void ws_function(WsCompiler *compiler, WsBool externp, char *name,
WsUInt32 line, WsList *params, WsList *block)
{
WsFunctionHash *hash;
WsFunction *f = ws_realloc(compiler->functions,
((compiler->num_functions + 1)
* sizeof(WsFunction)));
if (f == NULL) {
ws_free(name);
ws_error_memory(compiler);
return;
}
if (externp)
compiler->num_extern_functions++;
else
compiler->num_local_functions++;
compiler->functions = f;
f = &compiler->functions[compiler->num_functions];
f->findex = compiler->num_functions++;
f->externp = externp;
f->name = name;
f->line = line;
f->params = params;
f->block = block;
/* Update the function name hash. */
hash = ws_function_hash(compiler, name);
if (hash == NULL) {
ws_error_memory(compiler);
return;
}
if (hash->defined) {
ws_src_error(compiler, line, "redefinition of `%s'", name);
ws_src_error(compiler,
compiler->functions[hash->findex].line,
"`%s' previously defined here", name);
return;
}
hash->defined = WS_TRUE;
hash->findex = f->findex;
}
void ws_pragma_use(WsCompilerPtr compiler, WsUInt32 line, char *identifier,
WsUtf8String *url)
{
WsPragmaUse *u = ws_calloc(1, sizeof(*u));
WsPragmaUse *uold;
/* Do we already know this pragma? */
uold = ws_hash_get(compiler->pragma_use_hash, identifier);
if (uold) {
ws_src_error(compiler, line, "redefinition of pragma `%s'", identifier);
ws_src_error(compiler, uold->line, "`%s' previously defined here",
identifier);
goto error_cleanup;
}
if (u == NULL)
goto error;
u->line = line;
/* Insert the URL to the byte-code module. */
if (!ws_bc_add_const_utf8_string(compiler->bc, &u->urlindex, url->data,
url->len))
goto error;
/* Add it to the use pragma hash. */
if (!ws_hash_put(compiler->pragma_use_hash, identifier, u))
goto error;
/* Cleanup. */
ws_lexer_free_block(compiler, identifier);
ws_lexer_free_utf8(compiler, url);
return;
/* Error handling. */
error:
ws_error_memory(compiler);
error_cleanup:
ws_free(u);
ws_lexer_free_block(compiler, identifier);
ws_lexer_free_utf8(compiler, url);
}
static WsUInt32 buffer_to_int(WsCompilerPtr compiler, WsBuffer *buffer)
{
unsigned char *p;
unsigned long value;
/* Terminate the string. */
if (!ws_buffer_append_space(buffer, &p, 1)) {
ws_error_memory(compiler);
return 0;
}
p[0] = '\0';
/* Convert the buffer into an integer number. The base is taken
from the bufer. */
errno = 0;
value = strtoul((char *) ws_buffer_ptr(buffer), NULL, 0);
/* Check for overflow. We accept WS_INT32_MAX + 1 because we might
* be parsing the numeric part of '-2147483648'. */
if (errno == ERANGE || value > (WsUInt32) WS_INT32_MAX + 1)
ws_src_error(compiler, 0, "integer literal too large");
/* All done. */
return (WsUInt32) value;
}
WsNamespace *ws_variable_define(WsCompilerPtr compiler, WsUInt32 line,
WsBool variablep, char *name)
{
WsNamespace *ns;
/* Is the symbol already defined? */
ns = ws_hash_get(compiler->variables_hash, name);
if (ns) {
ws_src_error(compiler, line, "redeclaration of `%s'", name);
ws_src_error(compiler, ns->line, "`%s' previously declared here", name);
return NULL;
}
/* Can we still define more variables? */
if (compiler->next_vindex > 255) {
/* No we can't. */
ws_src_error(compiler, line, "too many local variables");
return NULL;
}
ns = ws_calloc(1, sizeof(*ns));
if (ns == NULL) {
ws_error_memory(compiler);
return NULL;
}
ns->line = line;
ns->vindex = compiler->next_vindex++;
if (!ws_hash_put(compiler->variables_hash, name, ns)) {
ws_free(ns);
ws_error_memory(compiler);
return NULL;
}
return ns;
}
void ws_expr_linearize(WsCompiler *compiler, WsExpression *expr)
{
WsListItem *li;
WsAsmIns *ins;
switch (expr->type) {
case WS_EXPR_COMMA:
/* Linearize left. */
ws_expr_linearize(compiler, expr->u.comma.left);
/* Pop its result. */
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line, WS_ASM_POP));
/* Linearize right */
ws_expr_linearize(compiler, expr->u.comma.right);
break;
case WS_EXPR_ASSIGN:
{
WsNamespace *ns = ws_variable_lookup(compiler,
expr->u.assign.identifier);
if (ns == NULL) {
/* Unknown identifier. */
ws_src_error(compiler, expr->line, "unknown variable `%s'",
expr->u.symbol);
return;
}
if (expr->u.assign.op == '=') {
/* Evaluate the expression. */
ws_expr_linearize(compiler, expr->u.assign.expr);
/* Store the value to the variable. */
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line,
WS_ASM_P_STORE_VAR, ns->vindex));
} else if (expr->u.assign.op == tADDA) {
/* Linearize the expression. */
ws_expr_linearize(compiler, expr->u.assign.expr);
/* Add it to the variable. */
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line,
WS_ASM_ADD_ASG, ns->vindex));
} else if (expr->u.assign.op == tSUBA) {
/* Linearize the expression. */
ws_expr_linearize(compiler, expr->u.assign.expr);
/* Substract it from the variable. */
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line,
WS_ASM_SUB_ASG, ns->vindex));
} else {
/* Load the old value from the variable. */
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line,
WS_ASM_P_LOAD_VAR, ns->vindex));
/* Evaluate the expression. */
ws_expr_linearize(compiler, expr->u.assign.expr);
/* Perform the operand. */
ins = NULL;
switch (expr->u.assign.op) {
case tMULA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_MUL);
break;
case tDIVA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_DIV);
break;
case tREMA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_REM);
break;
case tADDA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_ADD);
break;
case tSUBA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_SUB);
break;
case tLSHIFTA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_LSHIFT);
break;
case tRSSHIFTA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_RSSHIFT);
break;
case tRSZSHIFTA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_RSZSHIFT);
break;
case tANDA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_AND);
break;
case tXORA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_XOR);
break;
case tORA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_B_OR);
break;
case tIDIVA:
ins = ws_asm_ins(compiler, expr->line, WS_ASM_IDIV);
break;
default:
ws_fatal("ws_expr_linearize(): unknown assignment operand %x",
expr->u.assign.op);
break;
}
ws_asm_link(compiler, ins);
/* Store the value to the variable. */
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line,
WS_ASM_P_STORE_VAR, ns->vindex));
}
/* The value of the assignment expression is the value
assigned. So, we must load the value from the variable.
This would also be a good place for the `dup' operand but
we lose since we don't have it. */
ws_asm_link(compiler, ws_asm_variable(compiler, expr->line,
WS_ASM_P_LOAD_VAR, ns->vindex));
}
break;
case WS_EXPR_CONDITIONAL:
{
WsAsmIns *l_else = ws_asm_label(compiler, expr->line);
WsAsmIns *l_end = ws_asm_label(compiler, expr->line);
/* Linearize condition. */
ws_expr_linearize(compiler, expr->u.conditional.e_cond);
/* If the result if false, jump to the else-branch. */
ws_asm_link(compiler, ws_asm_branch(compiler, expr->line,
WS_ASM_P_TJUMP, l_else));
/* Linearize the then-expression and jump out. */
ws_expr_linearize(compiler, expr->u.conditional.e_then);
ws_asm_link(compiler, ws_asm_branch(compiler, expr->line,
WS_ASM_P_JUMP, l_end));
/* The else-branch. */
ws_asm_link(compiler, l_else);
ws_expr_linearize(compiler, expr->u.conditional.e_else);
/* Insert the end label. */
ws_asm_link(compiler, l_end);
}
break;
case WS_EXPR_LOGICAL:
{
WsAsmIns *l_out = ws_asm_label(compiler, expr->line);
/* Linearize the left-hand size expression. */
ws_expr_linearize(compiler, expr->u.logical.left);
/* Short-circuit check. The type of the logical expression is
the short-circuit byte-code operand. */
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
expr->u.logical.type));
ws_asm_link(compiler, ws_asm_branch(compiler, expr->line,
WS_ASM_P_TJUMP, l_out));
/* Linearize the right-hand size expression. */
ws_expr_linearize(compiler, expr->u.logical.right);
/* The result of a logical expression should be boolean.
* Control statements do automatic conversion, but typeof()
* does not. */
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
WS_ASM_TOBOOL));
/* Insert the end label. */
ws_asm_link(compiler, l_out);
}
break;
case WS_EXPR_BINARY:
/* Linearize left and right. */
ws_expr_linearize(compiler, expr->u.binary.left);
ws_expr_linearize(compiler, expr->u.binary.right);
/* The type of the binary expression is the byte-code opcode. */
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
expr->u.binary.type));
break;
case WS_EXPR_UNARY:
/* Linearize the expression. */
ws_expr_linearize(compiler, expr->u.unary.expr);
/* The type of the unary expression is the byte-code opcode. */
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
expr->u.unary.type));
break;
case WS_EXPR_UNARY_VAR:
{
WsNamespace *ns = ws_variable_lookup(compiler,
expr->u.unary_var.variable);
if (ns == NULL) {
/* An unknown identifier. */
ws_src_error(compiler, expr->line, "unknown variable `%s'",
expr->u.unary_var.variable);
return;
}
/* First, do the operation. */
if (expr->u.unary_var.addp)
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line, WS_ASM_P_INCR_VAR,
ns->vindex));
else
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line, WS_ASM_DECR_VAR,
ns->vindex));
/* Second, load the new value of the variable. */
ws_asm_link(compiler, ws_asm_variable(compiler, expr->line,
WS_ASM_P_LOAD_VAR, ns->vindex));
}
break;
case WS_EXPR_POSTFIX_VAR:
{
WsNamespace *ns = ws_variable_lookup(compiler,
expr->u.postfix_var.variable);
if (ns == NULL) {
/* An unknown identifier. */
ws_src_error(compiler, expr->line, "unknown variable `%s'",
expr->u.postfix_var.variable);
return;
}
/* First, load the old value of the variable. */
ws_asm_link(compiler, ws_asm_variable(compiler, expr->line,
WS_ASM_P_LOAD_VAR, ns->vindex));
/* Second, do the operation. */
if (expr->u.unary_var.addp)
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line, WS_ASM_P_INCR_VAR,
ns->vindex));
else
ws_asm_link(compiler,
ws_asm_variable(compiler, expr->line, WS_ASM_DECR_VAR,
ns->vindex));
}
break;
case WS_EXPR_CALL:
/* First, evaluate the arguments. */
for (li = expr->u.call.arguments->head; li; li = li->next)
ws_expr_linearize(compiler, li->data);
/* Second, emit the call instruction. */
switch (expr->u.call.type) {
case ' ': /* LocalScriptFunctionCall */
{
WsFunctionHash *f = ws_function_hash(compiler, expr->u.call.name);
if (f == NULL || !f->defined)
{
ws_src_error(compiler, expr->line,
"unknown local function `%s'",
expr->u.call.name);
return;
}
/* Check that the function is called with correct amount
of arguments. */
if (expr->u.call.arguments->num_items
!= compiler->functions[f->findex].params->num_items)
{
ws_src_error(compiler, expr->line,
"invalid amount of arguments for `%s': "
"expected %u, got %u",
expr->u.call.name,
compiler->functions[f->findex].params->num_items,
expr->u.call.arguments->num_items);
return;
}
/* Emit assembler. */
ws_asm_link(compiler, ws_asm_call(compiler, expr->line,
f->findex));
}
break;
case '#': /* ExternalScriptFunctionCall */
{
WsPragmaUse *use = ws_hash_get(compiler->pragma_use_hash,
expr->u.call.base);
WsUInt16 findex;
if (use == NULL)
{
ws_src_error(compiler, expr->line,
"unknown external compilation unit `%s'",
expr->u.call.base);
return;
}
/* Insert the function name to the byte-code pool. */
if (!ws_bc_add_const_utf8_string(
compiler->bc, &findex,
(unsigned char *) expr->u.call.name,
strlen(expr->u.call.name)))
{
ws_error_memory(compiler);
return;
}
/* Emit assembler. */
ws_asm_link(compiler,
ws_asm_call_url(compiler, expr->line,
findex, use->urlindex,
expr->u.call.arguments->num_items));
}
break;
case '.': /* LibraryFunctionCall */
{
WsUInt16 lindex;
WsUInt8 findex;
WsUInt8 num_args;
WsBool lindex_found;
WsBool findex_found;
if (!ws_stdlib_function(expr->u.call.base, expr->u.call.name,
&lindex, &findex, &num_args,
&lindex_found, &findex_found))
{
if (!lindex_found)
ws_src_error(compiler, expr->line,
"unknown system library `%s'",
expr->u.call.base);
else
ws_src_error(compiler, expr->line,
"unknown library function `%s.%s'",
expr->u.call.base, expr->u.call.name);
return;
}
/* Check the argument count. */
if (expr->u.call.arguments->num_items != num_args)
{
ws_src_error(compiler, expr->line,
"invalid amount of arguments for `%s.%s': "
"expected %u, got %u",
expr->u.call.base, expr->u.call.name,
num_args, expr->u.call.arguments->num_items);
return;
}
/* Emit assembler. */
ws_asm_link(compiler, ws_asm_call_lib(compiler, expr->line, findex,
lindex));
}
break;
default:
ws_fatal("ws_expr_linearize(): unknown call expression type %x",
expr->u.call.type);
break;
}
break;
case WS_EXPR_SYMBOL:
{
WsNamespace *ns = ws_variable_lookup(compiler, expr->u.symbol);
if (ns == NULL) {
/* An unknown identifier. */
ws_src_error(compiler, expr->line, "unknown variable `%s'",
expr->u.symbol);
return;
}
/* Create a load instruction for the variable. */
ws_asm_link(compiler, ws_asm_variable(compiler, expr->line,
WS_ASM_P_LOAD_VAR, ns->vindex));
}
break;
case WS_EXPR_CONST_INVALID:
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
WS_ASM_CONST_INVALID));
break;
case WS_EXPR_CONST_TRUE:
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
WS_ASM_CONST_TRUE));
break;
case WS_EXPR_CONST_FALSE:
ws_asm_link(compiler, ws_asm_ins(compiler, expr->line,
WS_ASM_CONST_FALSE));
break;
case WS_EXPR_CONST_INTEGER:
if (expr->u.integer.ival == 0)
ins = ws_asm_ins(compiler, expr->line, WS_ASM_CONST_0);
else if (expr->u.integer.ival == 1 && expr->u.integer.sign == 1)
ins = ws_asm_ins(compiler, expr->line, WS_ASM_CONST_1);
else {
WsUInt16 cindex;
WsInt32 ival;
if (expr->u.integer.sign >= 0) {
if (expr->u.integer.ival > (WsUInt32) WS_INT32_MAX)
ws_src_error(compiler, expr->line,
"integer literal too large");
ival = expr->u.integer.ival;
} else {
if (expr->u.integer.ival > (WsUInt32) WS_INT32_MAX + 1)
ws_src_error(compiler, expr->line, "integer too small");
ival = - (WsInt32) expr->u.integer.ival;
}
if (!ws_bc_add_const_int(compiler->bc, &cindex, ival)) {
ws_error_memory(compiler);
return;
}
ins = ws_asm_load_const(compiler, expr->line, cindex);
}
ws_asm_link(compiler, ins);
break;
case WS_EXPR_CONST_FLOAT:
{
WsUInt16 cindex;
if (!ws_bc_add_const_float(compiler->bc, &cindex, expr->u.fval)) {
ws_error_memory(compiler);
return;
}
ws_asm_link(compiler, ws_asm_load_const(compiler, expr->line, cindex));
}
break;
case WS_EXPR_CONST_STRING:
if (expr->u.string.len == 0)
ins = ws_asm_ins(compiler, expr->line, WS_ASM_CONST_ES);
else {
WsUInt16 cindex;
if (!ws_bc_add_const_utf8_string(compiler->bc, &cindex,
expr->u.string.data,
expr->u.string.len)) {
ws_error_memory(compiler);
return;
}
ins = ws_asm_load_const(compiler, expr->line, cindex);
}
ws_asm_link(compiler, ins);
break;
}
}
void ws_stmt_linearize(WsCompiler *compiler, WsStatement *stmt)
{
WsListItem *li;
WsAsmIns *ins;
switch (stmt->type) {
case WS_STMT_BLOCK:
for (li = stmt->u.block->head; li; li = li->next)
ws_stmt_linearize(compiler, li->data);
break;
case WS_STMT_VARIABLE:
linearize_variable_init(compiler, stmt->u.var, stmt->first_line);
break;
case WS_STMT_EMPTY:
/* Nothing here. */
break;
case WS_STMT_EXPR:
ws_expr_linearize(compiler, stmt->u.expr);
/* Pop the expressions result from the stack. Otherwise loops
could eventually cause stack overflows. */
ws_asm_link(compiler, ws_asm_ins(compiler, stmt->last_line, WS_ASM_POP));
break;
case WS_STMT_IF:
{
WsAsmIns *l_else = ws_asm_label(compiler,
(stmt->u.s_if.s_else
? stmt->u.s_if.s_else->first_line
: stmt->last_line));
WsAsmIns *l_end = ws_asm_label(compiler, stmt->last_line);
/* Linearize the expression. */
ws_expr_linearize(compiler, stmt->u.s_if.expr);
/* If the result is false, jump to the else-branch. */
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->first_line,
WS_ASM_P_TJUMP, l_else));
/* Else, execute the then-branch and jump to the end. */
ws_stmt_linearize(compiler, stmt->u.s_if.s_then);
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->last_line,
WS_ASM_P_JUMP, l_end));
/* Then else-branch. */
ws_asm_link(compiler, l_else);
/* Linearize the else-branch if it is present. */
if (stmt->u.s_if.s_else)
ws_stmt_linearize(compiler, stmt->u.s_if.s_else);
/* Insert the end label. */
ws_asm_link(compiler, l_end);
}
break;
case WS_STMT_FOR:
{
WsAsmIns *l_loop = ws_asm_label(compiler, stmt->first_line);
WsAsmIns *l_cont = ws_asm_label(compiler, stmt->first_line);
WsAsmIns *l_break = ws_asm_label(compiler, stmt->first_line);
WsContBreak *cb;
/* Store the labels to the compiler. */
cb = ws_f_calloc(compiler->pool_stree, 1, sizeof(*cb));
if (cb == NULL) {
ws_error_memory(compiler);
return;
}
cb->next = compiler->cont_break;
compiler->cont_break = cb;
cb->l_cont = l_cont;
cb->l_break = l_break;
/* Linearize the possible init code. */
if (stmt->u.s_for.init)
linearize_variable_init(compiler, stmt->u.s_for.init,
stmt->first_line);
else if (stmt->u.s_for.e1) {
/* Linearize the init. */
ws_expr_linearize(compiler, stmt->u.s_for.e1);
/* Pop the result. */
ws_asm_link(compiler, ws_asm_ins(compiler, stmt->first_line,
WS_ASM_POP));
}
/* Insert the loop label. */
ws_asm_link(compiler, l_loop);
/* Linearize the condition. */
if (stmt->u.s_for.e2) {
ws_expr_linearize(compiler, stmt->u.s_for.e2);
/* If false, jump out. */
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->first_line,
WS_ASM_P_TJUMP, l_break));
}
/* Linearize the body statement. */
ws_stmt_linearize(compiler, stmt->u.s_for.stmt);
/* Link the continue label. */
ws_asm_link(compiler, l_cont);
/* Linearize the update expression. */
if (stmt->u.s_for.e3) {
ws_expr_linearize(compiler, stmt->u.s_for.e3);
/* Pop the result. */
ws_asm_link(compiler, ws_asm_ins(compiler, stmt->first_line,
WS_ASM_POP));
}
/* Jump to the loop label to check the condition. */
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->last_line,
WS_ASM_P_JUMP, l_loop));
/* Insert the break label. */
ws_asm_link(compiler, l_break);
/* Pop the cont-break block. */
compiler->cont_break = compiler->cont_break->next;
}
break;
case WS_STMT_WHILE:
{
WsAsmIns *l_cont = ws_asm_label(compiler, stmt->first_line);
WsAsmIns *l_break = ws_asm_label(compiler,
stmt->u.s_while.stmt->last_line);
WsContBreak *cb;
/* Store the labels to the compiler. */
cb = ws_f_calloc(compiler->pool_stree, 1, sizeof(*cb));
if (cb == NULL) {
ws_error_memory(compiler);
return;
}
cb->next = compiler->cont_break;
compiler->cont_break = cb;
cb->l_cont = l_cont;
cb->l_break = l_break;
/* Insert the continue label. */
ws_asm_link(compiler, l_cont);
/* Linearize the expression. */
ws_expr_linearize(compiler, stmt->u.s_while.expr);
/* If false, jump out. */
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->first_line,
WS_ASM_P_TJUMP, l_break));
/* Linearize the body statement. */
ws_stmt_linearize(compiler, stmt->u.s_while.stmt);
/* And jump to the continue label to check the expression. */
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->last_line,
WS_ASM_P_JUMP, l_cont));
/* Insert the break label. */
ws_asm_link(compiler, l_break);
/* Pop the cont-break block. */
compiler->cont_break = compiler->cont_break->next;
}
break;
case WS_STMT_CONTINUE:
if (compiler->cont_break == NULL)
ws_src_error(compiler, stmt->first_line,
"continue statement not within a loop");
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->first_line,
WS_ASM_P_JUMP,
compiler->cont_break->l_cont));
break;
case WS_STMT_BREAK:
if (compiler->cont_break == NULL)
ws_src_error(compiler, stmt->first_line,
"break statement not within a loop");
ws_asm_link(compiler, ws_asm_branch(compiler, stmt->first_line,
WS_ASM_P_JUMP,
compiler->cont_break->l_break));
break;
case WS_STMT_RETURN:
if (stmt->u.expr) {
/* Linearize the return value and return it. */
ws_expr_linearize(compiler, stmt->u.expr);
ins = ws_asm_ins(compiler, stmt->first_line, WS_ASM_RETURN);
} else
/* Return an empty string. */
ins = ws_asm_ins(compiler, stmt->first_line, WS_ASM_RETURN_ES);
ws_asm_link(compiler, ins);
break;
}
}
static WsBool read_float_from_exp(WsCompiler *compiler, WsBuffer *buffer,
WsFloat *result)
{
WsUInt32 ch;
unsigned char *p;
int sign = '+';
unsigned char buf[4];
/* Do we have an exponent part. */
if (!ws_stream_getc(compiler->input, &ch))
goto done;
if (ch != 'e' && ch != 'E') {
/* No exponent part. */
ws_stream_ungetc(compiler->input, ch);
goto done;
}
/* Sign. */
if (!ws_stream_getc(compiler->input, &ch)) {
/* This is an error. */
ws_src_error(compiler, 0, "truncated float literal");
return WS_FALSE;
}
if (ch == '-')
sign = '-';
else if (ch == '+')
sign = '+';
else
ws_stream_ungetc(compiler->input, ch);
/* DecimalDigits. */
if (!ws_stream_getc(compiler->input, &ch)) {
ws_src_error(compiler, 0, "truncated float literal");
return WS_FALSE;
}
if (!WS_IS_DECIMAL_DIGIT(ch)) {
ws_src_error(compiler, 0, "no decimal digits in exponent part");
return WS_FALSE;
}
/* Append exponent part read so far. */
if (!ws_buffer_append_space(buffer, &p, 2)) {
ws_error_memory(compiler);
return WS_FALSE;
}
p[0] = 'e';
p[1] = sign;
/* Read decimal digits. */
while (WS_IS_DECIMAL_DIGIT(ch)) {
if (!ws_buffer_append_space(buffer, &p, 1)) {
ws_error_memory(compiler);
return WS_FALSE;
}
p[0] = (unsigned char) ch;
if (!ws_stream_getc(compiler->input, &ch))
/* EOF. This is ok. */
goto done;
}
/* Unget the extra character. */
ws_stream_ungetc(compiler->input, ch);
/* FALLTHROUGH */
done:
if (!ws_buffer_append_space(buffer, &p, 1)) {
ws_error_memory(compiler);
return WS_FALSE;
}
p[0] = 0;
/* Now the buffer contains a valid floating point number. */
*result = (WsFloat) strtod((char *) ws_buffer_ptr(buffer), NULL);
/* Check that the generated floating point number fits to
`float32'. */
if (*result == HUGE_VAL || *result == -HUGE_VAL
|| ws_ieee754_encode_single(*result, buf) != WS_IEEE754_OK)
ws_src_error(compiler, 0, "floating point literal too large");
return WS_TRUE;
}
int ws_yy_lex(YYSTYPE *yylval, YYLTYPE *yylloc, void *context)
{
WsCompiler *compiler = (WsCompiler *) context;
WsUInt32 ch, ch2;
WsBuffer buffer;
unsigned char *p;
WsBool success;
/* Just check that we get the correct amount of arguments. */
gw_assert(compiler->magic == COMPILER_MAGIC);
while (ws_stream_getc(compiler->input, &ch)) {
/* Save the token's line number. */
yylloc->first_line = compiler->linenum;
switch (ch) {
case '\t': /* Whitespace characters. */
case '\v':
case '\f':
case ' ':
continue;
case '\n': /* Line terminators. */
case '\r':
if (ch == '\r' && ws_stream_getc(compiler->input, &ch2)) {
if (ch2 != '\n')
ws_stream_ungetc(compiler->input, ch2);
}
compiler->linenum++;
continue;
case '!': /* !, != */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tNE;
ws_stream_ungetc(compiler->input, ch2);
}
return '!';
case '%': /* %, %= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tREMA;
ws_stream_ungetc(compiler->input, ch2);
}
return '%';
case '&': /* &, &&, &= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '&')
return tAND;
if (ch2 == '=')
return tANDA;
ws_stream_ungetc(compiler->input, ch2);
}
return '&';
case '*': /* *, *= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tMULA;
ws_stream_ungetc(compiler->input, ch2);
}
return '*';
case '+': /* +, ++, += */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '+')
return tPLUSPLUS;
if (ch2 == '=')
return tADDA;
ws_stream_ungetc(compiler->input, ch2);
}
return '+';
case '-': /* -, --, -= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '-')
return tMINUSMINUS;
if (ch2 == '=')
return tSUBA;
ws_stream_ungetc(compiler->input, ch2);
}
return '-';
case '.':
if (ws_stream_getc(compiler->input, &ch2)) {
if (WS_IS_DECIMAL_DIGIT(ch2)) {
/* DecimalFloatLiteral. */
ws_buffer_init(&buffer);
if (!ws_buffer_append_space(&buffer, &p, 2)) {
ws_error_memory(compiler);
ws_buffer_uninit(&buffer);
return EOF;
}
p[0] = '.';
p[1] = (unsigned char) ch2;
success = read_float_from_point(compiler, &buffer,
&yylval->vfloat);
ws_buffer_uninit(&buffer);
if (!success)
return EOF;
return tFLOAT;
}
ws_stream_ungetc(compiler->input, ch2);
}
return '.';
case '/': /* /, /=, block or a single line comment */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '*') {
/* Block comment. */
while (1) {
if (!ws_stream_getc(compiler->input, &ch)) {
ws_src_error(compiler, 0, "EOF in comment");
return EOF;
}
if (ch == '\n' || ch == '\r') {
/* Line terminators. */
if (ch == '\r' && ws_stream_getc(compiler->input,
&ch2)) {
if (ch2 != '\n')
ws_stream_ungetc(compiler->input, ch2);
}
compiler->linenum++;
/* Continue reading the block comment. */
continue;
}
if (ch == '*' && ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '/')
/* The end of the comment found. */
break;
ws_stream_ungetc(compiler->input, ch2);
}
}
/* Continue after the comment. */
continue;
}
if (ch2 == '/') {
/* Single line comment. */
while (1) {
if (!ws_stream_getc(compiler->input, &ch))
/* The end of input stream reached. We accept
this as a valid comment terminator. */
break;
if (ch == '\n' || ch == '\r') {
/* Line terminators. */
if (ch == '\r' && ws_stream_getc(compiler->input,
&ch2)) {
if (ch2 != '\n')
ws_stream_ungetc(compiler->input, ch2);
}
/* The end of the line (and the comment)
reached. */
compiler->linenum++;
break;
}
}
/* Continue after the comment. */
continue;
}
if (ch2 == '=')
return tDIVA;
ws_stream_ungetc(compiler->input, ch2);
}
return '/';
case '<': /* <, <<, <<=, <= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '<') {
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tLSHIFTA;
ws_stream_ungetc(compiler->input, ch2);
}
return tLSHIFT;
}
if (ch2 == '=')
return tLE;
ws_stream_ungetc(compiler->input, ch2);
}
return '<';
case '=': /* =, == */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tEQ;
ws_stream_ungetc(compiler->input, ch2);
}
return '=';
case '>': /* >, >=, >>, >>=, >>>, >>>= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '>') {
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '>') {
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tRSZSHIFTA;
ws_stream_ungetc(compiler->input, ch2);
}
return tRSZSHIFT;
}
if (ch2 == '=')
return tRSSHIFTA;
ws_stream_ungetc(compiler->input, ch2);
}
return tRSSHIFT;
}
if (ch2 == '=')
return tGE;
ws_stream_ungetc(compiler->input, ch2);
}
return '>';
case '^': /* ^, ^= */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tXORA;
ws_stream_ungetc(compiler->input, ch2);
}
return '^';
case '|': /* |, |=, || */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == '=')
return tORA;
if (ch2 == '|')
return tOR;
ws_stream_ungetc(compiler->input, ch2);
}
return '|';
case '#': /* The simple cases. */
case '(':
case ')':
case ',':
case ':':
case ';':
case '?':
case '{':
case '}':
case '~':
return (int) ch;
case '\'': /* String literals. */
case '"':
{
WsUInt32 string_end_ch = ch;
WsUtf8String *str = ws_utf8_alloc();
if (str == NULL) {
ws_error_memory(compiler);
return EOF;
}
while (1) {
if (!ws_stream_getc(compiler->input, &ch)) {
eof_in_string_literal:
ws_src_error(compiler, 0, "EOF in string literal");
ws_utf8_free(str);
return EOF;
}
if (ch == string_end_ch)
/* The end of string reached. */
break;
if (ch == '\\') {
/* An escape sequence. */
if (!ws_stream_getc(compiler->input, &ch))
goto eof_in_string_literal;
switch (ch) {
case '\'':
case '"':
case '\\':
case '/':
/* The character as-is. */
break;
case 'b':
ch = '\b';
break;
case 'f':
ch = '\f';
break;
case 'n':
ch = '\n';
break;
case 'r':
ch = '\r';
break;
case 't':
ch = '\t';
break;
case 'x':
case 'u':
{
int i, len;
int type = ch;
if (ch == 'x')
len = 2;
else
len = 4;
ch = 0;
for (i = 0; i < len; i++) {
if (!ws_stream_getc(compiler->input, &ch2))
goto eof_in_string_literal;
if (!WS_IS_HEX_DIGIT(ch2)) {
ws_src_error(compiler, 0,
"malformed `\\%c' escape in "
"string literal", (char) type);
ch = 0;
break;
}
ch *= 16;
ch += WS_HEX_TO_INT(ch2);
}
}
break;
default:
if (WS_IS_OCTAL_DIGIT(ch)) {
int i;
int limit = 3;
ch = WS_OCTAL_TO_INT(ch);
if (ch > 3)
limit = 2;
for (i = 1; i < limit; i++) {
if (!ws_stream_getc(compiler->input, &ch2))
goto eof_in_string_literal;
if (!WS_IS_OCTAL_DIGIT(ch2)) {
ws_stream_ungetc(compiler->input, ch2);
break;
}
ch *= 8;
ch += WS_OCTAL_TO_INT(ch2);
}
} else {
ws_src_error(compiler, 0,
"unknown escape sequence `\\%c' in "
"string literal", (char) ch);
ch = 0;
}
break;
}
/* FALLTHROUGH */
}
if (!ws_utf8_append_char(str, ch)) {
ws_error_memory(compiler);
ws_utf8_free(str);
return EOF;
}
}
if (!ws_lexer_register_utf8(compiler, str)) {
ws_error_memory(compiler);
ws_utf8_free(str);
return EOF;
}
gw_assert(str != NULL);
yylval->string = str;
return tSTRING;
}
break;
default:
/* Identifiers, keywords and number constants. */
if (WS_IS_IDENTIFIER_LETTER(ch)) {
WsBool got;
int token;
unsigned char *p;
unsigned char *np;
size_t len = 0;
/* An identifier or a keyword. We start with a 256
* bytes long buffer but it is expanded dynamically if
* needed. However, 256 should be enought for most
* cases since the byte-code format limits the function
* names to 255 characters. */
p = ws_malloc(256);
if (p == NULL) {
ws_error_memory(compiler);
return EOF;
}
do {
/* Add one extra for the possible terminator
character. */
np = ws_realloc(p, len + 2);
if (np == NULL) {
ws_error_memory(compiler);
ws_free(p);
return EOF;
}
p = np;
/* This is ok since the only valid identifier names
* can be written in 7 bit ASCII. */
p[len++] = (unsigned char) ch;
} while ((got = ws_stream_getc(compiler->input, &ch))
&& (WS_IS_IDENTIFIER_LETTER(ch)
|| WS_IS_DECIMAL_DIGIT(ch)));
if (got)
/* Put back the terminator character. */
ws_stream_ungetc(compiler->input, ch);
/* Is it a keyword? */
if (lookup_keyword((char *) p, len, &token)) {
/* Yes it is... */
ws_free(p);
/* ...except one case: `div='. */
if (token == tIDIV) {
if (ws_stream_getc(compiler->input, &ch)) {
if (ch == '=')
return tIDIVA;
ws_stream_ungetc(compiler->input, ch);
}
}
/* Return the token value. */
return token;
}
/* It is a normal identifier. Let's pad the name with a
null-character. We have already allocated space for
it. */
p[len] = '\0';
if (!ws_lexer_register_block(compiler, p)) {
ws_error_memory(compiler);
ws_free(p);
return EOF;
}
gw_assert(p != NULL);
yylval->identifier = (char *) p;
return tIDENTIFIER;
}
if (WS_IS_NON_ZERO_DIGIT(ch)) {
/* A decimal integer literal or a decimal float
literal. */
ws_buffer_init(&buffer);
if (!ws_buffer_append_space(&buffer, &p, 1)) {
number_error_memory:
ws_error_memory(compiler);
ws_buffer_uninit(&buffer);
return EOF;
}
p[0] = ch;
while (ws_stream_getc(compiler->input, &ch)) {
if (WS_IS_DECIMAL_DIGIT(ch)) {
if (!ws_buffer_append_space(&buffer, &p, 1))
goto number_error_memory;
p[0] = ch;
} else if (ch == '.' || ch == 'e' || ch == 'E') {
/* DecimalFloatLiteral. */
if (ch == '.') {
if (!ws_buffer_append_space(&buffer, &p, 1))
goto number_error_memory;
p[0] = '.';
success = read_float_from_point(compiler, &buffer,
&yylval->vfloat);
} else {
ws_stream_ungetc(compiler->input, ch);
success = read_float_from_exp(compiler, &buffer,
&yylval->vfloat);
}
ws_buffer_uninit(&buffer);
if (!success)
return EOF;
return tFLOAT;
} else {
ws_stream_ungetc(compiler->input, ch);
break;
}
}
/* Now the buffer contains an integer number as a
string. Let's convert it to an integer number. */
yylval->integer = buffer_to_int(compiler, &buffer);
ws_buffer_uninit(&buffer);
/* Read a DecimalIntegerLiteral. */
return tINTEGER;
}
if (ch == '0') {
/* The integer constant 0, an octal number or a
HexIntegerLiteral. */
if (ws_stream_getc(compiler->input, &ch2)) {
if (ch2 == 'x' || ch2 == 'X') {
/* HexIntegerLiteral. */
ws_buffer_init(&buffer);
if (!ws_buffer_append_space(&buffer, &p, 2))
goto number_error_memory;
p[0] = '0';
p[1] = 'x';
while (ws_stream_getc(compiler->input, &ch)) {
if (WS_IS_HEX_DIGIT(ch)) {
if (!ws_buffer_append_space(&buffer, &p, 1))
goto number_error_memory;
p[0] = ch;
} else {
ws_stream_ungetc(compiler->input, ch);
break;
}
}
if (ws_buffer_len(&buffer) == 2) {
ws_buffer_uninit(&buffer);
ws_src_error(compiler, 0,
"numeric constant with no digits");
yylval->integer = 0;
return tINTEGER;
}
/* Now the buffer contains an integer number as
* a string. Let's convert it to an integer
* number. */
yylval->integer = buffer_to_int(compiler, &buffer);
ws_buffer_uninit(&buffer);
/* Read a HexIntegerLiteral. */
return tINTEGER;
}
if (WS_IS_OCTAL_DIGIT(ch2)) {
/* OctalIntegerLiteral. */
ws_buffer_init(&buffer);
if (!ws_buffer_append_space(&buffer, &p, 2))
goto number_error_memory;
p[0] = '0';
p[1] = ch2;
while (ws_stream_getc(compiler->input, &ch)) {
if (WS_IS_OCTAL_DIGIT(ch)) {
if (!ws_buffer_append_space(&buffer, &p, 1))
goto number_error_memory;
p[0] = ch;
} else {
ws_stream_ungetc(compiler->input, ch);
break;
}
}
/* Convert the buffer into an intger number. */
yylval->integer = buffer_to_int(compiler, &buffer);
ws_buffer_uninit(&buffer);
/* Read an OctalIntegerLiteral. */
return tINTEGER;
}
if (ch2 == '.' || ch2 == 'e' || ch2 == 'E') {
/* DecimalFloatLiteral. */
ws_buffer_init(&buffer);
if (ch2 == '.') {
if (!ws_buffer_append_space(&buffer, &p, 1))
goto number_error_memory;
p[0] = '.';
success = read_float_from_point(compiler, &buffer,
&yylval->vfloat);
} else {
ws_stream_ungetc(compiler->input, ch);
success = read_float_from_exp(compiler, &buffer,
&yylval->vfloat);
}
ws_buffer_uninit(&buffer);
if (!success)
return EOF;
return tFLOAT;
}
ws_stream_ungetc(compiler->input, ch2);
}
/* Integer literal 0. */
yylval->integer = 0;
return tINTEGER;
}
/* Garbage found from the input stream. */
ws_src_error(compiler, 0,
"garbage found from the input stream: character=0x%x",
ch);
return EOF;
break;
}
}
return EOF;
}
