More work on de-vendoring pykakasi

COPYRIGHT

@@ -323,13 +323,6 @@ License: GPL-3
  The full text of the GPL is distributed as in
  /usr/share/common-licenses/GPL-3 on Debian systems.
 
-Files: src/calibre/ebooks/unihandecode/pykakasi/*
-Copyright: 2011, Hiroshi Miura <[email protected]>
-Copyright: 1992, Hironobu Takahashi
-License: GPL-2+
- The full text of the GPL is distributed as in
- /usr/share/common-licenses/GPL on Debian systems.
-
 Files: src/calibre/ebooks/unihandecode/*
 Copyright: 2010-2011, Hiroshi Miura <[email protected]>
 Copyright: 2009, John Schember

bypy/linux/__main__.py

@@ -85,11 +85,12 @@ def ignore_in_lib(base, items, ignored_dirs=None):
         ignored_dirs = {'.svn', '.bzr', '.git', 'test', 'tests', 'testing'}
     for name in items:
         path = j(base, name)
+        is_kakasi = 'pykakasi' in path
         if os.path.isdir(path):
-            if name != 'plugins' and (name in ignored_dirs or not is_package_dir(path)):
+            if name != 'plugins' and (name in ignored_dirs or not is_package_dir(path)) and not (is_kakasi and name == 'data'):
                 ans.append(name)
         else:
-            if name.rpartition('.')[-1] not in ('so', 'py'):
+            if name.rpartition('.')[-1] not in ('so', 'py') and not (is_kakasi and name.endswith('.db')):
                 ans.append(name)
     return ans
 

bypy/macos/__main__.py

@@ -614,12 +614,15 @@ def add_packages_from_dir(self, src):
 
     @flush
     def add_package_dir(self, x, dest=None):
+        is_kakasi = 'pykakasi' in x
+        allowed_exts = ('', '.py', '.so')
+        if is_kakasi:
+            allowed_exts += ('.db',)
         def ignore(root, files):
             ans = []
             for y in files:
                 ext = os.path.splitext(y)[1]
-                if ext not in ('', '.py', '.so') or \
-                        (not ext and not os.path.isdir(join(root, y))):
+                if ext not in allowed_exts or (not ext and not os.path.isdir(join(root, y))):
                     ans.append(y)
 
             return ans

bypy/sources.json

@@ -1043,6 +1043,24 @@
         }
     },
 
+    {
+        "name": "wrapt",
+        "unix": {
+            "filename": "wrapt-1.16.0-py3-none-any.whl",
+            "hash": "sha256:6906c4100a8fcbf2fa735f6059214bb13b97f75b1a61777fcf6432121ef12ef1",
+            "urls": ["pypi"]
+        }
+    },
+
+    {
+        "name": "deprecated",
+        "unix": {
+            "filename": "Deprecated-1.2.14-py2.py3-none-any.whl",
+            "hash": "sha256:6fac8b097794a90302bdbb17b9b815e732d3c4720583ff1b198499d78470466c",
+            "urls": ["pypi"]
+        }
+    },
+
     {
         "name": "pykakasi",
         "unix": {

setup/commands.py

@@ -11,7 +11,7 @@
     'gui',
     'git_version',
     'develop', 'install',
-    'kakasi', 'rapydscript', 'cacerts', 'recent_uas', 'resources',
+    'rapydscript', 'cacerts', 'recent_uas', 'resources',
     'check', 'test', 'test_rs', 'upgrade_source_code',
     'sdist', 'bootstrap', 'extdev',
     'manual', 'tag_release',
@@ -90,10 +90,9 @@
 test = Test()
 test_rs = TestRS()
 
-from setup.resources import CACerts, Kakasi, RapydScript, RecentUAs, Resources
+from setup.resources import CACerts, RapydScript, RecentUAs, Resources
 
 resources = Resources()
-kakasi = Kakasi()
 cacerts = CACerts()
 recent_uas = RecentUAs()
 rapydscript = RapydScript()

setup/resources.py

@@ -9,13 +9,11 @@
 import glob
 import json
 import os
-import re
 import shutil
 import zipfile
-from zlib import compress
 
-from polyglot.builtins import codepoint_to_chr, iteritems, itervalues, only_unicode_recursive
-from setup import Command, __appname__, basenames, download_securely, dump_json
+from polyglot.builtins import iteritems, itervalues, only_unicode_recursive
+from setup import Command, basenames, download_securely, dump_json
 
 
 def get_opts_from_parser(parser):
@@ -29,113 +27,6 @@ def do_opt(opt):
             yield from do_opt(o)
 
 
-class Kakasi(Command):  # {{{
-
-    description = 'Compile resources for unihandecode'
-
-    KAKASI_PATH = os.path.join(Command.SRC,  __appname__,
-            'ebooks', 'unihandecode', 'pykakasi')
-
-    def run(self, opts):
-        self.records = {}
-        src = self.j(self.KAKASI_PATH, 'kakasidict.utf8')
-        dest = self.j(self.RESOURCES, 'localization',
-                'pykakasi','kanwadict2.calibre_msgpack')
-        base = os.path.dirname(dest)
-        if not os.path.exists(base):
-            os.makedirs(base)
-
-        if self.newer(dest, src):
-            self.info('\tGenerating Kanwadict')
-
-            for line in open(src, "rb"):
-                self.parsekdict(line)
-            self.kanwaout(dest)
-
-        src = self.j(self.KAKASI_PATH, 'itaijidict.utf8')
-        dest = self.j(self.RESOURCES, 'localization',
-                'pykakasi','itaijidict2.calibre_msgpack')
-
-        if self.newer(dest, src):
-            self.info('\tGenerating Itaijidict')
-            self.mkitaiji(src, dest)
-
-        src = self.j(self.KAKASI_PATH, 'kanadict.utf8')
-        dest = self.j(self.RESOURCES, 'localization',
-                'pykakasi','kanadict2.calibre_msgpack')
-
-        if self.newer(dest, src):
-            self.info('\tGenerating kanadict')
-            self.mkkanadict(src, dest)
-
-    def mkitaiji(self, src, dst):
-        dic = {}
-        for line in open(src, "rb"):
-            line = line.decode('utf-8').strip()
-            if line.startswith(';;'):  # skip comment
-                continue
-            if re.match(r"^$",line):
-                continue
-            pair = re.sub(r'\\u([0-9a-fA-F]{4})', lambda x:codepoint_to_chr(int(x.group(1),16)), line)
-            dic[pair[0]] = pair[1]
-        from calibre.utils.serialize import msgpack_dumps
-        with open(dst, 'wb') as f:
-            f.write(msgpack_dumps(dic))
-
-    def mkkanadict(self, src, dst):
-        dic = {}
-        for line in open(src, "rb"):
-            line = line.decode('utf-8').strip()
-            if line.startswith(';;'):  # skip comment
-                continue
-            if re.match(r"^$",line):
-                continue
-            (alpha, kana) = line.split(' ')
-            dic[kana] = alpha
-        from calibre.utils.serialize import msgpack_dumps
-        with open(dst, 'wb') as f:
-            f.write(msgpack_dumps(dic))
-
-    def parsekdict(self, line):
-        line = line.decode('utf-8').strip()
-        if line.startswith(';;'):  # skip comment
-            return
-        (yomi, kanji) = line.split(' ')
-        if ord(yomi[-1:]) <= ord('z'):
-            tail = yomi[-1:]
-            yomi = yomi[:-1]
-        else:
-            tail = ''
-        self.updaterec(kanji, yomi, tail)
-
-    def updaterec(self, kanji, yomi, tail):
-        key = "%04x"%ord(kanji[0])
-        if key in self.records:
-            if kanji in self.records[key]:
-                rec = self.records[key][kanji]
-                rec.append((yomi,tail))
-                self.records[key].update({kanji: rec})
-            else:
-                self.records[key][kanji]=[(yomi, tail)]
-        else:
-            self.records[key] = {}
-            self.records[key][kanji]=[(yomi, tail)]
-
-    def kanwaout(self, out):
-        from calibre.utils.serialize import msgpack_dumps
-        with open(out, 'wb') as f:
-            dic = {}
-            for k, v in iteritems(self.records):
-                dic[k] = compress(msgpack_dumps(v))
-            f.write(msgpack_dumps(dic))
-
-    def clean(self):
-        kakasi = self.j(self.RESOURCES, 'localization', 'pykakasi')
-        if os.path.exists(kakasi):
-            shutil.rmtree(kakasi)
-# }}}
-
-
 class CACerts(Command):  # {{{
 
     description = 'Get updated mozilla CA certificate bundle'
@@ -213,7 +104,7 @@ def run(self, opts):
 class Resources(Command):  # {{{
 
     description = 'Compile various needed calibre resources'
-    sub_commands = ['kakasi', 'liberation_fonts', 'mathjax', 'rapydscript', 'hyphenation', 'piper_voices']
+    sub_commands = ['liberation_fonts', 'mathjax', 'rapydscript', 'hyphenation', 'piper_voices']
 
     def run(self, opts):
         from calibre.utils.serialize import msgpack_dumps
@@ -337,8 +228,6 @@ def clean(self):
             x = self.j(self.RESOURCES, x+'.pickle')
             if os.path.exists(x):
                 os.remove(x)
-        from setup.commands import kakasi
-        kakasi.clean()
         for x in ('builtin_recipes.xml', 'builtin_recipes.zip',
                 'template-functions.json', 'user-manual-translation-stats.json'):
             x = self.j(self.RESOURCES, x)

src/calibre/ebooks/unihandecode/jadecoder.py

@@ -16,15 +16,48 @@
 Copyright (c) 2010 Hiroshi Miura
 '''
 
+import pickle
 import re
+from importlib.resources import files
 
 from pykakasi import kakasi
+from pykakasi.kanji import Itaiji, Kanwa
+from pykakasi.properties import Configurations
+from pykakasi.scripts import Jisyo
 
 from calibre.ebooks.unihandecode.jacodepoints import CODEPOINTS as JACODES
 from calibre.ebooks.unihandecode.unicodepoints import CODEPOINTS
 from calibre.ebooks.unihandecode.unidecoder import Unidecoder
 
 
+# pykakasi uses paths for its dictionaries rather than using the
+# Traversable API of importlib.resources so we have to hack around it, sigh.
+def dictdata(dbfile: str):
+    t = files('pykakasi')
+    q = t.joinpath('data').joinpath(dbfile)
+    return q.read_bytes()
+
+
+def jisyo_init(self, dbname):
+    self._dict = pickle.loads(dictdata(dbname))
+
+
+def itaiji_init(self):
+    if self._itaijidict is None:
+        with self._lock:
+            if self._itaijidict is None:
+                self._itaijidict = pickle.loads(dictdata(Configurations.jisyo_itaiji))
+
+def kanwa_init(self):
+    if self._jisyo_table is None:
+        with self._lock:
+            if self._jisyo_table is None:
+                self._jisyo_table = pickle.loads(dictdata(Configurations.jisyo_kanwa))
+
+Jisyo.__init__ = jisyo_init
+Itaiji.__init__ = itaiji_init
+Kanwa.__init__ = kanwa_init
+
 class Jadecoder(Unidecoder):
 
     def __init__(self):