Switch from cchardet to uchardet

cchardet is not maintained anymore: https://github.com/PyYoshi/cChardet/issues/77

cchardet is based on uchardet with the addition of reporting encoding
detection confidence. We dont really need that, so moving to uchardet is
simplest.

See #1690 (Low effort port to charset_normalizer)

bypy/linux/__main__.py

@@ -45,7 +45,7 @@ def binary_includes():
             get_dll_path,
             ('usb-1.0 mtp expat sqlite3 ffi z lzma openjp2 poppler dbus-1 iconv xml2 xslt jpeg png16'
              ' webp webpmux webpdemux exslt ncursesw readline chm hunspell-1.7 hyphen'
-             ' icudata icui18n icuuc icuio stemmer gcrypt gpg-error'
+             ' icudata icui18n icuuc icuio stemmer gcrypt gpg-error uchardet'
              ' gobject-2.0 glib-2.0 gthread-2.0 gmodule-2.0 gio-2.0 dbus-glib-1').split()
         )) + [
             # debian/ubuntu for for some typical stupid reason use libpcre.so.3

bypy/macos/__main__.py

@@ -527,7 +527,7 @@ class Freeze:
     def add_misc_libraries(self):
         for x in (
             'usb-1.0.0', 'mtp.9', 'chm.0', 'sqlite3.0', 'hunspell-1.7.0',
-            'icudata.70', 'icui18n.70', 'icuio.70', 'icuuc.70', 'hyphen.0',
+            'icudata.70', 'icui18n.70', 'icuio.70', 'icuuc.70', 'hyphen.0', 'uchardet.0',
             'stemmer.0', 'xslt.1', 'exslt.0', 'xml2.2', 'z.1', 'unrar', 'lzma.5',
             'crypto.1.1', 'ssl.1.1', 'iconv.2',  # 'ltdl.7'
         ):

bypy/sources.json

@@ -700,11 +700,11 @@
     },
 
     {
-        "name": "cchardet",
+        "name": "uchardet",
         "unix": {
-            "filename": "cchardet-2.1.7.tar.gz",
-            "hash": "sha256:c428b6336545053c2589f6caf24ea32276c6664cb86db817e03a94c60afa0eaf",
-            "urls": ["pypi"]
+            "filename": "uchardet-0.0.7.tar.xz",
+            "hash": "sha256:3fc79408ae1d84b406922fa9319ce005631c95ca0f34b205fad867e8b30e45b1",
+            "urls": ["https://www.freedesktop.org/software/uchardet/releases/{filename}"]
         }
     },
 

setup/arch-ci.sh

@@ -5,7 +5,7 @@
 
 set -xe
 
-pacman -S --noconfirm --needed base-devel sudo git sip pyqt-builder cmake chmlib icu jxrlib hunspell libmtp libusb libwmf optipng podofo python-apsw python-beautifulsoup4 python-cssselect python-css-parser python-dateutil python-jeepney python-dnspython python-feedparser python-html2text python-html5-parser python-lxml python-markdown python-mechanize python-msgpack python-netifaces python-unrardll python-pillow python-psutil python-pygments python-pyqt6 python-regex python-zeroconf python-pyqt6-webengine qt6-svg qt6-imageformats udisks2 hyphen python-pychm python-pycryptodome speech-dispatcher python-sphinx python-urllib3 python-py7zr python-pip python-cchardet libstemmer poppler
+pacman -S --noconfirm --needed base-devel sudo git sip pyqt-builder cmake chmlib icu jxrlib hunspell libmtp libusb libwmf optipng podofo python-apsw python-beautifulsoup4 python-cssselect python-css-parser python-dateutil python-jeepney python-dnspython python-feedparser python-html2text python-html5-parser python-lxml python-markdown python-mechanize python-msgpack python-netifaces python-unrardll python-pillow python-psutil python-pygments python-pyqt6 python-regex python-zeroconf python-pyqt6-webengine qt6-svg qt6-imageformats udisks2 hyphen python-pychm python-pycryptodome speech-dispatcher python-sphinx python-urllib3 python-py7zr python-pip uchardet libstemmer poppler
 
 useradd -m ci
 chown -R ci:users $GITHUB_WORKSPACE

setup/build_environment.py

@@ -130,6 +130,7 @@ hunspell_inc_dirs = []
 hunspell_lib_dirs = []
 hyphen_inc_dirs = []
 hyphen_lib_dirs = []
+uchardet_inc_dirs, uchardet_lib_dirs, uchardet_libs = [], [], ['uchardet']
 openssl_inc_dirs, openssl_lib_dirs = [], []
 ICU = sw = ''
 
@@ -143,6 +144,8 @@ if iswindows:
     hyphen_lib_dirs = [sw_lib_dir]
     openssl_inc_dirs = [sw_inc_dir]
     openssl_lib_dirs = [sw_lib_dir]
+    uchardet_inc_dirs = [sw_inc_dir]
+    uchardet_lib_dirs = [sw_lib_dir]
     sqlite_inc_dirs = [sw_inc_dir]
     chmlib_inc_dirs = [sw_inc_dir]
     chmlib_lib_dirs = [sw_lib_dir]
@@ -165,6 +168,7 @@ elif ismacos:
     podofo_lib = sw_lib_dir
     ft_libs = ['freetype']
     ft_inc_dirs = [sw + '/include/freetype2']
+    uchardet_inc_dirs = [sw + '/include/uchardet']
     SSL = os.environ.get('OPENSSL_DIR', os.path.join(sw, 'private', 'ssl'))
     openssl_inc_dirs = [os.path.join(SSL, 'include')]
     openssl_lib_dirs = [os.path.join(SSL, 'lib')]
@@ -183,6 +187,9 @@ else:
     if not os.path.exists(podofo_inc + '/podofo.h'):
         podofo_inc = os.path.join(sw, 'include', 'podofo')
         podofo_lib = os.path.join(sw, 'lib')
+    uchardet_inc_dirs = pkgconfig_include_dirs('uchardet', '', '/usr/include/uchardet')
+    uchardet_lib_dirs = pkgconfig_lib_dirs('uchardet', '', '/usr/lib')
+    uchardet_libs = pkgconfig_libs('uchardet', '', '')
 
 
 podofo_lib = os.environ.get('PODOFO_LIB_DIR', podofo_lib)

setup/extensions.json

@@ -16,6 +16,13 @@
         "lib_dirs": "!hyphen_lib_dirs",
 		"needs_c99": true
     },
+    {
+        "name": "uchardet",
+        "sources": "calibre/ebooks/uchardet.c",
+        "libraries": "!uchardet_libs",
+        "inc_dirs": "!uchardet_inc_dirs",
+        "lib_dirs": "!uchardet_lib_dirs"
+    },
     {
         "name": "unicode_names",
         "headers": "unicode_names/names.h unicode_names/data-types.h",

src/calibre/constants.py

@@ -266,6 +266,7 @@ class ExtensionsImporter:
             'tokenizer',
             'certgen',
             'sqlite_extension',
+            'uchardet',
         )
         if iswindows:
             extra = ('winutil', 'wpd', 'winfonts', 'winsapi')

src/calibre/ebooks/chardet.py

@@ -103,16 +103,18 @@ _CHARSET_ALIASES = {"macintosh" : "mac-roman", "x-sjis" : "shift-jis"}
 
 
 def detect(bytestring):
-    from cchardet import detect as implementation
-    ans = implementation(bytestring)
-    enc = ans.get('encoding')
-    if enc:
-        ans['encoding'] = enc.lower()
-    elif enc is None:
-        ans['encoding'] = ''
-    if ans.get('confidence') is None:
-        ans['confidence'] = 0
-    return ans
+    if isinstance(bytestring, str):
+        bytestring = bytestring.encode('utf-8', 'replace')
+    try:
+        from calibre_extensions.uchardet import detect as implementation
+    except ImportError:
+        # People running from source without updated binaries
+        from cchardet import detect as cdi
+
+        def implementation(x):
+            return cdi(x).get('encoding') or ''
+    enc = implementation(bytestring).lower()
+    return {'encoding': enc, 'confidence': 1 if enc else 0}
 
 
 def force_encoding(raw, verbose, assume_utf8=False):

src/calibre/ebooks/uchardet.c

@@ -0,0 +1,65 @@
+/*
+ * uchardet.c
+ * Copyright (C) 2022 Kovid Goyal <kovid at kovidgoyal.net>
+ *
+ * Distributed under terms of the GPL3 license.
+ */
+
+#include "Python.h"
+#include <uchardet.h>
+
+#define CAPSULE_NAME "uchardet.detector_capsule"
+#define CAPSULE_ATTR "detector_capsule"
+
+static PyObject*
+detect(PyObject *self, PyObject *bytes) {
+    if (!PyBytes_Check(bytes)) { PyErr_SetString(PyExc_TypeError, "a byte string is required"); return NULL; }
+    PyObject *capsule = PyObject_GetAttrString(self, CAPSULE_ATTR);
+    if (!capsule) return NULL;
+    void *d = PyCapsule_GetPointer(capsule, CAPSULE_NAME);
+    if (!d) return NULL;
+    uchardet_reset(d);
+    uchardet_handle_data(d, PyBytes_AS_STRING(bytes), (size_t)PyBytes_GET_SIZE(bytes));
+    uchardet_data_end(d);
+    return PyUnicode_FromString(uchardet_get_charset(d));
+}
+
+static PyMethodDef methods[] = {
+    {"detect", detect, METH_O,
+    "detect(bytestring) -> encoding name\n\n"
+    		"Detect the encoding of the specified bytestring"
+    },
+    {NULL, NULL, 0, NULL}
+};
+
+
+static void
+free_detector(PyObject *capsule) {
+    void *d = PyCapsule_GetPointer(capsule, CAPSULE_NAME);
+    if (d) uchardet_delete(d);
+}
+
+static int
+exec_module(PyObject *module) {
+    uchardet_t detector = uchardet_new();
+    if (!detector) { PyErr_NoMemory(); return -1; }
+    PyObject *detector_capsule = PyCapsule_New(detector, CAPSULE_NAME, free_detector);
+    if (!detector_capsule) return -1;
+    int ret = PyModule_AddObjectRef(module, CAPSULE_ATTR, detector_capsule);
+    Py_DECREF(detector_capsule);
+    return ret;
+}
+
+static PyModuleDef_Slot slots[] = { {Py_mod_exec, exec_module}, {0, NULL} };
+
+static struct PyModuleDef module_def = {
+    .m_base     = PyModuleDef_HEAD_INIT,
+    .m_name     = "uchardet",
+    .m_doc      = "Detect the encoding of bytestring",
+    .m_methods  = methods,
+    .m_slots    = slots,
+};
+
+CALIBRE_MODINIT_FUNC PyInit_uchardet(void) {
+    return PyModuleDef_Init(&module_def);
+}

src/calibre/test_build.py

@@ -73,11 +73,10 @@ class BuildTest(unittest.TestCase):
         del CHMFile, chmlib
 
     def test_chardet(self):
-        from cchardet import detect
+        from calibre_extensions.uchardet import detect
         raw = 'mūsi Füße'.encode()
-        data = detect(raw)
-        self.assertEqual(data['encoding'].lower(), 'utf-8')
-        self.assertGreater(data['confidence'], 0.5)
+        enc = detect(raw).lower()
+        self.assertEqual(enc, 'utf-8')
         # The following is used by html5lib
         from chardet.universaldetector import UniversalDetector
         detector = UniversalDetector()