int filter_code_comments(char *filename, zval *retval)
{
zend_lex_state original_lex_state;
zend_file_handle file_handle = {0};
#if PHP_API_VERSION > 20090626
php_output_start_default(TSRMLS_C);
file_handle.type = ZEND_HANDLE_FILENAME;
file_handle.filename = filename;
file_handle.free_filename = 0;
file_handle.opened_path = NULL;
zend_save_lexical_state(&original_lex_state TSRMLS_CC);
if (open_file_for_scanning(&file_handle TSRMLS_CC) == FAILURE) {
zend_restore_lexical_state(&original_lex_state TSRMLS_CC);
php_output_end(TSRMLS_C);
return -1;
}
zend_strip(TSRMLS_C);
zend_destroy_file_handle(&file_handle TSRMLS_CC);
zend_restore_lexical_state(&original_lex_state TSRMLS_CC);
php_output_get_contents(retval TSRMLS_CC);
php_output_discard(TSRMLS_C);
#else
file_handle.type = ZEND_HANDLE_FILENAME;
file_handle.filename = filename;
file_handle.free_filename = 0;
file_handle.opened_path = NULL;
zend_save_lexical_state(&original_lex_state TSRMLS_CC);
if (open_file_for_scanning(&file_handle TSRMLS_CC) == FAILURE) {
zend_restore_lexical_state(&original_lex_state TSRMLS_CC);
return -1;
}
php_start_ob_buffer(NULL, 0, 1 TSRMLS_CC);
zend_strip(TSRMLS_C);
zend_destroy_file_handle(&file_handle TSRMLS_CC);
zend_restore_lexical_state(&original_lex_state TSRMLS_CC);
php_ob_get_buffer(retval TSRMLS_CC);
php_end_ob_buffer(0, 0 TSRMLS_CC);
#endif
return 0;
}
static zend_ast *get_ast(zend_string *code, zend_arena **ast_arena, char *filename) {
zval code_zv;
zend_bool original_in_compilation;
zend_lex_state original_lex_state;
zend_ast *ast;
ZVAL_STR_COPY(&code_zv, code);
original_in_compilation = CG(in_compilation);
CG(in_compilation) = 1;
zend_save_lexical_state(&original_lex_state);
if (zend_prepare_string_for_scanning(&code_zv, filename) == SUCCESS) {
CG(ast) = NULL;
CG(ast_arena) = zend_arena_create(1024 * 32);
LANG_SCNG(yy_state) = yycINITIAL;
if (zendparse() != 0) {
zend_ast_destroy(CG(ast));
zend_arena_destroy(CG(ast_arena));
CG(ast) = NULL;
}
}
/* restore_lexical_state changes CG(ast) and CG(ast_arena) */
ast = CG(ast);
*ast_arena = CG(ast_arena);
zend_restore_lexical_state(&original_lex_state);
CG(in_compilation) = original_in_compilation;
zval_dtor(&code_zv);
return ast;
}
static zend_bool tokenize_parse(zval *return_value, zend_string *source)
{
zval source_zval;
zend_lex_state original_lex_state;
zend_bool original_in_compilation;
zend_bool success;
ZVAL_STR_COPY(&source_zval, source);
original_in_compilation = CG(in_compilation);
CG(in_compilation) = 1;
zend_save_lexical_state(&original_lex_state);
if ((success = (zend_prepare_string_for_scanning(&source_zval, "") == SUCCESS))) {
zval token_stream;
array_init(&token_stream);
CG(ast) = NULL;
CG(ast_arena) = zend_arena_create(1024 * 32);
LANG_SCNG(yy_state) = yycINITIAL;
LANG_SCNG(on_event) = on_event;
LANG_SCNG(on_event_context) = &token_stream;
if((success = (zendparse() == SUCCESS))) {
ZVAL_COPY_VALUE(return_value, &token_stream);
} else {
zval_ptr_dtor(&token_stream);
}
zend_ast_destroy(CG(ast));
zend_arena_destroy(CG(ast_arena));
}
/* restore compiler and scanner global states */
zend_restore_lexical_state(&original_lex_state);
CG(in_compilation) = original_in_compilation;
zval_dtor(&source_zval);
return success;
}
static zend_bool tokenize(zval *return_value, zend_string *source)
{
zval source_zval;
zend_lex_state original_lex_state;
zval token;
zval keyword;
int token_type;
int token_line = 1;
int need_tokens = -1; /* for __halt_compiler lexing. -1 = disabled */
ZVAL_STR_COPY(&source_zval, source);
zend_save_lexical_state(&original_lex_state);
if (zend_prepare_string_for_scanning(&source_zval, "") == FAILURE) {
zend_restore_lexical_state(&original_lex_state);
return 0;
}
LANG_SCNG(yy_state) = yycINITIAL;
array_init(return_value);
ZVAL_UNDEF(&token);
while ((token_type = lex_scan(&token))) {
if (token_type == T_CLOSE_TAG && zendtext[zendleng - 1] != '>') {
CG(zend_lineno)++;
}
if (token_type >= 256) {
array_init(&keyword);
add_next_index_long(&keyword, token_type);
if (token_type == T_END_HEREDOC) {
if (CG(increment_lineno)) {
token_line = ++CG(zend_lineno);
CG(increment_lineno) = 0;
}
}
add_next_index_stringl(&keyword, (char *)zendtext, zendleng);
add_next_index_long(&keyword, token_line);
add_next_index_zval(return_value, &keyword);
} else {
add_next_index_stringl(return_value, (char *)zendtext, zendleng);
}
if (Z_TYPE(token) != IS_UNDEF) {
zval_dtor(&token);
ZVAL_UNDEF(&token);
}
/* after T_HALT_COMPILER collect the next three non-dropped tokens */
if (need_tokens != -1) {
if (token_type != T_WHITESPACE && token_type != T_OPEN_TAG
&& token_type != T_COMMENT && token_type != T_DOC_COMMENT
&& --need_tokens == 0
) {
/* fetch the rest into a T_INLINE_HTML */
if (zendcursor != zendlimit) {
array_init(&keyword);
add_next_index_long(&keyword, T_INLINE_HTML);
add_next_index_stringl(&keyword, (char *)zendcursor, zendlimit - zendcursor);
add_next_index_long(&keyword, token_line);
add_next_index_zval(return_value, &keyword);
}
break;
}
} else if (token_type == T_HALT_COMPILER) {
need_tokens = 3;
}
token_line = CG(zend_lineno);
}
zval_dtor(&source_zval);
zend_restore_lexical_state(&original_lex_state);
return 1;
}
