void DataSystem::query(const std::string q)
{
try {
// Start an enquire session.
Xapian::Enquire enquire(db);
Xapian::Query query = qp.parse_query(q);
printf("Parsed query is: %s", query.get_description().c_str());
// Find the top 10 results for the query.
enquire.set_query(query);
Xapian::MSet matches = enquire.get_mset(0, 10);
// Display the results.
printf("%d results found.\n", matches.get_matches_estimated());
printf("Matches 1-$d:\n", matches.size());
for (Xapian::MSetIterator i = matches.begin(); i != matches.end(); ++i)
{
printf("%d: %d\% docid= [%s]\n\n",
i.get_rank() + 1,
i.get_percent(),
i.get_document().get_data().c_str());
}
} catch (const Xapian::Error &e)
{
printf("DataSystem, query error: %s", e.get_description());
}
}
Xapian::Query XapianEngine::parseQuery(Xapian::Database *pIndex, const QueryProperties &queryProps,
const string &stemLanguage, DefaultOperator defaultOperator,
const string &limitQuery, string &correctedFreeQuery, bool minimal)
{
Xapian::QueryParser parser;
CJKVTokenizer tokenizer;
string freeQuery(queryProps.getFreeQuery());
unsigned int tokensCount = 1;
bool diacriticSensitive = queryProps.getDiacriticSensitive();
// Modifying the query is necessary if it's CJKV or diacritics are off
if ((tokenizer.has_cjkv(freeQuery) == true) ||
(diacriticSensitive == false))
{
QueryModifier handler(freeQuery,
diacriticSensitive,
tokenizer.get_ngram_size());
tokenizer.tokenize(freeQuery, handler, true);
tokensCount = handler.get_tokens_count();
// We can disable stemming and spelling correction for pure CJKV queries
string cjkvQuery(handler.get_modified_query(minimal));
#ifdef DEBUG
cout << "XapianEngine::parseQuery: CJKV query is " << cjkvQuery << endl;
#endif
// Do as if the user had given this as input
freeQuery = cjkvQuery;
}
else
{
string::size_type spacePos = freeQuery.find(' ');
while (spacePos != string::npos)
{
++tokensCount;
if (spacePos + 1 >= freeQuery.length())
{
break;
}
// Next
spacePos = freeQuery.find(' ', spacePos + 1);
}
}
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << tokensCount << " tokens" << endl;
#endif
if (pIndex != NULL)
{
// The database is required for wildcards and spelling
parser.set_database(*pIndex);
}
// Set things up
if ((minimal == false) &&
(stemLanguage.empty() == false))
{
parser.set_stemmer(m_stemmer);
parser.set_stemming_strategy(Xapian::QueryParser::STEM_SOME);
// Don't bother loading the stopwords list if there's only one token
if (tokensCount > 1)
{
FileStopper *pStopper = FileStopper::get_stopper(Languages::toCode(stemLanguage));
if ((pStopper != NULL) &&
(pStopper->get_stopwords_count() > 0))
{
parser.set_stopper(pStopper);
}
}
}
else
{
#ifdef DEBUG
cout << "XapianEngine::parseQuery: no stemming" << endl;
#endif
parser.set_stemming_strategy(Xapian::QueryParser::STEM_NONE);
}
// What's the default operator ?
if (defaultOperator == DEFAULT_OP_AND)
{
parser.set_default_op(Xapian::Query::OP_AND);
}
else
{
parser.set_default_op(Xapian::Query::OP_OR);
}
#if XAPIAN_NUM_VERSION >= 1000004
// Search across text body and title
parser.add_prefix("", "");
parser.add_prefix("", "S");
#endif
// X prefixes should always include a colon
parser.add_boolean_prefix("site", "H");
parser.add_boolean_prefix("file", "P");
parser.add_boolean_prefix("ext", "XEXT:");
parser.add_prefix("title", "S");
parser.add_boolean_prefix("url", "U");
parser.add_boolean_prefix("dir", "XDIR:");
parser.add_boolean_prefix("inurl", "XFILE:");
parser.add_prefix("path", "XPATH:");
parser.add_boolean_prefix("lang", "L");
parser.add_boolean_prefix("type", "T");
parser.add_boolean_prefix("class", "XCLASS:");
parser.add_boolean_prefix("label", "XLABEL:");
parser.add_boolean_prefix("tokens", "XTOK:");
// Any limit on what documents should be searched ?
if (limitQuery.empty() == false)
{
string limitedQuery(limitQuery);
limitedQuery += " AND ( ";
limitedQuery += freeQuery;
limitedQuery += " )";
freeQuery = limitedQuery;
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << freeQuery << endl;
#endif
}
// Date range
Xapian::DateValueRangeProcessor dateProcessor(0);
parser.add_valuerangeprocessor(&dateProcessor);
// Size with a "b" suffix, ie 1024..10240b
#if XAPIAN_NUM_VERSION >= 1001000
Xapian::NumberValueRangeProcessor sizeProcessor(2, "b", false);
parser.add_valuerangeprocessor(&sizeProcessor);
#elif XAPIAN_NUM_VERSION >= 1000002
// Xapian 1.02 is the bare minimum
Xapian::v102::NumberValueRangeProcessor sizeProcessor(2, "b", false);
parser.add_valuerangeprocessor(&sizeProcessor);
#endif
// Time range
TimeValueRangeProcessor timeProcessor(3);
parser.add_valuerangeprocessor(&timeProcessor);
// What type of query is this ?
QueryProperties::QueryType type = queryProps.getType();
if (type != QueryProperties::XAPIAN_QP)
{
map<string, string> fieldMapping;
// Bare minimum mapping between Xesam fields and our prefixes
fieldMapping["dc:title"] = "S";
XapianQueryBuilder builder(parser, fieldMapping);
XesamParser *pParser = NULL;
bool parsedQuery = false;
// Get a Xesam parser
if (type == QueryProperties::XESAM_QL)
{
pParser = new XesamQLParser();
}
#ifdef HAVE_BOOST_SPIRIT_CORE_HPP
else if (type == QueryProperties::XESAM_UL)
{
pParser = new XesamULParser();
}
#endif
if (pParser != NULL)
{
parsedQuery = pParser->parse(freeQuery, builder);
delete pParser;
}
if (parsedQuery == true)
{
return builder.get_query();
}
return Xapian::Query();
}
// Do some pre-processing : look for filters with quoted values
string::size_type escapedFilterEnd = 0;
string::size_type escapedFilterStart = freeQuery.find(":\"");
while ((escapedFilterStart != string::npos) &&
(escapedFilterStart < freeQuery.length() - 2))
{
escapedFilterEnd = freeQuery.find("\"", escapedFilterStart + 2);
if (escapedFilterEnd == string::npos)
{
break;
}
string filterValue = freeQuery.substr(escapedFilterStart + 2, escapedFilterEnd - escapedFilterStart - 2);
if (filterValue.empty() == false)
{
string escapedValue(Url::escapeUrl(filterValue));
bool escapeValue = false, hashValue = false;
// The value should be escaped and length-limited as done at indexing time
checkFilter(freeQuery, escapedFilterStart, escapeValue, hashValue);
if (escapeValue == false)
{
// No escaping
escapedValue = filterValue;
}
if (hashValue == true)
{
// Partially hash if necessary
escapedValue = XapianDatabase::limitTermLength(escapedValue, true);
}
else
{
escapedValue = XapianDatabase::limitTermLength(escapedValue);
}
freeQuery.replace(escapedFilterStart + 1, escapedFilterEnd - escapedFilterStart,
escapedValue);
escapedFilterEnd = escapedFilterEnd + escapedValue.length() - filterValue.length();
}
else
{
// No value !
freeQuery.replace(escapedFilterStart, escapedFilterEnd - escapedFilterStart + 1, ":");
escapedFilterEnd -= 2;
}
#ifdef DEBUG
cout << "XapianEngine::parseQuery: replaced filter: " << freeQuery << endl;
#endif
// Next
escapedFilterStart = freeQuery.find(":\"", escapedFilterEnd);
}
// Parse the query string with all necessary options
unsigned int flags = Xapian::QueryParser::FLAG_BOOLEAN|Xapian::QueryParser::FLAG_PHRASE|
Xapian::QueryParser::FLAG_LOVEHATE|Xapian::QueryParser::FLAG_PURE_NOT;
if (minimal == false)
{
flags |= Xapian::QueryParser::FLAG_WILDCARD;
#if ENABLE_XAPIAN_SPELLING_CORRECTION>0
flags |= Xapian::QueryParser::FLAG_SPELLING_CORRECTION;
#endif
}
Xapian::Query parsedQuery = parser.parse_query(freeQuery, flags);
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << parsedQuery.get_description() << endl;
#endif
if (minimal == false)
{
#if ENABLE_XAPIAN_SPELLING_CORRECTION>0
// Any correction ?
correctedFreeQuery = parser.get_corrected_query_string();
#ifdef DEBUG
if (correctedFreeQuery.empty() == false)
{
cout << "XapianEngine::parseQuery: corrected spelling to: " << correctedFreeQuery << endl;
}
#endif
#endif
}
return parsedQuery;
}
/// Runs a query; true if success.
bool XapianEngine::runQuery(QueryProperties& queryProps)
{
// Clear the results list
m_resultsList.clear();
XapianDatabase *pDatabase = XapianDatabaseFactory::getDatabase(m_databaseName, true);
if (pDatabase == NULL)
{
return false;
}
// Get the latest revision...
pDatabase->reopen();
Xapian::Database *pIndex = pDatabase->readLock();
try
{
string stemLanguage;
unsigned int searchStep = 1;
bool followOperators = true;
// Searches are run in this order :
// 1. follow operators and don't stem terms
// 2. if no results, follow operators and stem terms
// 3. if no results, don't follow operators and don't stem terms
// 4. if no results, don't follow operators and stem terms
// Steps 2 and 4 depend on a language being defined for the query
Xapian::Query fullQuery = parseQuery(pIndex, queryProps, "", followOperators);
while (fullQuery.empty() == false)
{
#ifdef DEBUG
cout << "XapianEngine::runQuery: " << fullQuery.get_description() << endl;
#endif
// Query the database
if (queryDatabase(pIndex, fullQuery) == false)
{
break;
}
if (m_resultsList.empty() == true)
{
// The search did succeed but didn't return anything
// Try the next step
switch (++searchStep)
{
case 2:
followOperators = true;
stemLanguage = queryProps.getLanguage();
if (stemLanguage.empty() == false)
{
break;
}
++searchStep;
case 3:
followOperators = false;
stemLanguage.clear();
break;
case 4:
followOperators = false;
stemLanguage = queryProps.getLanguage();
if (stemLanguage.empty() == false)
{
break;
}
++searchStep;
default:
pDatabase->unlock();
return true;
}
#ifdef DEBUG
cout << "XapianEngine::runQuery: trying step " << searchStep << endl;
#endif
fullQuery = parseQuery(pIndex, queryProps,
Languages::toEnglish(stemLanguage), followOperators);
continue;
}
pDatabase->unlock();
return true;
}
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::runQuery: " << error.get_type() << ": " << error.get_msg() << endl;
}
pDatabase->unlock();
return false;
}
int main(int argc, char **argv)
{
if(argc < 2) {
usage(argv);
return 1;
}
try {
char *action = argv[1];
char *db_path = argv[2];
if(!strcmp(action, "index")) {
Xapian::WritableDatabase db(db_path, Xapian::DB_CREATE_OR_OPEN);
Xapian::TermGenerator indexer;
Xapian::Stem stemmer("english");
indexer.set_stemmer(stemmer);
std::string doc_txt;
while(true) {
if(std::cin.eof()) break;
std::string line;
getline(std::cin, line);
doc_txt += line;
}
if(!doc_txt.empty()) {
Xapian::Document doc;
doc.set_data(doc_txt);
indexer.set_document(doc);
indexer.index_text(doc_txt);
db.add_document(doc);
std::cout << "Indexed: " << indexer.get_description() << std::endl;
}
db.commit();
} else if(!strcmp(action, "search")) {
if(argc < 4) {
std::cerr << "You must supply a query string" << std::endl;
return 1;
}
Xapian::Database db(db_path);
Xapian::Enquire enquire(db);
std::string query_str = argv[3];
argv+= 4;
while(*argv) {
query_str += ' ';
query_str += *argv++;
}
Xapian::QueryParser qp;
Xapian::Stem stemmer("english");
qp.set_stemmer(stemmer);
qp.set_database(db);
qp.set_stemming_strategy(Xapian::QueryParser::STEM_SOME);
Xapian::Query query = qp.parse_query(query_str);
std::cout << "Parsed query is: " << query.get_description() <<
std::endl;
enquire.set_query(query);
Xapian::MSet matches = enquire.get_mset(0, 10);
std::cout << matches.get_matches_estimated() << " results found.\n";
std::cout << "Matches 1-" << matches.size() << ":\n" << std::endl;
for (Xapian::MSetIterator i = matches.begin();
i != matches.end(); ++i) {
std::cout << i.get_rank() + 1 << ": " << i.get_percent() <<
"% docid=" << *i << " [" <<
i.get_document().get_data()<< "]" << std::endl <<
std::endl;
}
} else {
std::cerr << "Invalid action " << action << std::endl;
usage(argv);
return 1;
}
} catch (const Xapian::Error &error) {
std::cout << "Exception: " << error.get_msg() << std::endl;
}
}
Xapian::Query XapianEngine::parseQuery(Xapian::Database *pIndex, const QueryProperties &queryProps,
const string &stemLanguage, DefaultOperator defaultOperator,
const string &limitQuery, string &correctedFreeQuery, bool minimal)
{
Xapian::QueryParser parser;
Xapian::Stem stemmer;
CJKVTokenizer tokenizer;
string freeQuery(StringManip::replaceSubString(queryProps.getFreeQuery(), "\n", " "));
unsigned int minDay, minMonth, minYear = 0;
unsigned int maxDay, maxMonth, maxYear = 0;
if (tokenizer.has_cjkv_only(freeQuery) == true)
{
vector<string> tokens;
string cjkvQuery;
tokenizer.tokenize(freeQuery, tokens);
// Get the terms
for (vector<string>::const_iterator tokenIter = tokens.begin();
tokenIter != tokens.end(); ++tokenIter)
{
cjkvQuery += *tokenIter;
cjkvQuery += " ";
}
#ifdef DEBUG
cout << "XapianEngine::parseQuery: CJKV query is " << cjkvQuery << endl;
#endif
// Do as if the user had given this as input
freeQuery = cjkvQuery;
// We can disable stemming and spelling correction
minimal = true;
}
if (pIndex != NULL)
{
// The database is required for wildcards and spelling
parser.set_database(*pIndex);
}
// Set things up
if ((minimal == false) &&
(stemLanguage.empty() == false))
{
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << stemLanguage << " stemming" << endl;
#endif
try
{
stemmer = Xapian::Stem(StringManip::toLowerCase(stemLanguage));
}
catch (const Xapian::Error &error)
{
cerr << "Couldn't create stemmer: " << error.get_type() << ": " << error.get_msg() << endl;
}
parser.set_stemmer(stemmer);
parser.set_stemming_strategy(Xapian::QueryParser::STEM_SOME);
}
else
{
#ifdef DEBUG
cout << "XapianEngine::parseQuery: no stemming" << endl;
#endif
parser.set_stemming_strategy(Xapian::QueryParser::STEM_NONE);
}
// What's the default operator ?
if (defaultOperator == DEFAULT_OP_AND)
{
parser.set_default_op(Xapian::Query::OP_AND);
}
else
{
parser.set_default_op(Xapian::Query::OP_OR);
}
// X prefixes should always include a colon
parser.add_boolean_prefix("site", "H");
parser.add_boolean_prefix("file", "P");
parser.add_boolean_prefix("ext", "XEXT:");
parser.add_prefix("title", "S");
parser.add_boolean_prefix("url", "U");
parser.add_boolean_prefix("dir", "XDIR:");
parser.add_boolean_prefix("lang", "L");
parser.add_boolean_prefix("type", "T");
parser.add_boolean_prefix("class", "XCLASS:");
parser.add_boolean_prefix("label", "XLABEL:");
parser.add_boolean_prefix("tokens", "XTOK:");
// Any limit on what documents should be searched ?
if (limitQuery.empty() == false)
{
string limitedQuery(limitQuery);
limitedQuery += " AND ( ";
limitedQuery += freeQuery;
limitedQuery += " )";
freeQuery = limitedQuery;
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << freeQuery << endl;
#endif
}
// Date range
Xapian::DateValueRangeProcessor dateProcessor(0);
parser.add_valuerangeprocessor(&dateProcessor);
// Size with a "b" suffix, ie 1024..10240b
#if XAPIAN_NUM_VERSION >= 1001000
Xapian::NumberValueRangeProcessor sizeProcessor(2, "b", false);
parser.add_valuerangeprocessor(&sizeProcessor);
#elif XAPIAN_NUM_VERSION >= 1000002
// Xapian 1.02 is the bare minimum
Xapian::v102::NumberValueRangeProcessor sizeProcessor(2, "b", false);
parser.add_valuerangeprocessor(&sizeProcessor);
#endif
// Time range
TimeValueRangeProcessor timeProcessor(3);
parser.add_valuerangeprocessor(&timeProcessor);
// What type of query is this ?
QueryProperties::QueryType type = queryProps.getType();
if (type != QueryProperties::XAPIAN_QP)
{
map<string, string> fieldMapping;
// Bare minimum mapping between Xesam fields and our prefixes
fieldMapping["dc:title"] = "S";
XapianQueryBuilder builder(parser, fieldMapping);
XesamParser *pParser = NULL;
// Get a Xesam parser
if (type == QueryProperties::XESAM_QL)
{
pParser = new XesamQLParser();
}
#ifdef HAVE_BOOST_SPIRIT_CORE_HPP
else if (type == QueryProperties::XESAM_UL)
{
pParser = new XesamULParser();
}
#endif
if (pParser != NULL)
{
bool parsedQuery = pParser->parse(freeQuery, builder);
delete pParser;
if (parsedQuery == true)
{
return builder.get_query();
}
}
return Xapian::Query();
}
// Do some pre-processing : look for filters with quoted values
string::size_type escapedFilterEnd = 0;
string::size_type escapedFilterStart = freeQuery.find(":\"");
while ((escapedFilterStart != string::npos) &&
(escapedFilterStart < freeQuery.length() - 2))
{
escapedFilterEnd = freeQuery.find("\"", escapedFilterStart + 2);
if (escapedFilterEnd == string::npos)
{
break;
}
string filterValue = freeQuery.substr(escapedFilterStart + 2, escapedFilterEnd - escapedFilterStart - 2);
if (filterValue.empty() == false)
{
string escapedValue(Url::escapeUrl(filterValue));
bool escapeValue = false, hashValue = false;
// The value should be escaped and length-limited as done at indexing time
checkFilter(freeQuery, escapedFilterStart, escapeValue, hashValue);
if (escapeValue == false)
{
// No escaping
escapedValue = filterValue;
}
if (hashValue == true)
{
// Partially hash if necessary
escapedValue = XapianDatabase::limitTermLength(escapedValue, true);
}
else
{
escapedValue = XapianDatabase::limitTermLength(escapedValue);
}
freeQuery.replace(escapedFilterStart + 1, escapedFilterEnd - escapedFilterStart,
escapedValue);
escapedFilterEnd = escapedFilterEnd + escapedValue.length() - filterValue.length();
}
else
{
// No value !
freeQuery.replace(escapedFilterStart, escapedFilterEnd - escapedFilterStart + 1, ":");
escapedFilterEnd -= 2;
}
#ifdef DEBUG
cout << "XapianEngine::parseQuery: replaced filter: " << freeQuery << endl;
#endif
// Next
escapedFilterStart = freeQuery.find(":\"", escapedFilterEnd);
}
// Parse the query string with all necessary options
unsigned int flags = Xapian::QueryParser::FLAG_BOOLEAN|Xapian::QueryParser::FLAG_PHRASE|
Xapian::QueryParser::FLAG_LOVEHATE|Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE|
Xapian::QueryParser::FLAG_PURE_NOT;
if (minimal == false)
{
flags |= Xapian::QueryParser::FLAG_WILDCARD;
#if ENABLE_XAPIAN_SPELLING_CORRECTION>0
flags |= Xapian::QueryParser::FLAG_SPELLING_CORRECTION;
#endif
}
Xapian::Query parsedQuery = parser.parse_query(freeQuery, flags);
#ifdef DEBUG
cout << "XapianEngine::parseQuery: " << parsedQuery.get_description() << endl;
#endif
if (minimal == false)
{
#if ENABLE_XAPIAN_SPELLING_CORRECTION>0
// Any correction ?
correctedFreeQuery = parser.get_corrected_query_string();
#ifdef DEBUG
if (correctedFreeQuery.empty() == false)
{
cout << "XapianEngine::parseQuery: corrected spelling to: " << correctedFreeQuery << endl;
}
#endif
#endif
}
return parsedQuery;
}