Better parsing of <style>, enhanced support of @supports, SVG embedded images fix

crengine/include/lvxml.h

@@ -394,6 +394,7 @@ class LVXMLParser : public LVTextFileBase
     LVXMLParserCallback * m_callback;
     int  m_state;
     bool m_in_cdata;
+    bool m_in_html_style_tag;
     bool m_in_html_script_tag;
     bool m_trimspaces;
     bool SkipSpaces();

crengine/src/lvimg.cpp

@@ -2078,6 +2078,9 @@ static bool lunasvgDrawImageHelper(lunasvg::external_context_t * xcontext, const
     if ( doc ) {
         ldomNode * node = ((LVNodeImageSource *)xcontext->external_object)->GetSourceNode();
         img = doc->getObjectImageSource(Utf8ToUnicode(url), node);
+        if (img.isNull()) { // url may be url-encoded
+            img = doc->getObjectImageSource(DecodeHTMLUrlString(Utf8ToUnicode(url)), node);
+        }
     }
     else {
         // We may be used by frontends without a ldomDocument to render SVG, and

crengine/src/lvstsheet.cpp

@@ -1966,6 +1966,14 @@ class AtRuleLogicalConditionParser {
                 continue;
             }
             if ( *str == ')' ) {
+                if ( level <= 0 ) {
+                    // Closing ')' for a '(' we didn't account for: possibly
+                    // buggy CSS, or we stumbled on some unsupported syntax
+                    // using parens that we didn't parse well
+                    malformed = true;
+                    skip_to_next( str, 0, '{' );
+                    continue;
+                }
                 leaveLevel();
                 str++;
                 continue;
@@ -2000,15 +2008,112 @@ class AtRuleLogicalConditionParser {
 };
 
 // https://drafts.csswg.org/css-conditional/#at-supports
+// https://developer.mozilla.org/en-US/docs/Web/CSS/Reference/At-rules/@supports
 class AtSupportsLogicalConditionParser : public AtRuleLogicalConditionParser {
 public:
     AtSupportsLogicalConditionParser(lxmlDocBase * d)
         : AtRuleLogicalConditionParser(d, '{', 0) {}
 protected:
     virtual void parseCondition(const char * &str) {
-        // Use our regular declaration parser to see if supported
-        LVCssDeclaration tmp_decl;
-        setResult( tmp_decl.parseAndCheckIfSupported( str, doc ) );
+        skip_spaces( str );
+        // We may meet either some declaration: "property: value",
+        // or some function: "selector(a > b)".
+        // Actually, by specs, a declaration must be enclosed in (...),
+        // but a function may, as parens can also be used for grouping with
+        // logical operators. They are handled in AtRuleLogicalConditionParser:parse()
+        // as such grouping.
+        // So, we should try here to guess what kind it is.
+        const char * orig_pos = str;
+        char ident[16];
+        if ( !parse_ident(str, ident, 16) ) {
+            malformed = true;
+            skip_to_next( str, 0, '{' );
+            return;
+        }
+        skip_spaces( str );
+        if ( *str != '(' ) { // It's not a function
+            str = orig_pos;
+            // Use our regular declaration parser to see if supported
+            LVCssDeclaration tmp_decl;
+            setResult( tmp_decl.parseAndCheckIfSupported( str, doc ) );
+            return;
+        }
+        str++; // skip '(' - we'll have to consume the follow up ')' below
+        skip_spaces( str );
+        lString8 name(ident);
+        name.lowercase();
+        setResult(false);
+        if ( name == "selector" ) {
+            // Use our LVCssSelector parser, with for_functional_pseudo_class=true,
+            // which will have it stop on a ')' (but also on a ',' which is not
+            // per specs, and should not happen, so this should do)
+            LVCssSelector tmp_selector;
+            if ( tmp_selector.parse(str, doc, false, true) ) {
+                skip_spaces( str ); // should have stopped at a ')'
+                if ( *str == ')' ) {
+                    str++;
+                    setResult(true);
+                }
+                else {
+                    malformed = true;
+                    skip_to_next( str, 0, '{' );
+                }
+            }
+            else {
+                skip_to_next( str, ')', 0 );
+            }
+        }
+        else if ( name == "font-tech" ) {
+            // We don't support any (even if we do support some OpenType features
+            // with Harfbuzz, we don't support (yet) the CSS property 'font-feature',
+            // which is probably what would be used in a declaration following
+            // such a "@font-tech(features-opentype) {...}"
+            skip_to_next( str, ')', 0 );
+        }
+        else if ( name == "font-format" ) {
+            // Not sure we support "collection" (.otc, .ttc) for embedded fonts
+            if ( substr_icompare( "opentype", str ) ||
+                 substr_icompare( "truetype", str ) ||
+                 substr_icompare( "woff", str ) ||
+                 substr_icompare( "woff2", str ) ) {
+                skip_spaces( str );
+                if ( *str == ')' ) {
+                    str++;
+                    setResult(true);
+                }
+                else {
+                    skip_to_next( str, ')', 0 );
+                }
+            }
+            else {
+                malformed = true;
+                skip_to_next( str, 0, '{' );
+            }
+        }
+        else if ( name == "at-rule" ) {
+            // We only really support a few
+            if ( substr_icompare( "@media", str ) ||
+                 substr_icompare( "@import", str ) ||
+                 substr_icompare( "@font-face", str ) ||
+                 substr_icompare( "@supports", str ) ) {
+                skip_spaces( str );
+                if ( *str == ')' ) {
+                    str++;
+                    setResult(true);
+                }
+                else {
+                    skip_to_next( str, ')', 0 );
+                }
+            }
+            else {
+                malformed = true;
+                skip_to_next( str, 0, '{' );
+            }
+        }
+        else { // Not a known function
+            malformed = true;
+            skip_to_next( str, 0, '{' );
+        }
     }
 };
 

crengine/src/lvtextfm.cpp

@@ -1851,8 +1851,7 @@ class LVFormatter {
             FriBidiLevel newBidiLevel;
         #endif
         #if (USE_HARFBUZZ==1)
-            bool checkIfHarfbuzz = true;
-            bool usingHarfbuzz = false;
+            bool usingHarfbuzz = m_kerning_mode == KERNING_MODE_HARFBUZZ;
             // Unicode script change (note: hb_script_t is uint32_t)
             lUInt32 prevScript = HB_SCRIPT_COMMON;
             hb_unicode_funcs_t* _hb_unicode_funcs = hb_unicode_funcs_get_default();
@@ -1874,15 +1873,6 @@ class LVFormatter {
                 isObject = m_flags[i] & LCHAR_IS_OBJECT; // image, float or inline box
                 newFont = isObject ? NULL : (LVFont *)newSrc->t.font;
                 newLetterSpacing = newSrc->letter_spacing; // 0 for objects
-                #if (USE_HARFBUZZ==1)
-                    // Check if we are using Harfbuzz kerning with the first font met
-                    if ( checkIfHarfbuzz && newFont ) {
-                        if ( m_kerning_mode == KERNING_MODE_HARFBUZZ ) {
-                            usingHarfbuzz = true;
-                        }
-                        checkIfHarfbuzz = false;
-                    }
-                #endif
             }
             if (i > 0)
                 prevCharIsObject = m_flags[i-1] & LCHAR_IS_OBJECT; // image, float or inline box

crengine/src/lvxml.cpp

@@ -2894,6 +2894,7 @@ void LVXMLParser::Reset()
     LVTextFileBase::Reset();
     m_state = ps_bof;
     m_in_cdata = false;
+    m_in_html_style_tag = false;
     m_in_html_script_tag = false;
 }
 
@@ -2902,6 +2903,7 @@ LVXMLParser::LVXMLParser( LVStreamRef stream, LVXMLParserCallback * callback, bo
     , m_callback(callback)
     , m_state(0)
     , m_in_cdata(false)
+    , m_in_html_style_tag(false)
     , m_in_html_script_tag(false)
     , m_trimspaces(true)
     , m_citags(false)
@@ -3106,6 +3108,18 @@ bool LVXMLParser::Parse()
 //                    CRLog::trace("<%s>", LCSTR(tagname) );
                 if (!bodyStarted && tagname == "body")
                     bodyStarted = true;
+                else if ( tagname.length() == 5 &&
+                        ( tagname[0] == U'S' || tagname[0] == U's') &&
+                        ( tagname[1] == U'T' || tagname[1] == U't') &&
+                        ( tagname[2] == U'Y' || tagname[2] == U'y') &&
+                        ( tagname[3] == U'L' || tagname[3] == U'l') &&
+                        ( tagname[4] == U'E' || tagname[4] == U'e') ) {
+                    // Handle content as text, but don't stop just at any '<',
+                    // only at '</style>', as <style> may contain comments
+                    // (which can contain anything, so tags) and even tags
+                    // (eg. "background-image:url('data:image/svg+xml;utf8,<svg><path...")
+                    m_in_html_style_tag = true;
+                }
                 else if ( tagname.length() == 6 &&
                         ( tagname[0] == U'S' || tagname[0] == U's') &&
                         ( tagname[1] == U'C' || tagname[1] == U'c') &&
@@ -3133,6 +3147,8 @@ bool LVXMLParser::Parse()
                     // end of tag
                     if ( ch!='>' ) { // '/' in '<hr/>' : self closing tag
                         m_callback->OnTagClose(tagns.c_str(), tagname.c_str(), true);
+                        if ( m_in_html_style_tag )
+                            m_in_html_style_tag = false;
                         if ( m_in_html_script_tag )
                             m_in_html_script_tag = false;
                     }
@@ -3241,6 +3257,9 @@ bool LVXMLParser::Parse()
                     // regular text after ']]>' until the next '<tag>'
                     m_state = ps_text;
                 }
+                else if ( m_in_html_style_tag ) {
+                    m_in_html_style_tag = false;
+                }
                 else if ( m_in_html_script_tag ) {
                     m_in_html_script_tag = false;
                 }
@@ -5776,7 +5795,27 @@ bool LVXMLParser::ReadText()
                 }
             }
             else if ( ch=='<' ) {
-                if ( m_in_html_script_tag ) { // we're done only when we meet </script>
+                if ( m_in_html_style_tag ) { // we're done only when we meet </style>
+                    if ( ptr + 1 < end ) {
+                        if ( ptr[1] == '/' ) {
+                            if ( ptr + 6 < end ) {
+                                const lChar32 * buf = ptr + 2;
+                                lString32 tag(buf, 5);
+                                if ( tag.lowercase() == U"style" ) {
+                                    nbCharToSkipOnFlgBreak = 1;
+                                    goto end_of_node;
+                                }
+                            }
+                            else if ( !hasNoMoreData ) {
+                                break;
+                            }
+                        }
+                    }
+                    else if ( !hasNoMoreData ) {
+                        break;
+                    }
+                }
+                else if ( m_in_html_script_tag ) { // we're done only when we meet </script>
                     if ( ptr + 1 < end ) {
                         if ( ptr[1] == '/' ) {
                             if ( ptr + 7 < end ) {