xml解析器获取过程

在上篇文章我们知道ResourceTypes.h中定义了ResXMLParser类但没有实现其方法，在头文件中声明，在具体的cpp文件中定义。它的实现在ResourceTypes.cpp中：

ResXMLParser::ResXMLParser(const ResXMLTree& tree)
    : mTree(tree), mEventCode(BAD_DOCUMENT)
{
}

void ResXMLParser::restart()
{
    mCurNode = NULL;
    mEventCode = mTree.mError == NO_ERROR ? START_DOCUMENT : BAD_DOCUMENT;
}
const ResStringPool& ResXMLParser::getStrings() const
{
    return mTree.mStrings;
}

ResXMLParser::event_code_t ResXMLParser::getEventType() const
{
    return mEventCode;
}

ResXMLParser::event_code_t ResXMLParser::next()
{
    if (mEventCode == START_DOCUMENT) {
        mCurNode = mTree.mRootNode;
        mCurExt = mTree.mRootExt;
        return (mEventCode=mTree.mRootCode);
    } else if (mEventCode >= FIRST_CHUNK_CODE) {
        return nextNode();
    }
    return mEventCode;
}

int32_t ResXMLParser::getCommentID() const
{
    return mCurNode != NULL ? dtohl(mCurNode->comment.index) : -1;
}

const char16_t* ResXMLParser::getComment(size_t* outLen) const
{
    int32_t id = getCommentID();
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

uint32_t ResXMLParser::getLineNumber() const
{
    return mCurNode != NULL ? dtohl(mCurNode->lineNumber) : -1;
}

int32_t ResXMLParser::getTextID() const
{
    if (mEventCode == TEXT) {
        return dtohl(((const ResXMLTree_cdataExt*)mCurExt)->data.index);
    }
    return -1;
}

const char16_t* ResXMLParser::getText(size_t* outLen) const
{
    int32_t id = getTextID();
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

ssize_t ResXMLParser::getTextValue(Res_value* outValue) const
{
    if (mEventCode == TEXT) {
        outValue->copyFrom_dtoh(((const ResXMLTree_cdataExt*)mCurExt)->typedData);
        return sizeof(Res_value);
    }
    return BAD_TYPE;
}

int32_t ResXMLParser::getNamespacePrefixID() const
{
    if (mEventCode == START_NAMESPACE || mEventCode == END_NAMESPACE) {
        return dtohl(((const ResXMLTree_namespaceExt*)mCurExt)->prefix.index);
    }
    return -1;
}

const char16_t* ResXMLParser::getNamespacePrefix(size_t* outLen) const
{
    int32_t id = getNamespacePrefixID();
    //printf("prefix=%d  event=%p\n", id, mEventCode);
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

int32_t ResXMLParser::getNamespaceUriID() const
{
    if (mEventCode == START_NAMESPACE || mEventCode == END_NAMESPACE) {
        return dtohl(((const ResXMLTree_namespaceExt*)mCurExt)->uri.index);
    }
    return -1;
}

const char16_t* ResXMLParser::getNamespaceUri(size_t* outLen) const
{
    int32_t id = getNamespaceUriID();
    //printf("uri=%d  event=%p\n", id, mEventCode);
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

int32_t ResXMLParser::getElementNamespaceID() const
{
    if (mEventCode == START_TAG) {
        return dtohl(((const ResXMLTree_attrExt*)mCurExt)->ns.index);
    }
    if (mEventCode == END_TAG) {
        return dtohl(((const ResXMLTree_endElementExt*)mCurExt)->ns.index);
    }
    return -1;
}

const char16_t* ResXMLParser::getElementNamespace(size_t* outLen) const
{
    int32_t id = getElementNamespaceID();
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

int32_t ResXMLParser::getElementNameID() const
{
    if (mEventCode == START_TAG) {
        return dtohl(((const ResXMLTree_attrExt*)mCurExt)->name.index);
    }
    if (mEventCode == END_TAG) {
        return dtohl(((const ResXMLTree_endElementExt*)mCurExt)->name.index);
    }
    return -1;
}

const char16_t* ResXMLParser::getElementName(size_t* outLen) const
{
    int32_t id = getElementNameID();
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

size_t ResXMLParser::getAttributeCount() const
{
    if (mEventCode == START_TAG) {
        return dtohs(((const ResXMLTree_attrExt*)mCurExt)->attributeCount);
    }
    return 0;
}

int32_t ResXMLParser::getAttributeNamespaceID(size_t idx) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            return dtohl(attr->ns.index);
        }
    }
    return -2;
}

const char16_t* ResXMLParser::getAttributeNamespace(size_t idx, size_t* outLen) const
{
    int32_t id = getAttributeNamespaceID(idx);
    //printf("attribute namespace=%d  idx=%d  event=%p\n", id, idx, mEventCode);
    if (kDebugXMLNoisy) {
        printf("getAttributeNamespace 0x%zx=0x%x\n", idx, id);
    }
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

const char* ResXMLParser::getAttributeNamespace8(size_t idx, size_t* outLen) const
{
    int32_t id = getAttributeNamespaceID(idx);
    //printf("attribute namespace=%d  idx=%d  event=%p\n", id, idx, mEventCode);
    if (kDebugXMLNoisy) {
        printf("getAttributeNamespace 0x%zx=0x%x\n", idx, id);
    }
    return id >= 0 ? mTree.mStrings.string8At(id, outLen) : NULL;
}

int32_t ResXMLParser::getAttributeNameID(size_t idx) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            return dtohl(attr->name.index);
        }
    }
    return -1;
}

const char16_t* ResXMLParser::getAttributeName(size_t idx, size_t* outLen) const
{
    int32_t id = getAttributeNameID(idx);
    //printf("attribute name=%d  idx=%d  event=%p\n", id, idx, mEventCode);
    if (kDebugXMLNoisy) {
        printf("getAttributeName 0x%zx=0x%x\n", idx, id);
    }
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

const char* ResXMLParser::getAttributeName8(size_t idx, size_t* outLen) const
{
    int32_t id = getAttributeNameID(idx);
    //printf("attribute name=%d  idx=%d  event=%p\n", id, idx, mEventCode);
    if (kDebugXMLNoisy) {
        printf("getAttributeName 0x%zx=0x%x\n", idx, id);
    }
    return id >= 0 ? mTree.mStrings.string8At(id, outLen) : NULL;
}

uint32_t ResXMLParser::getAttributeNameResID(size_t idx) const
{
    int32_t id = getAttributeNameID(idx);
    if (id >= 0 && (size_t)id < mTree.mNumResIds) {
        uint32_t resId = dtohl(mTree.mResIds[id]);
        if (mTree.mDynamicRefTable != NULL) {
            mTree.mDynamicRefTable->lookupResourceId(&resId);
        }
        return resId;
    }
    return 0;
}

int32_t ResXMLParser::getAttributeValueStringID(size_t idx) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            return dtohl(attr->rawValue.index);
        }
    }
    return -1;
}

const char16_t* ResXMLParser::getAttributeStringValue(size_t idx, size_t* outLen) const
{
    int32_t id = getAttributeValueStringID(idx);
    if (kDebugXMLNoisy) {
        printf("getAttributeValue 0x%zx=0x%x\n", idx, id);
    }
    return id >= 0 ? mTree.mStrings.stringAt(id, outLen) : NULL;
}

int32_t ResXMLParser::getAttributeDataType(size_t idx) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            uint8_t type = attr->typedValue.dataType;
            if (type != Res_value::TYPE_DYNAMIC_REFERENCE) {
                return type;
            }

            // This is a dynamic reference. We adjust those references
            // to regular references at this level, so lie to the caller.
            return Res_value::TYPE_REFERENCE;
        }
    }
    return Res_value::TYPE_NULL;
}

int32_t ResXMLParser::getAttributeData(size_t idx) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            if (attr->typedValue.dataType != Res_value::TYPE_DYNAMIC_REFERENCE ||
                    mTree.mDynamicRefTable == NULL) {
                return dtohl(attr->typedValue.data);
            }

            uint32_t data = dtohl(attr->typedValue.data);
            if (mTree.mDynamicRefTable->lookupResourceId(&data) == NO_ERROR) {
                return data;
            }
        }
    }
    return 0;
}

ssize_t ResXMLParser::getAttributeValue(size_t idx, Res_value* outValue) const
{
    if (mEventCode == START_TAG) {
        const ResXMLTree_attrExt* tag = (const ResXMLTree_attrExt*)mCurExt;
        if (idx < dtohs(tag->attributeCount)) {
            const ResXMLTree_attribute* attr = (const ResXMLTree_attribute*)
                (((const uint8_t*)tag)
                 + dtohs(tag->attributeStart)
                 + (dtohs(tag->attributeSize)*idx));
            outValue->copyFrom_dtoh(attr->typedValue);
            if (mTree.mDynamicRefTable != NULL &&
                    mTree.mDynamicRefTable->lookupResourceValue(outValue) != NO_ERROR) {
                return BAD_TYPE;
            }
            return sizeof(Res_value);
        }
    }
    return BAD_TYPE;
}

ssize_t ResXMLParser::indexOfAttribute(const char* ns, const char* attr) const
{
    String16 nsStr(ns != NULL ? ns : "");
    String16 attrStr(attr);
    return indexOfAttribute(ns ? nsStr.string() : NULL, ns ? nsStr.size() : 0,
                            attrStr.string(), attrStr.size());
}

ssize_t ResXMLParser::indexOfAttribute(const char16_t* ns, size_t nsLen,
                                       const char16_t* attr, size_t attrLen) const
{
    if (mEventCode == START_TAG) {
        if (attr == NULL) {
            return NAME_NOT_FOUND;
        }
        const size_t N = getAttributeCount();
        if (mTree.mStrings.isUTF8()) {
            String8 ns8, attr8;
            if (ns != NULL) {
                ns8 = String8(ns, nsLen);
            }
            attr8 = String8(attr, attrLen);
            if (kDebugStringPoolNoisy) {
                ALOGI("indexOfAttribute UTF8 %s (%zu) / %s (%zu)", ns8.string(), nsLen,
                        attr8.string(), attrLen);
            }
            for (size_t i=0; i<N; i++) {
                size_t curNsLen = 0, curAttrLen = 0;
                const char* curNs = getAttributeNamespace8(i, &curNsLen);
                const char* curAttr = getAttributeName8(i, &curAttrLen);
                if (kDebugStringPoolNoisy) {
                    ALOGI("  curNs=%s (%zu), curAttr=%s (%zu)", curNs, curNsLen, curAttr, curAttrLen);
                }
                if (curAttr != NULL && curNsLen == nsLen && curAttrLen == attrLen
                        && memcmp(attr8.string(), curAttr, attrLen) == 0) {
                    if (ns == NULL) {
                        if (curNs == NULL) {
                            if (kDebugStringPoolNoisy) {
                                ALOGI("  FOUND!");
                            }
                            return i;
                        }
                    } else if (curNs != NULL) {
                        //printf(" --> ns=%s, curNs=%s\n",
                        //       String8(ns).string(), String8(curNs).string());
                        if (memcmp(ns8.string(), curNs, nsLen) == 0) {
                            if (kDebugStringPoolNoisy) {
                                ALOGI("  FOUND!");
                            }
                            return i;
                        }
                    }
                }
            }
        } else {
            if (kDebugStringPoolNoisy) {
                ALOGI("indexOfAttribute UTF16 %s (%zu) / %s (%zu)",
                        String8(ns, nsLen).string(), nsLen,
                        String8(attr, attrLen).string(), attrLen);
            }
            for (size_t i=0; i<N; i++) {
                size_t curNsLen = 0, curAttrLen = 0;
                const char16_t* curNs = getAttributeNamespace(i, &curNsLen);
                const char16_t* curAttr = getAttributeName(i, &curAttrLen);
                if (kDebugStringPoolNoisy) {
                    ALOGI("  curNs=%s (%zu), curAttr=%s (%zu)",
                            String8(curNs, curNsLen).string(), curNsLen,
                            String8(curAttr, curAttrLen).string(), curAttrLen);
                }
                if (curAttr != NULL && curNsLen == nsLen && curAttrLen == attrLen
                        && (memcmp(attr, curAttr, attrLen*sizeof(char16_t)) == 0)) {
                    if (ns == NULL) {
                        if (curNs == NULL) {
                            if (kDebugStringPoolNoisy) {
                                ALOGI("  FOUND!");
                            }
                            return i;
                        }
                    } else if (curNs != NULL) {
                        //printf(" --> ns=%s, curNs=%s\n",
                        //       String8(ns).string(), String8(curNs).string());
                        if (memcmp(ns, curNs, nsLen*sizeof(char16_t)) == 0) {
                            if (kDebugStringPoolNoisy) {
                                ALOGI("  FOUND!");
                            }
                            return i;
                        }
                    }
                }
            }
        }
    }

    return NAME_NOT_FOUND;
}

ssize_t ResXMLParser::indexOfID() const
{
    if (mEventCode == START_TAG) {
        const ssize_t idx = dtohs(((const ResXMLTree_attrExt*)mCurExt)->idIndex);
        if (idx > 0) return (idx-1);
    }
    return NAME_NOT_FOUND;
}

ssize_t ResXMLParser::indexOfClass() const
{
    if (mEventCode == START_TAG) {
        const ssize_t idx = dtohs(((const ResXMLTree_attrExt*)mCurExt)->classIndex);
        if (idx > 0) return (idx-1);
    }
    return NAME_NOT_FOUND;
}

ssize_t ResXMLParser::indexOfStyle() const
{
    if (mEventCode == START_TAG) {
        const ssize_t idx = dtohs(((const ResXMLTree_attrExt*)mCurExt)->styleIndex);
        if (idx > 0) return (idx-1);
    }
    return NAME_NOT_FOUND;
}

ResXMLParser::event_code_t ResXMLParser::nextNode()
{
    if (mEventCode < 0) {
        return mEventCode;
    }

    do {
        const ResXMLTree_node* next = (const ResXMLTree_node*)
            (((const uint8_t*)mCurNode) + dtohl(mCurNode->header.size));
        if (kDebugXMLNoisy) {
            ALOGI("Next node: prev=%p, next=%p\n", mCurNode, next);
        }

        if (((const uint8_t*)next) >= mTree.mDataEnd) {
            mCurNode = NULL;
            return (mEventCode=END_DOCUMENT);
        }

        if (mTree.validateNode(next) != NO_ERROR) {
            mCurNode = NULL;
            return (mEventCode=BAD_DOCUMENT);
        }

        mCurNode = next;
        const uint16_t headerSize = dtohs(next->header.headerSize);
        const uint32_t totalSize = dtohl(next->header.size);
        mCurExt = ((const uint8_t*)next) + headerSize;
        size_t minExtSize = 0;
        event_code_t eventCode = (event_code_t)dtohs(next->header.type);
        switch ((mEventCode=eventCode)) {
            case RES_XML_START_NAMESPACE_TYPE:
            case RES_XML_END_NAMESPACE_TYPE:
                minExtSize = sizeof(ResXMLTree_namespaceExt);
                break;
            case RES_XML_START_ELEMENT_TYPE:
                minExtSize = sizeof(ResXMLTree_attrExt);
                break;
            case RES_XML_END_ELEMENT_TYPE:
                minExtSize = sizeof(ResXMLTree_endElementExt);
                break;
            case RES_XML_CDATA_TYPE:
                minExtSize = sizeof(ResXMLTree_cdataExt);
                break;
            default:
                ALOGW("Unknown XML block: header type %d in node at %d\n",
                     (int)dtohs(next->header.type),
                     (int)(((const uint8_t*)next)-((const uint8_t*)mTree.mHeader)));
                continue;
        }

        if ((totalSize-headerSize) < minExtSize) {
            ALOGW("Bad XML block: header type 0x%x in node at 0x%x has size %d, need %d\n",
                 (int)dtohs(next->header.type),
                 (int)(((const uint8_t*)next)-((const uint8_t*)mTree.mHeader)),
                 (int)(totalSize-headerSize), (int)minExtSize);
            return (mEventCode=BAD_DOCUMENT);
        }

        //printf("CurNode=%p, CurExt=%p, headerSize=%d, minExtSize=%d\n",
        //       mCurNode, mCurExt, headerSize, minExtSize);

        return eventCode;
    } while (true);
}

void ResXMLParser::getPosition(ResXMLParser::ResXMLPosition* pos) const
{
    pos->eventCode = mEventCode;
    pos->curNode = mCurNode;
    pos->curExt = mCurExt;
}

void ResXMLParser::setPosition(const ResXMLParser::ResXMLPosition& pos)
{
    mEventCode = pos.eventCode;
    mCurNode = pos.curNode;
    mCurExt = pos.curExt;
}

这些操作都是对ResXMLTree进行的，那么ResXMLTree的实例是什么时候生成的呢？以XmlResourceParser为例来看看:

@NonNull
XmlResourceParser loadXmlResourceParser(@NonNull String file, @AnyRes int id, int assetCookie,
        @NonNull String type)
        throws NotFoundException {
    if (id != 0) {
        try {
            synchronized (mCachedXmlBlocks) {
                final int[] cachedXmlBlockCookies = mCachedXmlBlockCookies;
                final String[] cachedXmlBlockFiles = mCachedXmlBlockFiles;
                final XmlBlock[] cachedXmlBlocks = mCachedXmlBlocks;
                // First see if this block is in our cache.
                final int num = cachedXmlBlockFiles.length;
                for (int i = 0; i < num; i++) {
                    if (cachedXmlBlockCookies[i] == assetCookie && cachedXmlBlockFiles[i] != null
                            && cachedXmlBlockFiles[i].equals(file)) {
                        return cachedXmlBlocks[i].newParser();
                    }
                }

                // Not in the cache, create a new block and put it at
                // the next slot in the cache.
                final XmlBlock block = mAssets.openXmlBlockAsset(assetCookie, file);
                if (block != null) {
                    final int pos = (mLastCachedXmlBlockIndex + 1) % num;
                    mLastCachedXmlBlockIndex = pos;
                    final XmlBlock oldBlock = cachedXmlBlocks[pos];
                    if (oldBlock != null) {
                        oldBlock.close();
                    }
                    cachedXmlBlockCookies[pos] = assetCookie;
                    cachedXmlBlockFiles[pos] = file;
                    cachedXmlBlocks[pos] = block;
                    return block.newParser();
                }
            }
        } catch (Exception e) {
            final NotFoundException rnf = new NotFoundException("File " + file
                    + " from xml type " + type + " resource ID #0x" + Integer.toHexString(id));
            rnf.initCause(e);
            throw rnf;
        }
    }

    throw new NotFoundException("File " + file + " from xml type " + type + " resource ID #0x"
            + Integer.toHexString(id));
}

先由 final XmlBlock block = mAssets.openXmlBlockAsset(assetCookie, file)，然后返回block.newParser()，先来看看newParser方法中：

public XmlResourceParser newParser() {
    synchronized (this) {
        if (mNative != 0) {
            return new Parser(nativeCreateParseState(mNative), this);
        }
        return null;
    }
}

private static final native long nativeCreateParseState(long obj);

这个mNative是什么呢？很有可能是c层的XmlBlock实例的指针，在android_util_XmlBlock.cpp中：

static jlong android_content_XmlBlock_nativeCreateParseState(JNIEnv* env, jobject clazz,
                                                          jlong token)
{
    ResXMLTree* osb = reinterpret_cast<ResXMLTree*>(token);
    if (osb == NULL) {
        jniThrowNullPointerException(env, NULL);
        return 0;
    }

    ResXMLParser* st = new ResXMLParser(*osb);
    if (st == NULL) {
        jniThrowException(env, "java/lang/OutOfMemoryError", NULL);
        return 0;
    }

    st->restart();

    return reinterpret_cast<jlong>(st);
}

可以看到传入的值为jlong token,是ResXMLTree的指针，也就是说，在这之前，ResXMLTree的实例已经存了，并且指针保留在mNative中，所以我们来看看mAssets.openXmlBlockAsset(assetCookie, file)中有什么事情发生：

/*package*/ final XmlBlock openXmlBlockAsset(int cookie, String fileName)
    throws IOException {
    synchronized (this) {
        if (!mOpen) {
            throw new RuntimeException("Assetmanager has been closed");
        }
        long xmlBlock = openXmlAssetNative(cookie, fileName);
        if (xmlBlock != 0) {
            XmlBlock res = new XmlBlock(this, xmlBlock);
            incRefsLocked(res.hashCode());
            return res;
        }
    }
    throw new FileNotFoundException("Asset XML file: " + fileName);
}

private native final long openXmlAssetNative(int cookie, String fileName);

在android_util_AssetManager.cpp中

static jlong android_content_AssetManager_openXmlAssetNative(JNIEnv* env, jobject clazz,
                                                         jint cookie,
                                                         jstring fileName)
{
    AssetManager* am = assetManagerForJavaObject(env, clazz);
    if (am == NULL) {
        return 0;
    }

    ALOGV("openXmlAsset in %p (Java object %p)\n", am, clazz);

    ScopedUtfChars fileName8(env, fileName);
    if (fileName8.c_str() == NULL) {
        return 0;
    }

    int32_t assetCookie = static_cast<int32_t>(cookie);
    Asset* a = assetCookie
        ? am->openNonAsset(assetCookie, fileName8.c_str(), Asset::ACCESS_BUFFER)
        : am->openNonAsset(fileName8.c_str(), Asset::ACCESS_BUFFER, &assetCookie);

    if (a == NULL) {
        jniThrowException(env, "java/io/FileNotFoundException", fileName8.c_str());
        return 0;
    }

    const DynamicRefTable* dynamicRefTable =
            am->getResources().getDynamicRefTableForCookie(assetCookie);
    ResXMLTree* block = new ResXMLTree(dynamicRefTable);
    status_t err = block->setTo(a->getBuffer(true), a->getLength(), true);
    a->close();
    delete a;

    if (err != NO_ERROR) {
        jniThrowException(env, "java/io/FileNotFoundException", "Corrupt XML binary file");
        return 0;
    }

    return reinterpret_cast<jlong>(block);
}

到这返回了block指针。