518 lines
12 KiB
C
518 lines
12 KiB
C
/* $NetBSD: for.c,v 1.142 2021/04/03 11:08:40 rillig Exp $ */
|
|
|
|
/*
|
|
* Copyright (c) 1992, The Regents of the University of California.
|
|
* All rights reserved.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
* notice, this list of conditions and the following disclaimer.
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
* documentation and/or other materials provided with the distribution.
|
|
* 3. Neither the name of the University nor the names of its contributors
|
|
* may be used to endorse or promote products derived from this software
|
|
* without specific prior written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
* SUCH DAMAGE.
|
|
*/
|
|
|
|
/*
|
|
* Handling of .for/.endfor loops in a makefile.
|
|
*
|
|
* For loops have the form:
|
|
*
|
|
* .for <varname...> in <value...>
|
|
* # the body
|
|
* .endfor
|
|
*
|
|
* When a .for line is parsed, the following lines are copied to the body of
|
|
* the .for loop, until the corresponding .endfor line is reached. In this
|
|
* phase, the body is not yet evaluated. This also applies to any nested
|
|
* .for loops.
|
|
*
|
|
* After reaching the .endfor, the values from the .for line are grouped
|
|
* according to the number of variables. For each such group, the unexpanded
|
|
* body is scanned for variable expressions, and those that match the variable
|
|
* names are replaced with expressions of the form ${:U...} or $(:U...).
|
|
* After that, the body is treated like a file from an .include directive.
|
|
*
|
|
* Interface:
|
|
* For_Eval Evaluate the loop in the passed line.
|
|
*
|
|
* For_Run Run accumulated loop
|
|
*/
|
|
|
|
#include "make.h"
|
|
|
|
/* "@(#)for.c 8.1 (Berkeley) 6/6/93" */
|
|
MAKE_RCSID("$NetBSD: for.c,v 1.142 2021/04/03 11:08:40 rillig Exp $");
|
|
|
|
|
|
/* One of the variables to the left of the "in" in a .for loop. */
|
|
typedef struct ForVar {
|
|
char *name;
|
|
size_t nameLen;
|
|
} ForVar;
|
|
|
|
typedef struct ForLoop {
|
|
Buffer body; /* Unexpanded body of the loop */
|
|
Vector /* of ForVar */ vars; /* Iteration variables */
|
|
Words items; /* Substitution items */
|
|
Buffer curBody; /* Expanded body of the current iteration */
|
|
/* Is any of the names 1 character long? If so, when the variable values
|
|
* are substituted, the parser must handle $V expressions as well, not
|
|
* only ${V} and $(V). */
|
|
bool short_var;
|
|
unsigned int sub_next; /* Where to continue iterating */
|
|
} ForLoop;
|
|
|
|
|
|
static ForLoop *accumFor; /* Loop being accumulated */
|
|
static int forLevel = 0; /* Nesting level */
|
|
|
|
|
|
static ForLoop *
|
|
ForLoop_New(void)
|
|
{
|
|
ForLoop *f = bmake_malloc(sizeof *f);
|
|
|
|
Buf_Init(&f->body);
|
|
Vector_Init(&f->vars, sizeof(ForVar));
|
|
f->items.words = NULL;
|
|
f->items.freeIt = NULL;
|
|
Buf_Init(&f->curBody);
|
|
f->short_var = false;
|
|
f->sub_next = 0;
|
|
|
|
return f;
|
|
}
|
|
|
|
static void
|
|
ForLoop_Free(ForLoop *f)
|
|
{
|
|
Buf_Done(&f->body);
|
|
|
|
while (f->vars.len > 0) {
|
|
ForVar *var = Vector_Pop(&f->vars);
|
|
free(var->name);
|
|
}
|
|
Vector_Done(&f->vars);
|
|
|
|
Words_Free(f->items);
|
|
Buf_Done(&f->curBody);
|
|
|
|
free(f);
|
|
}
|
|
|
|
static void
|
|
ForLoop_AddVar(ForLoop *f, const char *name, size_t len)
|
|
{
|
|
ForVar *var = Vector_Push(&f->vars);
|
|
var->name = bmake_strldup(name, len);
|
|
var->nameLen = len;
|
|
}
|
|
|
|
static bool
|
|
ForLoop_ParseVarnames(ForLoop *f, const char **pp)
|
|
{
|
|
const char *p = *pp;
|
|
|
|
for (;;) {
|
|
size_t len;
|
|
|
|
cpp_skip_whitespace(&p);
|
|
if (*p == '\0') {
|
|
Parse_Error(PARSE_FATAL, "missing `in' in for");
|
|
return false;
|
|
}
|
|
|
|
/*
|
|
* XXX: This allows arbitrary variable names;
|
|
* see directive-for.mk.
|
|
*/
|
|
for (len = 1; p[len] != '\0' && !ch_isspace(p[len]); len++)
|
|
continue;
|
|
|
|
if (len == 2 && p[0] == 'i' && p[1] == 'n') {
|
|
p += 2;
|
|
break;
|
|
}
|
|
if (len == 1)
|
|
f->short_var = true;
|
|
|
|
ForLoop_AddVar(f, p, len);
|
|
p += len;
|
|
}
|
|
|
|
if (f->vars.len == 0) {
|
|
Parse_Error(PARSE_FATAL, "no iteration variables in for");
|
|
return false;
|
|
}
|
|
|
|
*pp = p;
|
|
return true;
|
|
}
|
|
|
|
static bool
|
|
ForLoop_ParseItems(ForLoop *f, const char *p)
|
|
{
|
|
char *items;
|
|
|
|
cpp_skip_whitespace(&p);
|
|
|
|
if (Var_Subst(p, SCOPE_GLOBAL, VARE_WANTRES, &items) != VPR_OK) {
|
|
Parse_Error(PARSE_FATAL, "Error in .for loop items");
|
|
return false;
|
|
}
|
|
|
|
f->items = Str_Words(items, false);
|
|
free(items);
|
|
|
|
if (f->items.len == 1 && f->items.words[0][0] == '\0')
|
|
f->items.len = 0; /* .for var in ${:U} */
|
|
|
|
if (f->items.len != 0 && f->items.len % f->vars.len != 0) {
|
|
Parse_Error(PARSE_FATAL,
|
|
"Wrong number of words (%u) in .for "
|
|
"substitution list with %u variables",
|
|
(unsigned)f->items.len, (unsigned)f->vars.len);
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
static bool
|
|
IsFor(const char *p)
|
|
{
|
|
return p[0] == 'f' && p[1] == 'o' && p[2] == 'r' && ch_isspace(p[3]);
|
|
}
|
|
|
|
static bool
|
|
IsEndfor(const char *p)
|
|
{
|
|
return p[0] == 'e' && strncmp(p, "endfor", 6) == 0 &&
|
|
(p[6] == '\0' || ch_isspace(p[6]));
|
|
}
|
|
|
|
/*
|
|
* Evaluate the for loop in the passed line. The line looks like this:
|
|
* .for <varname...> in <value...>
|
|
*
|
|
* Input:
|
|
* line Line to parse
|
|
*
|
|
* Results:
|
|
* 0: Not a .for statement, parse the line
|
|
* 1: We found a for loop
|
|
* -1: A .for statement with a bad syntax error, discard.
|
|
*/
|
|
int
|
|
For_Eval(const char *line)
|
|
{
|
|
ForLoop *f;
|
|
const char *p;
|
|
|
|
p = line + 1; /* skip the '.' */
|
|
cpp_skip_whitespace(&p);
|
|
|
|
if (!IsFor(p)) {
|
|
if (IsEndfor(p)) {
|
|
Parse_Error(PARSE_FATAL, "for-less endfor");
|
|
return -1;
|
|
}
|
|
return 0;
|
|
}
|
|
p += 3;
|
|
|
|
f = ForLoop_New();
|
|
|
|
if (!ForLoop_ParseVarnames(f, &p)) {
|
|
ForLoop_Free(f);
|
|
return -1;
|
|
}
|
|
|
|
if (!ForLoop_ParseItems(f, p)) {
|
|
/* Continue parsing the .for loop, but don't iterate. */
|
|
f->items.len = 0;
|
|
}
|
|
|
|
accumFor = f;
|
|
forLevel = 1;
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* Add another line to the .for loop that is being built up.
|
|
* Returns false when the matching .endfor is reached.
|
|
*/
|
|
bool
|
|
For_Accum(const char *line)
|
|
{
|
|
const char *p = line;
|
|
|
|
if (*p == '.') {
|
|
p++;
|
|
cpp_skip_whitespace(&p);
|
|
|
|
if (IsEndfor(p)) {
|
|
DEBUG1(FOR, "For: end for %d\n", forLevel);
|
|
if (--forLevel <= 0)
|
|
return false;
|
|
} else if (IsFor(p)) {
|
|
forLevel++;
|
|
DEBUG1(FOR, "For: new loop %d\n", forLevel);
|
|
}
|
|
}
|
|
|
|
Buf_AddStr(&accumFor->body, line);
|
|
Buf_AddByte(&accumFor->body, '\n');
|
|
return true;
|
|
}
|
|
|
|
|
|
static size_t
|
|
for_var_len(const char *var)
|
|
{
|
|
char ch, var_start, var_end;
|
|
int depth;
|
|
size_t len;
|
|
|
|
var_start = *var;
|
|
if (var_start == '\0')
|
|
/* just escape the $ */
|
|
return 0;
|
|
|
|
if (var_start == '(')
|
|
var_end = ')';
|
|
else if (var_start == '{')
|
|
var_end = '}';
|
|
else
|
|
return 1; /* Single char variable */
|
|
|
|
depth = 1;
|
|
for (len = 1; (ch = var[len++]) != '\0';) {
|
|
if (ch == var_start)
|
|
depth++;
|
|
else if (ch == var_end && --depth == 0)
|
|
return len;
|
|
}
|
|
|
|
/* Variable end not found, escape the $ */
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* The .for loop substitutes the items as ${:U<value>...}, which means
|
|
* that characters that break this syntax must be backslash-escaped.
|
|
*/
|
|
static bool
|
|
NeedsEscapes(const char *value, char endc)
|
|
{
|
|
const char *p;
|
|
|
|
for (p = value; *p != '\0'; p++) {
|
|
if (*p == ':' || *p == '$' || *p == '\\' || *p == endc)
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/*
|
|
* While expanding the body of a .for loop, write the item in the ${:U...}
|
|
* expression, escaping characters as needed.
|
|
*
|
|
* The result is later unescaped by ApplyModifier_Defined.
|
|
*/
|
|
static void
|
|
Buf_AddEscaped(Buffer *cmds, const char *item, char endc)
|
|
{
|
|
char ch;
|
|
|
|
if (!NeedsEscapes(item, endc)) {
|
|
Buf_AddStr(cmds, item);
|
|
return;
|
|
}
|
|
|
|
/* Escape ':', '$', '\\' and 'endc' - these will be removed later by
|
|
* :U processing, see ApplyModifier_Defined. */
|
|
while ((ch = *item++) != '\0') {
|
|
if (ch == '$') {
|
|
size_t len = for_var_len(item);
|
|
if (len != 0) {
|
|
Buf_AddBytes(cmds, item - 1, len + 1);
|
|
item += len;
|
|
continue;
|
|
}
|
|
Buf_AddByte(cmds, '\\');
|
|
} else if (ch == ':' || ch == '\\' || ch == endc)
|
|
Buf_AddByte(cmds, '\\');
|
|
Buf_AddByte(cmds, ch);
|
|
}
|
|
}
|
|
|
|
/*
|
|
* While expanding the body of a .for loop, replace the variable name of an
|
|
* expression like ${i} or ${i:...} or $(i) or $(i:...) with ":Uvalue".
|
|
*/
|
|
static void
|
|
ForLoop_SubstVarLong(ForLoop *f, const char **pp, const char *bodyEnd,
|
|
char endc, const char **inout_mark)
|
|
{
|
|
size_t i;
|
|
const char *p = *pp;
|
|
|
|
for (i = 0; i < f->vars.len; i++) {
|
|
ForVar *forVar = Vector_Get(&f->vars, i);
|
|
char *varname = forVar->name;
|
|
size_t varnameLen = forVar->nameLen;
|
|
|
|
if (varnameLen >= (size_t)(bodyEnd - p))
|
|
continue;
|
|
if (memcmp(p, varname, varnameLen) != 0)
|
|
continue;
|
|
/* XXX: why test for backslash here? */
|
|
if (p[varnameLen] != ':' && p[varnameLen] != endc &&
|
|
p[varnameLen] != '\\')
|
|
continue;
|
|
|
|
/*
|
|
* Found a variable match. Skip over the variable name and
|
|
* instead add ':U<value>' to the current body.
|
|
*/
|
|
Buf_AddBytesBetween(&f->curBody, *inout_mark, p);
|
|
Buf_AddStr(&f->curBody, ":U");
|
|
Buf_AddEscaped(&f->curBody,
|
|
f->items.words[f->sub_next + i], endc);
|
|
|
|
p += varnameLen;
|
|
*inout_mark = p;
|
|
*pp = p;
|
|
return;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* While expanding the body of a .for loop, replace single-character
|
|
* variable expressions like $i with their ${:U...} expansion.
|
|
*/
|
|
static void
|
|
ForLoop_SubstVarShort(ForLoop *f, const char *p, const char **inout_mark)
|
|
{
|
|
const char ch = *p;
|
|
ForVar *vars;
|
|
size_t i;
|
|
|
|
/* Skip $$ and stupid ones. */
|
|
if (!f->short_var || strchr("}):$", ch) != NULL)
|
|
return;
|
|
|
|
vars = Vector_Get(&f->vars, 0);
|
|
for (i = 0; i < f->vars.len; i++) {
|
|
const char *varname = vars[i].name;
|
|
if (varname[0] == ch && varname[1] == '\0')
|
|
goto found;
|
|
}
|
|
return;
|
|
|
|
found:
|
|
/* Replace $<ch> with ${:U<value>} */
|
|
Buf_AddBytesBetween(&f->curBody, *inout_mark, p), *inout_mark = p + 1;
|
|
Buf_AddStr(&f->curBody, "{:U");
|
|
Buf_AddEscaped(&f->curBody, f->items.words[f->sub_next + i], '}');
|
|
Buf_AddByte(&f->curBody, '}');
|
|
}
|
|
|
|
/*
|
|
* Compute the body for the current iteration by copying the unexpanded body,
|
|
* replacing the expressions for the iteration variables on the way.
|
|
*
|
|
* Using variable expressions ensures that the .for loop can't generate
|
|
* syntax, and that the later parsing will still see a variable.
|
|
* This code assumes that the variable with the empty name will never be
|
|
* defined, see unit-tests/varname-empty.mk for more details.
|
|
*
|
|
* The detection of substitutions of the loop control variables is naive.
|
|
* Many of the modifiers use '\' to escape '$' (not '$'), so it is possible
|
|
* to contrive a makefile where an unwanted substitution happens.
|
|
*/
|
|
static void
|
|
ForLoop_SubstBody(ForLoop *f)
|
|
{
|
|
const char *p, *bodyEnd;
|
|
const char *mark; /* where the last replacement left off */
|
|
|
|
Buf_Empty(&f->curBody);
|
|
|
|
mark = f->body.data;
|
|
bodyEnd = f->body.data + f->body.len;
|
|
for (p = mark; (p = strchr(p, '$')) != NULL;) {
|
|
if (p[1] == '{' || p[1] == '(') {
|
|
p += 2;
|
|
ForLoop_SubstVarLong(f, &p, bodyEnd,
|
|
p[-1] == '{' ? '}' : ')', &mark);
|
|
} else if (p[1] != '\0') {
|
|
ForLoop_SubstVarShort(f, p + 1, &mark);
|
|
p += 2;
|
|
} else
|
|
break;
|
|
}
|
|
|
|
Buf_AddBytesBetween(&f->curBody, mark, bodyEnd);
|
|
}
|
|
|
|
/*
|
|
* Compute the body for the current iteration by copying the unexpanded body,
|
|
* replacing the expressions for the iteration variables on the way.
|
|
*/
|
|
static char *
|
|
ForReadMore(void *v_arg, size_t *out_len)
|
|
{
|
|
ForLoop *f = v_arg;
|
|
|
|
if (f->sub_next == f->items.len) {
|
|
/* No more iterations */
|
|
ForLoop_Free(f);
|
|
return NULL;
|
|
}
|
|
|
|
ForLoop_SubstBody(f);
|
|
DEBUG1(FOR, "For: loop body:\n%s", f->curBody.data);
|
|
f->sub_next += (unsigned int)f->vars.len;
|
|
|
|
*out_len = f->curBody.len;
|
|
return f->curBody.data;
|
|
}
|
|
|
|
/* Run the .for loop, imitating the actions of an include file. */
|
|
void
|
|
For_Run(int lineno)
|
|
{
|
|
ForLoop *f = accumFor;
|
|
accumFor = NULL;
|
|
|
|
if (f->items.len == 0) {
|
|
/*
|
|
* Nothing to expand - possibly due to an earlier syntax
|
|
* error.
|
|
*/
|
|
ForLoop_Free(f);
|
|
return;
|
|
}
|
|
|
|
Parse_SetInput(NULL, lineno, -1, ForReadMore, f);
|
|
}
|