Skip to content

Commit

Permalink
Postpone position assignment until the end.
Browse files Browse the repository at this point in the history 
During AST expansion, it is possible for new literal nodes to be created
and given the same position as existing literals.  The code attempts to
reposition nodes when this happens but does not always succeed.  The
result is a short-circuit in the TNFA which can manifest as an infinite
loop, e.g. a bounded iteration being treated as an unbounded one.  This
reshuffling can also leave positions unfilled, wasting memory.

Instead of trying to assign positions to literal nodes on creation, and
then having to renumber them on the fly during AST expansion, wait until
the end and move position assignment into `tre_compute_nfl()`, which we
rename to `tre_compute_npfl()`.
  • Loading branch information
@dag-erling
dag-erling committed Jul 24, 2024
1 parent fdb68b3 commit 2e5030a
Show file tree
Hide file tree
Showing 5 changed files with 75 additions and 103 deletions.
4 changes: 2 additions & 2 deletions lib/tre-ast.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ tre_ast_new_node(tre_mem_t mem, tre_ast_type_t type, size_t size)
}

tre_ast_node_t *
tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position)
tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max)
{
tre_ast_node_t *node;
tre_literal_t *lit;
Expand All @@ -44,7 +44,7 @@ tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position)
lit = node->obj;
lit->code_min = code_min;
lit->code_max = code_max;
lit->position = position;
lit->position = -1;

return node;
}
Expand Down
2 changes: 1 addition & 1 deletion lib/tre-ast.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -101,7 +101,7 @@ tre_ast_node_t *
tre_ast_new_node(tre_mem_t mem, tre_ast_type_t type, size_t size);

tre_ast_node_t *
tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position);
tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max);

tre_ast_node_t *
tre_ast_new_iter(tre_mem_t mem, tre_ast_node_t *arg, int min, int max,
Expand Down
106 changes: 50 additions & 56 deletions lib/tre-compile.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ tre_add_tag_left(tre_mem_t mem, tre_ast_node_t *node, int tag_id)
c = tre_mem_alloc(mem, sizeof(*c));
if (c == NULL)
return REG_ESPACE;
c->left = tre_ast_new_literal(mem, TAG, tag_id, -1);
c->left = tre_ast_new_literal(mem, TAG, tag_id);
if (c->left == NULL)
return REG_ESPACE;
c->right = tre_mem_alloc(mem, sizeof(tre_ast_node_t));
Expand Down Expand Up @@ -78,7 +78,7 @@ tre_add_tag_right(tre_mem_t mem, tre_ast_node_t *node, int tag_id)
c = tre_mem_alloc(mem, sizeof(*c));
if (c == NULL)
return REG_ESPACE;
c->right = tre_ast_new_literal(mem, TAG, tag_id, -1);
c->right = tre_ast_new_literal(mem, TAG, tag_id);
if (c->right == NULL)
return REG_ESPACE;
c->left = tre_mem_alloc(mem, sizeof(tre_ast_node_t));
Expand Down Expand Up @@ -711,7 +711,7 @@ tre_copy_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
tag_directions[max] = TRE_TAG_MAXIMIZE;
first_tag = 0;
}
*result = tre_ast_new_literal(mem, min, max, pos);
*result = tre_ast_new_literal(mem, min, max);
if (*result == NULL)
status = REG_ESPACE;

Expand Down Expand Up @@ -798,8 +798,7 @@ typedef enum {
iteration count to a catenated sequence of copies of the node. */
static reg_errcode_t
tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
int *position, tre_tag_direction_t *tag_directions,
int *max_depth)
tre_tag_direction_t *tag_directions, int *max_depth)
{
reg_errcode_t status = REG_OK;
int bottom = tre_stack_num_objects(stack);
Expand Down Expand Up @@ -949,7 +948,7 @@ tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
seq2 = copy;
if (seq2 == NULL)
return REG_ESPACE;
tmp = tre_ast_new_literal(mem, EMPTY, -1, -1);
tmp = tre_ast_new_literal(mem, EMPTY, -1);
if (tmp == NULL)
return REG_ESPACE;
seq2 = tre_ast_new_union(mem, tmp, seq2);
Expand Down Expand Up @@ -983,12 +982,12 @@ tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
tre_ast_node_t *tmp_l, *tmp_r, *tmp_node, *node_copy;
int *old_params;

tmp_l = tre_ast_new_literal(mem, PARAMETER, 0, -1);
tmp_l = tre_ast_new_literal(mem, PARAMETER, 0);
if (!tmp_l)
return REG_ESPACE;
((tre_literal_t *)tmp_l->obj)->u.params = iter->params;
iter->params[TRE_PARAM_DEPTH] = params_depth + 1;
tmp_r = tre_ast_new_literal(mem, PARAMETER, 0, -1);
tmp_r = tre_ast_new_literal(mem, PARAMETER, 0);
if (!tmp_r)
return REG_ESPACE;
old_params = tre_mem_alloc(mem, sizeof(*old_params)
Expand Down Expand Up @@ -1028,19 +1027,9 @@ tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
}
}

*position += pos_add_total;

/* `max_pos' should never be larger than `*position' if the above
code works, but just an extra safeguard let's make sure
`*position' is set large enough so enough memory will be
allocated for the transition table. */
if (max_pos > *position)
*position = max_pos;

#ifdef TRE_DEBUG
DPRINT(("Expanded AST:\n"));
tre_ast_print(ast);
DPRINT(("*position %d, max_pos %d\n", *position, max_pos));
#endif

return status;
Expand Down Expand Up @@ -1305,33 +1294,36 @@ tre_match_empty(tre_stack_t *stack, tre_ast_node_t *node, int *tags,


typedef enum {
NFL_RECURSE,
NFL_POST_UNION,
NFL_POST_CATENATION,
NFL_POST_ITERATION
} tre_nfl_stack_symbol_t;
NPFL_RECURSE,
NPFL_POST_UNION,
NPFL_POST_CATENATION,
NPFL_POST_ITERATION
} tre_npfl_stack_symbol_t;


/* Computes and fills in the fields `nullable', `firstpos', and `lastpos' for
the nodes of the AST `tree'. */
/* Computes and fills in the fields `nullable', `position`, `firstpos',
and `lastpos' for the nodes of the AST `tree'; `nextpos' points to an
integer indicating the next available position, and will be updated on
return to reflect the number of additional positions assigned. */
static reg_errcode_t
tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
tre_compute_npfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree,
int *nextpos)
{
int bottom = tre_stack_num_objects(stack);

STACK_PUSHR(stack, voidptr, tree);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);

while (tre_stack_num_objects(stack) > bottom)
{
tre_nfl_stack_symbol_t symbol;
tre_npfl_stack_symbol_t symbol;
tre_ast_node_t *node;

symbol = (tre_nfl_stack_symbol_t)tre_stack_pop_int(stack);
symbol = (tre_npfl_stack_symbol_t)tre_stack_pop_int(stack);
node = tre_stack_pop_voidptr(stack);
switch (symbol)
{
case NFL_RECURSE:
case NPFL_RECURSE:
switch (node->type)
{
case LITERAL:
Expand All @@ -1342,6 +1334,7 @@ tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
/* Back references: nullable = false, firstpos = {i},
lastpos = {i}. */
node->nullable = 0;
lit->position = (*nextpos)++;
node->firstpos = tre_set_one(mem, lit->position, 0,
TRE_CHAR_MAX, 0, NULL, -1);
if (!node->firstpos)
Expand Down Expand Up @@ -1369,6 +1362,7 @@ tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
/* Literal at position i: nullable = false, firstpos = {i},
lastpos = {i}. */
node->nullable = 0;
lit->position = (*nextpos)++;
node->firstpos =
tre_set_one(mem, lit->position, (int)lit->code_min,
(int)lit->code_max, 0, NULL, -1);
Expand All @@ -1389,36 +1383,36 @@ tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
/* Compute the attributes for the two subtrees, and after that
for this node. */
STACK_PUSHR(stack, voidptr, node);
STACK_PUSHR(stack, int, NFL_POST_UNION);
STACK_PUSHR(stack, int, NPFL_POST_UNION);
STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->right);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);
STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->left);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);
break;

case CATENATION:
/* Compute the attributes for the two subtrees, and after that
for this node. */
STACK_PUSHR(stack, voidptr, node);
STACK_PUSHR(stack, int, NFL_POST_CATENATION);
STACK_PUSHR(stack, int, NPFL_POST_CATENATION);
STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->right);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);
STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->left);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);
break;

case ITERATION:
/* Compute the attributes for the subtree, and after that for
this node. */
STACK_PUSHR(stack, voidptr, node);
STACK_PUSHR(stack, int, NFL_POST_ITERATION);
STACK_PUSHR(stack, int, NPFL_POST_ITERATION);
STACK_PUSHR(stack, voidptr, ((tre_iteration_t *)node->obj)->arg);
STACK_PUSHR(stack, int, NFL_RECURSE);
STACK_PUSHR(stack, int, NPFL_RECURSE);
break;
}
break; /* end case: NFL_RECURSE */
break; /* end case: NPFL_RECURSE */

case NFL_POST_UNION:
case NPFL_POST_UNION:
{
tre_union_t *uni = (tre_union_t *)node->obj;
node->nullable = uni->left->nullable || uni->right->nullable;
Expand All @@ -1433,7 +1427,7 @@ tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
break;
}

case NFL_POST_ITERATION:
case NPFL_POST_ITERATION:
{
tre_iteration_t *iter = (tre_iteration_t *)node->obj;

Expand All @@ -1446,7 +1440,7 @@ tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
break;
}

case NFL_POST_CATENATION:
case NPFL_POST_CATENATION:
{
int num_tags, *tags, assertions, params_seen;
int *params;
Expand Down Expand Up @@ -1839,7 +1833,6 @@ tre_ast_to_tnfa(tre_ast_node_t *node, tre_tnfa_transition_t *transitions,
return errcode;
}


#define ERROR_EXIT(err) \
do \
{ \
Expand All @@ -1864,6 +1857,7 @@ tre_compile(regex_t *preg, const tre_char_t *regex, size_t n, int cflags)
tre_tag_direction_t *tag_directions = NULL;
reg_errcode_t errcode;
tre_mem_t mem;
int numpos = 0;

/* Parse context. */
tre_parse_ctx_t parse_ctx;
Expand Down Expand Up @@ -1970,8 +1964,8 @@ tre_compile(regex_t *preg, const tre_char_t *regex, size_t n, int cflags)
}

/* Expand iteration nodes. */
errcode = tre_expand_ast(mem, stack, tree, &parse_ctx.position,
tag_directions, &tnfa->params_depth);
errcode = tre_expand_ast(mem, stack, tree, tag_directions,
&tnfa->params_depth);
if (errcode != REG_OK)
ERROR_EXIT(errcode);

Expand All @@ -1980,37 +1974,37 @@ tre_compile(regex_t *preg, const tre_char_t *regex, size_t n, int cflags)
for example "a*" or "ab*". Figure out a simple way to detect
this possibility. */
tmp_ast_l = tree;
tmp_ast_r = tre_ast_new_literal(mem, 0, 0, parse_ctx.position++);
tmp_ast_r = tre_ast_new_literal(mem, 0, 0);
if (tmp_ast_r == NULL)
ERROR_EXIT(REG_ESPACE);

tree = tre_ast_new_catenation(mem, tmp_ast_l, tmp_ast_r);
if (tree == NULL)
ERROR_EXIT(REG_ESPACE);

errcode = tre_compute_npfl(mem, stack, tree, &numpos);
if (errcode != REG_OK)
ERROR_EXIT(errcode);

#ifdef TRE_DEBUG
tre_ast_print(tree);
DPRINT(("Number of states: %d\n", parse_ctx.position));
DPRINT(("Number of states: %d\n", numpos));
#endif /* TRE_DEBUG */

errcode = tre_compute_nfl(mem, stack, tree);
if (errcode != REG_OK)
ERROR_EXIT(errcode);

counts = xmalloc(sizeof(int) * parse_ctx.position);
counts = xmalloc(sizeof(int) * numpos);
if (counts == NULL)
ERROR_EXIT(REG_ESPACE);

offs = xmalloc(sizeof(int) * parse_ctx.position);
offs = xmalloc(sizeof(int) * numpos);
if (offs == NULL)
ERROR_EXIT(REG_ESPACE);

for (i = 0; i < parse_ctx.position; i++)
for (i = 0; i < numpos; i++)
counts[i] = 0;
tre_ast_to_tnfa(tree, NULL, counts, NULL);

add = 0;
for (i = 0; i < parse_ctx.position; i++)
for (i = 0; i < numpos; i++)
{
offs[i] = add;
add += counts[i] + 1;
Expand Down Expand Up @@ -2148,7 +2142,7 @@ tre_compile(regex_t *preg, const tre_char_t *regex, size_t n, int cflags)

tnfa->num_transitions = add;
tnfa->final = transitions + offs[tree->lastpos[0].position];
tnfa->num_states = parse_ctx.position;
tnfa->num_states = numpos;
tnfa->cflags = cflags;

DPRINT(("final state %p\n", (void *)tnfa->final));
Expand Down
Loading

0 comments on commit 2e5030a

Please sign in to comment.