selectolax 0.3.17__tar.gz → 0.3.18__tar.gz

{selectolax-0.3.17 → selectolax-0.3.18}/CHANGES.rst

@@ -1,6 +1,14 @@
 selectolax Changelog
 ====================
 
+Version 0.3.15
+-------------
+
+Released
+
+- Fix memory leak when using CSS selectors, `lexbor` backend
+
+
 Version 0.3.17
 -------------
 

{selectolax-0.3.17/selectolax.egg-info → selectolax-0.3.18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: selectolax
-Version: 0.3.17
+Version: 0.3.18
 Summary: Fast HTML5 parser with CSS selectors.
 Home-page: https://github.com/rushter/selectolax
 Author: Artem Golubin

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/CMakeLists.txt

@@ -12,7 +12,7 @@ cmake_minimum_required(VERSION 2.8.12...3.27)
 #                                         Each module will have its own shared library
 #    LEXBOR_BUILD_EXAMPLES               default: OFF; Build all examples
 #    LEXBOR_BUILD_TESTS                  default: OFF; Build tests
-#    LEXBOR_BUILD_TESTS_CPP              default: ON; Build C++ tests.
+#    LEXBOR_BUILD_TESTS_CPP              default: OFF; Build C++ tests.
 #                                         Used with LEXBOR_BUILD_TESTS
 #    LEXBOR_BUILD_UTILS                  default: OFF; Build utils/helpers for project.
 #    LEXBOR_BUILD_WITH_ASAN              default: OFF; Build with address sanitizer if possible
@@ -47,7 +47,7 @@ option(LEXBOR_BUILD_STATIC "Build static library" ON)
 option(LEXBOR_BUILD_SEPARATELY "Build modules separately" OFF)
 option(LEXBOR_BUILD_EXAMPLES "Build examples" OFF)
 option(LEXBOR_BUILD_TESTS "Build tests" OFF)
-option(LEXBOR_BUILD_TESTS_CPP "Build C++ tests" ON)
+option(LEXBOR_BUILD_TESTS_CPP "Build C++ tests" OFF)
 option(LEXBOR_BUILD_UTILS "Build utils" OFF)
 option(LEXBOR_BUILD_WITH_ASAN "Build with address sanitizer" OFF)
 option(LEXBOR_BUILD_WITH_FUZZER "Build with fuzzer" OFF)
@@ -60,7 +60,10 @@ IF(NOT LEXBOR_MAKE_DISTRO_NUM)
 ENDIF()
 
 enable_language(C)
-enable_language(CXX)
+
+IF(LEXBOR_BUILD_TESTS_CPP)
+    enable_language(CXX)
+ENDIF()
 
 ################
 ## Version and path
@@ -117,10 +120,19 @@ GET_LEXBOR_VERSION(LEXBOR_VERSION_MAJOR LEXBOR_VERSION_MINOR
 message(STATUS "Lexbor version: ${LEXBOR_VERSION_STRING}")
 
 IF(CMAKE_VERSION VERSION_LESS "3.0")
-    PROJECT("${PROJECT_NAME}")
+    IF(LEXBOR_BUILD_TESTS_CPP)
+        PROJECT("${PROJECT_NAME}")
+    ELSE()
+        PROJECT("${PROJECT_NAME}" "C")
+    ENDIF()
 ELSE()
     cmake_policy(SET CMP0048 NEW)
-    PROJECT("${PROJECT_NAME}" VERSION "${LEXBOR_VERSION_STRING}")
+
+    IF(LEXBOR_BUILD_TESTS_CPP)
+        PROJECT("${PROJECT_NAME}" VERSION "${LEXBOR_VERSION_STRING}")
+    ELSE()
+        PROJECT("${PROJECT_NAME}" VERSION "${LEXBOR_VERSION_STRING}" LANGUAGES "C")
+    ENDIF()
 ENDIF()
 
 ################

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/README.md

@@ -76,6 +76,10 @@ Binaries are available for:
 Currently for `x86_64` architecture.
 If you need any other architecture, please, write to [support@lexbor.com](mailto:support@lexbor.com).
 
+### vcpkg
+
+For vcpkg users there is a `lexbor` [port](https://github.com/microsoft/vcpkg/tree/master/ports/lexbor) that can be installed via `vcpkg install lexbor` or by adding it to `dependencies` section of your `vcpkg.json` file.
+
 ### macOS
 
 #### Homebrew

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/css/base.h

@@ -18,7 +18,7 @@ extern "C" {
 
 
 #define LXB_CSS_VERSION_MAJOR 1
-#define LXB_CSS_VERSION_MINOR 1
+#define LXB_CSS_VERSION_MINOR 2
 #define LXB_CSS_VERSION_PATCH 0
 
 #define LXB_CSS_VERSION_STRING                                                 \

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/css/selectors/base.h

@@ -16,7 +16,7 @@ extern "C" {
 
 
 #define LXB_CSS_SELECTORS_VERSION_MAJOR 1
-#define LXB_CSS_SELECTORS_VERSION_MINOR 0
+#define LXB_CSS_SELECTORS_VERSION_MINOR 1
 #define LXB_CSS_SELECTORS_VERSION_PATCH 0
 
 #define LXB_CSS_SELECTORS_VERSION_STRING                                       \

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/css/selectors/state.c

@@ -742,12 +742,14 @@ again:
             goto unexpected;
     }
 
+    if (status == LXB_STATUS_OK) {
+        return true;
+    }
+
     if (status == LXB_STATUS_ERROR_MEMORY_ALLOCATION) {
         return lxb_css_parser_memory_fail(parser);
     }
 
-    return true;
-
 unexpected:
 
     (void) lxb_css_parser_states_to_root(parser);
@@ -827,11 +829,15 @@ lxb_css_selectors_state_compound_sub(lxb_css_parser_t *parser,
             return lxb_css_parser_states_set_back(parser);
     }
 
+    if (status == LXB_STATUS_OK) {
+        return true;
+    }
+
     if (status == LXB_STATUS_ERROR_MEMORY_ALLOCATION) {
         return lxb_css_parser_memory_fail(parser);
     }
 
-    return true;
+    return lxb_css_parser_unexpected(parser);
 }
 
 static bool
@@ -874,11 +880,15 @@ lxb_css_selectors_state_compound_pseudo(lxb_css_parser_t *parser,
                                        lxb_css_selectors_state_compound_pseudo);
     }
 
+    if (status == LXB_STATUS_OK) {
+        return true;
+    }
+
     if (status == LXB_STATUS_ERROR_MEMORY_ALLOCATION) {
         return lxb_css_parser_memory_fail(parser);
     }
 
-    return true;
+    return lxb_css_parser_unexpected(parser);
 }
 
 /*
@@ -1015,12 +1025,14 @@ again:
             goto unexpected;
     }
 
+    if (status == LXB_STATUS_OK) {
+        return true;
+    }
+
     if (status == LXB_STATUS_ERROR_MEMORY_ALLOCATION) {
         return lxb_css_parser_memory_fail(parser);
     }
 
-    return true;
-
 unexpected:
 
     (void) lxb_css_parser_states_set_back(parser);
@@ -1224,7 +1236,10 @@ lxb_css_selectors_state_attribute(lxb_css_parser_t *parser)
             lxb_css_parser_token_status_m(parser, token);
 
             if (token->type != LXB_CSS_SYNTAX_TOKEN_IDENT) {
-                goto failed;
+                attribute = &selector->u.attribute;
+                attribute->match = LXB_CSS_SELECTOR_MATCH_DASH;
+
+                goto assignment;
             }
 
             selector->ns = selector->name;
@@ -1282,20 +1297,23 @@ lxb_css_selectors_state_attribute(lxb_css_parser_t *parser)
             lxb_css_syntax_parser_consume(parser);
             lxb_css_parser_token_status_m(parser, token);
 
-            if (token->type != LXB_CSS_SYNTAX_TOKEN_DELIM
-                || lxb_css_syntax_token_delim_char(token) != '=')
-            {
-                goto failed;
-            }
-
-            lxb_css_syntax_parser_consume(parser);
-            lxb_css_parser_token_status_wo_ws_m(parser, token);
             break;
 
         default:
             goto failed;
     }
 
+assignment:
+
+    if (token->type != LXB_CSS_SYNTAX_TOKEN_DELIM
+        || lxb_css_syntax_token_delim_char(token) != '=')
+    {
+        goto failed;
+    }
+
+    lxb_css_syntax_parser_consume(parser);
+    lxb_css_parser_token_status_wo_ws_m(parser, token);
+
 string_or_ident:
 
     if (token->type != LXB_CSS_SYNTAX_TOKEN_STRING
@@ -1861,11 +1879,6 @@ lxb_css_selectors_state_list_end(lxb_css_parser_t *parser,
     lxb_css_parser_state_t *states;
     lxb_css_selectors_t *selectors = parser->selectors;
 
-    if (token->type == LXB_CSS_SYNTAX_TOKEN_WHITESPACE) {
-        lxb_css_syntax_parser_consume(parser);
-        lxb_css_parser_token_status_m(parser, token);
-    }
-
     if (lxb_css_parser_is_failed(parser)) {
         token = lxb_css_selectors_state_function_error(parser, token);
         if (token == NULL) {
@@ -1873,6 +1886,10 @@ lxb_css_selectors_state_list_end(lxb_css_parser_t *parser,
                                        LXB_STATUS_ERROR_MEMORY_ALLOCATION);
         }
     }
+    else if (token->type == LXB_CSS_SYNTAX_TOKEN_WHITESPACE) {
+        lxb_css_syntax_parser_consume(parser);
+        lxb_css_parser_token_status_m(parser, token);
+    }
 
     if (selectors->parent != NULL && selectors->list_last &&
         selectors->list_last->prev != NULL)

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/html/base.h

@@ -15,7 +15,7 @@ extern "C" {
 
 
 #define LXB_HTML_VERSION_MAJOR 2
-#define LXB_HTML_VERSION_MINOR 3
+#define LXB_HTML_VERSION_MINOR 4
 #define LXB_HTML_VERSION_PATCH 0
 
 #define LXB_HTML_VERSION_STRING LEXBOR_STRINGIZE(LXB_HTML_VERSION_MAJOR) "."   \

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/html/encoding.c

@@ -488,6 +488,8 @@ spaces_state:
         return data;
     }
 
+    data += 1;
+
 value_state:
 
     data = lxb_html_encoding_skip_spaces(data, end);

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/selectors/selectors.c

@@ -1479,6 +1479,8 @@ lxb_selectors_pseudo_class_function(lxb_selectors_t *selectors,
 
                         if (found) {
                             index++;
+                        } else if (index == 0) {
+                            return false;
                         }
 
                         found = false;
@@ -1523,6 +1525,8 @@ lxb_selectors_pseudo_class_function(lxb_selectors_t *selectors,
 
                         if (found) {
                             index++;
+                        } else if (index == 0) {
+                            return false;
                         }
 
                         found = false;

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/url/base.h

@@ -17,7 +17,7 @@ extern "C" {
 
 
 #define LXB_URL_VERSION_MAJOR 0
-#define LXB_URL_VERSION_MINOR 1
+#define LXB_URL_VERSION_MINOR 2
 #define LXB_URL_VERSION_PATCH 0
 
 #define LXB_URL_VERSION_STRING LEXBOR_STRINGIZE(LXB_URL_VERSION_MAJOR) "."    \

{selectolax-0.3.17 → selectolax-0.3.18}/lexbor/source/lexbor/url/url.c

@@ -3249,6 +3249,49 @@ lxb_url_percent_decode(const lxb_char_t *data, const lxb_char_t *end,
 lxb_url_t *
 lxb_url_destroy(lxb_url_t *url)
 {
+    lexbor_str_t *str;
+
+    if (url->scheme.name.data != NULL) {
+        lexbor_str_destroy(&url->scheme.name, url->mraw, false);
+    }
+
+    switch (url->host.type) {
+        case LXB_URL_HOST_TYPE_DOMAIN:
+        case LXB_URL_HOST_TYPE_OPAQUE:
+            lexbor_str_destroy(&url->host.u.domain, url->mraw, false);
+            break;
+
+        default:
+            break;
+    }
+
+    if (url->username.data != NULL) {
+        lexbor_str_destroy(&url->username, url->mraw, false);
+    }
+
+    if (url->password.data != NULL) {
+        lexbor_str_destroy(&url->password, url->mraw, false);
+    }
+
+    if (url->path.list != NULL) {
+        for (size_t i = 0; i < url->path.length; i++) {
+            str = url->path.list[i];
+
+            lexbor_str_destroy(str, url->mraw, false);
+            lexbor_mraw_free(url->mraw, str);
+        }
+
+        lexbor_mraw_free(url->mraw, url->path.list);
+    }
+
+    if (url->query.data != NULL) {
+        lexbor_str_destroy(&url->query, url->mraw, false);
+    }
+
+    if (url->fragment.data != NULL) {
+        lexbor_str_destroy(&url->fragment, url->mraw, false);
+    }
+
     return lexbor_mraw_free(url->mraw, url);
 }
 

selectolax-0.3.18/lexbor/version

@@ -0,0 +1 @@
+LEXBOR_VERSION=2.4.0

{selectolax-0.3.17 → selectolax-0.3.18}/selectolax/__init__.py

@@ -3,7 +3,7 @@
 
 __author__ = """Artem Golubin"""
 __email__ = 'me@rushter.com'
-__version__ = '0.3.17'
+__version__ = '0.3.18'
 
 from . import parser
 from . import lexbor

{selectolax-0.3.17 → selectolax-0.3.18}/selectolax/lexbor/selection.pxi

@@ -72,6 +72,7 @@ cdef class LexborCSSSelector:
             raise SelectolaxError("Can't parse CSS selector.")
         result = bool(self.results)
         self.results = []
+        lxb_css_selector_list_destroy_memory(selectors_list)
         return result