diff --git a/.gitignore b/.gitignore index f6b8d6a..c005981 100644 --- a/.gitignore +++ b/.gitignore @@ -6,7 +6,7 @@ out/ src/*.o tmp/ .aiignore -dev.txt +*.txt fun fun_test test_opcodes diff --git a/examples/builtins_maps_and_more.fun b/examples/builtins_maps_and_more.fun new file mode 100755 index 0000000..3cc0f57 --- /dev/null +++ b/examples/builtins_maps_and_more.fun @@ -0,0 +1,60 @@ +#!/usr/bin/env fun + +// Maps, map/filter/reduce, labeled break/continue (depth), and file I/O + +// Map literal and indexing +m = { "a": 1, "b": 2 } +print(m) // -> {"a": 1, "b": 2} (order may vary) +print(m["a"]) // -> 1 +m["c"] = 5 +print(has(m, "c")) // -> 1 +print(keys(m)) // -> ["a", "b", "c"] (order may vary) +print(values(m)) // -> [1, 2, 5] (order may vary) + +// map/filter/reduce +nums = [1, 2, 3, 4] + +fun double(x) + return x + x + +fun is_odd(x) + return x % 2 == 1 + +fun sum2(a, b) + return a + b + +print(map(nums, double)) // -> [2, 4, 6, 8] +print(filter(nums, is_odd)) // -> [1, 3] +print(reduce(nums, 0, sum2)) // -> 10 + +// labeled break/continue by depth using nested loops +for i in range(0, 3) + for j in range(0, 5) + if j == 1 + continue 1 // skip to next j + if i == 1 && j == 2 + break 2 // break outer and inner + print(i * 10 + j) + +// file I/O (writes then reads) +path = "tmp_io_example.txt" +ok = write_file(path, "hello-io") +print(ok) // -> 1 +content = read_file(path) +print(content) // -> hello-io + +/* Expected output: +{"a": 1, "b": 2} +1 +1 +["a", "b", "c"] +[1, 2, 5] +[2, 4, 6, 8] +[1, 3] +10 +0 +2 +3 +1 +hello-io +*/ diff --git a/src/bytecode.c b/src/bytecode.c index c480569..31a601b 100644 --- a/src/bytecode.c +++ b/src/bytecode.c @@ -100,6 +100,12 @@ static const char *opcode_name(OpCode op) { case OP_POW: return "POW"; case OP_RANDOM_SEED: return "RANDOM_SEED"; case OP_RANDOM_INT: return "RANDOM_INT"; + case OP_MAKE_MAP: return "MAKE_MAP"; + case OP_KEYS: return "KEYS"; + case OP_VALUES: return "VALUES"; + case OP_HAS_KEY: return "HAS_KEY"; + case OP_READ_FILE: return "READ_FILE"; + case OP_WRITE_FILE: return "WRITE_FILE"; default: return "???"; } } diff --git a/src/bytecode.h b/src/bytecode.h index 595d8b3..9b7ebc9 100644 --- a/src/bytecode.h +++ b/src/bytecode.h @@ -85,7 +85,17 @@ typedef enum { OP_ABS, // pops x; pushes |x| OP_POW, // pops b, a; pushes a^b OP_RANDOM_SEED, // pops seed; sets RNG seed; pushes nothing (we'll push 0) - OP_RANDOM_INT // pops hi, lo; pushes random int in [lo, hi) + OP_RANDOM_INT, // pops hi, lo; pushes random int in [lo, hi) + + // maps + OP_MAKE_MAP, // operand = pair count; pops 2*n (key,value)..., pushes map + OP_KEYS, // pops map; pushes array of keys + OP_VALUES, // pops map; pushes array of values + OP_HAS_KEY, // pops key, map; pushes 1/0 + + // file I/O + OP_READ_FILE, // pops path string; pushes content string (or "") + OP_WRITE_FILE // pops data string, path string; pushes 1/0 } OpCode; typedef struct { diff --git a/src/parser.c b/src/parser.c index c151021..2038785 100644 --- a/src/parser.c +++ b/src/parser.c @@ -401,6 +401,37 @@ static int emit_primary(Bytecode *bc, const char *src, size_t len, size_t *pos) return 1; } + /* map literal: { "key": expr, ... } */ + skip_spaces(src, len, pos); + if (*pos < len && src[*pos] == '{') { + (*pos)++; /* '{' */ + int pairs = 0; + skip_spaces(src, len, pos); + if (*pos < len && src[*pos] != '}') { + for (;;) { + /* key must be a string literal */ + char *k = parse_string_literal_any_quote(src, len, pos); + if (!k) { parser_fail(*pos, "Expected string key in map literal"); return 0; } + int kci = bytecode_add_constant(bc, make_string(k)); + free(k); + bytecode_add_instruction(bc, OP_LOAD_CONST, kci); + skip_spaces(src, len, pos); + if (!consume_char(src, len, pos, ':')) { parser_fail(*pos, "Expected ':' after map key"); return 0; } + if (!emit_expression(bc, src, len, pos)) { parser_fail(*pos, "Expected value expression in map literal"); return 0; } + pairs++; + skip_spaces(src, len, pos); + if (*pos < len && src[*pos] == ',') { (*pos)++; skip_spaces(src, len, pos); continue; } + break; + } + } + if (!consume_char(src, len, pos, '}')) { + parser_fail(*pos, "Expected '}' to close map literal"); + return 0; + } + bytecode_add_instruction(bc, OP_MAKE_MAP, pairs); + return 1; + } + /* number */ int ok = 0; size_t save = *pos; @@ -535,6 +566,43 @@ static int emit_primary(Bytecode *bc, const char *src, size_t len, size_t *pos) free(name); return 1; } + if (strcmp(name, "keys") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "keys expects 1 arg"); free(name); return 0; } + bytecode_add_instruction(bc, OP_KEYS, 0); + free(name); + return 1; + } + if (strcmp(name, "values") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "values expects 1 arg"); free(name); return 0; } + bytecode_add_instruction(bc, OP_VALUES, 0); + free(name); + return 1; + } + if (strcmp(name, "has") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "has expects (map, key)"); free(name); return 0; } + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "has expects (map, key)"); free(name); return 0; } + bytecode_add_instruction(bc, OP_HAS_KEY, 0); + free(name); + return 1; + } + if (strcmp(name, "read_file") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "read_file expects 1 arg"); free(name); return 0; } + bytecode_add_instruction(bc, OP_READ_FILE, 0); + free(name); + return 1; + } + if (strcmp(name, "write_file") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "write_file expects 2 args"); free(name); return 0; } + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "write_file expects 2 args"); free(name); return 0; } + bytecode_add_instruction(bc, OP_WRITE_FILE, 0); + free(name); + return 1; + } /* string ops */ if (strcmp(name, "split") == 0) { (*pos)++; /* '(' */ @@ -616,6 +684,238 @@ static int emit_primary(Bytecode *bc, const char *src, size_t len, size_t *pos) free(name); return 1; } + if (strcmp(name, "map") == 0) { + (*pos)++; /* '(' */ + /* arr */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "map expects (array, function)"); free(name); return 0; } + /* store arr -> __map_arr */ + char tarr[64]; snprintf(tarr, sizeof(tarr), "__map_arr_%d", g_temp_counter++); + int larr = -1, garr = -1; + if (g_locals) { larr = local_add(tarr); bytecode_add_instruction(bc, OP_STORE_LOCAL, larr); } + else { garr = sym_index(tarr); bytecode_add_instruction(bc, OP_STORE_GLOBAL, garr); } + /* func */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "map expects (array, function)"); free(name); return 0; } + char tfn[64]; snprintf(tfn, sizeof(tfn), "__map_fn_%d", g_temp_counter++); + int lfn = -1, gfn = -1; + if (g_locals) { lfn = local_add(tfn); bytecode_add_instruction(bc, OP_STORE_LOCAL, lfn); } + else { gfn = sym_index(tfn); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gfn); } + /* res array */ + bytecode_add_instruction(bc, OP_MAKE_ARRAY, 0); + char tres[64]; snprintf(tres, sizeof(tres), "__map_res_%d", g_temp_counter++); + int lres = -1, gres = -1; + if (g_locals) { lres = local_add(tres); bytecode_add_instruction(bc, OP_STORE_LOCAL, lres); } + else { gres = sym_index(tres); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gres); } + /* i=0 */ + int c0 = bytecode_add_constant(bc, make_int(0)); + bytecode_add_instruction(bc, OP_LOAD_CONST, c0); + char ti[64]; snprintf(ti, sizeof(ti), "__map_i_%d", g_temp_counter++); + int li = -1, gi = -1; + if (g_locals) { li = local_add(ti); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { gi = sym_index(ti); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + /* loop start */ + int loop_start = bc->instr_count; + /* i < len(arr) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); } + bytecode_add_instruction(bc, OP_LEN, 0); + bytecode_add_instruction(bc, OP_LT, 0); + int jf = bytecode_add_instruction(bc, OP_JUMP_IF_FALSE, 0); + /* elem = arr[i] */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_INDEX_GET, 0); + /* call fn(elem) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lfn); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gfn); } + /* reorder: we need fn below arg -> push fn first then arg already on stack? We currently have elem on stack; push fn now results top=fn. We want top args then function; OP_CALL expects fn then pops args? Our OP_CALL pops function after args; earlier compile of calls: they push function then args then OP_CALL. So we need to swap: push fn, then swap to make function below arg */ + bytecode_add_instruction(bc, OP_SWAP, 0); + bytecode_add_instruction(bc, OP_CALL, 1); + /* Append to result via indexed assignment: res[len(res)] = value */ + /* Store computed value to a temp */ + char tv[64]; snprintf(tv, sizeof(tv), "__map_v_%d", g_temp_counter++); + int lv = -1, gv = -1; + if (g_locals) { lv = local_add(tv); bytecode_add_instruction(bc, OP_STORE_LOCAL, lv); } + else { gv = sym_index(tv); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gv); } + + /* Push array (for INDEX_SET we need stack: value, index, array; we will build array, index, then value) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + + /* Compute index = len(res) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + bytecode_add_instruction(bc, OP_LEN, 0); + + /* Load value back on top */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lv); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gv); } + + /* Append via insert: res.insert(index=len(res), value) */ + bytecode_add_instruction(bc, OP_ARR_INSERT, 0); + /* discard returned new length */ + bytecode_add_instruction(bc, OP_POP, 0); + + /* i++ */ + int c1 = bytecode_add_constant(bc, make_int(1)); + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_JUMP, loop_start); + bytecode_set_operand(bc, jf, bc->instr_count); + /* result value on stack */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + free(name); + return 1; + } + if (strcmp(name, "filter") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "filter expects (array, function)"); free(name); return 0; } + char tarr[64]; snprintf(tarr, sizeof(tarr), "__flt_arr_%d", g_temp_counter++); + int larr = -1, garr = -1; + if (g_locals) { larr = local_add(tarr); bytecode_add_instruction(bc, OP_STORE_LOCAL, larr); } + else { garr = sym_index(tarr); bytecode_add_instruction(bc, OP_STORE_GLOBAL, garr); } + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "filter expects (array, function)"); free(name); return 0; } + char tfn[64]; snprintf(tfn, sizeof(tfn), "__flt_fn_%d", g_temp_counter++); + int lfn = -1, gfn = -1; + if (g_locals) { lfn = local_add(tfn); bytecode_add_instruction(bc, OP_STORE_LOCAL, lfn); } + else { gfn = sym_index(tfn); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gfn); } + bytecode_add_instruction(bc, OP_MAKE_ARRAY, 0); + char tres[64]; snprintf(tres, sizeof(tres), "__flt_res_%d", g_temp_counter++); + int lres = -1, gres = -1; + if (g_locals) { lres = local_add(tres); bytecode_add_instruction(bc, OP_STORE_LOCAL, lres); } + else { gres = sym_index(tres); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gres); } + int c0 = bytecode_add_constant(bc, make_int(0)); + bytecode_add_instruction(bc, OP_LOAD_CONST, c0); + char ti[64]; snprintf(ti, sizeof(ti), "__flt_i_%d", g_temp_counter++); + int li = -1, gi = -1; + if (g_locals) { li = local_add(ti); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { gi = sym_index(ti); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + int loop_start = bc->instr_count; + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); } + bytecode_add_instruction(bc, OP_LEN, 0); + bytecode_add_instruction(bc, OP_LT, 0); + int jf = bytecode_add_instruction(bc, OP_JUMP_IF_FALSE, 0); + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_INDEX_GET, 0); + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lfn); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gfn); } + bytecode_add_instruction(bc, OP_SWAP, 0); + bytecode_add_instruction(bc, OP_CALL, 1); + /* if truthy then push elem to res */ + int jskip = bytecode_add_instruction(bc, OP_JUMP_IF_FALSE, 0); + /* Append element to result: res[len(res)] = elem */ + /* Reload element into a temp */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_INDEX_GET, 0); + + char tvf[64]; snprintf(tvf, sizeof(tvf), "__flt_v_%d", g_temp_counter++); + int lvf = -1, gvf = -1; + if (g_locals) { lvf = local_add(tvf); bytecode_add_instruction(bc, OP_STORE_LOCAL, lvf); } + else { gvf = sym_index(tvf); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gvf); } + + /* Push array */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + + /* index = len(res) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + bytecode_add_instruction(bc, OP_LEN, 0); + + /* value */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lvf); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gvf); } + + /* Append via insert: res.insert(index=len(res), value) */ + bytecode_add_instruction(bc, OP_ARR_INSERT, 0); + /* discard returned new length */ + bytecode_add_instruction(bc, OP_POP, 0); + + int c1 = bytecode_add_constant(bc, make_int(1)); + /* patch skip over append */ + bytecode_set_operand(bc, jskip, bc->instr_count); + /* i++ */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_JUMP, loop_start); + bytecode_set_operand(bc, jf, bc->instr_count); + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lres); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gres); } + free(name); + return 1; + } + if (strcmp(name, "reduce") == 0) { + (*pos)++; /* '(' */ + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "reduce expects (array, init, function)"); free(name); return 0; } + char tarr[64]; snprintf(tarr, sizeof(tarr), "__red_arr_%d", g_temp_counter++); + int larr = -1, garr = -1; + if (g_locals) { larr = local_add(tarr); bytecode_add_instruction(bc, OP_STORE_LOCAL, larr); } + else { garr = sym_index(tarr); bytecode_add_instruction(bc, OP_STORE_GLOBAL, garr); } + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ',')) { parser_fail(*pos, "reduce expects (array, init, function)"); free(name); return 0; } + char tacc[64]; snprintf(tacc, sizeof(tacc), "__red_acc_%d", g_temp_counter++); + int lacc = -1, gacc = -1; + if (g_locals) { lacc = local_add(tacc); bytecode_add_instruction(bc, OP_STORE_LOCAL, lacc); } + else { gacc = sym_index(tacc); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gacc); } + if (!emit_expression(bc, src, len, pos) || !consume_char(src, len, pos, ')')) { parser_fail(*pos, "reduce expects (array, init, function)"); free(name); return 0; } + char tfn[64]; snprintf(tfn, sizeof(tfn), "__red_fn_%d", g_temp_counter++); + int lfn = -1, gfn = -1; + if (g_locals) { lfn = local_add(tfn); bytecode_add_instruction(bc, OP_STORE_LOCAL, lfn); } + else { gfn = sym_index(tfn); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gfn); } + /* loop */ + int c0 = bytecode_add_constant(bc, make_int(0)); + bytecode_add_instruction(bc, OP_LOAD_CONST, c0); + char ti[64]; snprintf(ti, sizeof(ti), "__red_i_%d", g_temp_counter++); + int li = -1, gi = -1; + if (g_locals) { li = local_add(ti); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { gi = sym_index(ti); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + int loop_start = bc->instr_count; + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); } + bytecode_add_instruction(bc, OP_LEN, 0); + bytecode_add_instruction(bc, OP_LT, 0); + int jf = bytecode_add_instruction(bc, OP_JUMP_IF_FALSE, 0); + /* elem */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, larr); bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, garr); bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_INDEX_GET, 0); + + /* Store elem to a temp so we can build stack as: fn, acc, elem */ + char telem[64]; snprintf(telem, sizeof(telem), "__red_elem_%d", g_temp_counter++); + int lelem = -1, gelem = -1; + if (g_locals) { lelem = local_add(telem); bytecode_add_instruction(bc, OP_STORE_LOCAL, lelem); } + else { gelem = sym_index(telem); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gelem); } + + /* push function */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lfn); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gfn); } + + /* push accumulator (arg1) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lacc); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gacc); } + + /* push element (arg2) */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lelem); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gelem); } + + /* call fn(acc, elem) -> result */ + bytecode_add_instruction(bc, OP_CALL, 2); + /* store to acc */ + if (g_locals) { bytecode_add_instruction(bc, OP_STORE_LOCAL, lacc); } + else { bytecode_add_instruction(bc, OP_STORE_GLOBAL, gacc); } + int c1 = bytecode_add_constant(bc, make_int(1)); + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, li); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_LOCAL, li); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gi); bytecode_add_instruction(bc, OP_LOAD_CONST, c1); bytecode_add_instruction(bc, OP_ADD, 0); bytecode_add_instruction(bc, OP_STORE_GLOBAL, gi); } + bytecode_add_instruction(bc, OP_JUMP, loop_start); + bytecode_set_operand(bc, jf, bc->instr_count); + /* result = acc on stack */ + if (g_locals) { bytecode_add_instruction(bc, OP_LOAD_LOCAL, lacc); } + else { bytecode_add_instruction(bc, OP_LOAD_GLOBAL, gacc); } + free(name); + return 1; + } if (strcmp(name, "zip") == 0) { (*pos)++; /* '(' */ if (!emit_expression(bc, src, len, pos)) { parser_fail(*pos, "zip expects first array"); free(name); return 0; } diff --git a/src/value.c b/src/value.c index e88dd0b..19fff47 100644 --- a/src/value.c +++ b/src/value.c @@ -9,6 +9,14 @@ typedef struct Array { Value *items; /* owns items; each item owned by array */ } Array; +typedef struct Map { + int refcount; + int count; + int cap; + char **keys; /* each key owned here */ + Value *vals; /* each value owned here */ +} Map; + Value make_int(int64_t v) { Value val; val.type = VAL_INT; @@ -37,6 +45,100 @@ Value make_nil(void) { return v; } +Value make_map_empty(void) { + Map *m = (Map*)malloc(sizeof(Map)); + if (!m) return make_nil(); + m->refcount = 1; + m->count = 0; + m->cap = 0; + m->keys = NULL; + m->vals = NULL; + Value v; + v.type = VAL_MAP; + v.map = (struct Map*)m; + return v; +} + +static int map_ensure_cap(Map *m, int need) { + if (m->cap >= need) return 1; + int ncap = m->cap == 0 ? 4 : m->cap * 2; + while (ncap < need) ncap *= 2; + char **nkeys = (char**)realloc(m->keys, sizeof(char*) * ncap); + Value *nvals = (Value*)realloc(m->vals, sizeof(Value) * ncap); + if (!nkeys || !nvals) return 0; + m->keys = nkeys; + m->vals = nvals; + m->cap = ncap; + return 1; +} + +int map_set(Value *vm, const char *key, Value v) { + if (!vm || vm->type != VAL_MAP || !vm->map || !key) { free_value(v); return 0; } + Map *m = (Map*)vm->map; + /* replace if exists */ + for (int i = 0; i < m->count; ++i) { + if (strcmp(m->keys[i], key) == 0) { + free_value(m->vals[i]); + m->vals[i] = v; + return 1; + } + } + if (!map_ensure_cap(m, m->count + 1)) { free_value(v); return 0; } + m->keys[m->count] = strdup(key); + m->vals[m->count] = v; + m->count++; + return 1; +} + +int map_get_copy(const Value *vm, const char *key, Value *out) { + if (!vm || vm->type != VAL_MAP || !vm->map || !key) return 0; + Map *m = (Map*)vm->map; + for (int i = 0; i < m->count; ++i) { + if (strcmp(m->keys[i], key) == 0) { + if (out) *out = copy_value(&m->vals[i]); + return 1; + } + } + return 0; +} + +int map_has(const Value *vm, const char *key) { + if (!vm || vm->type != VAL_MAP || !vm->map || !key) return 0; + Map *m = (Map*)vm->map; + for (int i = 0; i < m->count; ++i) { + if (strcmp(m->keys[i], key) == 0) return 1; + } + return 0; +} + +Value map_keys_array(const Value *vm) { + if (!vm || vm->type != VAL_MAP || !vm->map) return make_array_from_values(NULL, 0); + Map *m = (Map*)vm->map; + if (m->count <= 0) return make_array_from_values(NULL, 0); + Value *tmp = (Value*)malloc(sizeof(Value) * m->count); + for (int i = 0; i < m->count; ++i) { + tmp[i] = make_string(m->keys[i]); + } + Value arr = make_array_from_values(tmp, m->count); + for (int i = 0; i < m->count; ++i) free_value(tmp[i]); + free(tmp); + return arr; +} + +Value map_values_array(const Value *vm) { + if (!vm || vm->type != VAL_MAP || !vm->map) return make_array_from_values(NULL, 0); + Map *m = (Map*)vm->map; + if (m->count <= 0) return make_array_from_values(NULL, 0); + Value *tmp = (Value*)malloc(sizeof(Value) * m->count); + for (int i = 0; i < m->count; ++i) { + tmp[i] = copy_value(&m->vals[i]); + } + Value arr = make_array_from_values(tmp, m->count); + for (int i = 0; i < m->count; ++i) free_value(tmp[i]); + free(tmp); + return arr; +} + Value make_array_from_values(const Value *vals, int count) { if (count < 0) count = 0; Array *arr = (Array*)malloc(sizeof(Array)); @@ -212,6 +314,12 @@ Value copy_value(const Value *v) { if (a) a->refcount++; break; } + case VAL_MAP: { + Map *m = (Map*)v->map; + out.map = (struct Map*)m; + if (m) m->refcount++; + break; + } case VAL_NIL: default: break; @@ -240,13 +348,22 @@ Value deep_copy_value(const Value *v) { tmp[i] = deep_copy_value(&a->items[i]); } Value out = make_array_from_values(tmp, a->count); - /* free temporaries (we created new deep copies already moved into out) */ for (int i = 0; i < a->count; ++i) { free_value(tmp[i]); } free(tmp); return out; } + case VAL_MAP: { + const Map *m = (const Map*)v->map; + if (!m || m->count <= 0) return make_map_empty(); + Value out = make_map_empty(); + for (int i = 0; i < m->count; ++i) { + Value dv = deep_copy_value(&m->vals[i]); + map_set(&out, m->keys[i], dv); + } + return out; + } case VAL_NIL: default: return make_nil(); @@ -265,6 +382,17 @@ void free_value(Value v) { free(a->items); free(a); } + } else if (v.type == VAL_MAP && v.map) { + Map *m = (Map*)v.map; + if (--m->refcount == 0) { + for (int i = 0; i < m->count; ++i) { + if (m->keys[i]) free(m->keys[i]); + free_value(m->vals[i]); + } + free(m->keys); + free(m->vals); + free(m); + } } /* VAL_FUNCTION: we *do not* free the Bytecode here (caller frees it) */ } @@ -292,6 +420,19 @@ void print_value(const Value *v) { printf("]"); break; } + case VAL_MAP: { + const Map *m = (const Map*)v->map; + printf("{"); + if (m) { + for (int i = 0; i < m->count; ++i) { + if (i > 0) printf(", "); + printf("\"%s\": ", m->keys[i] ? m->keys[i] : ""); + print_value(&m->vals[i]); + } + } + printf("}"); + break; + } case VAL_NIL: default: printf("nil"); diff --git a/src/value.h b/src/value.h index 46aaef2..34a2713 100644 --- a/src/value.h +++ b/src/value.h @@ -5,12 +5,14 @@ struct Bytecode; /* forward */ struct Array; /* forward */ +struct Map; /* forward */ typedef enum { VAL_INT, VAL_STRING, VAL_FUNCTION, VAL_ARRAY, + VAL_MAP, VAL_NIL } ValueType; @@ -21,6 +23,7 @@ typedef struct { char *s; struct Bytecode *fn; struct Array *arr; + struct Map *map; }; } Value; @@ -42,9 +45,17 @@ int array_remove(Value *v, int index, Value *out); /* returns 1 on suc Value array_slice(const Value *v, int start, int end); /* negative end means till end */ Value array_concat(const Value *a, const Value *b); /* returns new array */ +/* maps (string keys) */ +Value make_map_empty(void); /* new empty map */ +int map_set(Value *m, const char *key, Value v); /* 1 on ok (takes ownership of v) */ +int map_get_copy(const Value *m, const char *key, Value *out);/* 1 on found, out=copy */ +int map_has(const Value *m, const char *key); /* 1/0 */ +Value map_keys_array(const Value *m); /* array of strings */ +Value map_values_array(const Value *m); /* array of values (copies) */ + /* copy/free */ -Value copy_value(const Value *v); /* deep for strings, RC for arrays, shallow fn */ -Value deep_copy_value(const Value *v); /* deep copy including arrays */ +Value copy_value(const Value *v); /* deep for strings, RC for arrays/maps, shallow fn */ +Value deep_copy_value(const Value *v); /* deep copy including arrays/maps */ void free_value(Value v); /* frees owned resources */ /* utilities */ diff --git a/src/vm.c b/src/vm.c index 63dac9e..c08c68d 100644 --- a/src/vm.c +++ b/src/vm.c @@ -517,45 +517,56 @@ void vm_run(VM *vm, Bytecode *entry) { case OP_INDEX_GET: { Value idx = pop_value(vm); - Value arr = pop_value(vm); - if (arr.type != VAL_ARRAY) { - fprintf(stderr, "Runtime type error: INDEX_GET expects array\n"); + Value container = pop_value(vm); + if (container.type == VAL_ARRAY) { + if (idx.type != VAL_INT) { fprintf(stderr, "INDEX_GET index must be int for array\n"); exit(1); } + Value elem; + if (!array_get_copy(&container, (int)idx.i, &elem)) { + fprintf(stderr, "Runtime error: index out of range\n"); exit(1); + } + free_value(container); + free_value(idx); + push_value(vm, elem); + } else if (container.type == VAL_MAP) { + if (idx.type != VAL_STRING) { fprintf(stderr, "INDEX_GET key must be string for map\n"); exit(1); } + Value out; + if (!map_get_copy(&container, idx.s ? idx.s : "", &out)) { + /* missing -> nil */ + out = make_nil(); + } + free_value(container); + free_value(idx); + push_value(vm, out); + } else { + fprintf(stderr, "Runtime type error: INDEX_GET expects array or map\n"); exit(1); } - if (idx.type != VAL_INT) { - fprintf(stderr, "Runtime type error: INDEX_GET index must be int\n"); - exit(1); - } - Value elem; - if (!array_get_copy(&arr, (int)idx.i, &elem)) { - fprintf(stderr, "Runtime error: index out of range\n"); - exit(1); - } - free_value(arr); - free_value(idx); - push_value(vm, elem); break; } case OP_INDEX_SET: { Value v = pop_value(vm); Value idx = pop_value(vm); - Value arr = pop_value(vm); - if (arr.type != VAL_ARRAY) { - fprintf(stderr, "Runtime type error: INDEX_SET expects array\n"); + Value container = pop_value(vm); + if (container.type == VAL_ARRAY) { + if (idx.type != VAL_INT) { fprintf(stderr, "INDEX_SET index must be int for array\n"); exit(1); } + if (!array_set(&container, (int)idx.i, v)) { + fprintf(stderr, "Runtime error: index out of range\n"); exit(1); + } + /* v moved into array */ + free_value(container); + free_value(idx); + } else if (container.type == VAL_MAP) { + if (idx.type != VAL_STRING) { fprintf(stderr, "INDEX_SET key must be string for map\n"); exit(1); } + if (!map_set(&container, idx.s ? idx.s : "", v)) { + fprintf(stderr, "Runtime error: map set failed\n"); exit(1); + } + free_value(container); + free_value(idx); + } else { + fprintf(stderr, "Runtime type error: INDEX_SET expects array or map\n"); exit(1); } - if (idx.type != VAL_INT) { - fprintf(stderr, "Runtime type error: INDEX_SET index must be int\n"); - exit(1); - } - if (!array_set(&arr, (int)idx.i, v)) { - fprintf(stderr, "Runtime error: index out of range\n"); - exit(1); - } - /* arr modified in place; do not free v (ownership moved) */ - free_value(arr); - free_value(idx); break; } @@ -970,6 +981,92 @@ void vm_run(VM *vm, Bytecode *entry) { break; } + case OP_MAKE_MAP: { + int pairs = inst.operand; + if (pairs < 0) { fprintf(stderr, "MAKE_MAP invalid pair count\n"); exit(1); } + Value m = make_map_empty(); + for (int i = 0; i < pairs; ++i) { + Value val = pop_value(vm); + Value key = pop_value(vm); + if (key.type != VAL_STRING) { fprintf(stderr, "Map literal keys must be strings\n"); exit(1); } + if (!map_set(&m, key.s ? key.s : "", val)) { + fprintf(stderr, "Map literal set failed\n"); exit(1); + } + free_value(key); + /* val ownership moved */ + } + push_value(vm, m); + break; + } + + case OP_KEYS: { + Value m = pop_value(vm); + if (m.type != VAL_MAP) { fprintf(stderr, "KEYS expects map\n"); exit(1); } + Value arr = map_keys_array(&m); + free_value(m); + push_value(vm, arr); + break; + } + + case OP_VALUES: { + Value m = pop_value(vm); + if (m.type != VAL_MAP) { fprintf(stderr, "VALUES expects map\n"); exit(1); } + Value arr = map_values_array(&m); + free_value(m); + push_value(vm, arr); + break; + } + + case OP_HAS_KEY: { + Value key = pop_value(vm); + Value m = pop_value(vm); + if (m.type != VAL_MAP || key.type != VAL_STRING) { fprintf(stderr, "HAS_KEY expects (map, string)\n"); exit(1); } + int ok = map_has(&m, key.s ? key.s : ""); + free_value(m); free_value(key); + push_value(vm, make_int(ok ? 1 : 0)); + break; + } + + case OP_READ_FILE: { + Value path = pop_value(vm); + if (path.type != VAL_STRING) { fprintf(stderr, "READ_FILE expects string\n"); exit(1); } + const char *p = path.s ? path.s : ""; + FILE *f = fopen(p, "rb"); + if (!f) { free_value(path); push_value(vm, make_string("")); break; } + if (fseek(f, 0, SEEK_END) != 0) { fclose(f); free_value(path); push_value(vm, make_string("")); break; } + long sz = ftell(f); + if (sz < 0) { fclose(f); free_value(path); push_value(vm, make_string("")); break; } + rewind(f); + char *buf = (char*)malloc((size_t)sz + 1); + size_t n = buf ? fread(buf, 1, (size_t)sz, f) : 0; + fclose(f); + if (!buf) { free_value(path); push_value(vm, make_string("")); break; } + buf[n] = '\0'; + Value out = make_string(buf); + free(buf); + free_value(path); + push_value(vm, out); + break; + } + + case OP_WRITE_FILE: { + Value data = pop_value(vm); + Value path = pop_value(vm); + if (path.type != VAL_STRING || data.type != VAL_STRING) { fprintf(stderr, "WRITE_FILE expects (string, string)\n"); exit(1); } + const char *p = path.s ? path.s : ""; + FILE *f = fopen(p, "wb"); + int ok = 0; + if (f) { + size_t len = data.s ? strlen(data.s) : 0; + ok = (fwrite(data.s ? data.s : "", 1, len, f) == len); + fclose(f); + } + free_value(path); + free_value(data); + push_value(vm, make_int(ok ? 1 : 0)); + break; + } + case OP_LOAD_GLOBAL: { int idx = inst.operand; if (idx < 0 || idx >= VM_MAX_GLOBALS) { diff --git a/src/vm.h b/src/vm.h index 958731b..c187a9f 100644 --- a/src/vm.h +++ b/src/vm.h @@ -21,7 +21,9 @@ static const char *opcode_names[] = { "SPLIT","JOIN","SUBSTR","FIND", "CONTAINS","INDEX_OF","CLEAR", "ENUMERATE","ZIP", - "MIN","MAX","CLAMP","ABS","POW","RANDOM_SEED","RANDOM_INT" + "MIN","MAX","CLAMP","ABS","POW","RANDOM_SEED","RANDOM_INT", + "MAKE_MAP","KEYS","VALUES","HAS_KEY", + "READ_FILE","WRITE_FILE" }; typedef struct { @@ -62,7 +64,7 @@ void vm_dump_globals(VM *vm); void vm_run(VM *vm, Bytecode *entry); static inline int opcode_is_valid(int op) { - return op >= OP_NOP && op <= OP_RANDOM_INT; // all current opcodes + return op >= OP_NOP && op <= OP_WRITE_FILE; // all current opcodes } #endif