andrew/akbasic
1
0
Fork 0
Code Issues 27 Pull Requests 1 Actions Packages Projects Releases Wiki Activity
Files
5f1d1cb2e0295c9a8f3e6e1d866fa3a04f39ebcd
akbasic/src/runtime_functions.c

654 lines
25 KiB
C
Raw Normal View History

Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
/**
* @file runtime_functions.c
* @brief The built-in function implementations.
*
* Ported from basicruntime_functions.go. The reference bootstraps its function
* table by running a BASIC program of DEF statements through the interpreter at
* startup and then nulling out the expressions so the native handlers take over
* -- except MOD, SPC and STR, which stay as BASIC expressions. None of that is
* reproduced: the signatures are data in the dispatch table and all three of
* those are ordinary native handlers here, which removes the need to run the
* interpreter before the interpreter is ready.
*/
#include <inttypes.h>
#include <math.h>
#include <stdio.h>
#include <akerror.h>
Delete src/convert.c: libakstdlib 0.2.0 grew the contract it existed for That file wrapped strtoll and strtod because libakstdlib's aksl_ato* family could not report a conversion failure at all -- atoi("not a number") returned success with 0, turning four diagnosable errors into wrong answers. Its own note said what to do when that changed: "When it grows it, delete src/convert.c and switch the call sites over." 0.2.0 grew it, so it is gone. Six call sites go straight to the library now: both literal constructors in src/grammar.c, the line number in src/scanner.c, FunctionVAL in src/runtime_functions.c, INPUT in src/runtime_commands.c, and GSHAPE's handle in src/runtime_graphics.c. aksl_strtoll(str, NULL, base, &dest) rather than aksl_atoll wherever a base is involved, because the ato* forms are base 10 and grammar.c picks base 8 or 16 off the lexeme's prefix; the NULL endptr is what makes trailing junk an error rather than a stopping point. The raised status changed from AKBASIC_ERR_VALUE to AKERR_VALUE, and the message with it -- VAL("garbage") now says `no digits in "garbage"`. Section 1.8 makes message text part of the acceptance contract, so that was checked against the corpus before touching anything: no golden file contained a conversion message, which is also why section 1.9 had been asking for one. Two exist now, so the next change to that text has to move a golden file, and one of them pins the reference's octal-literal defect (section 6 item 10) while it is still deliberately reproduced. tests/convert.c became tests/numeric_contract.c rather than being deleted with the code. The assertions did not stop being worth making when the wrapper went away -- they became assertions about a contract this port depends on and no longer owns, and a regression in it would make four things quietly return zero. Repoints the CI mutation job, which was bounded to src/convert.c and src/symtab.c. src/symtab.c alone measures 74.1% against the gate of 65. src/audio_tables.c was measured as a replacement second file and scored 64.7%: that is a real gap rather than a reason to skip it, since almost every survivor is in akbasic_audio_state_init, where nothing asserts a freshly initialised audio state is actually zeroed -- the same gap this job's history records closing for symtab. Recorded in TODO.md so the file can earn its place back. One thing worth knowing before reading any score here, and now written down: the harness's ICR operator only rewrites the constants 0 and 1, so a lookup table of other values produces no mutants and a high score over one says nothing about whether its entries are right. tests/audio_verbs.c now asserts all 32 ADSR entries against the datasheet anyway, because a hand-edit typo is the real failure mode there -- but that could not and did not move the mutation number, and claiming otherwise would be the kind of thing this file exists to stop. 72/72 core, 73/73 with libakgl, 72/72 under ASan+UBSan, coverage 93.6% line and 97.8% function, clean under -Wall -Wextra, doxygen clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-31 10:28:04 -04:00
#include <akstdlib.h>
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
#include <akbasic/error.h>
#include <akbasic/runtime.h>
#include "verbs.h"
/*
* Every builtin starts the same way: find the argument list, evaluate the first
* argument, and hand back a fresh value to write the answer into. Arity is
* already guaranteed by the parser against the table's arity column.
*/
static akerr_ErrorContext *first_arg(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, const char *fname, akbasic_ASTLeaf **argleaf, akbasic_Value **argvalue, akbasic_Value **out)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *arg = NULL;
FAIL_ZERO_RETURN(errctx, (expr != NULL), AKERR_NULLPOINTER, "NIL leaf");
arg = akbasic_leaf_first_argument(expr);
FAIL_ZERO_RETURN(errctx, (arg != NULL), AKBASIC_ERR_SYNTAX,
"%s expected an argument", fname);
if ( argvalue != NULL ) {
PASS(errctx, akbasic_runtime_evaluate(obj, arg, argvalue));
}
if ( out != NULL ) {
PASS(errctx, akbasic_environment_new_value(obj->environment, out));
PASS(errctx, akbasic_value_zero(*out));
}
if ( argleaf != NULL ) {
*argleaf = arg;
}
SUCCEED_RETURN(errctx);
}
/* Coerce an integer-or-float argument to double, the shape every trig call wants. */
static akerr_ErrorContext *arg_as_double(akbasic_Value *value, const char *fname, double *dest)
{
PREPARE_ERROR(errctx);
if ( value->valuetype == AKBASIC_TYPE_INTEGER ) {
*dest = (double)value->intval;
SUCCEED_RETURN(errctx);
}
if ( value->valuetype == AKBASIC_TYPE_FLOAT ) {
*dest = value->floatval;
SUCCEED_RETURN(errctx);
}
FAIL_RETURN(errctx, AKBASIC_ERR_TYPE, "%s expected integer or float", fname);
}
/* The trig family differs only by the libm call, so it is one shape. */
#define DEFINE_MATH_FUNCTION(__cname, __basicname, __call) \
akerr_ErrorContext *__cname(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, \
akbasic_Value *lval, akbasic_Value *rval, \
akbasic_Value **dest) \
{ \
PREPARE_ERROR(errctx); \
akbasic_Value *arg = NULL; \
akbasic_Value *out = NULL; \
double input = 0.0; \
\
(void)lval; (void)rval; \
PASS(errctx, first_arg(obj, expr, __basicname, NULL, &arg, &out)); \
PASS(errctx, arg_as_double(arg, __basicname, &input)); \
out->valuetype = AKBASIC_TYPE_FLOAT; \
out->floatval = __call(input); \
*dest = out; \
SUCCEED_RETURN(errctx); \
}
DEFINE_MATH_FUNCTION(akbasic_fn_atn, "ATN", atan)
DEFINE_MATH_FUNCTION(akbasic_fn_cos, "COS", cos)
DEFINE_MATH_FUNCTION(akbasic_fn_sin, "SIN", sin)
DEFINE_MATH_FUNCTION(akbasic_fn_tan, "TAN", tan)
DEFINE_MATH_FUNCTION(akbasic_fn_log, "LOG", log)
akerr_ErrorContext *akbasic_fn_abs(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "ABS", NULL, &arg, &out));
if ( arg->valuetype == AKBASIC_TYPE_INTEGER ) {
out->valuetype = AKBASIC_TYPE_INTEGER;
out->intval = (arg->intval < 0 ? -arg->intval : arg->intval);
} else if ( arg->valuetype == AKBASIC_TYPE_FLOAT ) {
out->valuetype = AKBASIC_TYPE_FLOAT;
out->floatval = fabs(arg->floatval);
} else {
FAIL_RETURN(errctx, AKBASIC_ERR_TYPE, "ABS expected integer or float");
}
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_rad(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
double input = 0.0;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "RAD", NULL, &arg, &out));
PASS(errctx, arg_as_double(arg, "RAD", &input));
out->valuetype = AKBASIC_TYPE_FLOAT;
out->floatval = input * (M_PI / 180.0);
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_sgn(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
double input = 0.0;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "SGN", NULL, &arg, &out));
PASS(errctx, arg_as_double(arg, "SGN", &input));
out->valuetype = AKBASIC_TYPE_INTEGER;
out->intval = (input < 0.0 ? -1 : (input > 0.0 ? 1 : 0));
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_chr(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
int64_t codepoint = 0;
int written = 0;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "CHR", NULL, &arg, &out));
FAIL_NONZERO_RETURN(errctx, (arg->valuetype != AKBASIC_TYPE_INTEGER), AKBASIC_ERR_TYPE,
"CHR expected an integer codepoint");
codepoint = arg->intval;
FAIL_ZERO_RETURN(errctx, (codepoint >= 0 && codepoint <= 0x10FFFF), AKBASIC_ERR_VALUE,
"CHR codepoint %" PRId64 " is outside Unicode", codepoint);
/* UTF-8 encode, matching the reference's string(rune(x)). */
out->valuetype = AKBASIC_TYPE_STRING;
if ( codepoint < 0x80 ) {
out->stringval[written++] = (char)codepoint;
} else if ( codepoint < 0x800 ) {
out->stringval[written++] = (char)(0xC0 | (codepoint >> 6));
out->stringval[written++] = (char)(0x80 | (codepoint & 0x3F));
} else if ( codepoint < 0x10000 ) {
out->stringval[written++] = (char)(0xE0 | (codepoint >> 12));
out->stringval[written++] = (char)(0x80 | ((codepoint >> 6) & 0x3F));
out->stringval[written++] = (char)(0x80 | (codepoint & 0x3F));
} else {
out->stringval[written++] = (char)(0xF0 | (codepoint >> 18));
out->stringval[written++] = (char)(0x80 | ((codepoint >> 12) & 0x3F));
out->stringval[written++] = (char)(0x80 | ((codepoint >> 6) & 0x3F));
out->stringval[written++] = (char)(0x80 | (codepoint & 0x3F));
}
out->stringval[written] = '\0';
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_hex(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
int written = 0;
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "HEX", NULL, &arg, &out));
FAIL_NONZERO_RETURN(errctx, (arg->valuetype != AKBASIC_TYPE_INTEGER), AKBASIC_ERR_TYPE,
"HEX expected an integer");
out->valuetype = AKBASIC_TYPE_STRING;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
/*
* aksl_snprintf refuses to truncate, which costs nothing here: sixteen hex
* digits is the longest an int64_t can be and a string value holds far more.
*/
PASS(errctx, aksl_snprintf(&written, out->stringval, sizeof(out->stringval),
"%" PRIx64, arg->intval));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_str(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "STR", NULL, &arg, &out));
/*
* The reference defines this as `"" + X#`, so it renders exactly the way
* string concatenation does -- an integer as %d and a float as %f.
*/
PASS(errctx, akbasic_value_to_string(arg, out->stringval, sizeof(out->stringval)));
out->valuetype = AKBASIC_TYPE_STRING;
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_spc(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
int64_t i = 0;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "SPC", NULL, &arg, &out));
FAIL_NONZERO_RETURN(errctx, (arg->valuetype != AKBASIC_TYPE_INTEGER), AKBASIC_ERR_TYPE,
"SPC expected an integer");
FAIL_ZERO_RETURN(errctx, (arg->intval >= 0 && arg->intval < AKBASIC_MAX_STRING_LENGTH),
AKBASIC_ERR_VALUE,
"SPC count %" PRId64 " is outside 0..%d",
arg->intval, AKBASIC_MAX_STRING_LENGTH - 1);
out->valuetype = AKBASIC_TYPE_STRING;
for ( i = 0; i < arg->intval; i++ ) {
out->stringval[i] = ' ';
}
out->stringval[arg->intval] = '\0';
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_val(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "VAL", NULL, &arg, &out));
FAIL_NONZERO_RETURN(errctx, (arg->valuetype != AKBASIC_TYPE_STRING), AKBASIC_ERR_TYPE,
"VAL expected a string");
out->valuetype = AKBASIC_TYPE_FLOAT;
/*
* Through the strict converter, never aksl_atof: that family cannot report a
* failure, so VAL("garbage") would quietly answer 0.0 instead of raising.
* See TODO.md section 1.9.
*/
Delete src/convert.c: libakstdlib 0.2.0 grew the contract it existed for That file wrapped strtoll and strtod because libakstdlib's aksl_ato* family could not report a conversion failure at all -- atoi("not a number") returned success with 0, turning four diagnosable errors into wrong answers. Its own note said what to do when that changed: "When it grows it, delete src/convert.c and switch the call sites over." 0.2.0 grew it, so it is gone. Six call sites go straight to the library now: both literal constructors in src/grammar.c, the line number in src/scanner.c, FunctionVAL in src/runtime_functions.c, INPUT in src/runtime_commands.c, and GSHAPE's handle in src/runtime_graphics.c. aksl_strtoll(str, NULL, base, &dest) rather than aksl_atoll wherever a base is involved, because the ato* forms are base 10 and grammar.c picks base 8 or 16 off the lexeme's prefix; the NULL endptr is what makes trailing junk an error rather than a stopping point. The raised status changed from AKBASIC_ERR_VALUE to AKERR_VALUE, and the message with it -- VAL("garbage") now says `no digits in "garbage"`. Section 1.8 makes message text part of the acceptance contract, so that was checked against the corpus before touching anything: no golden file contained a conversion message, which is also why section 1.9 had been asking for one. Two exist now, so the next change to that text has to move a golden file, and one of them pins the reference's octal-literal defect (section 6 item 10) while it is still deliberately reproduced. tests/convert.c became tests/numeric_contract.c rather than being deleted with the code. The assertions did not stop being worth making when the wrapper went away -- they became assertions about a contract this port depends on and no longer owns, and a regression in it would make four things quietly return zero. Repoints the CI mutation job, which was bounded to src/convert.c and src/symtab.c. src/symtab.c alone measures 74.1% against the gate of 65. src/audio_tables.c was measured as a replacement second file and scored 64.7%: that is a real gap rather than a reason to skip it, since almost every survivor is in akbasic_audio_state_init, where nothing asserts a freshly initialised audio state is actually zeroed -- the same gap this job's history records closing for symtab. Recorded in TODO.md so the file can earn its place back. One thing worth knowing before reading any score here, and now written down: the harness's ICR operator only rewrites the constants 0 and 1, so a lookup table of other values produces no mutants and a high score over one says nothing about whether its entries are right. tests/audio_verbs.c now asserts all 32 ADSR entries against the datasheet anyway, because a hand-edit typo is the real failure mode there -- but that could not and did not move the mutation number, and claiming otherwise would be the kind of thing this file exists to stop. 72/72 core, 73/73 with libakgl, 72/72 under ASan+UBSan, coverage 93.6% line and 97.8% function, clean under -Wall -Wextra, doxygen clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-31 10:28:04 -04:00
PASS(errctx, aksl_atof(arg->stringval, &out->floatval));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_len(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *argleaf = NULL;
akbasic_Value *strval = NULL;
akbasic_Value *out = NULL;
akbasic_Variable *variable = NULL;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
size_t length = 0;
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
(void)lval; (void)rval;
/*
* LEN must not evaluate a non-string argument. `LEN(A#)` on a
* multi-dimensional array is legal and asks for the element count, but
* evaluating a bare identifier supplies the single subscript {0}, which a
* two-dimensional variable rightly rejects. So: inspect the leaf, and only
* evaluate when it is a string.
*/
PASS(errctx, first_arg(obj, expr, "LEN", &argleaf, NULL, NULL));
FAIL_ZERO_RETURN(errctx,
(akbasic_leaf_is_identifier(argleaf) || akbasic_leaf_is_literal(argleaf)),
AKBASIC_ERR_SYNTAX, "Expected identifier or string literal");
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_INTEGER;
if ( argleaf->leaftype == AKBASIC_LEAF_LITERAL_STRING ||
argleaf->leaftype == AKBASIC_LEAF_IDENTIFIER_STRING ) {
PASS(errctx, akbasic_runtime_evaluate(obj, argleaf, &strval));
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_strlen(strval->stringval, &length));
out->intval = (int64_t)length;
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
} else {
PASS(errctx, akbasic_environment_get(obj->environment, argleaf->identifier, &variable));
FAIL_ZERO_RETURN(errctx, (variable != NULL), AKBASIC_ERR_UNDEFINED,
"Identifier %s is undefined", argleaf->identifier);
out->intval = variable->valuecount;
}
*dest = out;
SUCCEED_RETURN(errctx);
}
/* Fetch argument N (zero-based) of a call, already evaluated. */
static akerr_ErrorContext *nth_arg(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, const char *fname, int n, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *arg = NULL;
int i = 0;
FAIL_ZERO_RETURN(errctx, (expr != NULL), AKERR_NULLPOINTER, "NIL leaf");
arg = akbasic_leaf_first_argument(expr);
for ( i = 0; i < n && arg != NULL; i++ ) {
Give argument lists their own link, not each argument's .right An identifier's subscript list, a unary's operand and a binary's right-hand side all lived in the same field the argument chain used, so ABS(-9), MOD(A#, B# + 12) and any array reference in a parameter list were counted as extra arguments and refused. Subscript lists move to .expr and arguments now chain through a dedicated .next. This is the third time the same collision was fixed; the first two moved it along rather than removing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-31 11:58:15 -04:00
arg = arg->next;
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
}
FAIL_ZERO_RETURN(errctx, (arg != NULL), AKBASIC_ERR_SYNTAX,
"%s is missing argument %d", fname, n + 1);
PASS(errctx, akbasic_runtime_evaluate(obj, arg, dest));
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_instr(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *haystack = NULL;
akbasic_Value *needle = NULL;
akbasic_Value *out = NULL;
char *hit = NULL;
(void)lval; (void)rval;
PASS(errctx, nth_arg(obj, expr, "INSTR", 0, &haystack));
PASS(errctx, nth_arg(obj, expr, "INSTR", 1, &needle));
FAIL_ZERO_RETURN(errctx,
(haystack->valuetype == AKBASIC_TYPE_STRING &&
needle->valuetype == AKBASIC_TYPE_STRING),
AKBASIC_ERR_TYPE, "INSTR expected two strings");
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_INTEGER;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
/* Not finding it is a successful NULL, which is exactly the -1 answer below. */
PASS(errctx, aksl_strstr(haystack->stringval, needle->stringval, &hit));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
out->intval = (hit == NULL ? -1 : (int64_t)(hit - haystack->stringval));
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_left(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *source = NULL;
akbasic_Value *count = NULL;
akbasic_Value *out = NULL;
int64_t take = 0;
size_t sourcelen = 0;
(void)lval; (void)rval;
PASS(errctx, nth_arg(obj, expr, "LEFT", 0, &source));
PASS(errctx, nth_arg(obj, expr, "LEFT", 1, &count));
FAIL_NONZERO_RETURN(errctx, (source->valuetype != AKBASIC_TYPE_STRING), AKBASIC_ERR_TYPE,
"LEFT expected a string");
FAIL_NONZERO_RETURN(errctx, (count->valuetype != AKBASIC_TYPE_INTEGER), AKBASIC_ERR_TYPE,
"LEFT expected an integer count");
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_strlen(source->stringval, &sourcelen));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
take = count->intval;
if ( take < 0 ) {
take = 0;
}
if ( (size_t)take > sourcelen ) {
take = (int64_t)sourcelen; /* clamped to LEN, as the README says */
}
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_STRING;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_memcpy(out->stringval, source->stringval, (size_t)take));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
out->stringval[take] = '\0';
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_right(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *source = NULL;
akbasic_Value *count = NULL;
akbasic_Value *out = NULL;
int64_t take = 0;
size_t sourcelen = 0;
(void)lval; (void)rval;
PASS(errctx, nth_arg(obj, expr, "RIGHT", 0, &source));
PASS(errctx, nth_arg(obj, expr, "RIGHT", 1, &count));
FAIL_NONZERO_RETURN(errctx, (source->valuetype != AKBASIC_TYPE_STRING), AKBASIC_ERR_TYPE,
"RIGHT expected a string");
FAIL_NONZERO_RETURN(errctx, (count->valuetype != AKBASIC_TYPE_INTEGER), AKBASIC_ERR_TYPE,
"RIGHT expected an integer count");
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_strlen(source->stringval, &sourcelen));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
take = count->intval;
if ( take < 0 ) {
take = 0;
}
if ( (size_t)take > sourcelen ) {
take = (int64_t)sourcelen;
}
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_STRING;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_memcpy(out->stringval, source->stringval + (sourcelen - (size_t)take),
(size_t)take));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
out->stringval[take] = '\0';
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_mid(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *source = NULL;
akbasic_Value *startval = NULL;
akbasic_Value *lengthval = NULL;
akbasic_Value *out = NULL;
int64_t start = 0;
int64_t length = 0;
size_t sourcelen = 0;
(void)lval; (void)rval;
PASS(errctx, nth_arg(obj, expr, "MID", 0, &source));
PASS(errctx, nth_arg(obj, expr, "MID", 1, &startval));
PASS(errctx, nth_arg(obj, expr, "MID", 2, &lengthval));
FAIL_NONZERO_RETURN(errctx, (source->valuetype != AKBASIC_TYPE_STRING), AKBASIC_ERR_TYPE,
"MID expected a string");
FAIL_ZERO_RETURN(errctx,
(startval->valuetype == AKBASIC_TYPE_INTEGER &&
lengthval->valuetype == AKBASIC_TYPE_INTEGER),
AKBASIC_ERR_TYPE, "MID expected integer start and length");
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_strlen(source->stringval, &sourcelen));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
start = startval->intval;
length = lengthval->intval;
FAIL_ZERO_RETURN(errctx, (start >= 0 && (size_t)start <= sourcelen), AKBASIC_ERR_BOUNDS,
"MID start %" PRId64 " is outside 0..%zu", start, sourcelen);
if ( length < 0 ) {
length = 0;
}
if ( (size_t)(start + length) > sourcelen ) {
length = (int64_t)sourcelen - start;
}
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_STRING;
Port onto libakstdlib 2b79aca and convert the eight bool predicates akbasic's src/ now calls libakstdlib 313 times and raw libc 7 -- 2.2% bypassed, against 86.4% on the same tree before this. The submodule bump 669b2b3 -> 2b79aca needed no source change of its own: the release is drop-in for what akbasic already used. Seven of the eight sites the earlier port left on raw libc change their own signature rather than swallowing an error, per andrew's ruling on libakstdlib#38. word_is, the is_waiting_for pair, the scanner's is_at_end, peek, peek_next and match_next_char, format.c's overflow, and sink_akgl's scroll/newline/putchar_at/echo_line/edit_key chain all return an akerr_ErrorContext * and hand the answer back through an out parameter. is_waiting_for and is_waiting_for_any are a public header change; every call site that used one as a term in a condition hoists it into a statement first. verb_compare is the eighth and stays on strcmp. bsearch(3) fixes the comparator's signature, so there is no out parameter to report through -- which is what libakstdlib#38 concluded. It carries a comment saying so and saying why the bypass is safe there. Six snprintf sites stay raw because they want truncation as an answer rather than an error, and aksl_snprintf cannot express that until libakstdlib#34 hands the required length back. Each of the six says so at the site. Two of them, in host.c, are a latent defect rather than a decision: a host type name over 31 characters truncates silently and two sharing a prefix then collide, where structtype.c refuses the same case. DLOAD leaked a file descriptor. Its read loop sat inside an ATTEMPT and the PASS in it returned past CLEANUP, so a scan error left the file open. Hoisting the loop into its own helper to convert fgets fixes it. Refs libakstdlib#26, libakstdlib#38 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-03 15:41:49 -04:00
PASS(errctx, aksl_memcpy(out->stringval, source->stringval + start, (size_t)length));
Port the BASIC interpreter from Go to C Reproduces deps/basicinterpret in C, in the idiom of the ak* libraries. All 41 .bas files in the reference's corpus produce byte-identical stdout, including the trailing double newline on an error line -- that comes from basicError building a string ending in \n and handing it to Println, and array_outofbounds.txt encodes it. The corpus is driven in place from the submodule as 41 individual CTest cases rather than copied, so it cannot drift from upstream. Eighteen unit tests cover what the corpus cannot reach. Three structural changes carry most of the work. Go's three reflection lookups (Command*, Function*, ParseCommand*) become one sorted dispatch table in src/verbs.c searched with bsearch; adding a verb is a row and two functions. The five Go maps become one fixed open-addressed table over aksl_strhash_djb2. And run(), which owned the process until MODE_QUIT, splits into step() plus a bounded run() -- goal 3 requires a host game to be able to bound execution, and nothing in the library now terminates the process or touches SDL. Output goes through an akbasic_TextSink vtable. src/sink_stdio.c is what makes the corpus runnable with no SDL present; the akgl-backed sink is still to come and is blocked on libakgl having no text-measurement call. src/convert.c exists because libakstdlib's aksl_ato* family cannot report a conversion failure (its TODO.md 2.1.5). The reference checks strconv's error at four sites and turns it into a BASIC error; routing those through aksl_atoi would have turned four diagnosable errors into wrong answers, with VAL("garbage") quietly returning 0. TODO.md 1.9 records which libakstdlib calls are cleared for use here and which are not. Reference defects are reproduced, not fixed: the golden files encode the observed behaviour and a silent correction is a behaviour change. TODO.md section 6 lists sixteen, and tests/known_reference_defects.c asserts the *correct* contract for six of them under AKBASIC_KNOWN_FAILING_TESTS, so a fix shows up as "unexpectedly passed". Five of the sixteen were found by this port and are new: subtraction stops after one operator so 1-2-3 computes 1-2 and abandons the rest of the line (a wrong answer, not a refused one); a unary-minus argument inflates a function's arity so ABS(-9) is rejected; a comparison operator in a line's final column is dropped; hex literals never survive the scanner; and the "Reserved word in variable name" check is dead code. Where the reference reaches undefined behaviour by a route that is defined in Go -- an out-of-range shift, a negative string multiplier, integer division by zero -- this raises instead of inheriting the UB. No golden case exercises any of them. The top-level CMakeLists shadows add_test, set_tests_properties and add_custom_target around all three add_subdirectory calls. Without it libakerror's tests land in our suite as Not Run, and its un-namespaced `coverage` target stops a coverage build from configuring at all. Test targets are akbasic_test_<name>: bare test_<name> collides with libakstdlib's, which is what broke libakgl's configure in c2b16d3. ctest 59/59; ASan+UBSan 59/59; 92.3% line and 96.9% function coverage; no warnings under -Wall -Wextra. Branch coverage is not a target, for the reason libakstdlib and libakgl both record: the akerror macros expand into large branch trees at every call site. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
2026-07-30 23:53:56 -04:00
out->stringval[length] = '\0';
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_mod(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *left = NULL;
akbasic_Value *right = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, nth_arg(obj, expr, "MOD", 0, &left));
PASS(errctx, nth_arg(obj, expr, "MOD", 1, &right));
FAIL_ZERO_RETURN(errctx,
(left->valuetype == AKBASIC_TYPE_INTEGER &&
right->valuetype == AKBASIC_TYPE_INTEGER),
AKBASIC_ERR_TYPE, "MOD expected two integers");
FAIL_ZERO_RETURN(errctx, (right->intval != 0), AKBASIC_ERR_VALUE, "DIVISION BY ZERO");
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_INTEGER;
/*
* The reference defines MOD as `X% - (Y% * (X% / Y%))` in BASIC, which with
* truncating integer division is exactly C's %.
*/
out->intval = left->intval - (right->intval * (left->intval / right->intval));
*dest = out;
SUCCEED_RETURN(errctx);
}
/* SHL, SHR and XOR share a shape: two integers in, one integer out. */
static akerr_ErrorContext *two_integers(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, const char *fname, int64_t *a, int64_t *b, akbasic_Value **out)
{
PREPARE_ERROR(errctx);
akbasic_Value *left = NULL;
akbasic_Value *right = NULL;
PASS(errctx, nth_arg(obj, expr, fname, 0, &left));
PASS(errctx, nth_arg(obj, expr, fname, 1, &right));
FAIL_ZERO_RETURN(errctx,
(left->valuetype == AKBASIC_TYPE_INTEGER &&
right->valuetype == AKBASIC_TYPE_INTEGER),
AKBASIC_ERR_TYPE, "%s expected two integers", fname);
*a = left->intval;
*b = right->intval;
PASS(errctx, akbasic_environment_new_value(obj->environment, out));
PASS(errctx, akbasic_value_zero(*out));
(*out)->valuetype = AKBASIC_TYPE_INTEGER;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_shl(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *out = NULL;
int64_t value = 0;
int64_t bits = 0;
(void)lval; (void)rval;
PASS(errctx, two_integers(obj, expr, "SHL", &value, &bits, &out));
FAIL_ZERO_RETURN(errctx, (bits >= 0 && bits < 64), AKBASIC_ERR_VALUE,
"SHL shift count %" PRId64 " is outside 0..63", bits);
out->intval = (int64_t)((uint64_t)value << bits);
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_shr(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *out = NULL;
int64_t value = 0;
int64_t bits = 0;
(void)lval; (void)rval;
PASS(errctx, two_integers(obj, expr, "SHR", &value, &bits, &out));
FAIL_ZERO_RETURN(errctx, (bits >= 0 && bits < 64), AKBASIC_ERR_VALUE,
"SHR shift count %" PRId64 " is outside 0..63", bits);
out->intval = value >> bits;
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_xor(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_Value *out = NULL;
int64_t left = 0;
int64_t right = 0;
(void)lval; (void)rval;
PASS(errctx, two_integers(obj, expr, "XOR", &left, &right, &out));
out->intval = left ^ right;
*dest = out;
SUCCEED_RETURN(errctx);
}
/*
* PEEK, POINTER and POINTERVAR reach real memory. In Go these went through
* unsafe.Pointer; in C they are ordinary casts and are simpler, not harder.
*
* They assume a little-endian host: tests/language/functions/pointer.bas sets
* A# = 255 and expects PEEK(POINTER(A#)) to be 255, which only holds if the low
* byte of the int64_t comes first.
*/
akerr_ErrorContext *akbasic_fn_peek(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *argleaf = NULL;
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
const uint8_t *source = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "PEEK", &argleaf, NULL, NULL));
FAIL_ZERO_RETURN(errctx,
(argleaf->leaftype == AKBASIC_LEAF_LITERAL_INT ||
argleaf->leaftype == AKBASIC_LEAF_IDENTIFIER_INT),
AKBASIC_ERR_TYPE, "PEEK expected INTEGER or INTEGER VARIABLE");
PASS(errctx, akbasic_runtime_evaluate(obj, argleaf, &arg));
FAIL_ZERO_RETURN(errctx, (arg->valuetype == AKBASIC_TYPE_INTEGER && arg->intval != 0),
AKBASIC_ERR_VALUE, "PEEK got NIL pointer or uninitialized variable");
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
source = (const uint8_t *)(uintptr_t)arg->intval;
out->valuetype = AKBASIC_TYPE_INTEGER;
out->intval = (int64_t)(*source);
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_pointer(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *argleaf = NULL;
akbasic_Value *arg = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "POINTER", &argleaf, NULL, NULL));
FAIL_ZERO_RETURN(errctx, akbasic_leaf_is_identifier(argleaf), AKBASIC_ERR_TYPE,
"POINTER expected IDENTIFIER");
/* The address of the live storage, not of a copy of it. */
obj->eval_clone_identifiers = false;
ATTEMPT {
CATCH(errctx, akbasic_runtime_evaluate(obj, argleaf, &arg));
} CLEANUP {
obj->eval_clone_identifiers = true;
} PROCESS(errctx) {
} FINISH(errctx, true);
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_INTEGER;
switch ( arg->valuetype ) {
case AKBASIC_TYPE_INTEGER:
out->intval = (int64_t)(uintptr_t)&arg->intval;
break;
case AKBASIC_TYPE_FLOAT:
out->intval = (int64_t)(uintptr_t)&arg->floatval;
break;
case AKBASIC_TYPE_STRING:
out->intval = (int64_t)(uintptr_t)&arg->stringval;
break;
default:
FAIL_RETURN(errctx, AKBASIC_ERR_TYPE, "POINTER expects a INT, FLOAT or STRING variable");
}
*dest = out;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *akbasic_fn_pointervar(akbasic_Runtime *obj, akbasic_ASTLeaf *expr, akbasic_Value *lval, akbasic_Value *rval, akbasic_Value **dest)
{
PREPARE_ERROR(errctx);
akbasic_ASTLeaf *argleaf = NULL;
akbasic_Variable *variable = NULL;
akbasic_Value *out = NULL;
(void)lval; (void)rval;
PASS(errctx, first_arg(obj, expr, "POINTERVAR", &argleaf, NULL, NULL));
FAIL_ZERO_RETURN(errctx, akbasic_leaf_is_identifier(argleaf), AKBASIC_ERR_TYPE,
"POINTERVAR expected IDENTIFIER");
PASS(errctx, akbasic_environment_get(obj->environment, argleaf->identifier, &variable));
FAIL_ZERO_RETURN(errctx, (variable != NULL), AKBASIC_ERR_UNDEFINED,
"Identifier %s is undefined", argleaf->identifier);
PASS(errctx, akbasic_environment_new_value(obj->environment, &out));
PASS(errctx, akbasic_value_zero(out));
out->valuetype = AKBASIC_TYPE_INTEGER;
out->intval = (int64_t)(uintptr_t)variable;
*dest = out;
SUCCEED_RETURN(errctx);
}
Reference in New Issue Copy Permalink