data: performance optimization NameHierarchy
* For performance reasons nodes now store a serialized version of their name hierarchy directly. Removed NameHierarchy table from database.
This commit is contained in:
malte_langkabel
+32
-118
@@ -100,9 +100,9 @@ void Storage::clearFileElements(const std::vector<FilePath>& filePaths)
|
||||
}
|
||||
}
|
||||
|
||||
void Storage::removeUnusedNames()
|
||||
void Storage::removeUnusedNames() // maybe rename this function. look for callers first.
|
||||
{
|
||||
m_sqliteStorage.removeUnusedNameHierarchyElements();
|
||||
// m_sqliteStorage.removeUnusedNameHierarchyElements();
|
||||
|
||||
clearCaches();
|
||||
}
|
||||
@@ -145,7 +145,6 @@ void Storage::logStats() const
|
||||
ss << "\t" << m_tokenIndex.getCharCount() << " Characters\n";
|
||||
ss << "\t" << m_tokenIndex.getWordCount() << " Words\n";
|
||||
ss << "\t" << m_tokenIndex.getNodeCount() << " SearchNodes\n";
|
||||
ss << "\t" << m_sqliteStorage.getNameHierarchyElementCount() << " NameHierarchyElements\n";
|
||||
|
||||
ss << "\nCode:\n";
|
||||
ss << "\t" << m_sqliteStorage.getFileCount() << " Files\n";
|
||||
@@ -685,27 +684,19 @@ Id Storage::onFileParsed(const FileInfo& fileInfo)
|
||||
{
|
||||
log("file", fileInfo.path.str(), ParseLocation());
|
||||
|
||||
const std::string fileName = fileInfo.path.fileName();
|
||||
|
||||
Id nameHierarchyElementId = m_sqliteStorage.getNameHierarchyElementIdByName(fileName);
|
||||
if (nameHierarchyElementId == 0)
|
||||
{
|
||||
nameHierarchyElementId = m_sqliteStorage.addNameHierarchyElement(fileName);
|
||||
}
|
||||
|
||||
Id fileNodeId = getFileNodeId(fileInfo.path);
|
||||
if (fileNodeId == 0)
|
||||
{
|
||||
NameHierarchy nameHierarchy;
|
||||
nameHierarchy.push(std::make_shared<NameElement>(fileInfo.path.fileName()));
|
||||
|
||||
fileNodeId = m_sqliteStorage.addFile(
|
||||
nameHierarchyElementId,
|
||||
NameHierarchy::serialize(nameHierarchy),
|
||||
fileInfo.path.str(),
|
||||
utility::timeToString(fileInfo.lastWriteTime)
|
||||
);
|
||||
}
|
||||
|
||||
NameHierarchy nameHierarchy;
|
||||
nameHierarchy.push(std::make_shared<NameElement>(fileName));
|
||||
|
||||
return fileNodeId;
|
||||
}
|
||||
|
||||
@@ -765,24 +756,13 @@ Id Storage::onCommentParsed(const ParseLocation& location)
|
||||
|
||||
Id Storage::getIdForNodeWithNameHierarchy(const NameHierarchy& nameHierarchy) const
|
||||
{
|
||||
Id currentId = 0;
|
||||
|
||||
for (size_t i = 0; i < nameHierarchy.size(); i++)
|
||||
{
|
||||
currentId = m_sqliteStorage.getNameHierarchyElementIdByName(nameHierarchy[i]->getFullName(), currentId);
|
||||
}
|
||||
|
||||
std::vector<StorageNode> nodes = m_sqliteStorage.getNodesByNameId(currentId);
|
||||
if (nodes.size() > 0) // TODO: make it impossible that one name id referrs to n nodes.
|
||||
{
|
||||
return nodes[0].id;
|
||||
}
|
||||
return 0;
|
||||
return m_sqliteStorage.getNodeBySerializedName(NameHierarchy::serialize(nameHierarchy)).id;
|
||||
}
|
||||
|
||||
Id Storage::getIdForEdge(
|
||||
Edge::EdgeType type, const NameHierarchy& fromNameHierarchy, const NameHierarchy& toNameHierarchy
|
||||
) const {
|
||||
) const
|
||||
{
|
||||
Id sourceId = getIdForNodeWithNameHierarchy(fromNameHierarchy);
|
||||
Id targetId = getIdForNodeWithNameHierarchy(toNameHierarchy);
|
||||
return m_sqliteStorage.getEdgeBySourceTargetType(sourceId, targetId, type).id;
|
||||
@@ -795,9 +775,7 @@ Id Storage::getIdForFirstNode() const
|
||||
|
||||
NameHierarchy Storage::getNameHierarchyForNodeWithId(Id nodeId) const
|
||||
{
|
||||
Id nameHierarchyElementId = m_sqliteStorage.getNameHierarchyElementIdByNodeId(nodeId);
|
||||
return m_sqliteStorage.getNameHierarchyById(nameHierarchyElementId);
|
||||
// return m_tokenIndex.getNameHierarchyForTokenId(nodeId).getFullName();
|
||||
return NameHierarchy::deserialize(m_sqliteStorage.getNodeById(nodeId).serializedName);
|
||||
}
|
||||
|
||||
Node::NodeType Storage::getNodeTypeForNodeWithId(Id nodeId) const
|
||||
@@ -1053,7 +1031,7 @@ std::vector<Id> Storage::getTokenIdsForMatches(const std::vector<SearchMatch>& m
|
||||
|
||||
Id Storage::getTokenIdForFileNode(const FilePath& filePath) const
|
||||
{
|
||||
return m_sqliteStorage.getFileByName(filePath.fileName()).id;
|
||||
return m_sqliteStorage.getFileByPath(filePath.str()).id;
|
||||
}
|
||||
|
||||
std::vector<Id> Storage::getTokenIdsForAggregationEdge(Id sourceId, Id targetId) const
|
||||
@@ -1306,73 +1284,34 @@ Id Storage::addNodeHierarchy(Node::NodeType nodeType, NameHierarchy nameHierarch
|
||||
return 0;
|
||||
}
|
||||
|
||||
std::vector<Id> nameIds = addNameHierarchyElements(nameHierarchy);
|
||||
|
||||
Id parentNodeId = 0;
|
||||
for (size_t i = 0; i < nameIds.size(); i++)
|
||||
{
|
||||
Id nameId = nameIds[i];
|
||||
const bool isLastElement = (i == nameHierarchy.size() - 1);
|
||||
const std::string& signature = nameHierarchy[i]->getFullSignature();
|
||||
const bool hasSignature = (signature.size() > 0);
|
||||
bool nodeMayExist = true;
|
||||
NameHierarchy currentNameHierarchy;
|
||||
|
||||
for (size_t i = 0; i < nameHierarchy.size(); i++)
|
||||
{
|
||||
currentNameHierarchy.push(nameHierarchy[i]);
|
||||
const bool isLastElement = (i == nameHierarchy.size() - 1);
|
||||
Node::NodeType type = (isLastElement ? nodeType : Node::NODE_UNDEFINED);
|
||||
|
||||
StorageNode node(0, 0, 0, false);
|
||||
if (hasSignature)
|
||||
StorageNode node(0, 0, "", false);
|
||||
|
||||
if (nodeMayExist)
|
||||
{
|
||||
std::vector<Id> potentialIds = m_sqliteStorage.getNodeIdsBySignature(signature);
|
||||
if (parentNodeId != 0)
|
||||
{
|
||||
for (size_t i = 0; i < potentialIds.size(); i++)
|
||||
{
|
||||
if (m_sqliteStorage.getEdgeBySourceTargetType(parentNodeId, potentialIds[i], Edge::EDGE_MEMBER).id != 0)
|
||||
{
|
||||
node = m_sqliteStorage.getNodeById(potentialIds[i]);
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
else if (potentialIds.size() > 0)
|
||||
{
|
||||
node = m_sqliteStorage.getNodeById(potentialIds[0]);
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
std::vector<StorageNode> potentialIds = m_sqliteStorage.getNodesByNameId(nameId);
|
||||
if (parentNodeId != 0)
|
||||
{
|
||||
for (size_t i = 0; i < potentialIds.size(); i++)
|
||||
{
|
||||
if (m_sqliteStorage.getEdgeBySourceTargetType(parentNodeId, potentialIds[i].id, Edge::EDGE_MEMBER).id != 0)
|
||||
{
|
||||
node = potentialIds[i];
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
else if (potentialIds.size() > 0)
|
||||
{
|
||||
node = potentialIds[0];
|
||||
}
|
||||
node = m_sqliteStorage.getNodeBySerializedName(NameHierarchy::serialize(currentNameHierarchy));
|
||||
}
|
||||
|
||||
Id nodeId = node.id;
|
||||
|
||||
if (nodeId && !node.defined && isLastElement && defined)
|
||||
if (nodeId && !node.defined && isLastElement && defined) // todo: move this down!
|
||||
{
|
||||
m_sqliteStorage.setNodeDefined(true, nodeId);
|
||||
}
|
||||
|
||||
if (nodeId == 0)
|
||||
{
|
||||
nodeId = m_sqliteStorage.addNode(Node::typeToInt(type), nameId, isLastElement && defined);
|
||||
|
||||
if (hasSignature)
|
||||
{
|
||||
m_sqliteStorage.addSignature(nodeId, signature);
|
||||
}
|
||||
nodeMayExist = false;
|
||||
nodeId = m_sqliteStorage.addNode(Node::typeToInt(type), NameHierarchy::serialize(currentNameHierarchy), isLastElement && defined);
|
||||
|
||||
if (parentNodeId != 0)
|
||||
{
|
||||
@@ -1399,35 +1338,6 @@ Id Storage::addNodeHierarchy(Node::NodeType type, const ParseFunction& function,
|
||||
return addNodeHierarchy(type, function.nameHierarchy, defined);
|
||||
}
|
||||
|
||||
std::vector<Id> Storage::addNameHierarchyElements(NameHierarchy nameHierarchy)
|
||||
{
|
||||
std::vector<Id> nameIds;
|
||||
Id parentId = 0;
|
||||
bool nodeMayExist = true;
|
||||
|
||||
for (size_t i = 0; i < nameHierarchy.size(); i++)
|
||||
{
|
||||
const std::string elementName = nameHierarchy[i]->getFullName();
|
||||
Id nodeId = 0;
|
||||
|
||||
if (nodeMayExist)
|
||||
{
|
||||
nodeId = m_sqliteStorage.getNameHierarchyElementIdByName(elementName, parentId);
|
||||
}
|
||||
|
||||
if (nodeId == 0)
|
||||
{
|
||||
nodeId = m_sqliteStorage.addNameHierarchyElement(elementName, parentId);
|
||||
nodeMayExist = false;
|
||||
}
|
||||
|
||||
nameIds.push_back(nodeId);
|
||||
parentId = nodeId;
|
||||
}
|
||||
|
||||
return nameIds;
|
||||
}
|
||||
|
||||
Id Storage::addSourceLocation(Id elementNodeId, const ParseLocation &location, bool isScope)
|
||||
{
|
||||
if (!location.isValid())
|
||||
@@ -1681,9 +1591,13 @@ void Storage::addNodesToGraph(const std::vector<Id> nodeIds, Graph* graph) const
|
||||
|
||||
if (type == Node::NODE_FUNCTION || type == Node::NODE_METHOD)
|
||||
{
|
||||
node->addComponentSignature(
|
||||
std::make_shared<TokenComponentSignature>(m_sqliteStorage.getSignatureByNodeId(storageNode.id))
|
||||
);
|
||||
std::shared_ptr<NameElement> lastElement = NameHierarchy::deserialize(storageNode.serializedName).back();
|
||||
if (lastElement)
|
||||
{
|
||||
node->addComponentSignature(
|
||||
std::make_shared<TokenComponentSignature>(lastElement->getFullSignature())
|
||||
);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -1786,7 +1700,7 @@ void Storage::buildSearchIndex()
|
||||
{
|
||||
for (StorageNode node: m_sqliteStorage.getAllNodes())
|
||||
{
|
||||
m_tokenIndex.addTokenId(m_tokenIndex.addNode(m_sqliteStorage.getNameHierarchyById(node.nameId)), node.id);
|
||||
m_tokenIndex.addTokenId(m_tokenIndex.addNode(NameHierarchy::deserialize(node.serializedName)), node.id);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user