Skip to content

Commit a613df7

Browse files
totalspectrumtotalspectrum
committed
Added brainf*ck frontend
1 parent 411bb69 commit a613df7

File tree

6 files changed

+357
-2
lines changed

6 files changed

+357
-2
lines changed

‎Makefile

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -123,7 +123,7 @@ CFLAGS = $(OPT) -Wall -fwrapv -Wc++-compat $(INC) $(DEFS)
123123
LIBS = -lm
124124
RM = rm -rf
125125

126-
VPATH=.:util:frontends:frontends/basic:frontends/spin:frontends/c:backends:backends/asm:backends/cpp:backends/bytecode:backends/dat:backends/nucode:backends/objfile:backends/zip:backends/compress:backends/compress/lz4:mcpp
126+
VPATH=.:util:frontends:frontends/basic:frontends/spin:frontends/c:backends:frontends/bf:backends/asm:backends/cpp:backends/bytecode:backends/dat:backends/nucode:backends/objfile:backends/zip:backends/compress:backends/compress/lz4:mcpp
127127

128128
LEXHEADERS = $(BUILD)/spin.tab.h $(BUILD)/basic.tab.h $(BUILD)/cgram.tab.h ast.h frontends/common.h
129129

@@ -140,7 +140,7 @@ NUBACK = outnu.c nuir.c nupeep.c
140140
CPPBACK = outcpp.c cppfunc.c outgas.c cppexpr.c cppbuiltin.c
141141
COMPBACK = compress.c lz4.c lz4hc.c
142142
ZIPBACK = outzip.c zip.c
143-
SPINSRCS = common.c case.c spinc.c $(LEXSRCS) functions.c cse.c loops.c hloptimize.c hltransform.c types.c pasm.c outdat.c outlst.c outobj.c spinlang.c basiclang.c clang.c $(PASMBACK) $(BCBACK) $(NUBACK) $(CPPBACK) $(COMPBACK) $(ZIPBACK) $(MCPP) version.c becommon.c brkdebug.c printdebug.c
143+
SPINSRCS = common.c case.c spinc.c $(LEXSRCS) functions.c cse.c loops.c hloptimize.c hltransform.c types.c pasm.c outdat.c outlst.c outobj.c spinlang.c basiclang.c clang.c bflang.c $(PASMBACK) $(BCBACK) $(NUBACK) $(CPPBACK) $(COMPBACK) $(ZIPBACK) $(MCPP) version.c becommon.c brkdebug.c printdebug.c
144144

145145
LEXOBJS = $(LEXSRCS:%.c=$(BUILD)/%.o)
146146
SPINOBJS = $(SPINSRCS:%.c=$(BUILD)/%.o)

‎frontends/bf/bflang.c

Lines changed: 339 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,339 @@
1+
/*
2+
* BF to C or PASM translater
3+
* Copyright 2025 Total Spectrum Software and Collabora, Ltd.
4+
*
5+
* SPDX-License-Identifier: MIT
6+
*/
7+
8+
/*
9+
* A parser for a really simple language to the flexspin internal API
10+
* Intended more as an example for implementors than as something
11+
* to actually use in the real world
12+
*
13+
* The language implemented here is an esoteric language called
14+
* BrainF*ck or BF (or various other things) with only 8 commands,
15+
* each consisting of only a single character:
16+
* > Increment data pointer by 1
17+
* < Decrement data pointer by 1
18+
* + Increment the byte at the data pointer by 1
19+
* - Decrement the byte at the data pointer by 1
20+
* . Output the byte at the data pointer
21+
* , Read a byte and store it at the data pointer
22+
* [ If the byte at the data pointer is 0, jump to the command
23+
* after the next matching ]
24+
* ] If the byte at the data pointer is non-zero, jump back to
25+
* the previous matching [
26+
*
27+
* All other characters are ignored and considered to be comments.
28+
*
29+
* I/O is performed using the standard Spin SEND and RECV pointers, so
30+
* a BF program may be included as an object in a Spin program. If
31+
* SEND and/or RECV are not initialized, they are set to an appropriate serial
32+
* function.
33+
*/
34+
35+
#include <stdlib.h>
36+
#include <string.h>
37+
#include "spinc.h"
38+
39+
static AST *array_name; /* the name of the BF main BF data array */
40+
static AST *array_index; /* the name of the var to index that array */
41+
static AST *array_mask; /* a mask so we wrap around at end of array */
42+
static unsigned array_size; /* size of the array (based on processor) */
43+
44+
static AST *cur_pos_deref; /* a dereference to fetch/store the byte at the current position */
45+
46+
/*
47+
* fetch a single BF character
48+
* ignore any non-useful (comment) characters
49+
* returns -1 on EOF
50+
*/
51+
static int nextbfchar(LexStream *L) {
52+
int c;
53+
54+
do {
55+
c = lexgetc(L);
56+
switch (c) {
57+
case '[':
58+
case ']':
59+
case '+':
60+
case '-':
61+
case '<':
62+
case '>':
63+
case '.':
64+
case ',':
65+
/* a valid BF character, return to parser */
66+
return c;
67+
default:
68+
break;
69+
}
70+
} while (c > 0);
71+
return c;
72+
}
73+
74+
/*
75+
* add an expression wrapped in a statement list to
76+
* a list of AST's
77+
* returns the new value for the list
78+
*/
79+
static AST *
80+
append_expr_statement(AST *body, AST *expr)
81+
{
82+
AST *stmt = NewAST(AST_STMTLIST, expr, NULL);
83+
body = AddToList(body, stmt);
84+
return body;
85+
}
86+
87+
/*
88+
* Utility routine: read a sequence of characters ch,
89+
* count them, and then increment or decrement the
90+
* given variable reference by that many.
91+
* Optionally (if mask is non-NULL) applies the given
92+
* mask before storing the result back.
93+
*/
94+
static AST *
95+
incDecVar(LexStream *L, int ch, int incDecOp, AST *varref, AST *mask)
96+
{
97+
int n = 0;
98+
int c = ch;
99+
AST *ast;
100+
// count how many ch's in a row we see
101+
// strictly speaking we don't have to merge them like this,
102+
// but it is a tremendous optimization
103+
while (c == ch) {
104+
n++;
105+
c = nextbfchar(L);
106+
}
107+
// the last character we saw was not '+', so
108+
// save it back for the next parse
109+
lexungetc(L, c);
110+
111+
ast = AstOperator(incDecOp, varref, AstInteger(n));
112+
if (mask) {
113+
ast = AstOperator('&', ast, mask);
114+
}
115+
// create the statement to update the variable
116+
ast = AstAssign(varref, ast);
117+
return ast;
118+
}
119+
120+
/*
121+
* return a dereference of the current position
122+
*/
123+
static AST *CurPos(void) {
124+
return DupAST(cur_pos_deref);
125+
}
126+
127+
/*
128+
* parse a BF expression; this can be one of:
129+
* .: output current character
130+
* ,: read a character and write to current location
131+
* a sequence of N +'s: add N to current location
132+
* a sequence of N -'s: subtract N from current location
133+
* a sequence of N >'s: add N to current location pointer
134+
* a sequence of N <'s: subtract N from current location pointer
135+
* [: form a loop up until the next ]
136+
* ]: terminate a loop
137+
*
138+
* Returns an AST for the expression, or NULL
139+
* on EOF
140+
*/
141+
static AST *
142+
parseBFstream(LexStream *L)
143+
{
144+
int c;
145+
AST *loopbody = NULL;
146+
AST *ast;
147+
148+
for(;;) {
149+
c = nextbfchar(L);
150+
if (c == '+') {
151+
ast = incDecVar(L, c, '+', CurPos(), NULL);
152+
} else if (c == '-') {
153+
ast = incDecVar(L, c, '-', CurPos(), NULL);
154+
} else if (c == '>') {
155+
ast = incDecVar(L, c, '+', array_index, array_mask);
156+
} else if (c == '<') {
157+
ast = incDecVar(L, c, '-', array_index, array_mask);
158+
} else if (c == '.') {
159+
// print the byte at the current position
160+
AST *sendptr = AstIdentifier("__sendptr");
161+
ast = NewAST(AST_FUNCCALL,
162+
sendptr,
163+
NewAST(AST_EXPRLIST, CurPos(), NULL));
164+
} else if (c == ',') {
165+
// read something into the current position
166+
AST *recvptr = AstIdentifier("__recvptr");
167+
ast = AstAssign(CurPos(),
168+
NewAST(AST_FUNCCALL,
169+
recvptr,
170+
NULL));
171+
} else if (c == '[') {
172+
// open a loop
173+
// this should emit
174+
// if (*cur_pos_deref != 0) {
175+
// do {
176+
// <loopbody>
177+
// } while (*cur_pos_deref != 0);
178+
ast = parseBFstream(L);
179+
// create the do-while loop
180+
ast = NewAST(AST_DOWHILE,
181+
AstOperator(K_NE, CurPos(), AstInteger(0)),
182+
ast);
183+
// wrap it in the if
184+
ast = NewAST(AST_IF,
185+
AstOperator(K_NE, CurPos(), AstInteger(0)),
186+
NewAST(AST_THENELSE,
187+
NewAST(AST_STMTLIST, ast, NULL),
188+
NULL));
189+
} else if (c == ']') {
190+
ast = NULL;
191+
break;
192+
} else {
193+
if (c != -1) {
194+
ERROR(NULL, "unexpected character encountered");
195+
}
196+
break;
197+
}
198+
loopbody = AddToList(loopbody,
199+
NewAST(AST_STMTLIST, ast, NULL));
200+
}
201+
return loopbody;
202+
}
203+
204+
/*
205+
* initialize a BF parse
206+
* there will be only one function (the
207+
* main body) which will go in current->body
208+
* we need to create an array (8K for P1, 32K for P2)
209+
* Note that the parser is not re-entrant (none of
210+
* flexspin's parsers are) so we're using static variables
211+
* for everything
212+
*/
213+
214+
static void
215+
init_bf_parse(LexStream *L)
216+
{
217+
AST *ast;
218+
219+
// array size is 8K for P1 (too small, but the P1 only has 32K total)
220+
// and 32K for P2
221+
// must be a power of 2 so array_mask can work
222+
array_size = gl_p2 ? 32768 : 8192;
223+
224+
// create an array mask
225+
array_mask = AstInteger(array_size - 1);
226+
227+
// create the member variable to hold the BF data (an array of bytes)
228+
// this declaration is like
229+
// long bf_array[array_size]
230+
// in C
231+
array_name = AstIdentifier("bf_array");
232+
AST *array_decl = NewAST(AST_ARRAYDECL,
233+
array_name, AstInteger(array_size));
234+
// the base index of the array is put in d.ptr (it will default to 0,
235+
// but it doesn't hurt to make it explicit)
236+
array_decl->d.ptr = AstInteger(0);
237+
238+
// we put the array in the module's DAT section
239+
MaybeDeclareMemberVar(current, array_decl, ast_type_byte, 0, NORMAL_VAR);
240+
241+
// create a local variable declaration for the array_index
242+
array_index = AstIdentifier("bf_pos");
243+
// declare it and initialize it to 0
244+
AST *initZero = AstAssign(array_index, AstInteger(0));
245+
246+
// the variable declaration AST expects a list of variables to
247+
// initialize, hence the AST_LISTHOLDER
248+
AST *decl = NewAST(AST_DECLARE_VAR,
249+
ast_type_long,
250+
NewAST(AST_LISTHOLDER, initZero, NULL));
251+
252+
// add the variable declaration to the main program body
253+
current->body = append_expr_statement(current->body, decl);
254+
255+
// create a dereference bf_array[bf_pos]
256+
cur_pos_deref = NewAST(AST_ARRAYREF,
257+
array_name, array_index);
258+
259+
// This part below is somewhat technical; we want to do I/O by using
260+
// the standard Spin2 send() and recv() functions. But those may not
261+
// be initialized (if we're not using the BF program as an object).
262+
// So add some code to initialize them to the default _tx and _rx
263+
// functions
264+
AST *sendfunc = AstIdentifier("__sendptr");
265+
AST *recvfunc = AstIdentifier("__recvptr");
266+
267+
// check for sendfunc initialized
268+
ast = AstOperator(K_EQ, sendfunc, AstInteger(0));
269+
ast = NewAST(AST_IF, ast,
270+
NewAST(AST_THENELSE,
271+
NewAST(AST_STMTLIST,
272+
AstAssign(sendfunc,
273+
NewAST(AST_ADDROF,
274+
AstIdentifier("_tx"),
275+
NULL)),
276+
NULL),
277+
NULL));
278+
current->body = append_expr_statement(current->body, ast);
279+
280+
// check for recvfunc initialized
281+
ast = AstOperator(K_EQ, recvfunc, AstInteger(0));
282+
ast = NewAST(AST_IF, ast,
283+
NewAST(AST_THENELSE,
284+
NewAST(AST_STMTLIST,
285+
AstAssign(recvfunc,
286+
NewAST(AST_ADDROF,
287+
AstIdentifier("_rx"),
288+
NULL)), NULL),
289+
NULL));
290+
current->body = append_expr_statement(current->body, ast);
291+
292+
}
293+
294+
/*
295+
* This uses the same prototype as a YACC generated parser, so it takes no
296+
* parameters and uses the lexer stream in the current module (a global
297+
* variable. This is ugly, but compatible.
298+
* returns 0 on success, -1 on failure
299+
*/
300+
int bfparse(void)
301+
{
302+
LexStream *L = current->Lptr;
303+
AST *ast;
304+
int c;
305+
306+
/* initialize the parser state */
307+
init_bf_parse(L);
308+
309+
ast = parseBFstream(L);
310+
current->body = AddToList(current->body, ast);
311+
312+
// we should be at EOF here
313+
c = lexpeekc(L);
314+
if (c > 0) {
315+
ERROR(NULL, "[] mismatch detected");
316+
return -1;
317+
}
318+
return 0;
319+
}
320+
321+
//
322+
// High level transformation on the intermediate AST
323+
// For BF, this doesn't really have to do anything, but
324+
// we may as well run the standard high level optimizations
325+
// and such
326+
//
327+
void
328+
BFTransform(Function *func)
329+
{
330+
InitGlobalFuncs();
331+
332+
// simplify assignments
333+
DoHLTransforms(func);
334+
335+
// Do type checking here. For BF this is
336+
// only needed because of the setup of send := @_tx
337+
// at initialization time
338+
CheckTypes(func->body);
339+
}

‎frontends/common.h

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -944,6 +944,9 @@ void initSpinLexer(int flags);
944944
#define LANG_CFAMILY_CPP 0x21
945945
#define IsCLang(lang) ((lang)>=LANG_CFAMILY_C && (lang)<=LANG_CFAMILY_CPP)
946946

947+
#define LANG_BF 0x30
948+
#define IsBFLang(lang) ((lang) == LANG_BF)
949+
947950
#define LANG_DEFAULT LANG_SPIN_SPIN1
948951
#define LANG_ANY (0xffff)
949952

‎functions.c

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -2332,6 +2332,8 @@ ProcessOneFunc(Function *pf)
23322332
CTransform(pf);
23332333
} else if (IsSpinLang(pf->language)) {
23342334
SpinTransform(pf);
2335+
} else if (IsBFLang(pf->language)) {
2336+
BFTransform(pf);
23352337
}
23362338

23372339
if (last_errors != gl_errors && gl_errors >= gl_max_errors) return;

0 commit comments

Comments
 (0)