xref: /freebsd/contrib/bc/include/bc.h (revision 8aac90f18aef7c9eea906c3ff9a001ca7b94f375)
1 /*
2  * *****************************************************************************
3  *
4  * SPDX-License-Identifier: BSD-2-Clause
5  *
6  * Copyright (c) 2018-2023 Gavin D. Howard and contributors.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions are met:
10  *
11  * * Redistributions of source code must retain the above copyright notice, this
12  *   list of conditions and the following disclaimer.
13  *
14  * * Redistributions in binary form must reproduce the above copyright notice,
15  *   this list of conditions and the following disclaimer in the documentation
16  *   and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
19  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
22  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28  * POSSIBILITY OF SUCH DAMAGE.
29  *
30  * *****************************************************************************
31  *
32  * Definitions for bc only.
33  *
34  */
35 
36 #ifndef BC_BC_H
37 #define BC_BC_H
38 
39 #if BC_ENABLED
40 
41 #include <limits.h>
42 #include <stdbool.h>
43 
44 #include <status.h>
45 #include <lex.h>
46 #include <parse.h>
47 
48 /**
49  * The main function for bc. It just sets variables and passes its arguments
50  * through to @a bc_vm_boot().
51  */
52 void
53 bc_main(int argc, char* argv[]);
54 
55 // These are references to the help text, the library text, and the "filename"
56 // for the library.
57 extern const char bc_help[];
58 extern const char bc_lib[];
59 extern const char* bc_lib_name;
60 
61 // These are references to the second math library and its "filename."
62 #if BC_ENABLE_EXTRA_MATH
63 extern const char bc_lib2[];
64 extern const char* bc_lib2_name;
65 #endif // BC_ENABLE_EXTRA_MATH
66 
67 /**
68  * A struct containing information about a bc keyword.
69  */
70 typedef struct BcLexKeyword
71 {
72 	/// Holds the length of the keyword along with a bit that, if set, means the
73 	/// keyword is used in POSIX bc.
74 	uchar data;
75 
76 	/// The keyword text.
77 	const char name[14];
78 } BcLexKeyword;
79 
80 /// Sets the most significant bit. Used for setting the POSIX bit in
81 /// BcLexKeyword's data field.
82 #define BC_LEX_CHAR_MSB(bit) ((bit) << (CHAR_BIT - 1))
83 
84 /// Returns non-zero if the keyword is POSIX, zero otherwise.
85 #define BC_LEX_KW_POSIX(kw) ((kw)->data & (BC_LEX_CHAR_MSB(1)))
86 
87 /// Returns the length of the keyword.
88 #define BC_LEX_KW_LEN(kw) ((size_t) ((kw)->data & ~(BC_LEX_CHAR_MSB(1))))
89 
90 /// A macro to easily build a keyword entry. See bc_lex_kws in src/data.c.
91 #define BC_LEX_KW_ENTRY(a, b, c)                                              \
92 	{                                                                         \
93 		.data = ((b) & ~(BC_LEX_CHAR_MSB(1))) | BC_LEX_CHAR_MSB(c), .name = a \
94 	}
95 
96 #if BC_ENABLE_EXTRA_MATH
97 
98 /// A macro for the number of keywords bc has. This has to be updated if any are
99 /// added. This is for the redefined_kws field of the BcVm struct.
100 #define BC_LEX_NKWS (37)
101 
102 #else // BC_ENABLE_EXTRA_MATH
103 
104 /// A macro for the number of keywords bc has. This has to be updated if any are
105 /// added. This is for the redefined_kws field of the BcVm struct.
106 #define BC_LEX_NKWS (33)
107 
108 #endif // BC_ENABLE_EXTRA_MATH
109 
110 // The array of keywords and its length.
111 extern const BcLexKeyword bc_lex_kws[];
112 extern const size_t bc_lex_kws_len;
113 
114 /**
115  * The @a BcLexNext function for bc. (See include/lex.h for a definition of
116  * @a BcLexNext.)
117  * @param l  The lexer.
118  */
119 void
120 bc_lex_token(BcLex* l);
121 
122 // The following section is for flags needed when parsing bc code. These flags
123 // are complicated, but necessary. Why you ask? Because bc's standard is awful.
124 //
125 // If you don't believe me, go read the bc Parsing section of the Development
126 // manual (manuals/development.md). Then come back.
127 //
128 // In other words, these flags are the sign declaring, "Here be dragons."
129 
130 /**
131  * This returns a pointer to the set of flags at the top of the flag stack.
132  * @a p is expected to be a BcParse pointer.
133  * @param p  The parser.
134  * @return   A pointer to the top flag set.
135  */
136 #define BC_PARSE_TOP_FLAG_PTR(p) ((uint16_t*) bc_vec_top(&(p)->flags))
137 
138 /**
139  * This returns the flag set at the top of the flag stack. @a p is expected to
140  * be a BcParse pointer.
141  * @param p  The parser.
142  * @return   The top flag set.
143  */
144 #define BC_PARSE_TOP_FLAG(p) (*(BC_PARSE_TOP_FLAG_PTR(p)))
145 
146 // After this point, all flag #defines are in sets of 2: one to define the flag,
147 // and one to define a way to grab the flag from the flag set at the top of the
148 // flag stack. All `p` arguments are pointers to a BcParse.
149 
150 // This flag is set if the parser has seen a left brace.
151 #define BC_PARSE_FLAG_BRACE (UINTMAX_C(1) << 0)
152 #define BC_PARSE_BRACE(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_BRACE)
153 
154 // This flag is set if the parser is parsing inside of the braces of a function
155 // body.
156 #define BC_PARSE_FLAG_FUNC_INNER (UINTMAX_C(1) << 1)
157 #define BC_PARSE_FUNC_INNER(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_FUNC_INNER)
158 
159 // This flag is set if the parser is parsing a function. It is different from
160 // the one above because it is set if it is parsing a function body *or* header,
161 // not just if it's parsing a function body.
162 #define BC_PARSE_FLAG_FUNC (UINTMAX_C(1) << 2)
163 #define BC_PARSE_FUNC(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_FUNC)
164 
165 // This flag is set if the parser is expecting to parse a body, whether of a
166 // function, an if statement, or a loop.
167 #define BC_PARSE_FLAG_BODY (UINTMAX_C(1) << 3)
168 #define BC_PARSE_BODY(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_BODY)
169 
170 // This flag is set if bc is parsing a loop. This is important because the break
171 // and continue keywords are only valid inside of a loop.
172 #define BC_PARSE_FLAG_LOOP (UINTMAX_C(1) << 4)
173 #define BC_PARSE_LOOP(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_LOOP)
174 
175 // This flag is set if bc is parsing the body of a loop. It is different from
176 // the one above the same way @a BC_PARSE_FLAG_FUNC_INNER is different from
177 // @a BC_PARSE_FLAG_FUNC.
178 #define BC_PARSE_FLAG_LOOP_INNER (UINTMAX_C(1) << 5)
179 #define BC_PARSE_LOOP_INNER(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_LOOP_INNER)
180 
181 // This flag is set if bc is parsing an if statement.
182 #define BC_PARSE_FLAG_IF (UINTMAX_C(1) << 6)
183 #define BC_PARSE_IF(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_IF)
184 
185 // This flag is set if bc is parsing an else statement. This is important
186 // because of "else if" constructions, among other things.
187 #define BC_PARSE_FLAG_ELSE (UINTMAX_C(1) << 7)
188 #define BC_PARSE_ELSE(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_ELSE)
189 
190 // This flag is set if bc just finished parsing an if statement and its body.
191 // It tells the parser that it can probably expect an else statement next. This
192 // flag is, thus, one of the most subtle.
193 #define BC_PARSE_FLAG_IF_END (UINTMAX_C(1) << 8)
194 #define BC_PARSE_IF_END(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_IF_END)
195 
196 /**
197  * This returns true if bc is in a state where it should not execute any code
198  * at all.
199  * @param p  The parser.
200  * @return   True if execution cannot proceed, false otherwise.
201  */
202 #define BC_PARSE_NO_EXEC(p) ((p)->flags.len != 1 || BC_PARSE_TOP_FLAG(p) != 0)
203 
204 /**
205  * This returns true if the token @a t is a statement delimiter, which is
206  * either a newline or a semicolon.
207  * @param t  The token to check.
208  * @return   True if t is a statement delimiter token; false otherwise.
209  */
210 #define BC_PARSE_DELIMITER(t) \
211 	((t) == BC_LEX_SCOLON || (t) == BC_LEX_NLINE || (t) == BC_LEX_EOF)
212 
213 /**
214  * This is poorly named, but it basically returns whether or not the current
215  * state is valid for the end of an else statement.
216  * @param f  The flag set to be checked.
217  * @return   True if the state is valid for the end of an else statement.
218  */
219 #define BC_PARSE_BLOCK_STMT(f) \
220 	((f) & (BC_PARSE_FLAG_ELSE | BC_PARSE_FLAG_LOOP_INNER))
221 
222 /**
223  * This returns the value of the data for an operator with precedence @a p and
224  * associativity @a l (true if left associative, false otherwise). This is used
225  * to construct an array of operators, bc_parse_ops, in src/data.c.
226  * @param p  The precedence.
227  * @param l  True if the operator is left associative, false otherwise.
228  * @return   The data for the operator.
229  */
230 #define BC_PARSE_OP(p, l) (((p) & ~(BC_LEX_CHAR_MSB(1))) | (BC_LEX_CHAR_MSB(l)))
231 
232 /**
233  * Returns the operator data for the lex token @a t.
234  * @param t  The token to return operator data for.
235  * @return   The operator data for @a t.
236  */
237 #define BC_PARSE_OP_DATA(t) bc_parse_ops[((t) -BC_LEX_OP_INC)]
238 
239 /**
240  * Returns non-zero if operator @a op is left associative, zero otherwise.
241  * @param op  The operator to test for associativity.
242  * @return    Non-zero if the operator is left associative, zero otherwise.
243  */
244 #define BC_PARSE_OP_LEFT(op) (BC_PARSE_OP_DATA(op) & BC_LEX_CHAR_MSB(1))
245 
246 /**
247  * Returns the precedence of operator @a op. Lower number means higher
248  * precedence.
249  * @param op  The operator to return the precedence of.
250  * @return    The precedence of @a op.
251  */
252 #define BC_PARSE_OP_PREC(op) (BC_PARSE_OP_DATA(op) & ~(BC_LEX_CHAR_MSB(1)))
253 
254 /**
255  * A macro to easily define a series of bits for whether a lex token is an
256  * expression token or not. It takes 8 expression bits, corresponding to the 8
257  * bits in a uint8_t. You can see this in use for bc_parse_exprs in src/data.c.
258  * @param e1  The first bit.
259  * @param e2  The second bit.
260  * @param e3  The third bit.
261  * @param e4  The fourth bit.
262  * @param e5  The fifth bit.
263  * @param e6  The sixth bit.
264  * @param e7  The seventh bit.
265  * @param e8  The eighth bit.
266  * @return    An expression entry for bc_parse_exprs[].
267  */
268 #define BC_PARSE_EXPR_ENTRY(e1, e2, e3, e4, e5, e6, e7, e8)               \
269 	((UINTMAX_C(e1) << 7) | (UINTMAX_C(e2) << 6) | (UINTMAX_C(e3) << 5) | \
270 	 (UINTMAX_C(e4) << 4) | (UINTMAX_C(e5) << 3) | (UINTMAX_C(e6) << 2) | \
271 	 (UINTMAX_C(e7) << 1) | (UINTMAX_C(e8) << 0))
272 
273 /**
274  * Returns true if token @a i is a token that belongs in an expression.
275  * @param i  The token to test.
276  * @return   True if i is an expression token, false otherwise.
277  */
278 #define BC_PARSE_EXPR(i) \
279 	(bc_parse_exprs[(((i) & (uchar) ~(0x07)) >> 3)] & (1 << (7 - ((i) & 0x07))))
280 
281 /**
282  * Returns the operator (by lex token) that is at the top of the operator
283  * stack.
284  * @param p  The parser.
285  * @return   The operator that is at the top of the operator stack, as a lex
286  *           token.
287  */
288 #define BC_PARSE_TOP_OP(p) (*((BcLexType*) bc_vec_top(&(p)->ops)))
289 
290 /**
291  * Returns true if bc has a "leaf" token. A "leaf" token is one that can stand
292  * alone in an expression. For example, a number by itself can be an expression,
293  * but a binary operator, while valid for an expression, cannot be alone in the
294  * expression. It must have an expression to the left and right of itself. See
295  * the documentation for @a bc_parse_expr_err() in src/bc_parse.c.
296  * @param prev      The previous token as an instruction.
297  * @param bin_last  True if that last operator was a binary operator, false
298  *                  otherwise.
299  * @param rparen    True if the last operator was a right paren.
300  * return           True if the last token was a leaf token, false otherwise.
301  */
302 #define BC_PARSE_LEAF(prev, bin_last, rparen) \
303 	(!(bin_last) && ((rparen) || bc_parse_inst_isLeaf(prev)))
304 
305 /**
306  * This returns true if the token @a t should be treated as though it's a
307  * variable. This goes for actual variables, array elements, and globals.
308  * @param t  The token to test.
309  * @return   True if @a t should be treated as though it's a variable, false
310  *           otherwise.
311  */
312 #if BC_ENABLE_EXTRA_MATH
313 #define BC_PARSE_INST_VAR(t) \
314 	((t) >= BC_INST_VAR && (t) <= BC_INST_SEED && (t) != BC_INST_ARRAY)
315 #else // BC_ENABLE_EXTRA_MATH
316 #define BC_PARSE_INST_VAR(t) \
317 	((t) >= BC_INST_VAR && (t) <= BC_INST_SCALE && (t) != BC_INST_ARRAY)
318 #endif // BC_ENABLE_EXTRA_MATH
319 
320 /**
321  * Returns true if the previous token @a p (in the form of a bytecode
322  * instruction) is a prefix operator. The fact that it is for bytecode
323  * instructions is what makes it different from @a BC_PARSE_OP_PREFIX below.
324  * @param p  The previous token.
325  * @return   True if @a p is a prefix operator.
326  */
327 #define BC_PARSE_PREV_PREFIX(p) ((p) >= BC_INST_NEG && (p) <= BC_INST_BOOL_NOT)
328 
329 /**
330  * Returns true if token @a t is a prefix operator.
331  * @param t  The token to test.
332  * @return   True if @a t is a prefix operator, false otherwise.
333  */
334 #define BC_PARSE_OP_PREFIX(t) ((t) == BC_LEX_OP_BOOL_NOT || (t) == BC_LEX_NEG)
335 
336 /**
337  * We can calculate the conversion between tokens and bytecode instructions by
338  * subtracting the position of the first operator in the lex enum and adding the
339  * position of the first in the instruction enum. Note: This only works for
340  * binary operators.
341  * @param t  The token to turn into an instruction.
342  * @return   The token as an instruction.
343  */
344 #define BC_PARSE_TOKEN_INST(t) ((uchar) ((t) -BC_LEX_NEG + BC_INST_NEG))
345 
346 /**
347  * Returns true if the token is a bc keyword.
348  * @param t  The token to check.
349  * @return   True if @a t is a bc keyword, false otherwise.
350  */
351 #define BC_PARSE_IS_KEYWORD(t) ((t) >= BC_LEX_KW_AUTO && (t) <= BC_LEX_KW_ELSE)
352 
353 /// A struct that holds data about what tokens should be expected next. There
354 /// are a few instances of these, all named because they are used in specific
355 /// cases. Basically, in certain situations, it's useful to use the same code,
356 /// but have a list of valid tokens.
357 ///
358 /// Obviously, @a len is the number of tokens in the @a tokens array. If more
359 /// than 4 is needed in the future, @a tokens will have to be changed.
360 typedef struct BcParseNext
361 {
362 	/// The number of tokens in the tokens array.
363 	uchar len;
364 
365 	/// The tokens that can be expected next.
366 	uchar tokens[4];
367 
368 } BcParseNext;
369 
370 /// A macro to construct an array literal of tokens from a parameter list.
371 #define BC_PARSE_NEXT_TOKENS(...) .tokens = { __VA_ARGS__ }
372 
373 /// A macro to generate a BcParseNext literal from BcParseNext data. See
374 /// src/data.c for examples.
375 #define BC_PARSE_NEXT(a, ...)                                 \
376 	{                                                         \
377 		.len = (uchar) (a), BC_PARSE_NEXT_TOKENS(__VA_ARGS__) \
378 	}
379 
380 /// A status returned by @a bc_parse_expr_err(). It can either return success or
381 /// an error indicating an empty expression.
382 typedef enum BcParseStatus
383 {
384 	BC_PARSE_STATUS_SUCCESS,
385 	BC_PARSE_STATUS_EMPTY_EXPR,
386 
387 } BcParseStatus;
388 
389 /**
390  * The @a BcParseExpr function for bc. (See include/parse.h for a definition of
391  * @a BcParseExpr.)
392  * @param p      The parser.
393  * @param flags  Flags that define the requirements that the parsed code must
394  *               meet or an error will result. See @a BcParseExpr for more info.
395  */
396 void
397 bc_parse_expr(BcParse* p, uint8_t flags);
398 
399 /**
400  * The @a BcParseParse function for bc. (See include/parse.h for a definition of
401  * @a BcParseParse.)
402  * @param p  The parser.
403  */
404 void
405 bc_parse_parse(BcParse* p);
406 
407 /**
408  * Ends a series of if statements. This is to ensure that full parses happen
409  * when a file finishes or before defining a function. Without this, bc thinks
410  * that it cannot parse any further. But if we reach the end of a file or a
411  * function definition, we know we can add an empty else clause.
412  * @param p  The parser.
413  */
414 void
415 bc_parse_endif(BcParse* p);
416 
417 /// References to the signal message and its length.
418 extern const char bc_sig_msg[];
419 extern const uchar bc_sig_msg_len;
420 
421 /// A reference to an array of bits that are set if the corresponding lex token
422 /// is valid in an expression.
423 extern const uint8_t bc_parse_exprs[];
424 
425 /// A reference to an array of bc operators.
426 extern const uchar bc_parse_ops[];
427 
428 // References to the various instances of BcParseNext's.
429 
430 /// A reference to what tokens are valid as next tokens when parsing normal
431 /// expressions. More accurately. these are the tokens that are valid for
432 /// *ending* the expression.
433 extern const BcParseNext bc_parse_next_expr;
434 
435 /// A reference to what tokens are valid as next tokens when parsing function
436 /// parameters (well, actually arguments).
437 extern const BcParseNext bc_parse_next_arg;
438 
439 /// A reference to what tokens are valid as next tokens when parsing a print
440 /// statement.
441 extern const BcParseNext bc_parse_next_print;
442 
443 /// A reference to what tokens are valid as next tokens when parsing things like
444 /// loop headers and builtin functions where the only thing expected is a right
445 /// paren.
446 ///
447 /// The name is an artifact of history, and is related to @a BC_PARSE_REL (see
448 /// include/parse.h). It refers to how POSIX only allows some operators as part
449 /// of the conditional of for loops, while loops, and if statements.
450 extern const BcParseNext bc_parse_next_rel;
451 
452 // What tokens are valid as next tokens when parsing an array element
453 // expression.
454 extern const BcParseNext bc_parse_next_elem;
455 
456 /// A reference to what tokens are valid as next tokens when parsing the first
457 /// two parts of a for loop header.
458 extern const BcParseNext bc_parse_next_for;
459 
460 /// A reference to what tokens are valid as next tokens when parsing a read
461 /// expression.
462 extern const BcParseNext bc_parse_next_read;
463 
464 /// A reference to what tokens are valid as next tokens when parsing a builtin
465 /// function with multiple arguments.
466 extern const BcParseNext bc_parse_next_builtin;
467 
468 #else // BC_ENABLED
469 
470 // If bc is not enabled, execution is always possible because dc has strict
471 // rules that ensure execution can always proceed safely.
472 #define BC_PARSE_NO_EXEC(p) (0)
473 
474 #endif // BC_ENABLED
475 
476 #endif // BC_BC_H
477