preloadscanner.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

                    m_state = BeforeAttributeName;
                    break;
                }
                if (cc >= 'A' && cc <= 'Z')
                    m_tagName.append(cc + 0x20);
                else
                    m_tagName.append(cc);
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case BeforeAttributeName:
            if (isWhitespace(cc))
                ;
            else if (cc == '>') {
                emitTag();
                m_state = Data;
            } else if (cc >= 'A' && cc <= 'Z') {
                m_attributeName.clear();
                m_attributeValue.clear();
                m_attributeName.append(cc + 0x20);
                m_state = AttributeName;
            } else if (cc == '/')
                ;
            else {
                m_attributeName.clear();
                m_attributeValue.clear();
                m_attributeName.append(cc);
                m_state = AttributeName;
            }
            break;
        case AttributeName:
            while (1) {
                if (isWhitespace(cc)) {
                    m_state = AfterAttributeName;
                    break;
                }
                if (cc == '=') {
                    m_state = BeforeAttributeValue;
                    break;
                }
                if (cc == '>') {
                    emitTag();
                    m_state = Data;
                    break;
                } 
                if (cc == '/') {
                    m_state = BeforeAttributeName;
                    break;
                }
                if (cc >= 'A' && cc <= 'Z')
                    m_attributeName.append(cc + 0x20);
                else
                    m_attributeName.append(cc);
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case AfterAttributeName:
            if (isWhitespace(cc))
                ;
            else if (cc == '=')
                m_state = BeforeAttributeValue; 
            else if (cc == '>') {
                emitTag();
                m_state = Data;
            } else if (cc >= 'A' && cc <= 'Z') {
                m_attributeName.clear();
                m_attributeValue.clear();
                m_attributeName.append(cc + 0x20);
                m_state = AttributeName;
            } else if (cc == '/')
                m_state = BeforeAttributeName;
            else {
                m_attributeName.clear();
                m_attributeValue.clear();
                m_attributeName.append(cc);
                m_state = AttributeName;
            }
            break;
        case BeforeAttributeValue:
            if (isWhitespace(cc))
                ;
            else if (cc == '"')
                m_state = AttributeValueDoubleQuoted;
            else if (cc == '&') {
                m_state = AttributeValueUnquoted;
                continue;
            } else if (cc == '\'')
                m_state = AttributeValueSingleQuoted;
            else if (cc == '>') {
                emitTag();
                m_state = Data;
            } else {
                m_attributeValue.append(cc);
                m_state = AttributeValueUnquoted;
            }
            break;
        case AttributeValueDoubleQuoted:
            while (1) {
                if (cc == '"') {
                    processAttribute();
                    m_state = BeforeAttributeName;
                    break;
                }
                if (cc == '&') {
                    m_stateBeforeEntityInAttributeValue = m_state;
                    m_state = EntityInAttributeValue;
                    break;
                } 
                m_attributeValue.append(cc);
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case AttributeValueSingleQuoted:
            while (1) {
                if (cc == '\'') {
                    processAttribute();
                    m_state = BeforeAttributeName;
                    break;
                }
                if (cc == '&') {
                    m_stateBeforeEntityInAttributeValue = m_state;
                    m_state = EntityInAttributeValue;
                    break;
                } 
                m_attributeValue.append(cc);
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case AttributeValueUnquoted:
            while (1) {
                if (isWhitespace(cc)) {
                    processAttribute();
                    m_state = BeforeAttributeName;
                    break;
                }
                if (cc == '&') {
                    m_stateBeforeEntityInAttributeValue = m_state;
                    m_state = EntityInAttributeValue;
                    break;
                }
                if (cc == '>') {
                    processAttribute();
                    emitTag();
                    m_state = Data;
                    break;
                }
                m_attributeValue.append(cc);
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case EntityInAttributeValue: 
            {
                bool notEnoughCharacters = false; 
                unsigned entity = consumeEntity(m_source, notEnoughCharacters);
                if (notEnoughCharacters)
                    return;
                if (entity > 0xFFFF) {
                    m_attributeValue.append(U16_LEAD(entity));
                    m_attributeValue.append(U16_TRAIL(entity));
                } else if (entity)
                    m_attributeValue.append(entity);
                else
                    m_attributeValue.append('&');
            }
            m_state = m_stateBeforeEntityInAttributeValue;
            continue;
        case BogusComment:
            while (1) {
                if (cc == '>') {
                    m_state = Data;
                    break;
                }
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case MarkupDeclarationOpen: {
            if (cc == '-') {
                if (m_source.length() < 2)
                    return;
                m_source.advance();
                cc = *m_source;
                if (cc == '-')
                    m_state = CommentStart;
                else {
                    m_state = BogusComment;
                    continue;
                }
            // If we cared about the DOCTYPE we would test to enter those states here
            } else {
                m_state = BogusComment;
                continue;
            }
            break;
        }
        case CommentStart:
            if (cc == '-')
                m_state = CommentStartDash;
            else if (cc == '>')
                m_state = Data;
            else
                m_state = Comment;
            break;
        case CommentStartDash:
            if (cc == '-')
                m_state = CommentEnd;
            else if (cc == '>')
                m_state = Data;
            else
                m_state = Comment;
            break;
        case Comment:
            while (1) {
                if (cc == '-') {
                    m_state = CommentEndDash;
                    break;
                }
                m_source.advance();
                if (m_source.isEmpty())
                    return;
                cc = *m_source;
            }
            break;
        case CommentEndDash:
            if (cc == '-')
                m_state = CommentEnd;
            else 
                m_state = Comment;
            break;
        case CommentEnd:
            if (cc == '>')
                m_state = Data;
            else if (cc == '-')
                ;
            else 
                m_state = Comment;
            break;
        }
        m_source.advance();
    }
}
    
void PreloadScanner::processAttribute()
{
    AtomicString tag = AtomicString(m_tagName.data(), m_tagName.size());
    AtomicString attribute = AtomicString(m_attributeName.data(), m_attributeName.size());
    
    String value(m_attributeValue.data(), m_attributeValue.size());
    if (tag == scriptTag || tag == imgTag) {
        if (attribute == srcAttr && m_urlToLoad.isEmpty())
            m_urlToLoad = parseURL(value);
        else if (attribute == charsetAttr)
            m_charset = value;
    } else if (tag == linkTag) {
        if (attribute == hrefAttr && m_urlToLoad.isEmpty())
            m_urlToLoad = parseURL(value);
        else if (attribute == relAttr) {
            bool styleSheet = false;
            bool alternate = false;
            bool icon = false;
            bool dnsPrefetch = false;
            HTMLLinkElement::tokenizeRelAttribute(value, styleSheet, alternate, icon, dnsPrefetch);
            m_linkIsStyleSheet = styleSheet && !alternate && !icon && !dnsPrefetch;
        } else if (attribute == charsetAttr)
            m_charset = value;
    }
}
    
inline void PreloadScanner::emitCharacter(UChar c)
{
    if (m_contentModel == CDATA && m_lastStartTag == styleTag) 
        tokenizeCSS(c);
}
    
inline void PreloadScanner::tokenizeCSS(UChar c)
{    
    // We are just interested in @import rules, no need for real tokenization here
    // Searching for other types of resources is probably low payoff
    switch (m_cssState) {
    case CSSInitial:
        if (c == '@')
            m_cssState = CSSRuleStart;
        else if (c == '/')
            m_cssState = CSSMaybeComment;
        break;
    case CSSMaybeComment:
        if (c == '*')
            m_cssState = CSSComment;
        else
            m_cssState = CSSInitial;
        break;
    case CSSComment:
        if (c == '*')
            m_cssState = CSSMaybeCommentEnd;
        break;
    case CSSMaybeCommentEnd:
        if (c == '/')
            m_cssState = CSSInitial;
        else if (c == '*')
            ;
        else
            m_cssState = CSSComment;
        break;
    case CSSRuleStart:
        if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z')) {
            m_cssRule.clear();
            m_cssRuleValue.clear();
            m_cssRule.append(c);
            m_cssState = CSSRule;
        } else
            m_cssState = CSSInitial;
        break;
    case CSSRule:
        if (isWhitespace(c))
            m_cssState = CSSAfterRule;
        else if (c == ';')
            m_cssState = CSSInitial;
        else
            m_cssRule.append(c);
        break;
    case CSSAfterRule:
        if (isWhitespace(c))
            ;
        else if (c == ';')
            m_cssState = CSSInitial;
        else {
            m_cssState = CSSRuleValue;
            m_cssRuleValue.append(c);
        }
        break;
    case CSSRuleValue:
        if (isWhitespace(c))
            m_cssState = CSSAfterRuleValue;
        else if (c == ';') {
            emitCSSRule();
            m_cssState = CSSInitial;
        } else 
            m_cssRuleValue.append(c);
        break;
    case CSSAfterRuleValue:
        if (isWhitespace(c))
            ;
        else if (c == ';') {
            emitCSSRule();
            m_cssState = CSSInitial;
        } else {
            // FIXME media rules
             m_cssState = CSSInitial;
        }
        break;
    }
}
    
void PreloadScanner::emitTag()
{
    if (m_closeTag) {
        m_contentModel = PCDATA;
        m_cssState = CSSInitial;
        clearLastCharacters();
        return;
    }
    
    AtomicString tag(m_tagName.data(), m_tagName.size());
    m_lastStartTag = tag;
    
    if (tag == textareaTag || tag == titleTag)
        m_contentModel = RCDATA;
    else if (tag == styleTag || tag == xmpTag || tag == scriptTag || tag == iframeTag || tag == noembedTag || tag == noframesTag)
        m_contentModel = CDATA;
    else if (tag == noscriptTag)
        // we wouldn't be here if scripts were disabled
        m_contentModel = CDATA;
    else if (tag == plaintextTag)
        m_contentModel = PLAINTEXT;
    else
        m_contentModel = PCDATA;
    
    if (tag == bodyTag)
        m_bodySeen = true;
    
    if (m_urlToLoad.isEmpty()) {
        m_linkIsStyleSheet = false;
        return;
    }
    
    if (tag == scriptTag)
        m_document->docLoader()->preload(CachedResource::Script, m_urlToLoad, m_charset, scanningBody());
    else if (tag == imgTag) 
        m_document->docLoader()->preload(CachedResource::ImageResource, m_urlToLoad, String(), scanningBody());
    else if (tag == linkTag && m_linkIsStyleSheet) 
        m_document->docLoader()->preload(CachedResource::CSSStyleSheet, m_urlToLoad, m_charset, scanningBody());

    m_urlToLoad = String();
    m_charset = String();
    m_linkIsStyleSheet = false;
}
    
void PreloadScanner::emitCSSRule()
{
    String rule(m_cssRule.data(), m_cssRule.size());
    if (equalIgnoringCase(rule, "import") && !m_cssRuleValue.isEmpty()) {
        String value(m_cssRuleValue.data(), m_cssRuleValue.size());
        String url = parseURL(value);
        if (!url.isEmpty())
            m_document->docLoader()->preload(CachedResource::CSSStyleSheet, url, String(), scanningBody());
    }
    m_cssRule.clear();
    m_cssRuleValue.clear();
}
                
}