Index: tests/apitest.cc =================================================================== --- tests/apitest.cc (revision 7293) +++ tests/apitest.cc (working copy) @@ -45,6 +45,13 @@ BackendManager backendmanager; Xapian::Database +get_database() +{ + vector dbnames; + return backendmanager.get_database(dbnames); +} + +Xapian::Database get_database(const string &dbname) { return backendmanager.get_database(dbname); @@ -101,6 +108,7 @@ RUNTESTS("inmemory", anydb); RUNTESTS("inmemory", specchar); RUNTESTS("inmemory", writabledb); + RUNTESTS("inmemory", writablelocaldb); RUNTESTS("inmemory", localdb); RUNTESTS("inmemory", positionaldb); RUNTESTS("inmemory", localpositionaldb); @@ -114,6 +122,7 @@ RUNTESTS("flint", anydb); RUNTESTS("flint", specchar); RUNTESTS("flint", writabledb); + RUNTESTS("flint", writablelocaldb); RUNTESTS("flint", localdb); RUNTESTS("flint", positionaldb); RUNTESTS("flint", localpositionaldb); @@ -129,6 +138,7 @@ RUNTESTS("quartz", anydb); RUNTESTS("quartz", specchar); RUNTESTS("quartz", writabledb); + RUNTESTS("quartz", writablelocaldb); RUNTESTS("quartz", localdb); RUNTESTS("quartz", positionaldb); RUNTESTS("quartz", localpositionaldb); Index: tests/apitest.h =================================================================== --- tests/apitest.h (revision 7293) +++ tests/apitest.h (working copy) @@ -26,6 +26,7 @@ #include +Xapian::Database get_database(); Xapian::Database get_database(const std::string &dbname); Xapian::Database get_database(const std::string &dbname, const std::string &dbname2); Index: tests/api_db.cc =================================================================== --- tests/api_db.cc (revision 7293) +++ tests/api_db.cc (working copy) @@ -1011,6 +1011,48 @@ return true; } +// tests all document postlists +static bool test_allpostlist1() +{ + Xapian::Database db(get_database("apitest_manydocs")); + Xapian::PostingIterator i = db.postlist_begin(""); + unsigned int j = 1; + while (i != db.postlist_end("")) { + TEST_EQUAL(*i, j); + i++; + j++; + } + TEST_EQUAL(j, 513); + return true; +} + +static void test_emptyterm1_helper(Xapian::Database & db) +{ + // Don't bother with postlist_begin() because allpostlist tests cover that. + TEST_EXCEPTION(Xapian::InvalidArgumentError, db.positionlist_begin(1, "")); + TEST_EQUAL(db.get_doccount(), db.get_termfreq("")); + TEST_EQUAL(db.get_doccount() != 0, db.term_exists("")); + TEST_EQUAL(db.get_doccount(), db.get_collection_freq("")); +} + +// tests results of passing an empty term to various methods +static bool test_emptyterm1() +{ + Xapian::Database db(get_database("apitest_manydocs")); + TEST_EQUAL(db.get_doccount(), 512); + test_emptyterm1_helper(db); + + db = get_database("apitest_onedoc"); + TEST_EQUAL(db.get_doccount(), 1); + test_emptyterm1_helper(db); + + db = get_database(); + TEST_EQUAL(db.get_doccount(), 0); + test_emptyterm1_helper(db); + + return true; +} + // tests collection frequency static bool test_collfreq1() { @@ -1379,6 +1421,8 @@ {"postlist4", test_postlist4}, {"postlist5", test_postlist5}, {"postlist6", test_postlist6}, + {"allpostlist1", test_allpostlist1}, + {"emptyterm1", test_emptyterm1}, {"termstats", test_termstats}, {"sortvalue1", test_sortvalue1}, // consistency1 will run on the remote backend, but it's particularly slow Index: tests/api_wrdb.cc =================================================================== --- tests/api_wrdb.cc (revision 7293) +++ tests/api_wrdb.cc (working copy) @@ -909,6 +909,76 @@ return true; } +// tests all document postlists +static bool test_allpostlist2() +{ + Xapian::WritableDatabase db(get_writable_database("apitest_manydocs")); + Xapian::PostingIterator i = db.postlist_begin(""); + unsigned int j = 1; + while (i != db.postlist_end("")) { + TEST_EQUAL(*i, j); + i++; + j++; + } + TEST_EQUAL(j, 513); + + db.delete_document(1); + db.delete_document(50); + db.delete_document(512); + + i = db.postlist_begin(""); + j = 2; + while (i != db.postlist_end("")) { + TEST_EQUAL(*i, j); + i++; + j++; + if (j == 50) j++; + } + TEST_EQUAL(j, 512); + + return true; +} + +static void test_emptyterm2_helper(Xapian::WritableDatabase & db) +{ + // Don't bother with postlist_begin() because allpostlist tests cover that. + TEST_EXCEPTION(Xapian::InvalidArgumentError, db.positionlist_begin(1, "")); + TEST_EQUAL(db.get_doccount(), db.get_termfreq("")); + TEST_EQUAL(db.get_doccount() != 0, db.term_exists("")); + TEST_EQUAL(db.get_doccount(), db.get_collection_freq("")); +} + +// tests results of passing an empty term to various methods +// equivalent of emptyterm1 for a writable database +static bool test_emptyterm2() +{ + Xapian::WritableDatabase db(get_writable_database("apitest_manydocs")); + TEST_EQUAL(db.get_doccount(), 512); + test_emptyterm2_helper(db); + db.delete_document(1); + TEST_EQUAL(db.get_doccount(), 511); + test_emptyterm2_helper(db); + db.delete_document(50); + TEST_EQUAL(db.get_doccount(), 510); + test_emptyterm2_helper(db); + db.delete_document(512); + TEST_EQUAL(db.get_doccount(), 509); + test_emptyterm2_helper(db); + + db = get_writable_database("apitest_onedoc"); + TEST_EQUAL(db.get_doccount(), 1); + test_emptyterm2_helper(db); + db.delete_document(1); + TEST_EQUAL(db.get_doccount(), 0); + test_emptyterm2_helper(db); + + db = get_writable_database(""); + TEST_EQUAL(db.get_doccount(), 0); + test_emptyterm2_helper(db); + + return true; +} + // Check that PHRASE/NEAR becomes AND if there's no positional info in the // database. static bool test_phraseorneartoand1() @@ -1035,7 +1105,14 @@ {"replacedoc3", test_replacedoc3}, {"replacedoc4", test_replacedoc4}, {"uniqueterm1", test_uniqueterm1}, + {"emptyterm2", test_emptyterm2}, {"phraseorneartoand1", test_phraseorneartoand1}, {"longpositionlist1", test_longpositionlist1}, {0, 0} }; + +/// The tests which use a writable, but local, backend +test_desc writablelocaldb_tests[] = { + {"allpostlist2", test_allpostlist2}, + {0, 0} +}; Index: tests/api_wrdb.h =================================================================== --- tests/api_wrdb.h (revision 7293) +++ tests/api_wrdb.h (working copy) @@ -27,5 +27,6 @@ #include "testsuite.h" extern test_desc writabledb_tests[]; +extern test_desc writablelocaldb_tests[]; #endif /* XAPIAN_HGUARD_API_WRDB_H */ Index: common/database.h =================================================================== --- common/database.h (revision 7293) +++ common/database.h (working copy) @@ -188,7 +188,7 @@ * use. */ LeafPostList * open_post_list(const string & tname) const { - if (!term_exists(tname)) { + if (!tname.empty() && !term_exists(tname)) { DEBUGLINE(MATCH, tname + " is not in database."); // Term doesn't exist in this database. However, we create // a (empty) postlist for it to help make distributed searching Index: api/omdatabase.cc =================================================================== --- api/omdatabase.cc (revision 7293) +++ api/omdatabase.cc (working copy) @@ -101,8 +101,6 @@ Database::postlist_begin(const string &tname) const { DEBUGAPICALL(PostingIterator, "Database::postlist_begin", tname); - if (tname.empty()) - throw InvalidArgumentError("Zero length terms are invalid"); // Don't bother checking that the term exists first. If it does, we // just end up doing more work, and if it doesn't, we save very little @@ -248,8 +246,9 @@ Database::get_termfreq(const string & tname) const { DEBUGAPICALL(Xapian::doccount, "Database::get_termfreq", tname); - if (tname.empty()) - throw InvalidArgumentError("Zero length terms are invalid"); + if (tname.empty()) { + return get_doccount(); + } Xapian::doccount tf = 0; vector >::const_iterator i; for (i = internal.begin(); i != internal.end(); i++) { @@ -262,8 +261,9 @@ Database::get_collection_freq(const string & tname) const { DEBUGAPICALL(Xapian::termcount, "Database::get_collection_freq", tname); - if (tname.empty()) - throw InvalidArgumentError("Zero length terms are invalid"); + if (tname.empty()) { + return get_doccount(); + } Xapian::termcount cf = 0; vector >::const_iterator i; @@ -303,8 +303,9 @@ bool Database::term_exists(const string & tname) const { - if (tname.empty()) - throw InvalidArgumentError("Zero length terms are invalid"); + if (tname.empty()) { + return get_doccount() != 0; + } vector >::const_iterator i; for (i = internal.begin(); i != internal.end(); ++i) { if ((*i)->term_exists(tname)) return true; Index: backends/inmemory/inmemory_database.cc =================================================================== --- backends/inmemory/inmemory_database.cc (revision 7293) +++ backends/inmemory/inmemory_database.cc (working copy) @@ -260,6 +260,89 @@ return Xapian::PositionIterator(db->open_position_list(did, (*pos).tname)); } +///////////////////////////// +// InMemoryAllDocsPostList // +///////////////////////////// + +InMemoryAllDocsPostList::InMemoryAllDocsPostList(Xapian::Internal::RefCntPtr db_) + : did(0), db(db_) +{ +} + +Xapian::doccount +InMemoryAllDocsPostList::get_termfreq() const +{ + return db->totdocs; +} + +Xapian::docid +InMemoryAllDocsPostList::get_docid() const +{ + Assert(did > 0); + Assert(did <= db->termlists.size()); + Assert(db->termlists[did - 1].is_valid); + return did; +} + +Xapian::doclength +InMemoryAllDocsPostList::get_doclength() const +{ + return db->get_doclength(did); +} + +Xapian::termcount +InMemoryAllDocsPostList::get_wdf() const +{ + return 1; +} + +PositionList * +InMemoryAllDocsPostList::read_position_list() +{ + throw Xapian::UnimplementedError("Can't open position list for all docs iterator"); +} + +PositionList * +InMemoryAllDocsPostList::open_position_list() const +{ + throw Xapian::UnimplementedError("Can't open position list for all docs iterator"); +} + +PostList * +InMemoryAllDocsPostList::next(Xapian::weight /*w_min*/) +{ + Assert(!at_end()); + do { + ++did; + } while (did <= db->termlists.size() && !db->termlists[did - 1].is_valid); + return NULL; +} + +PostList * +InMemoryAllDocsPostList::skip_to(Xapian::docid did_, Xapian::weight /*w_min*/) +{ + Assert(!at_end()); + if (did <= did_) { + did = did_; + while (did <= db->termlists.size() && !db->termlists[did - 1].is_valid) { + ++did; + } + } + return NULL; +} + +bool +InMemoryAllDocsPostList::at_end() const +{ + return (did > db->termlists.size()); +} + +string +InMemoryAllDocsPostList::get_description() const +{ + return "InMemoryAllDocsPostList" + om_tostring(did); +} + /////////////////////////// // Actual database class // /////////////////////////// @@ -279,7 +362,11 @@ LeafPostList * InMemoryDatabase::do_open_post_list(const string & tname) const { - Assert(tname.size() != 0); + if (tname.empty()) { + if (termlists.empty()) + return new EmptyPostList(); + return new InMemoryAllDocsPostList(Xapian::Internal::RefCntPtr(this)); + } map::const_iterator i = postlists.find(tname); if (i == postlists.end() || i->second.term_freq == 0) return new EmptyPostList(); Index: backends/inmemory/inmemory_database.h =================================================================== --- backends/inmemory/inmemory_database.h (revision 7293) +++ backends/inmemory/inmemory_database.h (working copy) @@ -159,6 +159,35 @@ string get_description() const; }; +/** A PostList over all docs in an inmemory database. + */ +class InMemoryAllDocsPostList : public LeafPostList { + friend class InMemoryDatabase; + private: + Xapian::docid did; + + Xapian::Internal::RefCntPtr db; + + InMemoryAllDocsPostList(Xapian::Internal::RefCntPtr db); + public: + Xapian::doccount get_termfreq() const; + + Xapian::docid get_docid() const; // Gets current docid + Xapian::doclength get_doclength() const; // Length of current document + Xapian::termcount get_wdf() const; // Within Document Frequency + PositionList * read_position_list(); + PositionList * open_position_list() const; + + PostList *next(Xapian::weight w_min); // Moves to next docid + + PostList *skip_to(Xapian::docid did, Xapian::weight w_min); // Moves to next docid >= specified docid + + // True if we're off the end of the list + bool at_end() const; + + string get_description() const; +}; + // Term List class InMemoryTermList : public LeafTermList { friend class InMemoryDatabase; @@ -193,6 +222,7 @@ * This is a prototype database, mainly used for debugging and testing. */ class InMemoryDatabase : public Xapian::Database::Internal { + friend class InMemoryAllDocsPostList; private: map postlists; vector termlists;