Merge pull request #25 from dwd/predicate-buffer-tr

Use buffer_ptr inside predicate skips
dwd · Nov 7, 2024 · 728c809 · 728c809
2 parents 28b12fe + d91229c
commit 728c809
Show file tree

Hide file tree

Showing 3 changed files with 61 additions and 26 deletions.
diff --git a/rapidxml.hpp b/rapidxml.hpp
@@ -1679,18 +1679,17 @@ namespace rapidxml
 
         template<int Flags>
         view_type decode_data_value_low(view_type const & v) {
-            auto * init = v.data();
-            auto * first = init;
+            buffer_ptr first{v};
             if (Flags & parse_normalize_whitespace) {
                 skip<text_pure_with_ws_pred,0>(first);
             } else {
                 skip<text_pure_no_ws_pred,0>(first);
             }
-            if (*first == '<') return v;
-            auto buf = this->allocate_span(v);
+            if (!*first) return v;
+            auto buf = this->allocate_string(v);
             auto * start = buf.data();
-            auto * tmp = start;
-            auto * end = (Flags & parse_normalize_whitespace) ?
+            buffer_ptr tmp{buf};
+            auto end = (Flags & parse_normalize_whitespace) ?
                     skip_and_expand_character_refs<text_pred,text_pure_with_ws_pred,Flags>(tmp) :
                     skip_and_expand_character_refs<text_pred,text_pure_no_ws_pred,Flags>(tmp);
             // Trim trailing whitespace if flag is set; leading was already trimmed by whitespace skip after >
@@ -1715,14 +1714,13 @@ namespace rapidxml
 
         template<Ch Q>
         view_type decode_attr_value_low(view_type const & v) {
-            Ch const * init = v.data();
-            Ch const * first = init;
+            buffer_ptr first{v};
             skip<attribute_value_pure_pred<Q>,0>(first);
-            if (*first == Q) return v;
-            auto buf = this->allocate_span(v);
-            Ch * start = buf.data();
-            Ch * tmp = start;
-            Ch * end = skip_and_expand_character_refs<attribute_value_pred<Q>,attribute_value_pure_pred<Q>,0>(tmp);
+            if (!*first || *first == Q) return v;
+            auto buf = this->allocate_string(v);
+            const Ch * start = buf.data();
+            buffer_ptr tmp{buf};
+            const Ch * end = skip_and_expand_character_refs<attribute_value_pred<Q>,attribute_value_pure_pred<Q>,0>(tmp);
             return {start, end};
         }
 
@@ -1922,24 +1920,24 @@ namespace rapidxml
         // Skip characters until predicate evaluates to true while doing the following:
         // - replacing XML character entity references with proper characters (&apos; &amp; &quot; &lt; &gt; &#...;)
         // - condensing whitespace sequences to single space character
-        template<class StopPred, class StopPredPure, int Flags>
-        static Ch *skip_and_expand_character_refs(Ch *&text)
+        template<class StopPred, class StopPredPure, int Flags, typename Chp>
+        static const Ch *skip_and_expand_character_refs(Chp text)
         {
             // If entity translation, whitespace condense and whitespace trimming is disabled, use plain skip
             if (Flags & parse_no_entity_translation &&
                 !(Flags & parse_normalize_whitespace) &&
                 !(Flags & parse_trim_whitespace))
             {
                 skip<StopPred, Flags>(text);
-                return text;
+                return &*text;
             }
 
             // Use simple skip until first modification is detected
             skip<StopPredPure, Flags>(text);
 
             // Use translation skip
-            Ch *src = text;
-            Ch *dest = src;
+            Chp src = text;
+            Ch * dest = const_cast<Ch *>(&*src);
             while (StopPred::test(*src))
             {
                 // If entity translation is enabled
@@ -2063,9 +2061,7 @@ namespace rapidxml
             }
 
             // Return new end
-            text = src;
             return dest;
-
         }
 
         ///////////////////////////////////////////////////////////////////////

diff --git a/test/low-level-parse.cpp b/test/low-level-parse.cpp
@@ -35,31 +35,31 @@ TEST(PredicateBuffer, Skip) {
 TEST(Predicates, SkipAndExpand) {
     std::string test_data{"&hello;<"};
     char * start = const_cast<char *>(test_data.c_str());
-    start = rapidxml::xml_document<>::skip_and_expand_character_refs<
+    auto end = rapidxml::xml_document<>::skip_and_expand_character_refs<
             rapidxml::xml_document<>::text_pred,
             rapidxml::xml_document<>::text_pure_with_ws_pred,
             rapidxml::parse_no_entity_translation>(start);
-    EXPECT_EQ(*start, '<');
+    EXPECT_EQ(*end, '<');
 }
 
 TEST(Predicates, SkipAndExpandShort) {
     std::string test_data{"&hello;"};
     char * start = const_cast<char *>(test_data.c_str());
-    start = rapidxml::xml_document<>::skip_and_expand_character_refs<
+    auto end = rapidxml::xml_document<>::skip_and_expand_character_refs<
             rapidxml::xml_document<>::text_pred,
             rapidxml::xml_document<>::text_pure_with_ws_pred,
             rapidxml::parse_no_entity_translation>(start);
-    EXPECT_EQ(*start, '\0');
+    EXPECT_EQ(*end, '\0');
 }
 
 TEST(Predicates, SkipAndExpandShorter) {
     std::string test_data{"&hell"};
     char * start = const_cast<char *>(test_data.c_str());
-    start = rapidxml::xml_document<>::skip_and_expand_character_refs<
+    auto end = rapidxml::xml_document<>::skip_and_expand_character_refs<
             rapidxml::xml_document<>::text_pred,
             rapidxml::xml_document<>::text_pure_with_ws_pred,
             rapidxml::parse_no_entity_translation>(start);
-    EXPECT_EQ(*start, '\0');
+    EXPECT_EQ(*end, '\0');
 }
 
 TEST(ParseFns, ParseBom) {

diff --git a/test/parse-simple.cpp b/test/parse-simple.cpp
@@ -264,3 +264,42 @@ TEST(ParseOptions, OpenOnlyFastest) {
         subdoc.validate();
     }
 }
+
+TEST(Parser_Emoji, Single) {
+    std::string foo{"<h>&apos;</h>"};
+    rapidxml::xml_document<> doc;
+    doc.parse<rapidxml::parse_default>(foo);
+    EXPECT_EQ("'", doc.first_node()->value());
+}
+
+TEST(Parser_Emoji, SingleUni) {
+    std::string foo{"<h>&#1234;</h>"};
+    rapidxml::xml_document<> doc;
+    doc.parse<rapidxml::parse_default>(foo);
+    EXPECT_EQ("\xD3\x92", doc.first_node()->value());
+}
+
+TEST(Parser_Emoji, SingleEmoji) {
+    std::string foo{"<h>&#128512;</h>"};
+    rapidxml::xml_document<> doc;
+    doc.parse<rapidxml::parse_default>(foo);
+    EXPECT_EQ("\xF0\x9F\x98\x80", doc.first_node()->value());
+    EXPECT_EQ(4, doc.first_node()->value().size());
+}
+
+TEST(Parser_Emoji, SingleEmojiReuse) {
+    std::string bar("<h>Sir I bear a rhyme excelling in mystic verse and magic spelling &#128512;</h>");
+    rapidxml::xml_document<> doc;
+    rapidxml::xml_document<> parent_doc;
+    parent_doc.parse<rapidxml::parse_default|rapidxml::parse_open_only>("<open>");
+    doc.parse<rapidxml::parse_default>(bar, &parent_doc);
+    EXPECT_EQ("Sir I bear a rhyme excelling in mystic verse and magic spelling \xF0\x9F\x98\x80", doc.first_node()->value());
+    auto doc_a = doc.first_node()->document();
+    doc.first_node()->value(doc_a->allocate_string("Sausages are the loneliest fruit, and are but one of the strange things I have witnessed in my long and interesting life."));
+    EXPECT_EQ("Sausages are the loneliest fruit, and are but one of the strange things I have witnessed in my long and interesting life.", doc.first_node()->value());
+    bar = "<h>&#128512;</h>";
+    doc.parse<rapidxml::parse_default>(bar, &parent_doc);
+    EXPECT_EQ("\xF0\x9F\x98\x80", doc.first_node()->value());
+    EXPECT_EQ(4, doc.first_node()->value().size());
+}
+