Add free_token_node grammar statement

This commit is contained in:
Josh Holtrop 2026-02-08 20:33:00 -05:00
parent addf27d837
commit cb426b4be1
5 changed files with 105 additions and 0 deletions

View File

@ -1163,6 +1163,9 @@ static void free_ast_node(ASTNode * node)
{
if (node->is_token)
{
<% if @grammar.free_token_node %>
<%= @grammar.free_token_node %>((<%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %> *) node);
<% end %>
/* TODO: free value_t */
}
else if (node->n_fields > 0u)

View File

@ -326,6 +326,36 @@ assert(itemsmore.pItem.pItem.pItem !is null);
assert(itemsmore.pItem.pItem.pItem.pToken1 !is null);
```
## Freeing user-allocated memory in token node `pvalue`: the `free_token_node` statement
If user lexer code block allocates memory to store in a token node's `pvalue`,
the `free_token_node` grammar statement can be used to specify the name of a
function which will be called during the `p_free_ast()` call to free the memory
associated with a token node.
Example:
```
<<
static void free_token(Token * token)
{
free(token->pvalue);
}
>>
ast;
free_token_node free_token;
ptype int *;
token a <<
$$ = (int *)malloc(sizeof(int));
*$$ = 1;
>>
token b <<
$$ = (int *)malloc(sizeof(int));
*$$ = 2;
>>
Start -> a:a b:b;
```
##> Specifying tokens - the `token` statement
The `token` statement allows defining a lexer token and a pattern to match that
@ -1098,6 +1128,20 @@ assert(code_point == 0x1F9E1u);
assert(code_point_length == 4u);
```
### `p_free_ast`
The `p_free_ast()` function can be used to free the memory used by the AST.
It should be passed the same value that is returned by `p_result()`.
The `p_free_ast()` function is only available for C/C++ output targets.
Note that if any lexer user code block allocates memory to store in a token's
`pvalue`, in order to properly free this memory a `free_token_node` function
should be specified in the grammar file.
If specified, the `free_token_node` function will be called during the
`p_free_ast()` process to allow user code to free any memory associated with
a token node's `pvalue`.
##> Data
### `p_token_names`

View File

@ -8,6 +8,7 @@ class Propane
attr_reader :ast
attr_reader :ast_prefix
attr_reader :ast_suffix
attr_reader :free_token_node
attr_reader :modulename
attr_reader :patterns
attr_reader :rules
@ -32,6 +33,7 @@ class Propane
@ast = false
@ast_prefix = ""
@ast_suffix = ""
@free_token_node = nil
parse_grammar!
end
@ -62,6 +64,7 @@ class Propane
elsif parse_ast_statement!
elsif parse_ast_prefix_statement!
elsif parse_ast_suffix_statement!
elsif parse_free_token_node_statement!
elsif parse_module_statement!
elsif parse_ptype_statement!
elsif parse_pattern_statement!
@ -112,6 +115,12 @@ class Propane
end
end
def parse_free_token_node_statement!
if md = consume!(/free_token_node\s+(\w+)\s*;/)
@free_token_node = md[1]
end
end
def parse_module_statement!
if consume!(/module\s+/)
md = consume!(/([\w.]+)\s*/, "expected module name")

View File

@ -1439,6 +1439,36 @@ EOF
expect(results.stderr).to eq ""
expect(results.status).to eq 0
end
if %w[c cpp].include?(language)
it "allows a user function to free token node memory in AST mode" do
write_grammar <<EOF
<<
static void free_token(Token * token)
{
free(token->pvalue);
}
>>
ast;
free_token_node free_token;
ptype int *;
token a <<
$$ = (int *)malloc(sizeof(int));
*$$ = 1;
>>
token b <<
$$ = (int *)malloc(sizeof(int));
*$$ = 2;
>>
Start -> a:a b:b;
EOF
run_propane(language: language)
compile("spec/test_free_ast_token_node_memory.#{language}", language: language)
results = run_test(language: language)
expect(results.stderr).to eq ""
expect(results.status).to eq 0
end
end
end
end
end

View File

@ -0,0 +1,19 @@
#include "testparser.h"
#include <assert.h>
#include <string.h>
#include "testutils.h"
int main()
{
char const * input = "ab";
p_context_t context;
p_context_init(&context, (uint8_t const *)input, strlen(input));
assert_eq(P_SUCCESS, p_parse(&context));
Start * start = p_result(&context);
assert(start->a != NULL);
assert(*start->a->pvalue == 1);
assert(start->b != NULL);
assert(*start->b->pvalue == 2);
p_free_ast(start);
}