Implement high performance C module for sqlite custom aggregators. To enable edit line 153 of library.sqlite. In my testing showed between 1 and 5% decrease in startup time (varies depending on how many multi-author books in the library). Disabled for now, pending more testing

setup/build_environment.py

@@ -90,11 +90,13 @@ fc_lib = '/usr/lib'
 podofo_inc = '/usr/include/podofo'
 podofo_lib = '/usr/lib'
 chmlib_inc_dirs = chmlib_lib_dirs = []
+sqlite_inc_dirs = []
 
 if iswindows:
     prefix  = r'C:\cygwin\home\kovid\sw'
     sw_inc_dir  = os.path.join(prefix, 'include')
     sw_lib_dir  = os.path.join(prefix, 'lib')
+    sqlite_inc_dirs = [sw_inc_dir]
     fc_inc = os.path.join(sw_inc_dir, 'fontconfig')
     fc_lib = sw_lib_dir
     chmlib_inc_dirs = consolidate('CHMLIB_INC_DIR', os.path.join(prefix,

setup/extensions.py

@@ -18,7 +18,7 @@ from setup.build_environment import fc_inc, fc_lib, chmlib_inc_dirs, \
         QMAKE, msvc, MT, win_inc, win_lib, png_inc_dirs, win_ddk, \
         magick_inc_dirs, magick_lib_dirs, png_lib_dirs, png_libs, \
         magick_error, magick_libs, ft_lib_dirs, ft_libs, jpg_libs, \
-        jpg_lib_dirs, chmlib_lib_dirs
+        jpg_lib_dirs, chmlib_lib_dirs, sqlite_inc_dirs
 MT
 isunix = islinux or isosx or isfreebsd
 
@@ -58,6 +58,11 @@ if iswindows:
 
 extensions = [
 
+    Extension('sqlite_custom',
+        ['calibre/library/sqlite_custom.c'],
+        inc_dirs=sqlite_inc_dirs
+        ),
+
     Extension('chmlib',
             ['calibre/utils/chm/swig_chm.c'],
             libraries=['ChmLib' if iswindows else 'chm'],

setup/installer/windows/notes.rst

@@ -32,6 +32,12 @@ Run the following command to install python dependencies::
 
 Install BeautifulSoup 3.0.x manually into site-packages (3.1.x parses broken HTML very poorly)
 
+
+SQLite
+---------
+
+Put sqlite3*.h from the sqlite windows amlgamation in ~/sw/include
+
 Qt
 --------
 

src/calibre/library/sqlite.py

@@ -7,7 +7,7 @@ __docformat__ = 'restructuredtext en'
 Wrapper for multi-threaded access to a single sqlite database connection. Serializes
 all calls.
 '''
-import sqlite3 as sqlite, traceback, time, uuid
+import sqlite3 as sqlite, traceback, time, uuid, sys, os
 from sqlite3 import IntegrityError, OperationalError
 from threading import Thread
 from Queue import Queue
@@ -19,6 +19,7 @@ from calibre.ebooks.metadata import title_sort, author_to_author_sort
 from calibre.utils.config import tweaks
 from calibre.utils.date import parse_date, isoformat
 from calibre import isbytestring
+from calibre.constants import iswindows, DEBUG
 
 global_lock = RLock()
 
@@ -114,6 +115,22 @@ def pynocase(one, two, encoding='utf-8'):
             pass
     return cmp(one.lower(), two.lower())
 
+
+def load_c_extensions(conn, debug=DEBUG):
+    try:
+        conn.enable_load_extension(True)
+        ext_path = os.path.join(sys.extensions_location, 'sqlite_custom.'+
+                ('pyd' if iswindows else 'so'))
+        conn.load_extension(ext_path)
+        conn.enable_load_extension(False)
+        return True
+    except Exception, e:
+        if debug:
+            print 'Failed to load high performance sqlite C extension'
+            print e
+    return False
+
+
 class DBThread(Thread):
 
     CLOSE = '-------close---------'
@@ -131,9 +148,12 @@ class DBThread(Thread):
     def connect(self):
         self.conn = sqlite.connect(self.path, factory=Connection,
                                    detect_types=sqlite.PARSE_DECLTYPES|sqlite.PARSE_COLNAMES)
+        self.conn.execute('pragma cache_size=5000')
         encoding = self.conn.execute('pragma encoding').fetchone()[0]
+        c_ext_loaded = False #load_c_extensions(self.conn)
         self.conn.row_factory = sqlite.Row if self.row_factory else  lambda cursor, row : list(row)
         self.conn.create_aggregate('concat', 1, Concatenate)
+        if not c_ext_loaded:
             self.conn.create_aggregate('sortconcat', 2, SortedConcatenate)
             self.conn.create_aggregate('sort_concat', 2, SafeSortedConcatenate)
         self.conn.create_collation('PYNOCASE', partial(pynocase,
@@ -263,3 +283,9 @@ def connect(dbpath, row_factory=None):
     if conn.proxy.unhandled_error[0] is not None:
         raise DatabaseException(*conn.proxy.unhandled_error)
     return conn
+
+def test():
+    c = sqlite.connect(':memory:')
+    if load_c_extensions(c, True):
+        print 'Loaded C extension successfully'
+

src/calibre/library/sqlite_custom.c

@@ -0,0 +1,173 @@
+#define UNICODE
+#include <Python.h>
+
+
+#include <stdlib.h>
+
+#include <sqlite3ext.h>
+SQLITE_EXTENSION_INIT1
+
+#ifdef _MSC_VER
+#define MYEXPORT __declspec(dllexport)
+#else
+#define MYEXPORT
+#endif
+
+// sortconcat {{{
+
+typedef struct {
+    unsigned char *val;
+    int index;
+    int length;
+} SortConcatItem;
+
+typedef struct {
+    SortConcatItem **vals;
+    int count;
+    int length;
+} SortConcatList;
+
+static void sort_concat_step(sqlite3_context *context, int argc, sqlite3_value **argv) {
+    const unsigned char *val;
+    int idx, sz;
+    SortConcatList *list;
+
+    assert(argc == 2);
+
+    list = (SortConcatList*) sqlite3_aggregate_context(context, sizeof(*list));
+    if (list == NULL) return;
+
+    if (list->vals == NULL) {
+        list->vals = (SortConcatItem**)calloc(100, sizeof(SortConcatItem*));
+        if (list->vals == NULL) return;
+        list->length = 100;
+        list->count = 0;
+    }
+
+    if (list->count == list->length) {
+        list->vals = (SortConcatItem**)realloc(list->vals, list->length + 100);
+        if (list->vals == NULL) return;
+        list->length = list->length + 100;
+    }
+
+    list->vals[list->count] = (SortConcatItem*)calloc(1, sizeof(SortConcatItem));
+    if (list->vals[list->count] == NULL) return;
+    
+    idx = sqlite3_value_int(argv[0]);
+    val = sqlite3_value_text(argv[1]);
+    sz  = sqlite3_value_bytes(argv[1]);
+    if (idx == 0 || val == NULL || sz == 0) {free(list->vals[list->count]); return;}
+
+
+
+    list->vals[list->count]->val = (unsigned char*)calloc(sz, sizeof(unsigned char));
+    if (list->vals[list->count]->val == NULL) 
+        {free(list->vals[list->count]); return;}
+    list->vals[list->count]->index = idx;
+    list->vals[list->count]->length = sz;
+    memcpy(list->vals[list->count]->val, val, sz);
+    list->count = list->count + 1;
+
+}
+
+static void sort_concat_free(SortConcatList *list) {
+    int i;
+    if (list == NULL) return;
+    for (i = 0; i < list->count; i++) {
+        free(list->vals[i]->val);
+        free(list->vals[i]);
+    }
+}
+
+static int sort_concat_cmp(const void *a_, const void *b_) {
+    return (*((SortConcatItem**)a_))->index - (*((SortConcatItem**)b_))->index;
+}
+
+static unsigned char* sort_concat_do_finalize(SortConcatList *list, const unsigned char join) {
+    unsigned char *ans, *pos;
+    int sz = 0, i;
+
+    for (i = 0; i < list->count; i++) { 
+        sz += list->vals[i]->length;
+    }
+    sz += list->count;
+
+    ans = (unsigned char *) calloc(sz, sizeof(unsigned char));
+    if (ans == NULL) return ans;
+
+    pos = ans;
+    for (i = 0; i < list->count; i++) {
+        if (list->vals[i]->length > 0) {
+            memcpy(pos, list->vals[i]->val, list->vals[i]->length);
+            pos += list->vals[i]->length;
+            if (i < list->count -1) { *pos = join; pos += 1; }
+        }
+    }
+
+    return ans;
+
+}
+
+static void sort_concat_finalize(sqlite3_context *context) {
+    SortConcatList *list;
+    unsigned char *ans;
+
+    list = (SortConcatList*) sqlite3_aggregate_context(context, sizeof(*list));
+
+    if (list != NULL && list->vals != NULL && list->count > 0) {
+        qsort(list->vals, list->count, sizeof(list->vals[0]), sort_concat_cmp);
+        ans = sort_concat_do_finalize(list, ',');
+        if (ans != NULL) sqlite3_result_text(context, (char*)ans, -1, SQLITE_TRANSIENT);
+        free(ans);
+        sort_concat_free(list);
+    }
+
+}
+
+static void sort_concat_finalize2(sqlite3_context *context) {
+    SortConcatList *list;
+    unsigned char *ans;
+
+    list = (SortConcatList*) sqlite3_aggregate_context(context, sizeof(*list));
+
+    if (list != NULL && list->vals != NULL && list->count > 0) {
+        qsort(list->vals, list->count, sizeof(list->vals[0]), sort_concat_cmp);
+        ans = sort_concat_do_finalize(list, '|');
+        if (ans != NULL) sqlite3_result_text(context, (char*)ans, -1, SQLITE_TRANSIENT);
+        free(ans);
+        sort_concat_free(list);
+    }
+
+}
+
+// }}}
+
+MYEXPORT int sqlite3_extension_init(
+    sqlite3 *db, char **pzErrMsg, const sqlite3_api_routines *pApi){
+  SQLITE_EXTENSION_INIT2(pApi);
+  sqlite3_create_function(db, "sortconcat", 2, SQLITE_UTF8, NULL, NULL, sort_concat_step, sort_concat_finalize);
+  sqlite3_create_function(db, "sort_concat", 2, SQLITE_UTF8, NULL, NULL, sort_concat_step, sort_concat_finalize2);
+  return 0;
+}
+
+static PyObject *
+sqlite_custom_init_funcs(PyObject *self, PyObject *args) {
+    Py_RETURN_NONE;
+}
+
+static PyMethodDef sqlite_custom_methods[] = {
+    {"init_funcs", sqlite_custom_init_funcs, METH_VARARGS,
+        "init_funcs()\n\nInitialize module."
+    },
+
+    {NULL, NULL, 0, NULL}
+};
+
+PyMODINIT_FUNC
+initsqlite_custom(void) {
+    PyObject *m;
+    m = Py_InitModule3("sqlite_custom", sqlite_custom_methods,
+    "Implementation of custom sqlite methods in C for speed."
+    );
+    if (m == NULL) return;
+}