void
find_clusters(int max_cluster_size_signed, SqlDatabase::TransactionPtr transaction)
{
assert(max_cluster_size_signed >= 0);
size_t max_cluster_size = max_cluster_size_signed;
SqlDatabase::StatementPtr insert_stmt = transaction->statement("insert into fr_ignored_function_pairs"
// 0 1 2
"(func1_id, func2_id, from_cluster_of_size)"
" values (?, ?, ?)");
//Get all vetexes and find the union
std::string _query_condition = "select func1_id, func2_id from fr_clone_pairs";
SqlDatabase::StatementPtr stmt = transaction->statement(_query_condition);
if (stmt->begin() == stmt->end())
return;
//Count how many vertices we have for boost graph
int VERTEX_COUNT = transaction->statement("select count(*) from semantic_functions")->execute_int();
typedef adjacency_list <vecS, vecS, undirectedS> Graph;
typedef graph_traits<Graph>::vertex_descriptor Vertex;
typedef graph_traits<Graph>::vertices_size_type VertexIndex;
Graph graph(VERTEX_COUNT);
std::vector<VertexIndex> rank(num_vertices(graph));
std::vector<Vertex> parent(num_vertices(graph));
typedef VertexIndex* Rank;
typedef Vertex* Parent;
disjoint_sets<Rank, Parent> ds(&rank[0], &parent[0]);
initialize_incremental_components(graph, ds);
incremental_components(graph, ds);
graph_traits<Graph>::edge_descriptor edge;
bool flag;
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int func1 = row.get<int>(0);
int func2 = row.get<int>(1);
boost::tie(edge, flag) = add_edge(func1, func2, graph);
ds.union_set(func1,func2);
}
typedef component_index<VertexIndex> Components;
Components components(parent.begin(), parent.end());
std::map<int,int> size_distribution;
// Iterate through the component indices
BOOST_FOREACH(VertexIndex current_index, components) {
std::vector<int> cluster_functions;
// Iterate through the child vertex indices for [current_index]
BOOST_FOREACH(VertexIndex child_index, components[current_index]) {
cluster_functions.push_back(child_index);
}
static void
list_assembly(const SqlDatabase::TransactionPtr &tx, int func_id)
{
Events events;
gather_events(tx, func_id);
load_events(tx, func_id, events);
SqlDatabase::StatementPtr stmt = tx->statement("select address, assembly from semantic_instructions where func_id = ?"
" order by position")->bind(0, func_id);
for (SqlDatabase::Statement::iterator insn=stmt->begin(); insn!=stmt->end(); ++insn) {
rose_addr_t addr = insn.get<rose_addr_t>(0);
std::string assembly = insn.get<std::string>(1);
Events::const_iterator ei=events.find(addr);
// Assembly line prefix
if (ei!=events.end() && ei->second.nexecuted>0) {
std::cout <<std::setw(9) <<std::right <<ei->second.nexecuted <<"x ";
} else {
std::cout <<std::string(11, ' ');
}
// Assembly instruction
std::cout <<"| " <<StringUtility::addrToString(addr) <<": " <<assembly <<"\n";
if (ei!=events.end())
show_events(ei->second);
}
}
void
computational_equivalent_classes(std::map<int,int>& norm_map)
{
SqlDatabase::StatementPtr stmt = transaction->statement("select func_id, equivalent_func_id from equivalent_classes");
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!= stmt->end(); ++row)
norm_map[row.get<int>(0)] = row.get<int>(1);
}
CallVec*
load_api_calls_for(int func_id, int igroup_id, bool ignore_no_compares, int call_depth, bool expand_ncalls)
{
SqlDatabase::StatementPtr stmt = transaction->statement("select distinct fio.pos, fio.callee_id, fio.ncalls"
" from semantic_fio_calls as fio"
" join tmp_interesting_funcs as f1"
// filter out functions with no compares
" on f1.func_id = fio.callee_id"
// filter on current parameters
" where fio.func_id = ? and fio.igroup_id = ?"
// filter out function not called directly
+ std::string(call_depth >= 0 ? " and fio.caller_id = ?" : "")
+" order by fio.pos");
stmt->bind(0, func_id);
stmt->bind(1, igroup_id);
if (call_depth >= 0)
stmt->bind(2, func_id);
CallVec* call_vec = new CallVec;
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int callee_id = row.get<int>(1);
int ncalls = row.get<int>(2);
if (expand_ncalls) {
for (int i = 0; i < ncalls; i++)
call_vec->push_back(callee_id);
} else {
call_vec->push_back(callee_id);
}
}
return call_vec;
}
static Dependencies
loadAllDependencies(const SqlDatabase::TransactionPtr &tx) {
Dependencies dependencies;
SqlDatabase::StatementPtr q = tx->statement("select name, value from dependencies where enabled <> 0");
for (SqlDatabase::Statement::iterator row = q->begin(); row != q->end(); ++row)
dependencies.insertMaybeDefault(row.get<std::string>(0)).push_back(row.get<std::string>(1));
return dependencies;
}
static void
load_source_code(const SqlDatabase::TransactionPtr &tx, Listing &listing/*in,out*/)
{
SqlDatabase::StatementPtr stmt = tx->statement("select file_id, linenum, line from tmp_src");
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int file_id = row.get<int>(0);
int linenum = row.get<int>(1);
SourcePosition srcpos(file_id, linenum);
listing[srcpos].source_code = row.get<std::string>(2);
}
}
CallVec*
load_function_api_calls_for(int func_id, bool reachability_graph)
{
SqlDatabase::StatementPtr stmt = transaction->statement("select distinct scg.callee from "
+ std::string(reachability_graph ? "semantic_rg" : "semantic_cg ") +
" as scg "
//" join tmp_interesting_funcs as tif on tif.func_id = scg.callee "
" where scg.caller=? ORDER BY scg.callee");
stmt->bind(0, func_id);
CallVec* call_vec = new CallVec;
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int callee_id = row.get<int>(0);
call_vec->push_back(callee_id);
}
return call_vec;
}
// List the errors ordered by how common they are.
static void
listErrors(const SqlDatabase::TransactionPtr &tx, const Settings &settings) {
std::vector<std::string> args;
SqlDatabase::StatementPtr q = tx->statement("select count(*) as n, status, test.first_error"
" from test_results test" +
sqlWhereClause(tx, settings, args) + " and"
" test.first_error is not null"
" group by status, test.first_error"
" order by n desc");
sqlBindArgs(q, args);
for (SqlDatabase::Statement::iterator row = q->begin(); row != q->end(); ++row) {
int count = row.get<int>(0);
std::string status = row.get<std::string>(1);
std::string mesg = row.get<std::string>(2);
printf("%6d %-16s %s\n", count, status.c_str(), oneLineEscaped(mesg).c_str());
}
}
// Generate the 'where' expression that limits what tests are being considered.
// Also appends variable values to the 'args' vector.
static std::string
sqlWhereClause(const SqlDatabase::TransactionPtr &tx, const Settings &settings, std::vector<std::string> &args /*in,out*/) {
std::vector<std::string> constraints;
if (settings.latestTests) {
// Constrain the tests to be only the latest version of ROSE present in the database.
SqlDatabase::StatementPtr q = tx->statement("select distinct rose, rose_date"
" from test_results"
" order by rose_date desc"
" limit 1");
SqlDatabase::Statement::iterator row = q->begin();
if (row != q->end()) {
constraints.push_back("rose = ?");
args.push_back(row.get<std::string>(0));
}
}
if (constraints.empty())
constraints.push_back("true");
return " where " + boost::join(constraints, " and ");
}
static DependencyNames
loadDependencyNames(const SqlDatabase::TransactionPtr &tx) {
DependencyNames retval;
SqlDatabase::StatementPtr q = tx->statement("select distinct name from dependencies");
for (SqlDatabase::Statement::iterator row=q->begin(); row!=q->end(); ++row) {
std::string key = row.get<std::string>(0);
retval.insert(key, "rmc_"+key);
}
// Additional key/column relationships
retval.insert("id", "test.id");
retval.insert("reporting_user", "auth_user.identity");
retval.insert("reporting_time", "test.reporting_time");
retval.insert("tester", "test.tester");
retval.insert("os", "test.os");
retval.insert("rose", "test.rose");
retval.insert("rose_date", "test.rose_date");
retval.insert("status", "test.status");
retval.insert("duration", "test.duration");
retval.insert("noutput", "test.noutput");
retval.insert("nwarnings", "test.nwarnings");
return retval;
}
static void
list_combined(const SqlDatabase::TransactionPtr &tx, int func_id, bool show_assembly)
{
CloneDetection::FilesTable files(tx);
Events events;
gather_events(tx, func_id);
load_events(tx, func_id, events/*out*/);
gather_instructions(tx, func_id, events);
Listing listing;
gather_source_code(tx);
load_source_code(tx, listing/*out*/);
// Get lines of assembly code and insert them into the correct place in the Listing.
if (show_assembly) {
SqlDatabase::StatementPtr stmt = tx->statement("select"
// 0 1 2 3
" insn.src_file_id, insn.src_line, insn.position, insn.address,"
// 4 5 6
" insn.assembly, func.id, func.name"
" from tmp_insns as insn"
" join semantic_functions as func on insn.func_id = func.id"
" order by position");
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int src_file_id = row.get<int>(0);
int src_line_num = row.get<int>(1);
SourcePosition srcpos(src_file_id, src_line_num);
int pos = row.get<int>(2);
rose_addr_t addr = row.get<rose_addr_t>(3);
std::string assembly = row.get<std::string>(4);
int func_id = row.get<int>(5);
std::string func_name = row.get<std::string>(6);
listing[srcpos].assembly_code.insert(std::make_pair(addr, AssemblyCode(pos, addr, assembly, func_id, func_name)));
}
// Listing header
std::cout <<"WARNING: This listing should be read cautiously. It is ordered according to the\n"
<<" source code with assembly lines following the source code line from which\n"
<<" they came. However, the compiler does not always generate machine\n"
<<" instructions in the same order as source code. When a discontinuity\n"
<<" occurs in the assembly instruction listing, it will be marked by a \"#\"\n"
<<" character. The assembly instructions are also numbered according to\n"
<<" their relative positions in the binary function.\n"
<<"\n"
<<" The prefix area contains either source location information or test trace\n"
<<" information. Note that trace information might be incomplete because\n"
<<" tracing was disabled or only partially enabled, or the trace includes\n"
<<" instructions that are not present in this function listing (e.g., when\n"
<<" execution follows a CALL instruction). The following notes are possible:\n"
<<" * \"Nx\" where N is an integer indicates that this instruction\n"
<<" was reached N times during testing. These notes are typically\n"
<<" only attached to the first instruction of a basic block and only\n"
<<" if the trace contains EV_REACHED events. Lack of an Nx notation\n"
<<" doesn't necessarily mean that the basic block was not reached, it\n"
<<" only means that there is no EV_REACHED event for that block.\n"
<<" * \"N<\" where N is an integer indicates that the instruction\n"
<<" on the previous line consumed N inputs. Information about the\n"
<<" inputs is listed on the right side of this line.\n"
<<" * \"N>\" where N is an integer indicates that the instruction\n"
<<" on the previous line produced N memory outputs. Information about the\n"
<<" outputs is listed on the right side of this line. Only the final\n"
<<" write to a memory address is considered a true output, and such\n"
<<" writes will be marked with the string \"final\".\n"
<<" * \"BR\" indicates that the instruction on the previous line is a\n"
<<" control flow branch point. The right side of the line shows more\n"
<<" detailed information about how many times the branch was taken.\n"
<<" * \"FAULT\" indicates that the test was terminated at the previous\n"
<<" instruction. The right side of the line shows the distribution of\n"
<<" faults that occurred here.\n"
<<"\n"
<<" /------------- Prefix area\n"
<<" /-------------/-------------- Source file ID or assembly function ID\n"
<<" | /------/--------------- Source line number or assembly instruction index\n"
<<" | | /-/---------------- Instruction out-of-order indicator\n"
<<" | | |/ /----------- Instruction virtual address\n"
<<" | | | |\n"
<<"vvvv vvvvv/| |\n"
<<"vvvvvvvvvv v vvvvvvvvvv\n";
}
// Show the listing
int prev_func_id = -1, prev_position = -1;
std::set<int> seen_files;
for (Listing::iterator li=listing.begin(); li!=listing.end(); ++li) {
int file_id = li->first.file_id;
if (seen_files.insert(file_id).second) {
if (file_id>=0) {
std::cout <<"\n" <<std::setw(4) <<std::right <<file_id <<".file |"
<<(opt.colorize?"\033[33;4m":"") <<files.name(file_id) <<(opt.colorize?"\033[m":"") <<"\n";
} else {
std::cout <<"\n" <<std::string(11, ' ') <<"|"
<<(opt.colorize?"\033[33;4m":"") <<"instructions not associated with a source file"
<<(opt.colorize?"\033[m":"") <<"\n";
}
}
if (file_id>=0) {
std::cout <<std::setw(4) <<std::right <<file_id <<"." <<std::setw(6) <<std::left <<li->first.line_num
<<"|"
<<(opt.colorize?"\033[34m":"")
<<StringUtility::untab(li->second.source_code)
<<(opt.colorize?"\033[m":"") <<"\n";
}
for (Instructions::iterator ii=li->second.assembly_code.begin(); ii!=li->second.assembly_code.end(); ++ii) {
const AssemblyCode assm = ii->second;
if (assm.func_id!=prev_func_id) {
std::cout <<std::string(11, ' ') <<"# "
<<(opt.colorize?"\033[33;4m":"") <<"function " <<StringUtility::numberToString(assm.func_id);
if (!assm.func_name.empty())
std::cout <<" <" <<assm.func_name <<">";
std::cout <<(opt.colorize?"\033[m":"") <<"\n";
}
Events::const_iterator ei=events.find(assm.addr);
std::cout <<std::setw(4) <<std::right <<assm.func_id <<"." <<std::setw(6) <<std::left <<assm.pos
<<(prev_func_id==assm.func_id && prev_position+1==assm.pos ? "|" : "#");
if (ei!=events.end() && ei->second.nexecuted>0) {
std::cout <<std::setw(9) <<std::right <<ei->second.nexecuted <<"x ";
} else {
std::cout <<std::string(11, ' ');
}
std::cout <<StringUtility::addrToString(assm.addr) <<": "
<<(opt.colorize?"\033[32m":"") <<assm.assembly <<(opt.colorize?"\033[m":"") <<"\n";
if (ei!=events.end())
show_events(ei->second);
prev_func_id = assm.func_id;
prev_position = assm.pos;
}
}
}
// Load all events into memory. Events are emitted for a particular function ID being analyzed, but if the 25-run-test
// --follow-calls was specified, then events for that function ID might be at instructions that are outside that function.
// We need to make note of those functions so that we can load all their instructions.
static void
load_events(const SqlDatabase::TransactionPtr &tx, int func_id, Events &events/*in,out*/)
{
int specimen_id = tx->statement("select specimen_id from semantic_functions where id = ?")
->bind(0, func_id)->execute_int();
SqlDatabase::StatementPtr stmt = tx->statement("select"
// 0 1 2 3 4
" event.addr, event.event_id, event.minor, event.val, func.id,"
// 5 6
" event.igroup_id, event.pos"
" from tmp_events as event"
" join semantic_instructions as insn on event.addr = insn.address"
" join semantic_functions as func on insn.func_id = func.id"
" where func.specimen_id = ?"
" order by igroup_id, pos");
stmt->bind(0, specimen_id);
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
rose_addr_t addr = row.get<rose_addr_t>(0);
int event_id = row.get<int>(1);
int minor = row.get<int>(2);
int64_t val = row.get<int64_t>(3);
events[addr].func_id = row.get<int>(4); // the hard-to-get ID, not the one stored in the events func_id column.
int igroup_id = row.get<int>(5);
int pos = row.get<int>(6);
switch (event_id) {
case CloneDetection::EV_REACHED: {
++events[addr].nexecuted;
break;
}
case CloneDetection::EV_BRANCHED: {
++events[addr].nbranches;
++events[addr].branches[val];
break;
}
case CloneDetection::EV_RETURNED: {
++events[addr].nreturns;
break;
}
case CloneDetection::EV_CONSUME_INPUT: {
++events[addr].ninputs;
assert(minor>=0);
if ((size_t)minor>=events[addr].inputs.size())
events[addr].inputs.resize(minor+1);
++events[addr].inputs[minor][val];
break;
}
case CloneDetection::EV_FAULT: {
CloneDetection::AnalysisFault::Fault fault = (CloneDetection::AnalysisFault::Fault)minor;
++events[addr].nfaults;
++events[addr].faults[fault];
break;
}
case CloneDetection::EV_MEM_WRITE: {
OutputEventKey output_key(igroup_id, val);
OutputEventValue output_val(pos, minor);
// Track final writes to each address
final_output_events[output_key] = output_val;
// Append event to the appropriate instruction
events[addr].outputs.push_back(std::make_pair(output_key, output_val));
}
default:
/*void*/
break;
}
}
}
/* Remove the functions from the compilation unit that is only available in one of the traces.
* - criteria complement of the functions from the files of the caller functions in the call trace is removed. */
std::pair<CallVec*, CallVec*>
remove_compilation_unit_complement(int func1_id, int func2_id, int igroup_id, int similarity, CallVec* func1_vec,
CallVec* func2_vec)
{
CallVec* new_func1_vec = new CallVec;
CallVec* new_func2_vec = new CallVec;
if (func1_vec->size() > 0 || func2_vec->size() > 0) {
// Find the set complement of functions called by the two functions
// - we are not interested in functions called by both
std::set<int> func1_vec_set;
std::set<int> func2_vec_set;
for (CallVec::iterator it = func1_vec->begin(); it != func1_vec->end(); ++it)
func1_vec_set.insert(*it);
for (CallVec::iterator it = func2_vec->begin(); it != func2_vec->end(); ++it)
func2_vec_set.insert(*it);
std::set<int> func1_func2_complement;
std::set_difference(func1_vec_set.begin(), func1_vec_set.end(), func2_vec_set.begin(), func2_vec_set.end(),
std::inserter(func1_func2_complement, func1_func2_complement.end()));
// Find the compilation units in question. A compilation unit is in our case a file.
SqlDatabase::StatementPtr func1_file_stmt = transaction->statement("select file_id from semantic_functions"
" where id = ?");
func1_file_stmt->bind(0, func1_id);
int func1_file_id = func1_file_stmt->execute_int();
SqlDatabase::StatementPtr func2_file_stmt = transaction->statement("select file_id from semantic_functions"
" where id = ?");
func2_file_stmt->bind(0, func2_id);
int func2_file_id = func2_file_stmt->execute_int();
// Find the functions that needs to be removed
// - all functions that has a clone in between the files
SqlDatabase::StatementPtr stmt = transaction->statement("select sem.func1_id, sem.func2_id from semantic_funcsim as sem"
" join semantic_functions as sf1 on sem.func1_id = sf1.id"
" join semantic_functions as sf2 on sem.func2_id = sf2.id"
" where similarity >= ? and sf1.file_id in (?,?)"
" and sf2.file_id in (?, ?) and sf1.file_id != sf2.file_id");
stmt->bind(0, similarity);
stmt->bind(1, func1_file_id);
stmt->bind(2, func2_file_id);
stmt->bind(3, func1_file_id);
stmt->bind(4, func2_file_id);
std::set<int> complement_functions;
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row) {
int clone_func1 = row.get<int>(0);
int clone_func2 = row.get<int>(1);
complement_functions.insert(clone_func1);
complement_functions.insert(clone_func2);
}
// Find the functions we want to remove
// - functions present with clones in between the files that is not part of both traces
std::set<int> remove_these;
std::set_intersection(complement_functions.begin(), complement_functions.end(), func1_func2_complement.begin(),
func1_func2_complement.end(), std::inserter(remove_these, remove_these.end()));
//prune functions to remove away from the call trace into new vectors
for (CallVec::iterator it = func1_vec->begin(); it != func1_vec->end(); ++it) {
if (remove_these.find(*it) == remove_these.end())
new_func1_vec->push_back(*it);
}
for (CallVec::iterator it = func2_vec->begin(); it != func2_vec->end(); ++it) {
if (remove_these.find(*it) == remove_these.end())
new_func2_vec->push_back(*it);
}
}
return std::pair<CallVec*, CallVec*>(new_func1_vec, new_func2_vec);
}
static void
postprocess(const SqlDatabase::TransactionPtr &tx)
{
int windowSize = tx->statement("select window_size from run_parameters limit 1")->execute_int();
int stride = tx->statement("select stride from run_parameters limit 1")->execute_int();
assert(windowSize != 0);
assert(stride != 0);
cerr << "About to delete from postprocessed_clusters" << endl;
tx->execute("delete from postprocessed_clusters");
cerr << "... done" << endl;
cerr << "About to postprocess" << endl;
SqlDatabase::StatementPtr cmd = tx->statement("select cluster, function_id, index_within_function, vectors_row"
" from clusters order by cluster, function_id, index_within_function");
SqlDatabase::StatementPtr insertCmd = tx->statement("insert into postprocessed_clusters"
" select * from clusters where row_number = ?");
const size_t numStridesThatMustBeDifferent = windowSize / (stride * 2);
string last_cluster = "";
string last_func_id = "";
size_t last_index_within_function = 0;
vector<string> rows_in_this_cluster;
bool first = true;
for (SqlDatabase::Statement::iterator postproc_reader=cmd->begin(); postproc_reader!=cmd->end(); ++postproc_reader) {
string cluster = postproc_reader.get<std::string>(0);
string function_id = postproc_reader.get<std::string>(1);
size_t index_within_function = postproc_reader.get<size_t>(2);
string cluster_row_number = postproc_reader.get<std::string>(3);
bool differentFunction = cluster != last_cluster || function_id != last_func_id;
bool endingCluster = differentFunction;
bool beginningNewCluster = first || differentFunction;
first = false;
if (endingCluster) {
if (rows_in_this_cluster.size() > 1) { // Skip clusters that have only one element left
for (size_t i = 0; i < rows_in_this_cluster.size(); ++i) {
insertCmd->bind(0, rows_in_this_cluster[i]);
insertCmd->execute();
}
}
}
if (beginningNewCluster) {
last_cluster = cluster;
last_func_id = function_id;
last_index_within_function = index_within_function;
rows_in_this_cluster.clear();
}
bool keep = beginningNewCluster || (index_within_function >= last_index_within_function + numStridesThatMustBeDifferent);
if (keep) {
last_index_within_function = index_within_function;
rows_in_this_cluster.push_back(cluster_row_number);
}
}
cerr << "... done" << endl;
}
int main(int argc, char* argv[])
{
std::string database;
//Timing
struct timeval before;
struct rusage ru_before;
gettimeofday(&before, NULL);
getrusage(RUSAGE_SELF, &ru_before);
try {
options_description desc("Allowed options");
desc.add_options()
("help", "produce a help message")
("database,q", value< string >()->composing(),
"the sqlite database that we are to use")
;
variables_map vm;
store(command_line_parser(argc, argv).options(desc)
.run(), vm);
if (vm.count("help")) {
cout << desc;
exit(0);
}
if (vm.count("database")!=1 ) {
std::cerr << "Missing options. Call as: findClones --database <database-name>"
<< std::endl;
exit(1);
}
database = vm["database"].as<string >();
cout << "database: " << database << std::endl;
}
catch(exception& e) {
cout << e.what() << "\n";
}
SqlDatabase::TransactionPtr tx = SqlDatabase::Connection::create(database)->transaction();
try {
tx->statement("create table IF NOT EXISTS largest_clones(row_number INTEGER PRIMARY KEY, function_id_A INTEGER, begin_index_within_function_A INTEGER, end_index_within_function_A INTEGER,"
"function_id_B INTEGER , begin_index_within_function_B INTEGER, end_index_within_function_B INTEGER )")->execute();
}
catch(exception &ex) {
cerr << "Exception Occurred: " << ex.what() << endl;
}
try {
tx->statement("delete from largest_clones")->execute();
}
catch(exception &ex) {
cerr << "Exception Occurred: " << ex.what() << endl;
}
string filen = database;
std::vector<std::pair <Element, Element> > listOfClonePairs;
std::list<std::pair<int,int> > listOfFunctionClonePairs;
int windowSize = 0;
int stride = 0;
try {
windowSize = tx->statement("select window_size from run_parameters limit 1")->execute_int();
} catch (exception& ex) {cerr << "Exception Occurred: " << ex.what() << endl;}
try {
stride = tx->statement("select stride from run_parameters limit 1")->execute_int();
} catch (exception& ex) {cerr << "Exception Occurred: " << ex.what() << endl;}
assert (windowSize != 0);
assert (stride != 0);
//Create set of clone pairs
try{
std::string selectSeparateDatasets ="SELECT cluster, function_id, index_within_function, vectors_row from clusters ORDER BY cluster, function_id, index_within_function";
SqlDatabase::StatementPtr cmd = tx->statement(selectSeparateDatasets);
int64_t thisClusterName=-1;
std::vector<Element> thisCluster;
for (SqlDatabase::Statement::iterator r=cmd->begin(); r!=cmd->end(); ++r) {
Element cur_elem;
int64_t cluster = r.get<int64_t>(0);
cur_elem.function_id = r.get<int64_t>(1);
cur_elem.index_within_function = r.get<int64_t>(2);
cur_elem.last_index_within_function = cur_elem.index_within_function;
cur_elem.vectors_row = r.get<int64_t>(3);
// cur_elem.line = boost::lexical_cast<int> ( datasets.getstring(5) );
// cur_elem.offset = boost::lexical_cast<int> ( datasets.getstring(6) );
if( cluster == thisClusterName )
{
for( std::vector<Element>::iterator iItr = thisCluster.begin();
iItr != thisCluster.end(); ++iItr )
{
if( cur_elem.function_id < iItr->function_id)
listOfFunctionClonePairs.push_back(std::pair<int,int>(cur_elem.function_id, iItr->function_id));
else
listOfFunctionClonePairs.push_back(std::pair<int,int>( iItr->function_id, cur_elem.function_id ));
if( cur_elem < *iItr )
{
listOfClonePairs.push_back(std::pair<Element,Element>( cur_elem,*iItr ) );
}
else
{
listOfClonePairs.push_back(std::pair<Element,Element>(*iItr,cur_elem ) );
}
};
}else{
thisCluster.clear();
thisClusterName = cluster;
}
thisCluster.push_back( cur_elem );
}
}catch(exception &ex) {
cerr << "Exception Occured: " << ex.what() << endl;
}
listOfFunctionClonePairs.sort();
listOfFunctionClonePairs.unique();
for(std::list<std::pair<int,int> >::iterator iItr = listOfFunctionClonePairs.begin();
iItr != listOfFunctionClonePairs.end(); iItr++ )
{
if(iItr->first != iItr->second)
std::cout << iItr->second << " " << iItr->first << std::endl;
}
return 0;
};
int
main(int argc, char *argv[])
{
std::ios::sync_with_stdio();
argv0 = argv[0];
{
size_t slash = argv0.rfind('/');
argv0 = slash==std::string::npos ? argv0 : argv0.substr(slash+1);
if (0==argv0.substr(0, 3).compare("lt-"))
argv0 = argv0.substr(3);
}
Switches opt;
int argno = 1;
for (/*void*/; argno<argc && '-'==argv[argno][0]; ++argno) {
if (!strcmp(argv[argno], "--")) {
++argno;
break;
} else if (!strcmp(argv[argno], "--help") || !strcmp(argv[argno], "-h")) {
::usage(0);
} else if (!strncmp(argv[argno], "--entry=", 8)) {
opt.entry_vas.insert(strtoull(argv[argno]+8, NULL, 0));
} else if (!strcmp(argv[argno], "--file=list") || !strcmp(argv[argno], "--files=list")) {
opt.list_files = true;
} else if (!strncmp(argv[argno], "--file=", 7) || !strncmp(argv[argno], "--files=", 8)) {
std::vector<std::string> ids = StringUtility::split(",", strchr(argv[argno], '=')+1, (size_t)-1, true);
for (size_t i=0; i<ids.size(); ++i) {
const char *s = ids[i].c_str();
char *rest;
errno = 0;
int id = strtoul(s, &rest, 0);
if (errno || rest==s || *rest) {
std::cerr <<argv0 <<": invalid file ID: " <<ids[i] <<"\n";
exit(1);
}
opt.files.insert(id);
}
} else if (!strncmp(argv[argno], "--function=", 11) || !strncmp(argv[argno], "--functions=", 12)) {
std::vector<std::string> ids = StringUtility::split(",", strchr(argv[argno], '=')+1, (size_t)-1, true);
if (ids.size()==1 && isalpha(ids[0][0]) && ids[0].find_first_of('.')!=std::string::npos) {
std::vector<std::string> words = StringUtility::split(".", ids[0]);
if (words.size()!=2 ||
!SqlDatabase::is_valid_table_name(words[0]) || !SqlDatabase::is_valid_table_name(words[1])) {
std::cerr <<argv0 <<": --function switch needs either IDs or a database TABLE.COLUMN\n";
exit(1);
}
opt.function_table = words[0];
opt.function_column = words[1];
} else {
for (size_t i=0; i<ids.size(); ++i) {
const char *s = ids[i].c_str();
char *rest;
errno = 0;
int id = strtoul(s, &rest, 0);
if (errno || rest==s || *rest) {
std::cerr <<argv0 <<": invalid function ID: " <<ids[i] <<"\n";
exit(1);
}
opt.functions.insert(id);
}
}
} else if (!strncmp(argv[argno], "--first-fuzz=", 13)) {
opt.first_fuzz = strtoul(argv[argno]+13, NULL, 0);
} else if (!strncmp(argv[argno], "--name=", 7)) {
opt.names.insert(argv[argno]+7);
} else if (!strncmp(argv[argno], "--nfuzz=", 8)) {
opt.nfuzz = strtoul(argv[argno]+8, NULL, 0);
opt.nfuzz_set = true;
} else if (!strncmp(argv[argno], "--size=", 7)) {
opt.ninsns = strtoul(argv[argno]+7, NULL, 0);
} else if (!strcmp(argv[argno], "--specimen=list") || !strcmp(argv[argno], "--specimens=list")) {
opt.list_specimens = true;
} else if (!strncmp(argv[argno], "--specimen=", 11) || !strncmp(argv[argno], "--specimens=", 12)) {
std::vector<std::string> ids = StringUtility::split(",", strchr(argv[argno], '=')+1, (size_t)-1, true);
for (size_t i=0; i<ids.size(); ++i) {
const char *s = ids[i].c_str();
char *rest;
errno = 0;
int id = strtoul(s, &rest, 0);
if (errno || rest==s || *rest) {
std::cerr <<argv0 <<": invalid specimen ID: " <<ids[i] <<"\n";
exit(1);
}
opt.specimens.insert(id);
}
} else {
std::cerr <<argv0 <<": unrecognized switch: " <<argv[argno] <<"\n"
<<"see \"" <<argv0 <<" --help\" for usage info.\n";
exit(1);
}
}
if (argno+1!=argc)
::usage(1);
SqlDatabase::TransactionPtr tx = SqlDatabase::Connection::create(argv[argno++])->transaction();
// List the ID numbers and names for all specimen files
if (opt.list_specimens) {
SqlDatabase::Table<int, std::string> specimens;
specimens.insert(tx->statement("select file.id, file.name"
" from (select distinct specimen_id as id from semantic_functions) as specimen"
" join semantic_files as file on specimen.id = file.id"
" order by file.name"));
specimens.headers("File ID", "Specimen Name");
specimens.print(std::cout);
return 0;
}
// List the ID numbers and names for all files containing functions
if (opt.list_files) {
SqlDatabase::Table<int, std::string> files;
files.insert(tx->statement("select file.id, file.name"
" from (select distinct file_id as id from semantic_functions) as used"
" join semantic_files as file on used.id = file.id"
" order by file.name"));
files.headers("File ID", "Binary File Name");
files.print(std::cout);
return 0;
}
// Sanity checks
if (!opt.functions.empty() && !opt.function_table.empty()) {
std::cerr <<argv0 <<": --function=ID and --function=TABLE are mutually exclusive\n";
exit(1);
}
if (0==tx->statement("select count(*) from semantic_functions")->execute_int()) {
std::cerr <<argv0 <<": database has no functions; nothing to test\n";
return 0;
}
if (0==tx->statement("select count(*) from semantic_inputvalues")->execute_int()) {
std::cerr <<argv0 <<": database has no input groups; nothing to test\n";
return 0;
}
// Create table tmp_functions containing IDs for selected functions and their specimen IDs
std::vector<std::string> constraints;
if (!opt.entry_vas.empty())
constraints.push_back("func.entry_va " + SqlDatabase::in(opt.entry_vas));
if (!opt.names.empty())
constraints.push_back("func.name " + SqlDatabase::in_strings(opt.names, tx->driver()));
if (!opt.specimens.empty())
constraints.push_back("func.specimen_id " + SqlDatabase::in(opt.specimens));
if (!opt.files.empty())
constraints.push_back("func.file_id " + SqlDatabase::in(opt.files));
if (!opt.functions.empty())
constraints.push_back("func.id " + SqlDatabase::in(opt.functions));
if (opt.ninsns>0)
constraints.push_back("func.ninsns >= " + StringUtility::numberToString(opt.ninsns));
std::string sql1 = "select func.id, func.specimen_id from semantic_functions as func";
if (!opt.function_table.empty())
sql1 += " join "+opt.function_table+" as flist on func.id = flist."+opt.function_column;
if (!constraints.empty())
sql1 += " where " + StringUtility::join(" and ", constraints);
tx->execute("create temporary table tmp_functions as " + sql1);
// Create table tmp_inputgroups containing IDs for selected input groups
std::string sql2 = "select distinct igroup_id from semantic_inputvalues where igroup_id >= " +
StringUtility::numberToString(opt.first_fuzz);
if (opt.nfuzz_set)
sql2 += " and igroup_id < " + StringUtility::numberToString(opt.first_fuzz+opt.nfuzz);
tx->execute("create temporary table tmp_inputgroups as " + sql2);
// Create tmp_pending as the cross product of functions and inputgroups except for those already tested
tx->execute("create temporary table tmp_pending as"
" select func.specimen_id as specimen_id, func.id as func_id, igroup.igroup_id as igroup_id"
" from tmp_functions as func"
" join tmp_inputgroups as igroup"
" on igroup.igroup_id is not null" // "on" clause and "is not null" (rather than "true") for portability
" except"
" select func.specimen_id, func.id, fio.igroup_id"
" from semantic_fio as fio"
" join semantic_functions as func on fio.func_id=func.id");
SqlDatabase::StatementPtr stmt = tx->statement("select distinct specimen_id, func_id, igroup_id"
" from tmp_pending"
" order by specimen_id, igroup_id, func_id");
for (SqlDatabase::Statement::iterator row=stmt->begin(); row!=stmt->end(); ++row)
std::cout <<row.get<int>(0) <<"\t" <<row.get<int>(1) <<"\t" <<row.get<int>(2) <<"\n";
// no need to commit, but if we change this in the future, be sure to add begin_command()/finish_command()
return 0;
}