Parse Accept-Language header

src/calibre/srv/http_request.py

@@ -154,6 +154,7 @@ class HTTPRequest(Connection):
 
     request_handler = None
     static_cache = None
+    translator_cache = None
 
     def __init__(self, *args, **kwargs):
         Connection.__init__(self, *args, **kwargs)

src/calibre/srv/http_response.py

@@ -19,7 +19,9 @@ from calibre.srv.loop import WRITE
 from calibre.srv.errors import HTTPSimpleResponse
 from calibre.srv.http_request import HTTPRequest, read_headers
 from calibre.srv.sendfile import file_metadata, sendfile_to_socket_async, CannotSendfile, SendfileInterrupted
-from calibre.srv.utils import MultiDict, http_date, HTTP1, HTTP11, socket_errors_socket_closed
+from calibre.srv.utils import (
+    MultiDict, http_date, HTTP1, HTTP11, socket_errors_socket_closed,
+    sort_q_values, get_translator_for_lang)
 from calibre.utils.monotonic import monotonic
 
 Range = namedtuple('Range', 'start stop size')
@@ -71,21 +73,19 @@ def parse_if_none_match(val):  # {{{
 # }}}
 
 def acceptable_encoding(val, allowed=frozenset({'gzip'})):  # {{{
-    def enc(x):
-        e, r = x.partition(';')[::2]
-        p, v = r.partition('=')[::2]
-        q = 1.0
-        if p == 'q' and v:
-            try:
-                q = float(v)
-            except Exception:
-                pass
-        return e.lower(), q
+    for x in sort_q_values(val):
+        x = x.lower()
+        if x in allowed:
+            return x
+# }}}
 
-    emap = dict(enc(x.strip()) for x in val.split(','))
-    acceptable = sorted(set(emap) & allowed, key=emap.__getitem__, reverse=True)
-    if acceptable:
-        return acceptable[0]
+def preferred_lang(val, get_translator_for_lang):  # {{{
+    for x in sort_q_values(val):
+        x = x.lower()
+        found, lang, translator = get_translator_for_lang(x)
+        if found:
+            return x
+    return 'en'
 # }}}
 
 def get_ranges(headervalue, content_length):  # {{{
@@ -180,9 +180,13 @@ def get_range_parts(ranges, content_type, content_length):  # {{{
 
 class RequestData(object):  # {{{
 
-    def __init__(self, method, path, query, inheaders, request_body_file, outheaders, response_protocol, static_cache, opts, remote_addr, remote_port):
-        self.method, self.path, self.query, self.inheaders, self.request_body_file, self.outheaders, self.response_protocol, self.static_cache = (
-            method, path, query, inheaders, request_body_file, outheaders, response_protocol, static_cache
+    def __init__(self, method, path, query, inheaders, request_body_file, outheaders, response_protocol,
+                 static_cache, opts, remote_addr, remote_port, translator_cache):
+
+        (self.method, self.path, self.query, self.inheaders, self.request_body_file, self.outheaders,
+         self.response_protocol, self.static_cache, self.translator_cache) = (
+            method, path, query, inheaders, request_body_file, outheaders,
+            response_protocol, static_cache, translator_cache
         )
         self.remote_addr, self.remote_port = remote_addr, remote_port
         self.opts = opts
@@ -196,6 +200,12 @@ class RequestData(object):  # {{{
 
     def read(self, size=-1):
         return self.request_body_file.read(size)
+
+    def get_translator(self, bcp_47_code):
+        return get_translator_for_lang(self.translator_cache, bcp_47_code)
+
+    def get_preferred_language(self):
+        return preferred_lang(self.outheaders.get('Accept-Language'), self.get_translator)
 # }}}
 
 class ReadableOutput(object):
@@ -322,7 +332,7 @@ class HTTPConnection(HTTPRequest):
         data = RequestData(
             self.method, self.path, self.query, inheaders, request_body_file,
             outheaders, self.response_protocol, self.static_cache, self.opts,
-            self.remote_addr, self.remote_port
+            self.remote_addr, self.remote_port, self.translator_cache
         )
         self.queue_job(self.run_request_handler, data)
 
@@ -545,10 +555,12 @@ class HTTPConnection(HTTPRequest):
 
 def create_http_handler(handler):
     static_cache = {}
+    translator_cache = {}
     @wraps(handler)
     def wrapper(*args, **kwargs):
         ans = HTTPConnection(*args, **kwargs)
         ans.request_handler = handler
         ans.static_cache = static_cache
+        ans.translator_cache = translator_cache
         return ans
     return wrapper

src/calibre/srv/tests/http.py

@@ -66,6 +66,18 @@ class TestHTTP(BaseTest):
         test('Priority', '1;q=0.5, 2;q=0.75, 3;q=1.0', '3', {'1', '2', '3'})
     # }}}
 
+    def test_accept_language(self):  # {{{
+        'Test parsing of Accept-Language'
+        from calibre.srv.http_response import preferred_lang
+        def test(name, val, ans):
+            self.ae(preferred_lang(val, lambda x:(True, x, None)), ans, name + ' failed')
+        test('Empty field', '', 'en')
+        test('Simple', 'de', 'de')
+        test('Case insensitive', 'Es', 'es')
+        test('Multiple', 'fr, es', 'fr')
+        test('Priority', 'en;q=0.1, de;q=0.7, fr;q=0.5', 'de')
+    # }}}
+
     def test_range_parsing(self):  # {{{
         'Test parsing of Range header'
         from calibre.srv.http_response import get_ranges

src/calibre/srv/utils.py

@@ -12,10 +12,12 @@ from urlparse import parse_qs
 import repr as reprlib
 from email.utils import formatdate
 from operator import itemgetter
+from future_builtins import map
 
 from calibre import prints
 from calibre.constants import iswindows
 from calibre.utils.filenames import atomic_rename
+from calibre.utils.localization import get_translator
 from calibre.utils.socket_inheritance import set_socket_inherit
 from calibre.utils.logging import ThreadSafeLog
 
@@ -169,6 +171,22 @@ def create_sock_pair(port=0):
 
     return client_sock, srv_sock
 
+def sort_q_values(header_val):
+    'Get sorted items from an HTTP header of type: a;q=0.5, b;q=0.7...'
+    if not header_val:
+        return []
+    def item(x):
+        e, r = x.partition(';')[::2]
+        p, v = r.partition('=')[::2]
+        q = 1.0
+        if p == 'q' and v:
+            try:
+                q = max(0.0, min(1.0, float(v.strip())))
+            except Exception:
+                pass
+        return e.strip(), q
+    return tuple(map(itemgetter(0), sorted(map(item, header_val.split(',')), key=itemgetter(1), reverse=True)))
+
 def eintr_retry_call(func, *args, **kwargs):
     while True:
         try:
@@ -178,6 +196,14 @@ def eintr_retry_call(func, *args, **kwargs):
                 continue
             raise
 
+def get_translator_for_lang(cache, bcp_47_code):
+    try:
+        return cache[bcp_47_code]
+    except KeyError:
+        pass
+    cache[bcp_47_code] = ans = get_translator(bcp_47_code)
+    return ans
+
 # Logging {{{
 
 class ServerLog(ThreadSafeLog):

src/calibre/utils/localization.py

@@ -107,6 +107,30 @@ def get_all_translators():
                 buf = cStringIO.StringIO(zf.read(mpath + '/messages.mo'))
                 yield lang, GNUTranslations(buf)
 
+def get_single_translator(mpath):
+    from zipfile import ZipFile
+    with ZipFile(P('localization/locales.zip', allow_user_override=False), 'r') as zf:
+        buf = cStringIO.StringIO(zf.read(mpath + '/messages.mo'))
+        return GNUTranslations(buf)
+
+def get_translator(bcp_47_code):
+    parts = bcp_47_code.replace('-', '_').split('_')[:2]
+    parts[0] = lang_as_iso639_1(parts[0].lower())
+    if len(parts) > 1:
+        parts[1] = parts[1].upper()
+    lang = '_'.join(parts)
+    lang = {'pt':'pt_BR', 'zh':'zh_CN'}.get(lang, lang)
+    available = available_translations()
+    found = True
+    if lang not in available:
+        lang = {'pt':'pt_BR', 'zh':'zh_CN'}.get(parts[0], parts[0])
+        if lang not in available:
+            lang = get_lang()
+            found = False
+    if lang == 'en':
+        return found, lang, NullTranslations()
+    return found, lang, get_single_translator(lang)
+
 lcdata = {
     u'abday': (u'Sun', u'Mon', u'Tue', u'Wed', u'Thu', u'Fri', u'Sat'),
     u'abmon': (u'Jan', u'Feb', u'Mar', u'Apr', u'May', u'Jun', u'Jul', u'Aug', u'Sep', u'Oct', u'Nov', u'Dec'),