int retrieve(option& opt, istream_type& is, ostream_type& os)
{
typedef std::basic_string<char_type> string_type;
typedef std::vector<string_type> strings_type;
typedef simstring::reader reader_type;
std::ostream& es = std::cerr;
// Open the database.
reader_type db;
if (!db.open(opt.name)) {
es << "ERROR: " << db.error() << std::endl;
return 1;
}
// Check the size of characters.
if (db.char_size() != sizeof(char_type)) {
es << "ERROR: Inconsistent character encoding " <<
"(DB:" << db.char_size() << ", " <<
"CUR:" << sizeof(char_type) << "): " << std::endl;
es << "This problem may be solved by specifying -u (--unicode) option." << std::endl;
return 1;
}
int num_queries = 0;
int num_retrieved = 0;
clock_t clk_total = 0;
for (;;) {
// Read a line.
string_type line;
std::getline(is, line);
if (is.eof()) {
break;
}
// Issue a query.
strings_type xstrs;
clock_t clk = std::clock();
db.retrieve(line, opt.measure, opt.threshold, std::back_inserter(xstrs));
clock_t elapsed = (std::clock() - clk);
// Update stats.
clk_total += elapsed;
num_retrieved += (int)xstrs.size();
++num_queries;
// Do not output results when the benchmarking flag is on.
if (!opt.benchmark) {
// Output the query string if necessary.
if (opt.echo_back) {
os << line << std::endl;
}
// Output the retrieved strings.
typename strings_type::const_iterator it;
for (it = xstrs.begin();it != xstrs.end();++it) {
os << os.widen('\t') << *it << std::endl;
}
os.flush();
}
// Do not output information when the quiet flag is on.
if (!opt.quiet) {
os <<
xstrs.size() <<
widen<char_type>(" strings retrieved (") <<
(std::clock() - clk) / (double)CLOCKS_PER_SEC <<
widen<char_type>(" sec)") << std::endl;
}
}
// Output the benchmark information if necessary.
if (opt.benchmark) {
os <<
widen<char_type>("Total number of queries: ") <<
num_queries << std::endl;
os <<
widen<char_type>("Seconds per query: ") <<
clk_total / (double)CLOCKS_PER_SEC / num_queries << std::endl;
os <<
widen<char_type>("Number of retrieved strings per query: ") <<
num_retrieved / (double)num_queries << std::endl;
}
return 0;
}
int build(option& opt, istream_type& is)
{
typedef std::basic_string<char_type> string_type;
typedef simstring::ngram_generator ngram_generator_type;
typedef simstring::writer_base<string_type, ngram_generator_type> writer_type;
std::ostream& os = std::cout;
std::ostream& es = std::cerr;
// Show the copyright information.
version(os);
// Show parameters for database construction.
os << "Constructing the database" << std::endl;
os << "Database name: " << opt.name << std::endl;
os << "N-gram length: " << opt.ngram_size << std::endl;
os << "Begin/end marks: " << std::boolalpha << opt.be << std::endl;
os << "Char type: " << typeid(char_type).name() << " (" << sizeof(char_type) << ")" << std::endl;
os.flush();
// Open the database for construction.
clock_t clk = std::clock();
ngram_generator_type gen(opt.ngram_size, opt.be);
writer_type db(gen, opt.name);
if (db.fail()) {
es << "ERROR: " << db.error() << std::endl;
return 1;
}
// Insert every string from STDIN into the database.
int n = 0;
for (;;) {
// Read a line.
string_type line;
std::getline(is, line);
if (is.eof()) {
break;
}
// Insert the string.
if (!db.insert(line)) {
es << "ERROR: " << db.error() << std::endl;
return 1;
}
// Progress report.
if (!opt.quiet && ++n % 10000 == 0) {
os << "Number of strings: " << n << std::endl;
os.flush();
}
}
os << "Number of strings: " << n << std::endl;
os << std::endl;
os.flush();
// Finalize the database.
os << "Flushing the database" << std::endl;
if (!db.close()) {
es << "ERROR: " << db.error() << std::endl;
return 1;
}
os << std::endl;
// Report the elaped time for construction.
os << "Total number of strings: " << n << std::endl;
os << "Seconds required: "
<< (std::clock() - clk) / (double)CLOCKS_PER_SEC << std::endl;
os << std::endl;
os.flush();
return 0;
}