = -1: rest */
2181 if (constant == 0) {
2182 if (s->v.If.orelse)
2183 VISIT_SEQ(c, stmt, s->v.If.orelse);
2184 } else if (constant == 1) {
2185 VISIT_SEQ(c, stmt, s->v.If.body);
2186 } else {
2187 VISIT(c, expr, s->v.If.test);
2188 ADDOP_JREL(c, JUMP_IF_FALSE, next);
2189 ADDOP(c, POP_TOP);
2190 VISIT_SEQ(c, stmt, s->v.If.body);
2191 ADDOP_JREL(c, JUMP_FORWARD, end);
2192 compiler_use_next_block(c, next);
2193 ADDOP(c, POP_TOP);
2194 if (s->v.If.orelse)
2195 VISIT_SEQ(c, stmt, s->v.If.orelse);
2196 }
2197 compiler_use_next_block(c, end);
2198 return 1;
2199}
2200
2201static int
2202compiler_for(struct compiler *c, stmt_ty s)
2203{
2204 basicblock *start, *cleanup, *end;
2205
2206 start = compiler_new_block(c);
2207 cleanup = compiler_new_block(c);
2208 end = compiler_new_block(c);
2209 if (start == NULL || end == NULL || cleanup == NULL)
2210 return 0;
2211 ADDOP_JREL(c, SETUP_LOOP, end);
2212 if (!compiler_push_fblock(c, LOOP, start))
2213 return 0;
2214 VISIT(c, expr, s->v.For.iter);
2215 ADDOP(c, GET_ITER);
2216 compiler_use_next_block(c, start);
2217 /* XXX(nnorwitz): is there a better way to handle this?
2218 for loops are special, we want to be able to trace them
2219 each time around, so we need to set an extra line number. */
2220 c->u->u_lineno_set = false;
2221 ADDOP_JREL(c, FOR_ITER, cleanup);
2222 VISIT(c, expr, s->v.For.target);
2223 VISIT_SEQ(c, stmt, s->v.For.body);
2224 ADDOP_JABS(c, JUMP_ABSOLUTE, start);
2225 compiler_use_next_block(c, cleanup);
2226 ADDOP(c, POP_BLOCK);
2227 compiler_pop_fblock(c, LOOP, start);
2228 VISIT_SEQ(c, stmt, s->v.For.orelse);
2229 compiler_use_next_block(c, end);
2230 return 1;
2231}
2232
2233static int
2234compiler_while(struct compiler *c, stmt_ty s)
2235{
2236 basicblock *loop, *orelse, *end, *anchor = NULL;
2237 int constant = expr_constant(s->v.While.test);
2238
2239 if (constant == 0)
2240 return 1;
2241 loop = compiler_new_block(c);
2242 end = compiler_new_block(c);
2243 if (constant == -1) {
2244 anchor = compiler_new_block(c);
2245 if (anchor == NULL)
2246 return 0;
2247 }
2248 if (loop == NULL || end == NULL)
2249 return 0;
2250 if (s->v.While.orelse) {
2251 orelse = compiler_new_block(c);
2252 if (orelse == NULL)
2253 return 0;
2254 }
2255 else
2256 orelse = NULL;
2257
2258 ADDOP_JREL(c, SETUP_LOOP, end);
2259 compiler_use_next_block(c, loop);
2260 if (!compiler_push_fblock(c, LOOP, loop))
2261 return 0;
2262 if (constant == -1) {
2263 VISIT(c, expr, s->v.While.test);
2264 ADDOP_JREL(c, JUMP_IF_FALSE, anchor);
2265 ADDOP(c, POP_TOP);
2266 }
2267 VISIT_SEQ(c, stmt, s->v.While.body);
2268 ADDOP_JABS(c, JUMP_ABSOLUTE, loop);
2269
2270 /* XXX should the two POP instructions be in a separate block
2271 if there is no else clause ?
2272 */
2273
2274 if (constant == -1) {
2275 compiler_use_next_block(c, anchor);
2276 ADDOP(c, POP_TOP);
2277 ADDOP(c, POP_BLOCK);
2278 }
2279 compiler_pop_fblock(c, LOOP, loop);
2280 if (orelse != NULL) /* what if orelse is just pass? */
2281 VISIT_SEQ(c, stmt, s->v.While.orelse);
2282 compiler_use_next_block(c, end);
2283
2284 return 1;
2285}
2286
2287static int
2288compiler_continue(struct compiler *c)
2289{
2290 static const char LOOP_ERROR_MSG[] = "'continue' not properly in loop";
2291 int i;
2292
2293 if (!c->u->u_nfblocks)
2294 return compiler_error(c, LOOP_ERROR_MSG);
2295 i = c->u->u_nfblocks - 1;
2296 switch (c->u->u_fblock[i].fb_type) {
2297 case LOOP:
2298 ADDOP_JABS(c, JUMP_ABSOLUTE, c->u->u_fblock[i].fb_block);
2299 break;
2300 case EXCEPT:
2301 case FINALLY_TRY:
2302 while (--i >= 0 && c->u->u_fblock[i].fb_type != LOOP)
2303 ;
2304 if (i == -1)
2305 return compiler_error(c, LOOP_ERROR_MSG);
2306 ADDOP_JABS(c, CONTINUE_LOOP, c->u->u_fblock[i].fb_block);
2307 break;
2308 case FINALLY_END:
2309 return compiler_error(c,
2310 "'continue' not supported inside 'finally' clause");
2311 }
2312
2313 return 1;
2314}
2315
2316/* Code generated for "try: <body> finally: <finalbody>" is as follows:
2317
2318 SETUP_FINALLY L
2319 <code for body>
2320 POP_BLOCK
2321 LOAD_CONST <None>
2322 L: <code for finalbody>
2323 END_FINALLY
2324
2325 The special instructions use the block stack. Each block
2326 stack entry contains the instruction that created it (here
2327 SETUP_FINALLY), the level of the value stack at the time the
2328 block stack entry was created, and a label (here L).
2329
2330 SETUP_FINALLY:
2331 Pushes the current value stack level and the label
2332 onto the block stack.
2333 POP_BLOCK:
2334 Pops en entry from the block stack, and pops the value
2335 stack until its level is the same as indicated on the
2336 block stack. (The label is ignored.)
2337 END_FINALLY:
2338 Pops a variable number of entries from the *value* stack
2339 and re-raises the exception they specify. The number of
2340 entries popped depends on the (pseudo) exception type.
2341
2342 The block stack is unwound when an exception is raised:
2343 when a SETUP_FINALLY entry is found, the exception is pushed
2344 onto the value stack (and the exception condition is cleared),
2345 and the interpreter jumps to the label gotten from the block
2346 stack.
2347*/
2348
2349static int
2350compiler_try_finally(struct compiler *c, stmt_ty s)
2351{
2352 basicblock *body, *end;
2353 body = compiler_new_block(c);
2354 end = compiler_new_block(c);
2355 if (body == NULL || end == NULL)
2356 return 0;
2357
2358 ADDOP_JREL(c, SETUP_FINALLY, end);
2359 compiler_use_next_block(c, body);
2360 if (!compiler_push_fblock(c, FINALLY_TRY, body))
2361 return 0;
2362 VISIT_SEQ(c, stmt, s->v.TryFinally.body);
2363 ADDOP(c, POP_BLOCK);
2364 compiler_pop_fblock(c, FINALLY_TRY, body);
2365
2366 ADDOP_O(c, LOAD_CONST, Py_None, consts);
2367 compiler_use_next_block(c, end);
2368 if (!compiler_push_fblock(c, FINALLY_END, end))
2369 return 0;
2370 VISIT_SEQ(c, stmt, s->v.TryFinally.finalbody);
2371 ADDOP(c, END_FINALLY);
2372 compiler_pop_fblock(c, FINALLY_END, end);
2373
2374 return 1;
2375}
2376
2377/*
2378 Code generated for "try: S except E1, V1: S1 except E2, V2: S2 ...":
2379 (The contents of the value stack is shown in [], with the top
2380 at the right; 'tb' is trace-back info, 'val' the exception's
2381 associated value, and 'exc' the exception.)
2382
2383 Value stack Label Instruction Argument
2384 [] SETUP_EXCEPT L1
2385 [] <code for S>
2386 [] POP_BLOCK
2387 [] JUMP_FORWARD L0
2388
2389 [tb, val, exc] L1: DUP )
2390 [tb, val, exc, exc] <evaluate E1> )
2391 [tb, val, exc, exc, E1] COMPARE_OP EXC_MATCH ) only if E1
2392 [tb, val, exc, 1-or-0] JUMP_IF_FALSE L2 )
2393 [tb, val, exc, 1] POP )
2394 [tb, val, exc] POP
2395 [tb, val] <assign to V1> (or POP if no V1)
2396 [tb] POP
2397 [] <code for S1>
2398 JUMP_FORWARD L0
2399
2400 [tb, val, exc, 0] L2: POP
2401 [tb, val, exc] DUP
2402 .............................etc.......................
2403
2404 [tb, val, exc, 0] Ln+1: POP
2405 [tb, val, exc] END_FINALLY # re-raise exception
2406
2407 [] L0: <next statement>
2408
2409 Of course, parts are not generated if Vi or Ei is not present.
2410*/
2411static int
2412compiler_try_except(struct compiler *c, stmt_ty s)
2413{
2414 basicblock *body, *orelse, *except, *end;
2415 int i, n;
2416
2417 body = compiler_new_block(c);
2418 except = compiler_new_block(c);
2419 orelse = compiler_new_block(c);
2420 end = compiler_new_block(c);
2421 if (body == NULL || except == NULL || orelse == NULL || end == NULL)
2422 return 0;
2423 ADDOP_JREL(c, SETUP_EXCEPT, except);
2424 compiler_use_next_block(c, body);
2425 if (!compiler_push_fblock(c, EXCEPT, body))
2426 return 0;
2427 VISIT_SEQ(c, stmt, s->v.TryExcept.body);
2428 ADDOP(c, POP_BLOCK);
2429 compiler_pop_fblock(c, EXCEPT, body);
2430 ADDOP_JREL(c, JUMP_FORWARD, orelse);
2431 n = asdl_seq_LEN(s->v.TryExcept.handlers);
2432 compiler_use_next_block(c, except);
2433 for (i = 0; i < n; i++) {
2434 excepthandler_ty handler = (excepthandler_ty)asdl_seq_GET(
2435 s->v.TryExcept.handlers, i);
2436 if (!handler->type && i < n-1)
2437 return compiler_error(c, "default 'except:' must be last");
2438 c->u->u_lineno_set = false;
2439 c->u->u_lineno = handler->lineno;
2440 except = compiler_new_block(c);
2441 if (except == NULL)
2442 return 0;
2443 if (handler->type) {
2444 ADDOP(c, DUP_TOP);
2445 VISIT(c, expr, handler->type);
2446 ADDOP_I(c, COMPARE_OP, PyCmp_EXC_MATCH);
2447 ADDOP_JREL(c, JUMP_IF_FALSE, except);
2448 ADDOP(c, POP_TOP);
2449 }
2450 ADDOP(c, POP_TOP);
2451 if (handler->name) {
2452 VISIT(c, expr, handler->name);
2453 }
2454 else {
2455 ADDOP(c, POP_TOP);
2456 }
2457 ADDOP(c, POP_TOP);
2458 VISIT_SEQ(c, stmt, handler->body);
2459 ADDOP_JREL(c, JUMP_FORWARD, end);
2460 compiler_use_next_block(c, except);
2461 if (handler->type)
2462 ADDOP(c, POP_TOP);
2463 }
2464 ADDOP(c, END_FINALLY);
2465 compiler_use_next_block(c, orelse);
2466 VISIT_SEQ(c, stmt, s->v.TryExcept.orelse);
2467 compiler_use_next_block(c, end);
2468 return 1;
2469}
2470
2471static int
2472compiler_import_as(struct compiler *c, identifier name, identifier asname)
2473{
2474 /* The IMPORT_NAME opcode was already generated. This function
2475 merely needs to bind the result to a name.
2476
2477 If there is a dot in name, we need to split it and emit a
2478 LOAD_ATTR for each name.
2479 */
2480 const char *src = PyString_AS_STRING(name);
2481 const char *dot = strchr(src, '.');
2482 if (dot) {
2483 /* Consume the base module name to get the first attribute */
2484 src = dot + 1;
2485 while (dot) {
2486 /* NB src is only defined when dot != NULL */
2487 PyObject *attr;
2488 dot = strchr(src, '.');
2489 attr = PyString_FromStringAndSize(src,
2490 dot ? dot - src : strlen(src));
2491 if (!attr)
2492 return -1;
2493 ADDOP_O(c, LOAD_ATTR, attr, names);
2494 Py_DECREF(attr);
2495 src = dot + 1;
2496 }
2497 }
2498 return compiler_nameop(c, asname, Store);
2499}
2500
2501static int
2502compiler_import(struct compiler *c, stmt_ty s)
2503{
2504 /* The Import node stores a module name like a.b.c as a single
2505 string. This is convenient for all cases except
2506 import a.b.c as d
2507 where we need to parse that string to extract the individual
2508 module names.
2509 XXX Perhaps change the representation to make this case simpler?
2510 */
2511 int i, n = asdl_seq_LEN(s->v.Import.names);
2512
2513 for (i = 0; i < n; i++) {
2514 alias_ty alias = (alias_ty)asdl_seq_GET(s->v.Import.names, i);
2515 int r;
2516 PyObject *level;
2517
2518 if (c->c_flags && (c->c_flags->cf_flags & CO_FUTURE_ABSOLUTE_IMPORT))
2519 level = PyInt_FromLong(0);
2520 else
2521 level = PyInt_FromLong(-1);
2522
2523 if (level == NULL)
2524 return 0;
2525
2526 ADDOP_O(c, LOAD_CONST, level, consts);
2527 Py_DECREF(level);
2528 ADDOP_O(c, LOAD_CONST, Py_None, consts);
2529 ADDOP_NAME(c, IMPORT_NAME, alias->name, names);
2530
2531 if (alias->asname) {
2532 r = compiler_import_as(c, alias->name, alias->asname);
2533 if (!r)
2534 return r;
2535 }
2536 else {
2537 identifier tmp = alias->name;
2538 const char *base = PyString_AS_STRING(alias->name);
2539 char *dot = strchr(base, '.');
2540 if (dot)
2541 tmp = PyString_FromStringAndSize(base,
2542 dot - base);
2543 r = compiler_nameop(c, tmp, Store);
2544 if (dot) {
2545 Py_DECREF(tmp);
2546 }
2547 if (!r)
2548 return r;
2549 }
2550 }
2551 return 1;
2552}
2553
2554static int
2555compiler_from_import(struct compiler *c, stmt_ty s)
2556{
2557 int i, n = asdl_seq_LEN(s->v.ImportFrom.names);
2558
2559 PyObject *names = PyTuple_New(n);
2560 PyObject *level;
2561
2562 if (!names)
2563 return 0;
2564
2565 if (s->v.ImportFrom.level == 0 && c->c_flags &&
2566 !(c->c_flags->cf_flags & CO_FUTURE_ABSOLUTE_IMPORT))
2567 level = PyInt_FromLong(-1);
2568 else
2569 level = PyInt_FromLong(s->v.ImportFrom.level);
2570
2571 if (!level) {
2572 Py_DECREF(names);
2573 return 0;
2574 }
2575
2576 /* build up the names */
2577 for (i = 0; i < n; i++) {
2578 alias_ty alias = (alias_ty)asdl_seq_GET(s->v.ImportFrom.names, i);
2579 Py_INCREF(alias->name);
2580 PyTuple_SET_ITEM(names, i, alias->name);
2581 }
2582
2583 if (s->lineno > c->c_future->ff_lineno) {
2584 if (!strcmp(PyString_AS_STRING(s->v.ImportFrom.module),
2585 "__future__")) {
2586 Py_DECREF(level);
2587 Py_DECREF(names);
2588 return compiler_error(c,
2589 "from __future__ imports must occur "
2590 "at the beginning of the file");
2591
2592 }
2593 }
2594
2595 ADDOP_O(c, LOAD_CONST, level, consts);
2596 Py_DECREF(level);
2597 ADDOP_O(c, LOAD_CONST, names, consts);
2598 Py_DECREF(names);
2599 ADDOP_NAME(c, IMPORT_NAME, s->v.ImportFrom.module, names);
2600 for (i = 0; i < n; i++) {
2601 alias_ty alias = (alias_ty)asdl_seq_GET(s->v.ImportFrom.names, i);
2602 identifier store_name;
2603
2604 if (i == 0 && *PyString_AS_STRING(alias->name) == '*') {
2605 assert(n == 1);
2606 ADDOP(c, IMPORT_STAR);
2607 return 1;
2608 }
2609
2610 ADDOP_NAME(c, IMPORT_FROM, alias->name, names);
2611 store_name = alias->name;
2612 if (alias->asname)
2613 store_name = alias->asname;
2614
2615 if (!compiler_nameop(c, store_name, Store)) {
2616 Py_DECREF(names);
2617 return 0;
2618 }
2619 }
2620 /* remove imported module */
2621 ADDOP(c, POP_TOP);
2622 return 1;
2623}
2624
2625static int
2626compiler_assert(struct compiler *c, stmt_ty s)
2627{
2628 static PyObject *assertion_error = NULL;
2629 basicblock *end;
2630
2631 if (Py_OptimizeFlag)
2632 return 1;
2633 if (assertion_error == NULL) {
2634 assertion_error = PyString_FromString("AssertionError");
2635 if (assertion_error == NULL)
2636 return 0;
2637 }
2638 VISIT(c, expr, s->v.Assert.test);
2639 end = compiler_new_block(c);
2640 if (end == NULL)
2641 return 0;
2642 ADDOP_JREL(c, JUMP_IF_TRUE, end);
2643 ADDOP(c, POP_TOP);
2644 ADDOP_O(c, LOAD_GLOBAL, assertion_error, names);
2645 if (s->v.Assert.msg) {
2646 VISIT(c, expr, s->v.Assert.msg);
2647 ADDOP_I(c, RAISE_VARARGS, 2);
2648 }
2649 else {
2650 ADDOP_I(c, RAISE_VARARGS, 1);
2651 }
2652 compiler_use_next_block(c, end);
2653 ADDOP(c, POP_TOP);
2654 return 1;
2655}
2656
2657static int
2658compiler_visit_stmt(struct compiler *c, stmt_ty s)
2659{
2660 int i, n;
2661
2662 /* Always assign a lineno to the next instruction for a stmt. */
2663 c->u->u_lineno = s->lineno;
2664 c->u->u_lineno_set = false;
2665
2666 switch (s->kind) {
2667 case FunctionDef_kind:
2668 return compiler_function(c, s);
2669 case ClassDef_kind:
2670 return compiler_class(c, s);
2671 case Return_kind:
2672 if (c->u->u_ste->ste_type != FunctionBlock)
2673 return compiler_error(c, "'return' outside function");
2674 if (s->v.Return.value) {
2675 VISIT(c, expr, s->v.Return.value);
2676 }
2677 else
2678 ADDOP_O(c, LOAD_CONST, Py_None, consts);
2679 ADDOP(c, RETURN_VALUE);
2680 break;
2681 case Delete_kind:
2682 VISIT_SEQ(c, expr, s->v.Delete.targets)
2683 break;
2684 case Assign_kind:
2685 n = asdl_seq_LEN(s->v.Assign.targets);
2686 VISIT(c, expr, s->v.Assign.value);
2687 for (i = 0; i < n; i++) {
2688 if (i < n - 1)
2689 ADDOP(c, DUP_TOP);
2690 VISIT(c, expr,
2691 (expr_ty)asdl_seq_GET(s->v.Assign.targets, i));
2692 }
2693 break;
2694 case AugAssign_kind:
2695 return compiler_augassign(c, s);
2696 case Print_kind:
2697 return compiler_print(c, s);
2698 case For_kind:
2699 return compiler_for(c, s);
2700 case While_kind:
2701 return compiler_while(c, s);
2702 case If_kind:
2703 return compiler_if(c, s);
2704 case Raise_kind:
2705 n = 0;
2706 if (s->v.Raise.type) {
2707 VISIT(c, expr, s->v.Raise.type);
2708 n++;
2709 if (s->v.Raise.inst) {
2710 VISIT(c, expr, s->v.Raise.inst);
2711 n++;
2712 if (s->v.Raise.tback) {
2713 VISIT(c, expr, s->v.Raise.tback);
2714 n++;
2715 }
2716 }
2717 }
2718 ADDOP_I(c, RAISE_VARARGS, n);
2719 break;
2720 case TryExcept_kind:
2721 return compiler_try_except(c, s);
2722 case TryFinally_kind:
2723 return compiler_try_finally(c, s);
2724 case Assert_kind:
2725 return compiler_assert(c, s);
2726 case Import_kind:
2727 return compiler_import(c, s);
2728 case ImportFrom_kind:
2729 return compiler_from_import(c, s);
2730 case Exec_kind:
2731 VISIT(c, expr, s->v.Exec.body);
2732 if (s->v.Exec.globals) {
2733 VISIT(c, expr, s->v.Exec.globals);
2734 if (s->v.Exec.locals) {
2735 VISIT(c, expr, s->v.Exec.locals);
2736 } else {
2737 ADDOP(c, DUP_TOP);
2738 }
2739 } else {
2740 ADDOP_O(c, LOAD_CONST, Py_None, consts);
2741 ADDOP(c, DUP_TOP);
2742 }
2743 ADDOP(c, EXEC_STMT);
2744 break;
2745 case Global_kind:
2746 break;
2747 case Expr_kind:
2748 if (c->c_interactive && c->c_nestlevel <= 1) {
2749 VISIT(c, expr, s->v.Expr.value);
2750 ADDOP(c, PRINT_EXPR);
2751 }
2752 else if (s->v.Expr.value->kind != Str_kind &&
2753 s->v.Expr.value->kind != Num_kind) {
2754 VISIT(c, expr, s->v.Expr.value);
2755 ADDOP(c, POP_TOP);
2756 }
2757 break;
2758 case Pass_kind:
2759 break;
2760 case Break_kind:
2761 if (!c->u->u_nfblocks)
2762 return compiler_error(c, "'break' outside loop");
2763 ADDOP(c, BREAK_LOOP);
2764 break;
2765 case Continue_kind:
2766 return compiler_continue(c);
2767 case With_kind:
2768 return compiler_with(c, s);
2769 }
2770 return 1;
2771}
2772
2773static int
2774unaryop(unaryop_ty op)
2775{
2776 switch (op) {
2777 case Invert:
2778 return UNARY_INVERT;
2779 case Not:
2780 return UNARY_NOT;
2781 case UAdd:
2782 return UNARY_POSITIVE;
2783 case USub:
2784 return UNARY_NEGATIVE;
2785 }
2786 return 0;
2787}
2788
2789static int
2790binop(struct compiler *c, operator_ty op)
2791{
2792 switch (op) {
2793 case Add:
2794 return BINARY_ADD;
2795 case Sub:
2796 return BINARY_SUBTRACT;
2797 case Mult:
2798 return BINARY_MULTIPLY;
2799 case Div:
2800 if (c->c_flags && c->c_flags->cf_flags & CO_FUTURE_DIVISION)
2801 return BINARY_TRUE_DIVIDE;
2802 else
2803 return BINARY_DIVIDE;
2804 case Mod:
2805 return BINARY_MODULO;
2806 case Pow:
2807 return BINARY_POWER;
2808 case LShift:
2809 return BINARY_LSHIFT;
2810 case RShift:
2811 return BINARY_RSHIFT;
2812 case BitOr:
2813 return BINARY_OR;
2814 case BitXor:
2815 return BINARY_XOR;
2816 case BitAnd:
2817 return BINARY_AND;
2818 case FloorDiv:
2819 return BINARY_FLOOR_DIVIDE;
2820 }
2821 return 0;
2822}
2823
2824static int
2825cmpop(cmpop_ty op)
2826{
2827 switch (op) {
2828 case Eq:
2829 return PyCmp_EQ;
2830 case NotEq:
2831 return PyCmp_NE;
2832 case Lt:
2833 return PyCmp_LT;
2834 case LtE:
2835 return PyCmp_LE;
2836 case Gt:
2837 return PyCmp_GT;
2838 case GtE:
2839 return PyCmp_GE;
2840 case Is:
2841 return PyCmp_IS;
2842 case IsNot:
2843 return PyCmp_IS_NOT;
2844 case In:
2845 return PyCmp_IN;
2846 case NotIn:
2847 return PyCmp_NOT_IN;
2848 }
2849 return PyCmp_BAD;
2850}
2851
2852static int
2853inplace_binop(struct compiler *c, operator_ty op)
2854{
2855 switch (op) {
2856 case Add:
2857 return INPLACE_ADD;
2858 case Sub:
2859 return INPLACE_SUBTRACT;
2860 case Mult:
2861 return INPLACE_MULTIPLY;
2862 case Div:
2863 if (c->c_flags && c->c_flags->cf_flags & CO_FUTURE_DIVISION)
2864 return INPLACE_TRUE_DIVIDE;
2865 else
2866 return INPLACE_DIVIDE;
2867 case Mod:
2868 return INPLACE_MODULO;
2869 case Pow:
2870 return INPLACE_POWER;
2871 case LShift:
2872 return INPLACE_LSHIFT;
2873 case RShift:
2874 return INPLACE_RSHIFT;
2875 case BitOr:
2876 return INPLACE_OR;
2877 case BitXor:
2878 return INPLACE_XOR;
2879 case BitAnd:
2880 return INPLACE_AND;
2881 case FloorDiv:
2882 return INPLACE_FLOOR_DIVIDE;
2883 }
2884 PyErr_Format(PyExc_SystemError,
2885 "inplace binary op %d should not be possible", op);
2886 return 0;
2887}
2888
2889static int
2890compiler_nameop(struct compiler *c, identifier name, expr_context_ty ctx)
2891{
2892 int op, scope, arg;
2893 enum { OP_FAST, OP_GLOBAL, OP_DEREF, OP_NAME } optype;
2894
2895 PyObject *dict = c->u->u_names;
2896 PyObject *mangled;
2897 /* XXX AugStore isn't used anywhere! */
2898
2899 /* First check for assignment to __debug__. Param? */
2900 if ((ctx == Store || ctx == AugStore || ctx == Del)
2901 && !strcmp(PyString_AS_STRING(name), "__debug__")) {
2902 return compiler_error(c, "can not assign to __debug__");
2903 }
2904
2905 mangled = _Py_Mangle(c->u->u_private, name);
2906 if (!mangled)
2907 return 0;
2908
2909 op = 0;
2910 optype = OP_NAME;
2911 scope = PyST_GetScope(c->u->u_ste, mangled);
2912 switch (scope) {
2913 case FREE:
2914 dict = c->u->u_freevars;
2915 optype = OP_DEREF;
2916 break;
2917 case CELL:
2918 dict = c->u->u_cellvars;
2919 optype = OP_DEREF;
2920 break;
2921 case LOCAL:
2922 if (c->u->u_ste->ste_type == FunctionBlock)
2923 optype = OP_FAST;
2924 break;
2925 case GLOBAL_IMPLICIT:
2926 if (c->u->u_ste->ste_type == FunctionBlock &&
2927 !c->u->u_ste->ste_unoptimized)
2928 optype = OP_GLOBAL;
2929 break;
2930 case GLOBAL_EXPLICIT:
2931 optype = OP_GLOBAL;
2932 break;
2933 default:
2934 /* scope can be 0 */
2935 break;
2936 }
2937
2938 /* XXX Leave assert here, but handle __doc__ and the like better */
2939 assert(scope || PyString_AS_STRING(name)[0] == '_');
2940
2941 switch (optype) {
2942 case OP_DEREF:
2943 switch (ctx) {
2944 case Load: op = LOAD_DEREF; break;
2945 case Store: op = STORE_DEREF; break;
2946 case AugLoad:
2947 case AugStore:
2948 break;
2949 case Del:
2950 PyErr_Format(PyExc_SyntaxError,
2951 "can not delete variable '%s' referenced "
2952 "in nested scope",
2953 PyString_AS_STRING(name));
2954 Py_DECREF(mangled);
2955 return 0;
2956 case Param:
2957 default:
2958 PyErr_SetString(PyExc_SystemError,
2959 "param invalid for deref variable");
2960 return 0;
2961 }
2962 break;
2963 case OP_FAST:
2964 switch (ctx) {
2965 case Load: op = LOAD_FAST; break;
2966 case Store: op = STORE_FAST; break;
2967 case Del: op = DELETE_FAST; break;
2968 case AugLoad:
2969 case AugStore:
2970 break;
2971 case Param:
2972 default:
2973 PyErr_SetString(PyExc_SystemError,
2974 "param invalid for local variable");
2975 return 0;
2976 }
2977 ADDOP_O(c, op, mangled, varnames);
2978 Py_DECREF(mangled);
2979 return 1;
2980 case OP_GLOBAL:
2981 switch (ctx) {
2982 case Load: op = LOAD_GLOBAL; break;
2983 case Store: op = STORE_GLOBAL; break;
2984 case Del: op = DELETE_GLOBAL; break;
2985 case AugLoad:
2986 case AugStore:
2987 break;
2988 case Param:
2989 default:
2990 PyErr_SetString(PyExc_SystemError,
2991 "param invalid for global variable");
2992 return 0;
2993 }
2994 break;
2995 case OP_NAME:
2996 switch (ctx) {
2997 case Load: op = LOAD_NAME; break;
2998 case Store: op = STORE_NAME; break;
2999 case Del: op = DELETE_NAME; break;
3000 case AugLoad:
3001 case AugStore:
3002 break;
3003 case Param:
3004 default:
3005 PyErr_SetString(PyExc_SystemError,
3006 "param invalid for name variable");
3007 return 0;
3008 }
3009 break;
3010 }
3011
3012 assert(op);
3013 arg = compiler_add_o(c, dict, mangled);
3014 Py_DECREF(mangled);
3015 if (arg < 0)
3016 return 0;
3017 return compiler_addop_i(c, op, arg);
3018}
3019
3020static int
3021compiler_boolop(struct compiler *c, expr_ty e)
3022{
3023 basicblock *end;
3024 int jumpi, i, n;
3025 asdl_seq *s;
3026
3027 assert(e->kind == BoolOp_kind);
3028 if (e->v.BoolOp.op == And)
3029 jumpi = JUMP_IF_FALSE;
3030 else
3031 jumpi = JUMP_IF_TRUE;
3032 end = compiler_new_block(c);
3033 if (end == NULL)
3034 return 0;
3035 s = e->v.BoolOp.values;
3036 n = asdl_seq_LEN(s) - 1;
3037 assert(n >= 0);
3038 for (i = 0; i < n; ++i) {
3039 VISIT(c, expr, (expr_ty)asdl_seq_GET(s, i));
3040 ADDOP_JREL(c, jumpi, end);
3041 ADDOP(c, POP_TOP)
3042 }
3043 VISIT(c, expr, (expr_ty)asdl_seq_GET(s, n));
3044 compiler_use_next_block(c, end);
3045 return 1;
3046}
3047
3048static int
3049compiler_list(struct compiler *c, expr_ty e)
3050{
3051 int n = asdl_seq_LEN(e->v.List.elts);
3052 if (e->v.List.ctx == Store) {
3053 ADDOP_I(c, UNPACK_SEQUENCE, n);
3054 }
3055 VISIT_SEQ(c, expr, e->v.List.elts);
3056 if (e->v.List.ctx == Load) {
3057 ADDOP_I(c, BUILD_LIST, n);
3058 }
3059 return 1;
3060}
3061
3062static int
3063compiler_tuple(struct compiler *c, expr_ty e)
3064{
3065 int n = asdl_seq_LEN(e->v.Tuple.elts);
3066 if (e->v.Tuple.ctx == Store) {
3067 ADDOP_I(c, UNPACK_SEQUENCE, n);
3068 }
3069 VISIT_SEQ(c, expr, e->v.Tuple.elts);
3070 if (e->v.Tuple.ctx == Load) {
3071 ADDOP_I(c, BUILD_TUPLE, n);
3072 }
3073 return 1;
3074}
3075
3076static int
3077compiler_compare(struct compiler *c, expr_ty e)
3078{
3079 int i, n;
3080 basicblock *cleanup = NULL;
3081
3082 /* XXX the logic can be cleaned up for 1 or multiple comparisons */
3083 VISIT(c, expr, e->v.Compare.left);
3084 n = asdl_seq_LEN(e->v.Compare.ops);
3085 assert(n > 0);
3086 if (n > 1) {
3087 cleanup = compiler_new_block(c);
3088 if (cleanup == NULL)
3089 return 0;
3090 VISIT(c, expr,
3091 (expr_ty)asdl_seq_GET(e->v.Compare.comparators, 0));
3092 }
3093 for (i = 1; i < n; i++) {
3094 ADDOP(c, DUP_TOP);
3095 ADDOP(c, ROT_THREE);
3096 ADDOP_I(c, COMPARE_OP,
3097 cmpop((cmpop_ty)(asdl_seq_GET(
3098 e->v.Compare.ops, i - 1))));
3099 ADDOP_JREL(c, JUMP_IF_FALSE, cleanup);
3100 NEXT_BLOCK(c);
3101 ADDOP(c, POP_TOP);
3102 if (i < (n - 1))
3103 VISIT(c, expr,
3104 (expr_ty)asdl_seq_GET(e->v.Compare.comparators, i));
3105 }
3106 VISIT(c, expr, (expr_ty)asdl_seq_GET(e->v.Compare.comparators, n - 1));
3107 ADDOP_I(c, COMPARE_OP,
3108 cmpop((cmpop_ty)(asdl_seq_GET(e->v.Compare.ops, n - 1))));
3109 if (n > 1) {
3110 basicblock *end = compiler_new_block(c);
3111 if (end == NULL)
3112 return 0;
3113 ADDOP_JREL(c, JUMP_FORWARD, end);
3114 compiler_use_next_block(c, cleanup);
3115 ADDOP(c, ROT_TWO);
3116 ADDOP(c, POP_TOP);
3117 compiler_use_next_block(c, end);
3118 }
3119 return 1;
3120}
3121#undef CMPCAST
3122
3123static int
3124compiler_call(struct compiler *c, expr_ty e)
3125{
3126 int n, code = 0;
3127
3128 VISIT(c, expr, e->v.Call.func);
3129 n = asdl_seq_LEN(e->v.Call.args);
3130 VISIT_SEQ(c, expr, e->v.Call.args);
3131 if (e->v.Call.keywords) {
3132 VISIT_SEQ(c, keyword, e->v.Call.keywords);
3133 n |= asdl_seq_LEN(e->v.Call.keywords) << 8;
3134 }
3135 if (e->v.Call.starargs) {
3136 VISIT(c, expr, e->v.Call.starargs);
3137 code |= 1;
3138 }
3139 if (e->v.Call.kwargs) {
3140 VISIT(c, expr, e->v.Call.kwargs);
3141 code |= 2;
3142 }
3143 switch (code) {
3144 case 0:
3145 ADDOP_I(c, CALL_FUNCTION, n);
3146 break;
3147 case 1:
3148 ADDOP_I(c, CALL_FUNCTION_VAR, n);
3149 break;
3150 case 2:
3151 ADDOP_I(c, CALL_FUNCTION_KW, n);
3152 break;
3153 case 3:
3154 ADDOP_I(c, CALL_FUNCTION_VAR_KW, n);
3155 break;
3156 }
3157 return 1;
3158}
3159
3160static int
3161compiler_listcomp_generator(struct compiler *c, PyObject *tmpname,
3162 asdl_seq *generators, int gen_index,
3163 expr_ty elt)
3164{
3165 /* generate code for the iterator, then each of the ifs,
3166 and then write to the element */
3167
3168 comprehension_ty l;
3169 basicblock *start, *anchor, *skip, *if_cleanup;
3170 int i, n;
3171
3172 start = compiler_new_block(c);
3173 skip = compiler_new_block(c);
3174 if_cleanup = compiler_new_block(c);
3175 anchor = compiler_new_block(c);
3176
3177 if (start == NULL || skip == NULL || if_cleanup == NULL ||
3178 anchor == NULL)
3179 return 0;
3180
3181 l = (comprehension_ty)asdl_seq_GET(generators, gen_index);
3182 VISIT(c, expr, l->iter);
3183 ADDOP(c, GET_ITER);
3184 compiler_use_next_block(c, start);
3185 ADDOP_JREL(c, FOR_ITER, anchor);
3186 NEXT_BLOCK(c);
3187 VISIT(c, expr, l->target);
3188
3189 /* XXX this needs to be cleaned up...a lot! */
3190 n = asdl_seq_LEN(l->ifs);
3191 for (i = 0; i < n; i++) {
3192 expr_ty e = (expr_ty)asdl_seq_GET(l->ifs, i);
3193 VISIT(c, expr, e);
3194 ADDOP_JREL(c, JUMP_IF_FALSE, if_cleanup);
3195 NEXT_BLOCK(c);
3196 ADDOP(c, POP_TOP);
3197 }
3198
3199 if (++gen_index < asdl_seq_LEN(generators))
3200 if (!compiler_listcomp_generator(c, tmpname,
3201 generators, gen_index, elt))
3202 return 0;
3203
3204 /* only append after the last for generator */
3205 if (gen_index >= asdl_seq_LEN(generators)) {
3206 if (!compiler_nameop(c, tmpname, Load))
3207 return 0;
3208 VISIT(c, expr, elt);
3209 ADDOP(c, LIST_APPEND);
3210
3211 compiler_use_next_block(c, skip);
3212 }
3213 for (i = 0; i < n; i++) {
3214 ADDOP_I(c, JUMP_FORWARD, 1);
3215 if (i == 0)
3216 compiler_use_next_block(c, if_cleanup);
3217 ADDOP(c, POP_TOP);
3218 }
3219 ADDOP_JABS(c, JUMP_ABSOLUTE, start);
3220 compiler_use_next_block(c, anchor);
3221 /* delete the append method added to locals */
3222 if (gen_index == 1)
3223 if (!compiler_nameop(c, tmpname, Del))
3224 return 0;
3225
3226 return 1;
3227}
3228
3229static int
3230compiler_listcomp(struct compiler *c, expr_ty e)
3231{
3232 identifier tmp;
3233 int rc = 0;
3234 static identifier append;
3235 asdl_seq *generators = e->v.ListComp.generators;
3236
3237 assert(e->kind == ListComp_kind);
3238 if (!append) {
3239 append = PyString_InternFromString("append");
3240 if (!append)
3241 return 0;
3242 }
3243 tmp = compiler_new_tmpname(c);
3244 if (!tmp)
3245 return 0;
3246 ADDOP_I(c, BUILD_LIST, 0);
3247 ADDOP(c, DUP_TOP);
3248 if (compiler_nameop(c, tmp, Store))
3249 rc = compiler_listcomp_generator(c, tmp, generators, 0,
3250 e->v.ListComp.elt);
3251 Py_DECREF(tmp);
3252 return rc;
3253}
3254
3255static int
3256compiler_genexp_generator(struct compiler *c,
3257 asdl_seq *generators, int gen_index,
3258 expr_ty elt)
3259{
3260 /* generate code for the iterator, then each of the ifs,
3261 and then write to the element */
3262
3263 comprehension_ty ge;
3264 basicblock *start, *anchor, *skip, *if_cleanup, *end;
3265 int i, n;
3266
3267 start = compiler_new_block(c);
3268 skip = compiler_new_block(c);
3269 if_cleanup = compiler_new_block(c);
3270 anchor = compiler_new_block(c);
3271 end = compiler_new_block(c);
3272
3273 if (start == NULL || skip == NULL || if_cleanup == NULL ||
3274 anchor == NULL || end == NULL)
3275 return 0;
3276
3277 ge = (comprehension_ty)asdl_seq_GET(generators, gen_index);
3278 ADDOP_JREL(c, SETUP_LOOP, end);
3279 if (!compiler_push_fblock(c, LOOP, start))
3280 return 0;
3281
3282 if (gen_index == 0) {
3283 /* Receive outermost iter as an implicit argument */
3284 c->u->u_argcount = 1;
3285 ADDOP_I(c, LOAD_FAST, 0);
3286 }
3287 else {
3288 /* Sub-iter - calculate on the fly */
3289 VISIT(c, expr, ge->iter);
3290 ADDOP(c, GET_ITER);
3291 }
3292 compiler_use_next_block(c, start);
3293 ADDOP_JREL(c, FOR_ITER, anchor);
3294 NEXT_BLOCK(c);
3295 VISIT(c, expr, ge->target);
3296
3297 /* XXX this needs to be cleaned up...a lot! */
3298 n = asdl_seq_LEN(ge->ifs);
3299 for (i = 0; i < n; i++) {
3300 expr_ty e = (expr_ty)asdl_seq_GET(ge->ifs, i);
3301 VISIT(c, expr, e);
3302 ADDOP_JREL(c, JUMP_IF_FALSE, if_cleanup);
3303 NEXT_BLOCK(c);
3304 ADDOP(c, POP_TOP);
3305 }
3306
3307 if (++gen_index < asdl_seq_LEN(generators))
3308 if (!compiler_genexp_generator(c, generators, gen_index, elt))
3309 return 0;
3310
3311 /* only append after the last 'for' generator */
3312 if (gen_index >= asdl_seq_LEN(generators)) {
3313 VISIT(c, expr, elt);
3314 ADDOP(c, YIELD_VALUE);
3315 ADDOP(c, POP_TOP);
3316
3317 compiler_use_next_block(c, skip);
3318 }
3319 for (i = 0; i < n; i++) {
3320 ADDOP_I(c, JUMP_FORWARD, 1);
3321 if (i == 0)
3322 compiler_use_next_block(c, if_cleanup);
3323
3324 ADDOP(c, POP_TOP);
3325 }
3326 ADDOP_JABS(c, JUMP_ABSOLUTE, start);
3327 compiler_use_next_block(c, anchor);
3328 ADDOP(c, POP_BLOCK);
3329 compiler_pop_fblock(c, LOOP, start);
3330 compiler_use_next_block(c, end);
3331
3332 return 1;
3333}
3334
3335static int
3336compiler_genexp(struct compiler *c, expr_ty e)
3337{
3338 static identifier name;
3339 PyCodeObject *co;
3340 expr_ty outermost_iter = ((comprehension_ty)
3341 (asdl_seq_GET(e->v.GeneratorExp.generators,
3342 0)))->iter;
3343
3344 if (!name) {
3345 name = PyString_FromString("<genexpr>");
3346 if (!name)
3347 return 0;
3348 }
3349
3350 if (!compiler_enter_scope(c, name, (void *)e, e->lineno))
3351 return 0;
3352 compiler_genexp_generator(c, e->v.GeneratorExp.generators, 0,
3353 e->v.GeneratorExp.elt);
3354 co = assemble(c, 1);
3355 compiler_exit_scope(c);
3356 if (co == NULL)
3357 return 0;
3358
3359 compiler_make_closure(c, co, 0);
3360 Py_DECREF(co);
3361
3362 VISIT(c, expr, outermost_iter);
3363 ADDOP(c, GET_ITER);
3364 ADDOP_I(c, CALL_FUNCTION, 1);
3365
3366 return 1;
3367}
3368
3369static int
3370compiler_visit_keyword(struct compiler *c, keyword_ty k)
3371{
3372 ADDOP_O(c, LOAD_CONST, k->arg, consts);
3373 VISIT(c, expr, k->value);
3374 return 1;
3375}
3376
3377/* Test whether expression is constant. For constants, report
3378 whether they are true or false.
3379
3380 Return values: 1 for true, 0 for false, -1 for non-constant.
3381 */
3382
3383static int
3384expr_constant(expr_ty e)
3385{
3386 switch (e->kind) {
3387 case Num_kind:
3388 return PyObject_IsTrue(e->v.Num.n);
3389 case Str_kind:
3390 return PyObject_IsTrue(e->v.Str.s);
3391 case Name_kind:
3392 /* __debug__ is not assignable, so we can optimize
3393 * it away in if and while statements */
3394 if (strcmp(PyString_AS_STRING(e->v.Name.id),
3395 "__debug__") == 0)
3396 return ! Py_OptimizeFlag;
3397 /* fall through */
3398 default:
3399 return -1;
3400 }
3401}
3402
3403/*
3404 Implements the with statement from PEP 343.
3405
3406 The semantics outlined in that PEP are as follows:
3407
3408 with EXPR as VAR:
3409 BLOCK
3410
3411 It is implemented roughly as:
3412
3413 context = EXPR
3414 exit = context.__exit__ # not calling it
3415 value = context.__enter__()
3416 try:
3417 VAR = value # if VAR present in the syntax
3418 BLOCK
3419 finally:
3420 if an exception was raised:
3421 exc = copy of (exception, instance, traceback)
3422 else:
3423 exc = (None, None, None)
3424 exit(*exc)
3425 */
3426static int
3427compiler_with(struct compiler *c, stmt_ty s)
3428{
3429 static identifier enter_attr, exit_attr;
3430 basicblock *block, *finally;
3431 identifier tmpexit, tmpvalue = NULL;
3432
3433 assert(s->kind == With_kind);
3434
3435 if (!enter_attr) {
3436 enter_attr = PyString_InternFromString("__enter__");
3437 if (!enter_attr)
3438 return 0;
3439 }
3440 if (!exit_attr) {
3441 exit_attr = PyString_InternFromString("__exit__");
3442 if (!exit_attr)
3443 return 0;
3444 }
3445
3446 block = compiler_new_block(c);
3447 finally = compiler_new_block(c);
3448 if (!block || !finally)
3449 return 0;
3450
3451 /* Create a temporary variable to hold context.__exit__ */
3452 tmpexit = compiler_new_tmpname(c);
3453 if (tmpexit == NULL)
3454 return 0;
3455 PyArena_AddPyObject(c->c_arena, tmpexit);
3456
3457 if (s->v.With.optional_vars) {
3458 /* Create a temporary variable to hold context.__enter__().
3459 We need to do this rather than preserving it on the stack
3460 because SETUP_FINALLY remembers the stack level.
3461 We need to do the assignment *inside* the try/finally
3462 so that context.__exit__() is called when the assignment
3463 fails. But we need to call context.__enter__() *before*
3464 the try/finally so that if it fails we won't call
3465 context.__exit__().
3466 */
3467 tmpvalue = compiler_new_tmpname(c);
3468 if (tmpvalue == NULL)
3469 return 0;
3470 PyArena_AddPyObject(c->c_arena, tmpvalue);
3471 }
3472
3473 /* Evaluate EXPR */
3474 VISIT(c, expr, s->v.With.context_expr);
3475
3476 /* Squirrel away context.__exit__ */
3477 ADDOP(c, DUP_TOP);
3478 ADDOP_O(c, LOAD_ATTR, exit_attr, names);
3479 if (!compiler_nameop(c, tmpexit, Store))
3480 return 0;
3481
3482 /* Call context.__enter__() */
3483 ADDOP_O(c, LOAD_ATTR, enter_attr, names);
3484 ADDOP_I(c, CALL_FUNCTION, 0);
3485
3486 if (s->v.With.optional_vars) {
3487 /* Store it in tmpvalue */
3488 if (!compiler_nameop(c, tmpvalue, Store))
3489 return 0;
3490 }
3491 else {
3492 /* Discard result from context.__enter__() */
3493 ADDOP(c, POP_TOP);
3494 }
3495
3496 /* Start the try block */
3497 ADDOP_JREL(c, SETUP_FINALLY, finally);
3498
3499 compiler_use_next_block(c, block);
3500 if (!compiler_push_fblock(c, FINALLY_TRY, block)) {
3501 return 0;
3502 }
3503
3504 if (s->v.With.optional_vars) {
3505 /* Bind saved result of context.__enter__() to VAR */
3506 if (!compiler_nameop(c, tmpvalue, Load) ||
3507 !compiler_nameop(c, tmpvalue, Del))
3508 return 0;
3509 VISIT(c, expr, s->v.With.optional_vars);
3510 }
3511
3512 /* BLOCK code */
3513 VISIT_SEQ(c, stmt, s->v.With.body);
3514
3515 /* End of try block; start the finally block */
3516 ADDOP(c, POP_BLOCK);
3517 compiler_pop_fblock(c, FINALLY_TRY, block);
3518
3519 ADDOP_O(c, LOAD_CONST, Py_None, consts);
3520 compiler_use_next_block(c, finally);
3521 if (!compiler_push_fblock(c, FINALLY_END, finally))
3522 return 0;
3523
3524 /* Finally block starts; push tmpexit and issue our magic opcode. */
3525 if (!compiler_nameop(c, tmpexit, Load) ||
3526 !compiler_nameop(c, tmpexit, Del))
3527 return 0;
3528 ADDOP(c, WITH_CLEANUP);
3529
3530 /* Finally block ends. */
3531 ADDOP(c, END_FINALLY);
3532 compiler_pop_fblock(c, FINALLY_END, finally);
3533 return 1;
3534}
3535
3536static int
3537compiler_visit_expr(struct compiler *c, expr_ty e)
3538{
3539 int i, n;
3540
3541 /* If expr e has a different line number than the last expr/stmt,
3542 set a new line number for the next instruction.
3543 */
3544 if (e->lineno > c->u->u_lineno) {
3545 c->u->u_lineno = e->lineno;
3546 c->u->u_lineno_set = false;
3547 }
3548 switch (e->kind) {
3549 case BoolOp_kind:
3550 return compiler_boolop(c, e);
3551 case BinOp_kind:
3552 VISIT(c, expr, e->v.BinOp.left);
3553 VISIT(c, expr, e->v.BinOp.right);
3554 ADDOP(c, binop(c, e->v.BinOp.op));
3555 break;
3556 case UnaryOp_kind:
3557 VISIT(c, expr, e->v.UnaryOp.operand);
3558 ADDOP(c, unaryop(e->v.UnaryOp.op));
3559 break;
3560 case Lambda_kind:
3561 return compiler_lambda(c, e);
3562 case IfExp_kind:
3563 return compiler_ifexp(c, e);
3564 case Dict_kind:
3565 /* XXX get rid of arg? */
3566 ADDOP_I(c, BUILD_MAP, 0);
3567 n = asdl_seq_LEN(e->v.Dict.values);
3568 /* We must arrange things just right for STORE_SUBSCR.
3569 It wants the stack to look like (value) (dict) (key) */
3570 for (i = 0; i < n; i++) {
3571 ADDOP(c, DUP_TOP);
3572 VISIT(c, expr,
3573 (expr_ty)asdl_seq_GET(e->v.Dict.values, i));
3574 ADDOP(c, ROT_TWO);
3575 VISIT(c, expr,
3576 (expr_ty)asdl_seq_GET(e->v.Dict.keys, i));
3577 ADDOP(c, STORE_SUBSCR);
3578 }
3579 break;
3580 case ListComp_kind:
3581 return compiler_listcomp(c, e);
3582 case GeneratorExp_kind:
3583 return compiler_genexp(c, e);
3584 case Yield_kind:
3585 if (c->u->u_ste->ste_type != FunctionBlock)
3586 return compiler_error(c, "'yield' outside function");
3587 /*
3588 for (i = 0; i < c->u->u_nfblocks; i++) {
3589 if (c->u->u_fblock[i].fb_type == FINALLY_TRY)
3590 return compiler_error(
3591 c, "'yield' not allowed in a 'try' "
3592 "block with a 'finally' clause");
3593 }
3594 */
3595 if (e->v.Yield.value) {
3596 VISIT(c, expr, e->v.Yield.value);
3597 }
3598 else {
3599 ADDOP_O(c, LOAD_CONST, Py_None, consts);
3600 }
3601 ADDOP(c, YIELD_VALUE);
3602 break;
3603 case Compare_kind:
3604 return compiler_compare(c, e);
3605 case Call_kind:
3606 return compiler_call(c, e);
3607 case Repr_kind:
3608 VISIT(c, expr, e->v.Repr.value);
3609 ADDOP(c, UNARY_CONVERT);
3610 break;
3611 case Num_kind:
3612 ADDOP_O(c, LOAD_CONST, e->v.Num.n, consts);
3613 break;
3614 case Str_kind:
3615 ADDOP_O(c, LOAD_CONST, e->v.Str.s, consts);
3616 break;
3617 /* The following exprs can be assignment targets. */
3618 case Attribute_kind:
3619 if (e->v.Attribute.ctx != AugStore)
3620 VISIT(c, expr, e->v.Attribute.value);
3621 switch (e->v.Attribute.ctx) {
3622 case AugLoad:
3623 ADDOP(c, DUP_TOP);
3624 /* Fall through to load */
3625 case Load:
3626 ADDOP_NAME(c, LOAD_ATTR, e->v.Attribute.attr, names);
3627 break;
3628 case AugStore:
3629 ADDOP(c, ROT_TWO);
3630 /* Fall through to save */
3631 case Store:
3632 ADDOP_NAME(c, STORE_ATTR, e->v.Attribute.attr, names);
3633 break;
3634 case Del:
3635 ADDOP_NAME(c, DELETE_ATTR, e->v.Attribute.attr, names);
3636 break;
3637 case Param:
3638 default:
3639 PyErr_SetString(PyExc_SystemError,
3640 "param invalid in attribute expression");
3641 return 0;
3642 }
3643 break;
3644 case Subscript_kind:
3645 switch (e->v.Subscript.ctx) {
3646 case AugLoad:
3647 VISIT(c, expr, e->v.Subscript.value);
3648 VISIT_SLICE(c, e->v.Subscript.slice, AugLoad);
3649 break;
3650 case Load:
3651 VISIT(c, expr, e->v.Subscript.value);
3652 VISIT_SLICE(c, e->v.Subscript.slice, Load);
3653 break;
3654 case AugStore:
3655 VISIT_SLICE(c, e->v.Subscript.slice, AugStore);
3656 break;
3657 case Store:
3658 VISIT(c, expr, e->v.Subscript.value);
3659 VISIT_SLICE(c, e->v.Subscript.slice, Store);
3660 break;
3661 case Del:
3662 VISIT(c, expr, e->v.Subscript.value);
3663 VISIT_SLICE(c, e->v.Subscript.slice, Del);
3664 break;
3665 case Param:
3666 default:
3667 PyErr_SetString(PyExc_SystemError,
3668 "param invalid in subscript expression");
3669 return 0;
3670 }
3671 break;
3672 case Name_kind:
3673 return compiler_nameop(c, e->v.Name.id, e->v.Name.ctx);
3674 /* child nodes of List and Tuple will have expr_context set */
3675 case List_kind:
3676 return compiler_list(c, e);
3677 case Tuple_kind:
3678 return compiler_tuple(c, e);
3679 }
3680 return 1;
3681}
3682
3683static int
3684compiler_augassign(struct compiler *c, stmt_ty s)
3685{
3686 expr_ty e = s->v.AugAssign.target;
3687 expr_ty auge;
3688
3689 assert(s->kind == AugAssign_kind);
3690
3691 switch (e->kind) {
3692 case Attribute_kind:
3693 auge = Attribute(e->v.Attribute.value, e->v.Attribute.attr,
3694 AugLoad, e->lineno, e->col_offset, c->c_arena);
3695 if (auge == NULL)
3696 return 0;
3697 VISIT(c, expr, auge);
3698 VISIT(c, expr, s->v.AugAssign.value);
3699 ADDOP(c, inplace_binop(c, s->v.AugAssign.op));
3700 auge->v.Attribute.ctx = AugStore;
3701 VISIT(c, expr, auge);
3702 break;
3703 case Subscript_kind:
3704 auge = Subscript(e->v.Subscript.value, e->v.Subscript.slice,
3705 AugLoad, e->lineno, e->col_offset, c->c_arena);
3706 if (auge == NULL)
3707 return 0;
3708 VISIT(c, expr, auge);
3709 VISIT(c, expr, s->v.AugAssign.value);
3710 ADDOP(c, inplace_binop(c, s->v.AugAssign.op));
3711 auge->v.Subscript.ctx = AugStore;
3712 VISIT(c, expr, auge);
3713 break;
3714 case Name_kind:
3715 if (!compiler_nameop(c, e->v.Name.id, Load))
3716 return 0;
3717 VISIT(c, expr, s->v.AugAssign.value);
3718 ADDOP(c, inplace_binop(c, s->v.AugAssign.op));
3719 return compiler_nameop(c, e->v.Name.id, Store);
3720 default:
3721 PyErr_Format(PyExc_SystemError,
3722 "invalid node type (%d) for augmented assignment",
3723 e->kind);
3724 return 0;
3725 }
3726 return 1;
3727}
3728
3729static int
3730compiler_push_fblock(struct compiler *c, enum fblocktype t, basicblock *b)
3731{
3732 struct fblockinfo *f;
3733 if (c->u->u_nfblocks >= CO_MAXBLOCKS)
3734 return 0;
3735 f = &c->u->u_fblock[c->u->u_nfblocks++];
3736 f->fb_type = t;
3737 f->fb_block = b;
3738 return 1;
3739}
3740
3741static void
3742compiler_pop_fblock(struct compiler *c, enum fblocktype t, basicblock *b)
3743{
3744 struct compiler_unit *u = c->u;
3745 assert(u->u_nfblocks > 0);
3746 u->u_nfblocks--;
3747 assert(u->u_fblock[u->u_nfblocks].fb_type == t);
3748 assert(u->u_fblock[u->u_nfblocks].fb_block == b);
3749}
3750
3751/* Raises a SyntaxError and returns 0.
3752 If something goes wrong, a different exception may be raised.
3753*/
3754
3755static int
3756compiler_error(struct compiler *c, const char *errstr)
3757{
3758 PyObject *loc;
3759 PyObject *u = NULL, *v = NULL;
3760
3761 loc = PyErr_ProgramText(c->c_filename, c->u->u_lineno);
3762 if (!loc) {
3763 Py_INCREF(Py_None);
3764 loc = Py_None;
3765 }
3766 u = Py_BuildValue("(ziOO)", c->c_filename, c->u->u_lineno,
3767 Py_None, loc);
3768 if (!u)
3769 goto exit;
3770 v = Py_BuildValue("(zO)", errstr, u);
3771 if (!v)
3772 goto exit;
3773 PyErr_SetObject(PyExc_SyntaxError, v);
3774 exit:
3775 Py_DECREF(loc);
3776 Py_XDECREF(u);
3777 Py_XDECREF(v);
3778 return 0;
3779}
3780
3781static int
3782compiler_handle_subscr(struct compiler *c, const char *kind,
3783 expr_context_ty ctx)
3784{
3785 int op = 0;
3786
3787 /* XXX this code is duplicated */
3788 switch (ctx) {
3789 case AugLoad: /* fall through to Load */
3790 case Load: op = BINARY_SUBSCR; break;
3791 case AugStore:/* fall through to Store */
3792 case Store: op = STORE_SUBSCR; break;
3793 case Del: op = DELETE_SUBSCR; break;
3794 case Param:
3795 PyErr_Format(PyExc_SystemError,
3796 "invalid %s kind %d in subscript\n",
3797 kind, ctx);
3798 return 0;
3799 }
3800 if (ctx == AugLoad) {
3801 ADDOP_I(c, DUP_TOPX, 2);
3802 }
3803 else if (ctx == AugStore) {
3804 ADDOP(c, ROT_THREE);
3805 }
3806 ADDOP(c, op);
3807 return 1;
3808}
3809
3810static int
3811compiler_slice(struct compiler *c, slice_ty s, expr_context_ty ctx)
3812{
3813 int n = 2;
3814 assert(s->kind == Slice_kind);
3815
3816 /* only handles the cases where BUILD_SLICE is emitted */
3817 if (s->v.Slice.lower) {
3818 VISIT(c, expr, s->v.Slice.lower);
3819 }
3820 else {
3821 ADDOP_O(c, LOAD_CONST, Py_None, consts);
3822 }
3823
3824 if (s->v.Slice.upper) {
3825 VISIT(c, expr, s->v.Slice.upper);
3826 }
3827 else {
3828 ADDOP_O(c, LOAD_CONST, Py_None, consts);
3829 }
3830
3831 if (s->v.Slice.step) {
3832 n++;
3833 VISIT(c, expr, s->v.Slice.step);
3834 }
3835 ADDOP_I(c, BUILD_SLICE, n);
3836 return 1;
3837}
3838
3839static int
3840compiler_simple_slice(struct compiler *c, slice_ty s, expr_context_ty ctx)
3841{
3842 int op = 0, slice_offset = 0, stack_count = 0;
3843
3844 assert(s->v.Slice.step == NULL);
3845 if (s->v.Slice.lower) {
3846 slice_offset++;
3847 stack_count++;
3848 if (ctx != AugStore)
3849 VISIT(c, expr, s->v.Slice.lower);
3850 }
3851 if (s->v.Slice.upper) {
3852 slice_offset += 2;
3853 stack_count++;
3854 if (ctx != AugStore)
3855 VISIT(c, expr, s->v.Slice.upper);
3856 }
3857
3858 if (ctx == AugLoad) {
3859 switch (stack_count) {
3860 case 0: ADDOP(c, DUP_TOP); break;
3861 case 1: ADDOP_I(c, DUP_TOPX, 2); break;
3862 case 2: ADDOP_I(c, DUP_TOPX, 3); break;
3863 }
3864 }
3865 else if (ctx == AugStore) {
3866 switch (stack_count) {
3867 case 0: ADDOP(c, ROT_TWO); break;
3868 case 1: ADDOP(c, ROT_THREE); break;
3869 case 2: ADDOP(c, ROT_FOUR); break;
3870 }
3871 }
3872
3873 switch (ctx) {
3874 case AugLoad: /* fall through to Load */
3875 case Load: op = SLICE; break;
3876 case AugStore:/* fall through to Store */
3877 case Store: op = STORE_SLICE; break;
3878 case Del: op = DELETE_SLICE; break;
3879 case Param:
3880 default:
3881 PyErr_SetString(PyExc_SystemError,
3882 "param invalid in simple slice");
3883 return 0;
3884 }
3885
3886 ADDOP(c, op + slice_offset);
3887 return 1;
3888}
3889
3890static int
3891compiler_visit_nested_slice(struct compiler *c, slice_ty s,
3892 expr_context_ty ctx)
3893{
3894 switch (s->kind) {
3895 case Ellipsis_kind:
3896 ADDOP_O(c, LOAD_CONST, Py_Ellipsis, consts);
3897 break;
3898 case Slice_kind:
3899 return compiler_slice(c, s, ctx);
3900 case Index_kind:
3901 VISIT(c, expr, s->v.Index.value);
3902 break;
3903 case ExtSlice_kind:
3904 default:
3905 PyErr_SetString(PyExc_SystemError,
3906 "extended slice invalid in nested slice");
3907 return 0;
3908 }
3909 return 1;
3910}
3911
3912
3913static int
3914compiler_visit_slice(struct compiler *c, slice_ty s, expr_context_ty ctx)
3915{
3916 char * kindname = NULL;
3917 switch (s->kind) {
3918 case Index_kind:
3919 kindname = "index";
3920 if (ctx != AugStore) {
3921 VISIT(c, expr, s->v.Index.value);
3922 }
3923 break;
3924 case Ellipsis_kind:
3925 kindname = "ellipsis";
3926 if (ctx != AugStore) {
3927 ADDOP_O(c, LOAD_CONST, Py_Ellipsis, consts);
3928 }
3929 break;
3930 case Slice_kind:
3931 kindname = "slice";
3932 if (!s->v.Slice.step)
3933 return compiler_simple_slice(c, s, ctx);
3934 if (ctx != AugStore) {
3935 if (!compiler_slice(c, s, ctx))
3936 return 0;
3937 }
3938 break;
3939 case ExtSlice_kind:
3940 kindname = "extended slice";
3941 if (ctx != AugStore) {
3942 int i, n = asdl_seq_LEN(s->v.ExtSlice.dims);
3943 for (i = 0; i < n; i++) {
3944 slice_ty sub = (slice_ty)asdl_seq_GET(
3945 s->v.ExtSlice.dims, i);
3946 if (!compiler_visit_nested_slice(c, sub, ctx))
3947 return 0;
3948 }
3949 ADDOP_I(c, BUILD_TUPLE, n);
3950 }
3951 break;
3952 default:
3953 PyErr_Format(PyExc_SystemError,
3954 "invalid subscript kind %d", s->kind);
3955 return 0;
3956 }
3957 return compiler_handle_subscr(c, kindname, ctx);
3958}
3959
3960/* do depth-first search of basic block graph, starting with block.
3961 post records the block indices in post-order.
3962
3963 XXX must handle implicit jumps from one block to next
3964*/
3965
3966static void
3967dfs(struct compiler *c, basicblock *b, struct assembler *a)
3968{
3969 int i;
3970 struct instr *instr = NULL;
3971
3972 if (b->b_seen)
3973 return;
3974 b->b_seen = 1;
3975 if (b->b_next != NULL)
3976 dfs(c, b->b_next, a);
3977 for (i = 0; i < b->b_iused; i++) {
3978 instr = &b->b_instr[i];
3979 if (instr->i_jrel || instr->i_jabs)
3980 dfs(c, instr->i_target, a);
3981 }
3982 a->a_postorder[a->a_nblocks++] = b;
3983}
3984
3985static int
3986stackdepth_walk(struct compiler *c, basicblock *b, int depth, int maxdepth)
3987{
3988 int i;
3989 struct instr *instr;
3990 if (b->b_seen || b->b_startdepth >= depth)
3991 return maxdepth;
3992 b->b_seen = 1;
3993 b->b_startdepth = depth;
3994 for (i = 0; i < b->b_iused; i++) {
3995 instr = &b->b_instr[i];
3996 depth += opcode_stack_effect(instr->i_opcode, instr->i_oparg);
3997 if (depth > maxdepth)
3998 maxdepth = depth;
3999 assert(depth >= 0); /* invalid code or bug in stackdepth() */
4000 if (instr->i_jrel || instr->i_jabs) {
4001 maxdepth = stackdepth_walk(c, instr->i_target,
4002 depth, maxdepth);
4003 if (instr->i_opcode == JUMP_ABSOLUTE ||
4004 instr->i_opcode == JUMP_FORWARD) {
4005 goto out; /* remaining code is dead */
4006 }
4007 }
4008 }
4009 if (b->b_next)
4010 maxdepth = stackdepth_walk(c, b->b_next, depth, maxdepth);
4011out:
4012 b->b_seen = 0;
4013 return maxdepth;
4014}
4015
4016/* Find the flow path that needs the largest stack. We assume that
4017 * cycles in the flow graph have no net effect on the stack depth.
4018 */
4019static int
4020stackdepth(struct compiler *c)
4021{
4022 basicblock *b, *entryblock;
4023 entryblock = NULL;
4024 for (b = c->u->u_blocks; b != NULL; b = b->b_list) {
4025 b->b_seen = 0;
4026 b->b_startdepth = INT_MIN;
4027 entryblock = b;
4028 }
4029 if (!entryblock)
4030 return 0;
4031 return stackdepth_walk(c, entryblock, 0, 0);
4032}
4033
4034static int
4035assemble_init(struct assembler *a, int nblocks, int firstlineno)
4036{
4037 memset(a, 0, sizeof(struct assembler));
4038 a->a_lineno = firstlineno;
4039 a->a_bytecode = PyString_FromStringAndSize(NULL, DEFAULT_CODE_SIZE);
4040 if (!a->a_bytecode)
4041 return 0;
4042 a->a_lnotab = PyString_FromStringAndSize(NULL, DEFAULT_LNOTAB_SIZE);
4043 if (!a->a_lnotab)
4044 return 0;
4045 a->a_postorder = (basicblock **)PyObject_Malloc(
4046 sizeof(basicblock *) * nblocks);
4047 if (!a->a_postorder) {
4048 PyErr_NoMemory();
4049 return 0;
4050 }
4051 return 1;
4052}
4053
4054static void
4055assemble_free(struct assembler *a)
4056{
4057 Py_XDECREF(a->a_bytecode);
4058 Py_XDECREF(a->a_lnotab);
4059 if (a->a_postorder)
4060 PyObject_Free(a->a_postorder);
4061}
4062
4063/* Return the size of a basic block in bytes. */
4064
4065static int
4066instrsize(struct instr *instr)
4067{
4068 if (!instr->i_hasarg)
4069 return 1;
4070 if (instr->i_oparg > 0xffff)
4071 return 6;
4072 return 3;
4073}
4074
4075static int
4076blocksize(basicblock *b)
4077{
4078 int i;
4079 int size = 0;
4080
4081 for (i = 0; i < b->b_iused; i++)
4082 size += instrsize(&b->b_instr[i]);
4083 return size;
4084}
4085
4086/* All about a_lnotab.
4087
4088c_lnotab is an array of unsigned bytes disguised as a Python string.
4089It is used to map bytecode offsets to source code line #s (when needed
4090for tracebacks).
4091
4092The array is conceptually a list of
4093 (bytecode offset increment, line number increment)
4094pairs. The details are important and delicate, best illustrated by example:
4095
4096 byte code offset source code line number
4097 0 1
4098 6 2
4099 50 7
4100 350 307
4101 361 308
4102
4103The first trick is that these numbers aren't stored, only the increments
4104from one row to the next (this doesn't really work, but it's a start):
4105
4106 0, 1, 6, 1, 44, 5, 300, 300, 11, 1
4107
4108The second trick is that an unsigned byte can't hold negative values, or
4109values larger than 255, so (a) there's a deep assumption that byte code
4110offsets and their corresponding line #s both increase monotonically, and (b)
4111if at least one column jumps by more than 255 from one row to the next, more
4112than one pair is written to the table. In case #b, there's no way to know
4113from looking at the table later how many were written. That's the delicate
4114part. A user of c_lnotab desiring to find the source line number
4115corresponding to a bytecode address A should do something like this
4116
4117 lineno = addr = 0
4118 for addr_incr, line_incr in c_lnotab:
4119 addr += addr_incr
4120 if addr > A:
4121 return lineno
4122 lineno += line_incr
4123
4124In order for this to work, when the addr field increments by more than 255,
4125the line # increment in each pair generated must be 0 until the remaining addr
4126increment is < 256. So, in the example above, assemble_lnotab (it used
4127to be called com_set_lineno) should not (as was actually done until 2.2)
4128expand 300, 300 to 255, 255, 45, 45,
4129 but to 255, 0, 45, 255, 0, 45.
4130*/
4131
4132static int
4133assemble_lnotab(struct assembler *a, struct instr *i)
4134{
4135 int d_bytecode, d_lineno;
4136 int len;
4137 unsigned char *lnotab;
4138
4139 d_bytecode = a->a_offset - a->a_lineno_off;
4140 d_lineno = i->i_lineno - a->a_lineno;
4141
4142 assert(d_bytecode >= 0);
4143 assert(d_lineno >= 0);
4144
4145 /* XXX(nnorwitz): is there a better way to handle this?
4146 for loops are special, we want to be able to trace them
4147 each time around, so we need to set an extra line number. */
4148 if (d_lineno == 0 && i->i_opcode != FOR_ITER)
4149 return 1;
4150
4151 if (d_bytecode > 255) {
4152 int j, nbytes, ncodes = d_bytecode / 255;
4153 nbytes = a->a_lnotab_off + 2 * ncodes;
4154 len = PyString_GET_SIZE(a->a_lnotab);
4155 if (nbytes >= len) {
4156 if (len * 2 < nbytes)
4157 len = nbytes;
4158 else
4159 len *= 2;
4160 if (_PyString_Resize(&a->a_lnotab, len) < 0)
4161 return 0;
4162 }
4163 lnotab = (unsigned char *)
4164 PyString_AS_STRING(a->a_lnotab) + a->a_lnotab_off;
4165 for (j = 0; j < ncodes; j++) {
4166 *lnotab++ = 255;
4167 *lnotab++ = 0;
4168 }
4169 d_bytecode -= ncodes * 255;
4170 a->a_lnotab_off += ncodes * 2;
4171 }
4172 assert(d_bytecode <= 255);
4173 if (d_lineno > 255) {
4174 int j, nbytes, ncodes = d_lineno / 255;
4175 nbytes = a->a_lnotab_off + 2 * ncodes;
4176 len = PyString_GET_SIZE(a->a_lnotab);
4177 if (nbytes >= len) {
4178 if (len * 2 < nbytes)
4179 len = nbytes;
4180 else
4181 len *= 2;
4182 if (_PyString_Resize(&a->a_lnotab, len) < 0)
4183 return 0;
4184 }
4185 lnotab = (unsigned char *)
4186 PyString_AS_STRING(a->a_lnotab) + a->a_lnotab_off;
4187 *lnotab++ = d_bytecode;
4188 *lnotab++ = 255;
4189 d_bytecode = 0;
4190 for (j = 1; j < ncodes; j++) {
4191 *lnotab++ = 0;
4192 *lnotab++ = 255;
4193 }
4194 d_lineno -= ncodes * 255;
4195 a->a_lnotab_off += ncodes * 2;
4196 }
4197
4198 len = PyString_GET_SIZE(a->a_lnotab);
4199 if (a->a_lnotab_off + 2 >= len) {
4200 if (_PyString_Resize(&a->a_lnotab, len * 2) < 0)
4201 return 0;
4202 }
4203 lnotab = (unsigned char *)
4204 PyString_AS_STRING(a->a_lnotab) + a->a_lnotab_off;
4205
4206 a->a_lnotab_off += 2;
4207 if (d_bytecode) {
4208 *lnotab++ = d_bytecode;
4209 *lnotab++ = d_lineno;
4210 }
4211 else { /* First line of a block; def stmt, etc. */
4212 *lnotab++ = 0;
4213 *lnotab++ = d_lineno;
4214 }
4215 a->a_lineno = i->i_lineno;
4216 a->a_lineno_off = a->a_offset;
4217 return 1;
4218}
4219
4220/* assemble_emit()
4221 Extend the bytecode with a new instruction.
4222 Update lnotab if necessary.
4223*/
4224
4225static int
4226assemble_emit(struct assembler *a, struct instr *i)
4227{
4228 int size, arg = 0, ext = 0;
4229 Py_ssize_t len = PyString_GET_SIZE(a->a_bytecode);
4230 char *code;
4231
4232 size = instrsize(i);
4233 if (i->i_hasarg) {
4234 arg = i->i_oparg;
4235 ext = arg >> 16;
4236 }
4237 if (i->i_lineno && !assemble_lnotab(a, i))
4238 return 0;
4239 if (a->a_offset + size >= len) {
4240 if (_PyString_Resize(&a->a_bytecode, len * 2) < 0)
4241 return 0;
4242 }
4243 code = PyString_AS_STRING(a->a_bytecode) + a->a_offset;
4244 a->a_offset += size;
4245 if (size == 6) {
4246 assert(i->i_hasarg);
4247 *code++ = (char)EXTENDED_ARG;
4248 *code++ = ext & 0xff;
4249 *code++ = ext >> 8;
4250 arg &= 0xffff;
4251 }
4252 *code++ = i->i_opcode;
4253 if (i->i_hasarg) {
4254 assert(size == 3 || size == 6);
4255 *code++ = arg & 0xff;
4256 *code++ = arg >> 8;
4257 }
4258 return 1;
4259}
4260
4261static void
4262assemble_jump_offsets(struct assembler *a, struct compiler *c)
4263{
4264 basicblock *b;
4265 int bsize, totsize, extended_arg_count, last_extended_arg_count = 0;
4266 int i;
4267
4268 /* Compute the size of each block and fixup jump args.
4269 Replace block pointer with position in bytecode. */
4270start:
4271 totsize = 0;
4272 for (i = a->a_nblocks - 1; i >= 0; i--) {
4273 b = a->a_postorder[i];
4274 bsize = blocksize(b);
4275 b->b_offset = totsize;
4276 totsize += bsize;
4277 }
4278 extended_arg_count = 0;
4279 for (b = c->u->u_blocks; b != NULL; b = b->b_list) {
4280 bsize = b->b_offset;
4281 for (i = 0; i < b->b_iused; i++) {
4282 struct instr *instr = &b->b_instr[i];
4283 /* Relative jumps are computed relative to
4284 the instruction pointer after fetching
4285 the jump instruction.
4286 */
4287 bsize += instrsize(instr);
4288 if (instr->i_jabs)
4289 instr->i_oparg = instr->i_target->b_offset;
4290 else if (instr->i_jrel) {
4291 int delta = instr->i_target->b_offset - bsize;
4292 instr->i_oparg = delta;
4293 }
4294 else
4295 continue;
4296 if (instr->i_oparg > 0xffff)
4297 extended_arg_count++;
4298 }
4299 }
4300
4301 /* XXX: This is an awful hack that could hurt performance, but
4302 on the bright side it should work until we come up
4303 with a better solution.
4304
4305 In the meantime, should the goto be dropped in favor
4306 of a loop?
4307
4308 The issue is that in the first loop blocksize() is called
4309 which calls instrsize() which requires i_oparg be set
4310 appropriately. There is a bootstrap problem because
4311 i_oparg is calculated in the second loop above.
4312
4313 So we loop until we stop seeing new EXTENDED_ARGs.
4314 The only EXTENDED_ARGs that could be popping up are
4315 ones in jump instructions. So this should converge
4316 fairly quickly.
4317 */
4318 if (last_extended_arg_count != extended_arg_count) {
4319 last_extended_arg_count = extended_arg_count;
4320 goto start;
4321 }
4322}
4323
4324static PyObject *
4325dict_keys_inorder(PyObject *dict, int offset)
4326{
4327 PyObject *tuple, *k, *v;
4328 Py_ssize_t i, pos = 0, size = PyDict_Size(dict);
4329
4330 tuple = PyTuple_New(size);
4331 if (tuple == NULL)
4332 return NULL;
4333 while (PyDict_Next(dict, &pos, &k, &v)) {
4334 i = PyInt_AS_LONG(v);
4335 k = PyTuple_GET_ITEM(k, 0);
4336 Py_INCREF(k);
4337 assert((i - offset) < size);
4338 assert((i - offset) >= 0);
4339 PyTuple_SET_ITEM(tuple, i - offset, k);
4340 }
4341 return tuple;
4342}
4343
4344static int
4345compute_code_flags(struct compiler *c)
4346{
4347 PySTEntryObject *ste = c->u->u_ste;
4348 int flags = 0, n;
4349 if (ste->ste_type != ModuleBlock)
4350 flags |= CO_NEWLOCALS;
4351 if (ste->ste_type == FunctionBlock) {
4352 if (!ste->ste_unoptimized)
4353 flags |= CO_OPTIMIZED;
4354 if (ste->ste_nested)
4355 flags |= CO_NESTED;
4356 if (ste->ste_generator)
4357 flags |= CO_GENERATOR;
4358 }
4359 if (ste->ste_varargs)
4360 flags |= CO_VARARGS;
4361 if (ste->ste_varkeywords)
4362 flags |= CO_VARKEYWORDS;
4363 if (ste->ste_generator)
4364 flags |= CO_GENERATOR;
4365
4366 /* (Only) inherit compilerflags in PyCF_MASK */
4367 flags |= (c->c_flags->cf_flags & PyCF_MASK);
4368
4369 n = PyDict_Size(c->u->u_freevars);
4370 if (n < 0)
4371 return -1;
4372 if (n == 0) {
4373 n = PyDict_Size(c->u->u_cellvars);
4374 if (n < 0)
4375 return -1;
4376 if (n == 0) {
4377 flags |= CO_NOFREE;
4378 }
4379 }
4380
4381 return flags;
4382}
4383
4384static PyCodeObject *
4385makecode(struct compiler *c, struct assembler *a)
4386{
4387 PyObject *tmp;
4388 PyCodeObject *co = NULL;
4389 PyObject *consts = NULL;
4390 PyObject *names = NULL;
4391 PyObject *varnames = NULL;
4392 PyObject *filename = NULL;
4393 PyObject *name = NULL;
4394 PyObject *freevars = NULL;
4395 PyObject *cellvars = NULL;
4396 PyObject *bytecode = NULL;
4397 int nlocals, flags;
4398
4399 tmp = dict_keys_inorder(c->u->u_consts, 0);
4400 if (!tmp)
4401 goto error;
4402 consts = PySequence_List(tmp); /* optimize_code requires a list */
4403 Py_DECREF(tmp);
4404
4405 names = dict_keys_inorder(c->u->u_names, 0);
4406 varnames = dict_keys_inorder(c->u->u_varnames, 0);
4407 if (!consts || !names || !varnames)
4408 goto error;
4409
4410 cellvars = dict_keys_inorder(c->u->u_cellvars, 0);
4411 if (!cellvars)
4412 goto error;
4413 freevars = dict_keys_inorder(c->u->u_freevars, PyTuple_Size(cellvars));
4414 if (!freevars)
4415 goto error;
4416 filename = PyString_FromString(c->c_filename);
4417 if (!filename)
4418 goto error;
4419
4420 nlocals = PyDict_Size(c->u->u_varnames);
4421 flags = compute_code_flags(c);
4422 if (flags < 0)
4423 goto error;
4424
4425 bytecode = optimize_code(a->a_bytecode, consts, names, a->a_lnotab);
4426 if (!bytecode)
4427 goto error;
4428
4429 tmp = PyList_AsTuple(consts); /* PyCode_New requires a tuple */
4430 if (!tmp)
4431 goto error;
4432 Py_DECREF(consts);
4433 consts = tmp;
4434
4435 co = PyCode_New(c->u->u_argcount, nlocals, stackdepth(c), flags,
4436 bytecode, consts, names, varnames,
4437 freevars, cellvars,
4438 filename, c->u->u_name,
4439 c->u->u_firstlineno,
4440 a->a_lnotab);
4441 error:
4442 Py_XDECREF(consts);
4443 Py_XDECREF(names);
4444 Py_XDECREF(varnames);
4445 Py_XDECREF(filename);
4446 Py_XDECREF(name);
4447 Py_XDECREF(freevars);
4448 Py_XDECREF(cellvars);
4449 Py_XDECREF(bytecode);
4450 return co;
4451}
4452
4453
4454/* For debugging purposes only */
4455#if 0
4456static void
4457dump_instr(const struct instr *i)
4458{
4459 const char *jrel = i->i_jrel ? "jrel " : "";
4460 const char *jabs = i->i_jabs ? "jabs " : "";
4461 char arg[128];
4462
4463 *arg = '\0';
4464 if (i->i_hasarg)
4465 sprintf(arg, "arg: %d ", i->i_oparg);
4466
4467 fprintf(stderr, "line: %d, opcode: %d %s%s%s\n",
4468 i->i_lineno, i->i_opcode, arg, jabs, jrel);
4469}
4470
4471static void
4472dump_basicblock(const basicblock *b)
4473{
4474 const char *seen = b->b_seen ? "seen " : "";
4475 const char *b_return = b->b_return ? "return " : "";
4476 fprintf(stderr, "used: %d, depth: %d, offset: %d %s%s\n",
4477 b->b_iused, b->b_startdepth, b->b_offset, seen, b_return);
4478 if (b->b_instr) {
4479 int i;
4480 for (i = 0; i < b->b_iused; i++) {
4481 fprintf(stderr, " [%02d] ", i);
4482 dump_instr(b->b_instr + i);
4483 }
4484 }
4485}
4486#endif
4487
4488static PyCodeObject *
4489assemble(struct compiler *c, int addNone)
4490{
4491 basicblock *b, *entryblock;
4492 struct assembler a;
4493 int i, j, nblocks;
4494 PyCodeObject *co = NULL;
4495
4496 /* Make sure every block that falls off the end returns None.
4497 XXX NEXT_BLOCK() isn't quite right, because if the last
4498 block ends with a jump or return b_next shouldn't set.
4499 */
4500 if (!c->u->u_curblock->b_return) {
4501 NEXT_BLOCK(c);
4502 if (addNone)
4503 ADDOP_O(c, LOAD_CONST, Py_None, consts);
4504 ADDOP(c, RETURN_VALUE);
4505 }
4506
4507 nblocks = 0;
4508 entryblock = NULL;
4509 for (b = c->u->u_blocks; b != NULL; b = b->b_list) {
4510 nblocks++;
4511 entryblock = b;
4512 }
4513
4514 /* Set firstlineno if it wasn't explicitly set. */
4515 if (!c->u->u_firstlineno) {
4516 if (entryblock && entryblock->b_instr)
4517 c->u->u_firstlineno = entryblock->b_instr->i_lineno;
4518 else
4519 c->u->u_firstlineno = 1;
4520 }
4521 if (!assemble_init(&a, nblocks, c->u->u_firstlineno))
4522 goto error;
4523 dfs(c, entryblock, &a);
4524
4525 /* Can't modify the bytecode after computing jump offsets. */
4526 assemble_jump_offsets(&a, c);
4527
4528 /* Emit code in reverse postorder from dfs. */
4529 for (i = a.a_nblocks - 1; i >= 0; i--) {
4530 b = a.a_postorder[i];
4531 for (j = 0; j < b->b_iused; j++)
4532 if (!assemble_emit(&a, &b->b_instr[j]))
4533 goto error;
4534 }
4535
4536 if (_PyString_Resize(&a.a_lnotab, a.a_lnotab_off) < 0)
4537 goto error;
4538 if (_PyString_Resize(&a.a_bytecode, a.a_offset) < 0)
4539 goto error;
4540
4541 co = makecode(c, &a);
4542 error:
4543 assemble_free(&a);
4544 return co;
4545}
Note: See TracBrowser for help on using the repository browser.