in FTS5, reject an APSW connection if amalgamation build is used

README.rst

@@ -72,7 +72,9 @@ Requirements
    * SQLite 3.10.2 and older versions do not have extra requirements. 2-arg fts3_tokenizer is always avaiable.
    * SQLite 3.12.0 and later vesrions do not have extra requirements. 2-arg fts3_tokenizer can be enabled dynamically.
 
-Note for APSW users: An APSW Amalgamation build does not expose SQLite APIs used in this module, so libsqlite3.so/sqlite3.dll is also required even it has no runtime library dependencies on SQLite. An APSW local build already depends on the shared library. Detail: sqlite3_db_config can be invoked via Connection.config, but it rejects SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER to register a new tokenizer. tested at APSW 3.21.0-r1.
+Note for APSW users:
+ * An APSW Amalgamation build does not expose SQLite APIs used in this module, so libsqlite3.so/sqlite3.dll is also required even it has no runtime library dependencies on SQLite. An APSW local build already depends on the shared library. Detail: sqlite3_db_config can be invoked via Connection.config, but it rejects SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER to register a new tokenizer. tested at APSW 3.21.0-r1.
+ * sqlitefts.fts5 does not support APSW Amalgamation build. see GH-14_
 
 Licence
 =======
@@ -98,3 +100,4 @@ Thanks
 .. |appveyor build status| image:: https://ci.appveyor.com/api/projects/status/github/hideaki-t/sqlite-fts-python?svg=true
 .. _appveyor build status: https://ci.appveyor.com/project/hideaki-t/sqlite-fts-python
 .. _APSW: https://github.com/rogerbinns/apsw
+.. _GH-14: https://github.com/hideaki-t/sqlite-fts-python/issues/14

sqlitefts/fts5.py

@@ -5,6 +5,7 @@
 import struct
 from typing import Callable, Any, Union, Optional, TYPE_CHECKING, Iterable, Tuple, Dict
 
+from .error import Error
 from .tokenizer import ffi, dll, get_db_from_connection, SQLITE_OK
 from .fts3 import Tokenizer as FTS3Tokenizer
 if TYPE_CHECKING:
@@ -132,6 +133,11 @@ def tokenize(self, text: str, flags: int=0) -> Iterable[Tuple[str, int, int]]:
 
 
 def fts5_api_from_db(c: 'Union[sqlite3.Connection, apsw.Connection]'):
+    if not hasattr(c, 'commit'):
+        # APSW doesn't have conn.commit/rollback
+        import apsw
+        if apsw.using_amalgamation:
+            raise Error('unable to get fts5_api')
     cur = c.cursor()
     try:
         cur.execute('SELECT sqlite_version()')

tests/test_apsw.py

@@ -6,6 +6,7 @@
 apsw = pytest.importorskip('apsw')
 
 import sqlitefts as fts
+from sqlitefts import fts5, fts5_aux
 
 
 class SimpleTokenizer(fts.Tokenizer):
@@ -20,6 +21,18 @@ def tokenize(self, text):
             yield t, p, p + l
 
 
+class SimpleFTS5Tokenizer(fts5.FTS5Tokenizer):
+    _p = re.compile(r'\w+', re.UNICODE)
+
+    def tokenize(self, text, flags):
+        for m in self._p.finditer(text):
+            s, e = m.span()
+            t = text[s:e]
+            l = len(t.encode('utf-8'))
+            p = len(text[:s].encode('utf-8'))
+            yield t, p, p + l
+
+
 def test_createtable():
     c = apsw.Connection(':memory:')
     name = 'simple'
@@ -258,3 +271,31 @@ def test_tokenizer_output():
                 "SELECT token, start, end, position "
                 "FROM tok1 WHERE input=?", [s]), expect):
             assert e == a
+
+
+@pytest.mark.xfail(apsw.using_amalgamation,
+                   reason='FTS5 with APSW+Amalgamation not supported')
+def test_fts5_api_from_db():
+    with apsw.Connection(':memory:') as c:
+        fts5api = fts5.fts5_api_from_db(c)
+        assert fts5api.iVersion == 2
+        assert fts5api.xCreateTokenizer
+
+
+@pytest.mark.xfail(apsw.using_amalgamation,
+                   reason='FTS5 with APSW+Amalgamation not supported',
+                   raises=fts.Error)
+def test_aux_and_tokenize():
+    c = apsw.Connection(':memory:')
+    try:
+        fts5_aux.register_aux_function(c, 'tokenize', fts5_aux.aux_tokenize)
+        cur = c.cursor()
+        cur.execute('CREATE VIRTUAL TABLE fts USING FTS5(content)')
+        cur.executemany('INSERT INTO fts VALUES(?)',
+                        (['hello world'], ['こんにちは 世界']))
+        cur.execute('SELECT COUNT(*) FROM fts')
+        assert 2 == cur.fetchone()[0]
+        cur.execute('SELECT tokenize(fts, 0) FROM fts')
+        assert [x[0] for x in cur.fetchall()] == ['hello, world', 'こんにちは, 世界']
+    finally:
+        c.close()