nokogumbo 1.1 → 1.1.1

checksums.yaml

@@ -1,15 +1,15 @@
 ---
 !binary "U0hBMQ==":
   metadata.gz: !binary |-
-    MDM0N2MwNTE2MzI0MmFjNTZmYzQ5YTk2MDMwNWQ1YzVlNTk5NTMzYQ==
+    ZDM4N2U1MTA3MjkzNGQ1ZmU1OGY4NjljY2NiNzdlMGEwNzYyYTkwMA==
   data.tar.gz: !binary |-
-    OTNlOWZkMDc0NTc3N2U5MjY1ODhlNTEwMmMxMmMxOTBiYTA0NTViZA==
+    YzQ1MTYzYTJjZjM1YjFkZTU2NThlZWFkMzhmODFjMTI4MDk2YzVlYw==
 !binary "U0hBNTEy":
   metadata.gz: !binary |-
-    YjI2NzhjYzE0ZjQ1OTA3MDRiNWJjOGZkNzRlZDNhOTQ2N2ViODZjNDFlOWE3
-    Zjk1MTY0MmQ2ZDgwMTQxYzczNDQxYTFmYjdiNGY3ZjAwZDYxNzVlNjE2MWJh
-    NDQyYzk5MGU5ZTZlNDdkZWUxMjE3ZjhmNWIxZWIzYzFkZDRkYTA=
+    ZDM3MDI2MjkwMzAyNTQ2ZmRhNDc3NmRlNjIzMjBlZjE5OGU2MGU3M2MyZDcw
+    MTgxOTBhOGUxZjkxYmZlZGZlN2I1NTVmYTBlM2E3NWZkNmFiODE4ZThjMjdj
+    N2E1M2RkNDljYjc4YWY0YmIxOGYyYTNjYTM1OWQzNDA0NjBiOTQ=
   data.tar.gz: !binary |-
-    YjE4NDhmMTIxNTI5MThlM2Q1MmRkNmVkNWRmYmFjNjllYzQ3YTFkMWJmNWI1
-    MWRkNDI2NzY0MWRhNjVlMjNjNmQ2NWZjOTQ3YzUwNzc3ZmNiMDBjNTBlZTBk
-    YTZkYzU2NTQ3NDNmMzIxNzA4NzQ3ZDI2NmFjY2Y3YWU3MmFmNTY=
+    NmFkYjA3MjA3Njk5YTJiMmIyZWEyM2EyNGQ0ZjQxYzAwMTY2NDQwZmM3YmUz
+    YWQwM2YzNjFlOGM2OTljNjc1MzZlMTZmNDM1ODUwYmZkZjZhNzcyZGE0ZGMy
+    N2Y4YzA5ZWI4NjZiY2M5NGI2Nzk5MjNiZDE5MzMzNmM4Y2M2OWE=

data/ext/nokogumboc/extconf.rb

@@ -28,6 +28,7 @@ unless have_library('gumbo', 'gumbo_parse')
     require 'fileutils'
     FileUtils.cp Dir["#{rakehome}/gumbo-parser/src/*"],
       "#{rakehome}/ext/nokogumboc"
+    $srcs = $objs = nil
   end
 end
 

data/gumbo-parser/src/parser.c

@@ -578,7 +578,7 @@ static GumboInsertionMode get_appropriate_insertion_mode(
 // This performs the actual "reset the insertion mode" loop.
 static void reset_insertion_mode_appropriately(GumboParser* parser) {
   const GumboVector* open_elements = &parser->_parser_state->_open_elements;
-  for (int i = open_elements->length - 1; i >= 0; --i) {
+  for (int i = open_elements->length; --i >= 0; ) {
     GumboInsertionMode mode =
         get_appropriate_insertion_mode(open_elements->data[i], i == 0);
     if (mode != GUMBO_INSERTION_MODE_INITIAL) {
@@ -762,7 +762,7 @@ static void foster_parent_element(GumboParser* parser, GumboNode* node) {
   GumboNode* foster_parent_element = open_elements->data[0];
   assert(foster_parent_element->type == GUMBO_NODE_ELEMENT);
   assert(node_tag_is(foster_parent_element, GUMBO_TAG_HTML));
-  for (int i = open_elements->length - 1; i > 1; --i) {
+  for (int i = open_elements->length; --i > 1; ) {
     GumboNode* table_element = open_elements->data[i];
     if (node_tag_is(table_element, GUMBO_TAG_TABLE)) {
       foster_parent_element = table_element->parent;
@@ -1062,7 +1062,7 @@ static void acknowledge_self_closing_tag(GumboParser* parser) {
 // elements, and fills in its index if so.
 static bool find_last_anchor_index(GumboParser* parser, int* anchor_index) {
   GumboVector* elements = &parser->_parser_state->_active_formatting_elements;
-  for (int i = elements->length - 1; i >= 0; --i) {
+  for (int i = elements->length; --i >= 0; ) {
     GumboNode* node = elements->data[i];
     if (node == &kActiveFormattingScopeMarker) {
       return false;
@@ -1085,7 +1085,7 @@ static int count_formatting_elements_of_tag(
   const GumboElement* desired_element = &desired_node->v.element;
   GumboVector* elements = &parser->_parser_state->_active_formatting_elements;
   int num_identical_elements = 0;
-  for (int i = elements->length - 1; i >= 0; --i) {
+  for (int i = elements->length; --i >= 0; ) {
     GumboNode* node = elements->data[i];
     if (node == &kActiveFormattingScopeMarker) {
       break;
@@ -1289,7 +1289,7 @@ static bool has_an_element_in_specific_scope(
   va_end(args);
 
   bool result = false;
-  for (int i = open_elements->length - 1; i >= 0; --i) {
+  for (int i = open_elements->length; --i >= 0; ) {
     const GumboNode* node = open_elements->data[i];
     if (node->type != GUMBO_NODE_ELEMENT) {
       continue;
@@ -1350,7 +1350,7 @@ static bool has_an_element_in_scope(GumboParser* parser, GumboTag tag) {
 // parameterize it.
 static bool has_node_in_scope(GumboParser* parser, const GumboNode* node) {
   GumboVector* open_elements = &parser->_parser_state->_open_elements;
-  for (int i = open_elements->length - 1; i >= 0; --i) {
+  for (int i = open_elements->length; --i >= 0; ) {
     const GumboNode* current = open_elements->data[i];
     if (current == node) {
       return true;
@@ -1586,7 +1586,7 @@ static void maybe_implicitly_close_list_tag(
     GumboParser* parser, GumboToken* token, bool is_li) {
   GumboParserState* state = parser->_parser_state;
   state->_frameset_ok = false;
-  for (int i = state->_open_elements.length - 1; i >= 0; --i) {
+  for (int i = state->_open_elements.length; --i >= 0; ) {
     const GumboNode* node = state->_open_elements.data[i];
     bool is_list_tag = is_li ?
         node_tag_is(node, GUMBO_TAG_LI) :
@@ -1761,7 +1761,7 @@ static bool adoption_agency_algorithm(
     // Step 4.
     GumboNode* formatting_node = NULL;
     int formatting_node_in_open_elements = -1;
-    for (int j = state->_active_formatting_elements.length - 1; j >= 0; --j) {
+    for (int j = state->_active_formatting_elements.length; --j >= 0; ) {
       GumboNode* current_node = state->_active_formatting_elements.data[j];
       if (current_node == &kActiveFormattingScopeMarker) {
         gumbo_debug("Broke on scope marker; aborting.\n");
@@ -2873,7 +2873,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
     // If we see a), implicitly close everything up to and including it.  If we
     // see b), then record a parse error, don't close anything (except the
     // implied end tags) and ignore the end tag token.
-    for (int i = state->_open_elements.length - 1; ; --i) {
+    for (int i = state->_open_elements.length; --i >= 0; ) {
       const GumboNode* node = state->_open_elements.data[i];
       if (node->v.element.tag_namespace == GUMBO_NAMESPACE_HTML &&
           node_tag_is(node, end_tag)) {
@@ -3686,8 +3686,8 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
       add_parse_error(parser, token);
       is_success = false;
     }
-    int i = parser->_parser_state->_open_elements.length - 1;
-    while (i > 0) {
+    int i = parser->_parser_state->_open_elements.length;
+    for( --i; i > 0; ) {
       // Here we move up the stack until we find an HTML element (in which
       // case we do nothing) or we find the element that we're about to
       // close (in which case we pop everything we've seen until that

data/gumbo-parser/src/utf8.c

@@ -115,7 +115,7 @@ static void read_char(Utf8Iterator* iter) {
   // we want.  If not, we set the iter_truncated flag, mark this as a bad
   // character, and adjust the current width so that it consumes the rest of the
   // iter.
-  int code_point = c & mask;
+  uint64_t code_point = c & mask;
   if (iter->_start + iter->_width > iter->_end) {
     iter->_width = iter->_end - iter->_start;
     add_error(iter, GUMBO_ERR_UTF8_TRUNCATED);
@@ -136,6 +136,7 @@ static void read_char(Utf8Iterator* iter) {
     }
     code_point = (code_point << 6) | (c & ~0x80);
   }
+  if (code_point > 0x10FFFF) is_bad_char = true;
 
   // If we had a decode error, set the current code point to the replacement
   // character and flip the flag indicating that a decode error occurred.

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: nokogumbo
 version: !ruby/object:Gem::Version
-  version: '1.1'
+  version: 1.1.1
 platform: ruby
 authors:
 - Sam Ruby
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-09-12 00:00:00.000000000 Z
+date: 2013-09-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri