//
// Groups results.
//
void ResultsTree::regroupResults(bool groupBySearchEngine)
{
ResultsModelColumns::ResultType currentType, newType;
#ifdef DEBUG
cout << "ResultsTree::regroupResults: called" << endl;
#endif
// What's the new grouping criteria ?
if (groupBySearchEngine == true)
{
// By search engine
currentType = ResultsModelColumns::RESULT_HOST;
newType = ResultsModelColumns::RESULT_ROOT;
}
else
{
// By host
currentType = ResultsModelColumns::RESULT_ROOT;
newType = ResultsModelColumns::RESULT_HOST;
}
// Go through tree rows
TreeModel::Children children = m_refStore->children();
if (children.empty() == true)
{
return;
}
// Clear the map
m_resultsGroups.clear();
// Unselect results
get_selection()->unselect_all();
TreeModel::Children::iterator iter = children.begin();
while (iter != children.end())
{
TreeModel::Row row = *iter;
#ifdef DEBUG
cout << "ResultsTree::groupBySearchEngine: looking at " << row[m_resultsColumns.m_text] << endl;
#endif
ResultsModelColumns::ResultType type = row[m_resultsColumns.m_type];
// Skip new type rows
if (type == newType)
{
iter++;
continue;
}
TreeModel::Children child = iter->children();
if (child.empty() == false)
{
TreeModel::Children::iterator childIter = child.begin();
// Type RESULT_TITLE
while (childIter != child.end())
{
TreeModel::Row childRow = *childIter;
TreeModel::iterator groupIter, newIter;
bool success = false;
// We will need the URL and engines columns in all cases
string url = from_utf8(childRow[m_resultsColumns.m_url]);
unsigned int engineIds = childRow[m_resultsColumns.m_engines];
unsigned int indexIds = childRow[m_resultsColumns.m_indexes];
// Get the name of the group this should go into
if (newType == ResultsModelColumns::RESULT_HOST)
{
Url urlObj(url);
#ifdef DEBUG
cout << "ResultsTree::groupBySearchEngine: row " << url << endl;
#endif
string groupName = urlObj.getHost();
// Add group
if (appendGroup(groupName, newType, groupIter) == true)
{
// Add result
success = appendResult(childRow[m_resultsColumns.m_text],
childRow[m_resultsColumns.m_url],
(float)atof(from_utf8(childRow[m_resultsColumns.m_score]).c_str()),
from_utf8(childRow[m_resultsColumns.m_language]),
childRow[m_resultsColumns.m_rankDiff],
from_utf8(childRow[m_resultsColumns.m_queryName]),
engineIds, indexIds, newIter, &(*groupIter), true);
}
}
else
{
// Look at the engines column and see which engines this result is for
set<string> engineNames;
m_settings.getEngineNames(engineIds, engineNames);
if (engineNames.empty() == false)
{
#ifdef DEBUG
cout << "ResultsTree::groupBySearchEngine: row is for " << engineNames.size() << endl;
#endif
// Are there indexes in the list ?
set<string>::iterator xapianIter = engineNames.find("Xapian");
if ((xapianIter != engineNames.end()) &&
(indexIds > 0))
{
// Erase this
engineNames.erase(xapianIter);
// Add entries for each index name so that we can loop once on engine names
set<string> indexNames;
m_settings.getIndexNames(indexIds, indexNames);
for (set<string>::iterator iter = indexNames.begin(); iter != indexNames.end(); ++iter)
{
string indexName = (*iter);
engineNames.insert(indexName);
#ifdef DEBUG
cout << "ResultsTree::groupBySearchEngine: row is for index " << indexName << endl;
#endif
}
}
for (set<string>::iterator iter = engineNames.begin(); iter != engineNames.end(); ++iter)
{
string engineName = (*iter);
unsigned int indexId = 0;
unsigned int engineId = m_settings.getEngineId(engineName);
if (engineId == 0)
{
// This is actually an index, not an engine...
indexId = m_settings.getIndexId(engineName);
if (indexId > 0)
{
engineId = m_settings.getEngineId("Xapian");
}
}
// Add group
if (appendGroup(engineName, newType, groupIter) == true)
{
// Add result
appendResult(childRow[m_resultsColumns.m_text],
childRow[m_resultsColumns.m_url],
(float)atof(from_utf8(childRow[m_resultsColumns.m_score]).c_str()),
from_utf8(childRow[m_resultsColumns.m_language]),
childRow[m_resultsColumns.m_rankDiff],
from_utf8(childRow[m_resultsColumns.m_queryName]),
engineId, indexId,
newIter, &(*groupIter), true);
#ifdef DEBUG
cout << "ResultsTree::groupBySearchEngine: row for " << *iter << endl;
#endif
}
}
// FIXME: make sure at least one row was added
success = true;
}
}
if (success == true)
{
// Delete it
m_refStore->erase(*childIter);
childIter = child.begin();
}
else
{
// Don't delete anything then, just go to the next child
childIter++;
}
}
}
// Erase this row
m_refStore->erase(*iter);
// Get the new first row, that way we don't have to worry about iterators validity
iter = children.begin();
}
for (std::map<string, TreeModel::iterator>::iterator mapIter = m_resultsGroups.begin();
mapIter != m_resultsGroups.end(); mapIter++)
{
TreeModel::iterator groupIter = mapIter->second;
updateGroup(groupIter);
}
onSelectionChanged();
}
//
// Adds a new row in the results tree.
//
bool ResultsTree::appendResult(const ustring &text, const ustring &url,
float score, const string &language, int rankDiff,
const string &queryName, unsigned int engineId, unsigned int indexId,
TreeModel::iterator &newRowIter, const TreeModel::Row *parentRow, bool noDuplicates)
{
if (parentRow == NULL)
{
newRowIter = m_refStore->append();
}
else
{
// Merge duplicates within groups ?
if (noDuplicates == true)
{
// Look for a row with the same URL and query. For instance, in group
// by host mode, if a page is returned by several search engines, it
// should appear only once
TreeModel::Children children = parentRow->children();
if (children.empty() == false)
{
TreeModel::Children::iterator childIter = children.begin();
for (; childIter != children.end(); ++childIter)
{
TreeModel::Row row = *childIter;
if ((row[m_resultsColumns.m_url] == to_utf8(url)) &&
(row[m_resultsColumns.m_queryName] == to_utf8(queryName)))
{
// Update the engines column...
row[m_resultsColumns.m_engines] = row[m_resultsColumns.m_engines] | engineId;
// ...and the indexes column too
row[m_resultsColumns.m_indexes] = row[m_resultsColumns.m_indexes] | engineId;
#ifdef DEBUG
cout << "ResultsTree::appendResult: merged " << text << " " << engineId << " (" << row[m_resultsColumns.m_engines] << "," << row[m_resultsColumns.m_indexes] << ")" << endl;
#endif
newRowIter = childIter;
return true;
}
}
}
}
newRowIter = m_refStore->append(parentRow->children());
#ifdef DEBUG
cout << "ResultsTree::appendResult: added " << text << ", " << score << " to "
<< (*parentRow)[m_resultsColumns.m_text] << endl;
#endif
}
XapianIndex index(m_settings.m_indexLocation);
ViewHistory viewHistory(m_settings.m_historyDatabase);
bool isIndexed = false;
// Is this document indexed ?
if ((index.isGood() == true) &&
(index.hasDocument(url) > 0))
{
isIndexed = true;
}
// Has it been already viewed ?
bool wasViewed = viewHistory.hasItem(url);
char scoreStr[128];
snprintf(scoreStr, 128, "%.f", score);
TreeModel::Row childRow = *newRowIter;
updateRow(childRow, text, url, scoreStr,
to_utf8(language), to_utf8(queryName), engineId, indexId,
ResultsModelColumns::RESULT_TITLE, isIndexed,
wasViewed, rankDiff);
return true;
}