py3: misc fixes for the server

20 out of 34 tests for the server now pass

src/calibre/srv/auth.py

@@ -68,12 +68,18 @@ def as_bytestring(x):
     return x
 
 
+def as_unicodestring(x):
+    if isinstance(x, bytes):
+        x = x.decode('utf-8')
+    return x
+
+
 def md5_hex(s):
-    return md5(as_bytestring(s)).hexdigest().decode('ascii')
+    return as_unicodestring(md5(as_bytestring(s)).hexdigest())
 
 
 def sha256_hex(s):
-    return sha256(as_bytestring(s)).hexdigest().decode('ascii')
+    return as_unicodestring(sha256(as_bytestring(s)).hexdigest())
 
 
 def base64_decode(s):

src/calibre/srv/content.py

@@ -9,7 +9,7 @@ __copyright__ = '2015, Kovid Goyal <kovid at kovidgoyal.net>'
 import os, errno
 from io import BytesIO
 from threading import Lock
-from polyglot.builtins import map
+from polyglot.builtins import map, unicode_type
 from functools import partial
 
 from calibre import fit_image, sanitize_file_name
@@ -167,7 +167,7 @@ def book_filename(rd, book_id, mi, fmt, as_encoded_unicode=False):
     if as_encoded_unicode:
         # See https://tools.ietf.org/html/rfc6266
         fname = sanitize_file_name(fname).encode('utf-8')
-        fname = quote(fname).decode('ascii')
+        fname = unicode_type(quote(fname))
     else:
         fname = ascii_filename(fname).replace('"', '_')
     return fname

src/calibre/srv/http_request.py

@@ -266,6 +266,7 @@ class HTTPRequest(Connection):
 
         try:
             method, uri, req_protocol = line.strip().split(b' ', 2)
+            req_protocol = req_protocol.decode('ascii')
             rp = int(req_protocol[5]), int(req_protocol[7])
             self.method = method.decode('ascii').upper()
         except Exception:

src/calibre/srv/http_response.py

@@ -13,7 +13,7 @@ from itertools import chain, repeat
 from operator import itemgetter
 from functools import wraps
 
-from polyglot.builtins import iteritems, itervalues, reraise, map, is_py3
+from polyglot.builtins import iteritems, itervalues, reraise, map, is_py3, unicode_type, string_or_bytes
 
 from calibre import guess_type, force_unicode
 from calibre.constants import __version__, plugins, ispy3
@@ -247,8 +247,7 @@ class RequestData(object):  # {{{
 
     def filesystem_file_with_custom_etag(self, output, *etag_parts):
         etag = hashlib.sha1()
-        string = type('')
-        tuple(map(lambda x:etag.update(string(x)), etag_parts))
+        tuple(map(lambda x:etag.update(unicode_type(x).encode('utf-8')), etag_parts))
         return ETaggedFile(output, etag.hexdigest())
 
     def filesystem_file_with_constant_etag(self, output, etag_as_hexencoded_string):
@@ -312,7 +311,10 @@ class ReadableOutput(object):
 def filesystem_file_output(output, outheaders, stat_result):
     etag = getattr(output, 'etag', None)
     if etag is None:
-        etag = hashlib.sha1(type('')(stat_result.st_mtime) + force_unicode(output.name or '')).hexdigest()
+        oname = output.name or ''
+        if not isinstance(oname, string_or_bytes):
+            oname = unicode_type(oname)
+        etag = hashlib.sha1((unicode_type(stat_result.st_mtime) + force_unicode(oname)).encode('utf-8')).hexdigest()
     else:
         output = output.output
     etag = '"%s"' % etag
@@ -356,7 +358,7 @@ class GeneratedOutput(object):
 class StaticOutput(object):
 
     def __init__(self, data):
-        if isinstance(data, type('')):
+        if isinstance(data, unicode_type):
             data = data.encode('utf-8')
         self.data = data
         self.etag = '"%s"' % hashlib.sha1(data).hexdigest()
@@ -649,12 +651,17 @@ class HTTPConnection(HTTPRequest):
         if stat_result is not None:
             output = filesystem_file_output(output, outheaders, stat_result)
             if 'Content-Type' not in outheaders:
-                mt = guess_type(output.name)[0]
+                output_name = output.name
+                if not isinstance(output_name, string_or_bytes):
+                    output_name = unicode_type(output_name)
+                mt = guess_type(output_name)[0]
                 if mt:
                     if mt in {'text/plain', 'text/html', 'application/javascript', 'text/css'}:
                         mt += '; charset=UTF-8'
                     outheaders['Content-Type'] = mt
-        elif isinstance(output, (bytes, type(''))):
+                else:
+                    outheaders['Content-Type'] = 'application/octet-stream'
+        elif isinstance(output, string_or_bytes):
             output = dynamic_output(output, outheaders)
         elif hasattr(output, 'read'):
             output = ReadableOutput(output)

src/calibre/srv/tests/ajax.py

@@ -168,7 +168,7 @@ class ContentTest(LibraryBaseTest):
             ae = self.assertEqual
 
             def a(filename, data=None, status=OK, method='POST', username='12', add_duplicates='n', job_id=1):
-                r, data = make_request(conn, '/cdb/add-book/{}/{}/{}'.format(job_id, add_duplicates, quote(filename.encode('utf-8')).decode('ascii')),
+                r, data = make_request(conn, '/cdb/add-book/{}/{}/{}'.format(job_id, add_duplicates, quote(filename.encode('utf-8'))),
                                        username=username, password='test', prefix='', method=method, data=data)
                 ae(status, r.status)
                 return data

src/calibre/srv/tests/content.py

@@ -43,8 +43,8 @@ class ContentTest(LibraryBaseTest):
                 missing('/%s/C:/out.html' % prefix, b'Naughty, naughty!')
 
             def test_response(r):
-                self.assertIn(b'max-age=', r.getheader('Cache-Control'))
-                self.assertIn(b'public', r.getheader('Cache-Control'))
+                self.assertIn('max-age=', r.getheader('Cache-Control'))
+                self.assertIn('public', r.getheader('Cache-Control'))
                 self.assertIsNotNone(r.getheader('Expires'))
                 self.assertIsNotNone(r.getheader('ETag'))
                 self.assertIsNotNone(r.getheader('Content-Type'))
@@ -182,7 +182,7 @@ class ContentTest(LibraryBaseTest):
             self.ae(r.status, http_client.OK)
             self.ae(data, db.cover(2))
             self.ae(r.getheader('Used-Cache'), 'no')
-            path = from_hex_unicode(r.getheader('Tempfile')).decode('utf-8')
+            path = from_hex_unicode(r.getheader('Tempfile'))
             f, fdata = share_open(path, 'rb'), data
             # Now force an update
             change_cover(1)
@@ -190,7 +190,7 @@ class ContentTest(LibraryBaseTest):
             self.ae(r.status, http_client.OK)
             self.ae(data, db.cover(2))
             self.ae(r.getheader('Used-Cache'), 'no')
-            path = from_hex_unicode(r.getheader('Tempfile')).decode('utf-8')
+            path = from_hex_unicode(r.getheader('Tempfile'))
             f2, f2data = share_open(path, 'rb'), data
             # Do it again
             change_cover(2)

src/calibre/srv/tests/http.py

@@ -175,7 +175,7 @@ class TestHTTP(BaseTest):
             self.ae(r.getheader('Content-Length'), str(len(body)))
             self.ae(r.getheader('Content-Type'), 'text/plain; charset=UTF-8')
             self.ae(len(r.getheaders()), 3)
-            self.ae(r.read(), '')
+            self.ae(r.read(), b'')
             conn.request('GET', '/choose')
             r = conn.getresponse()
             self.ae(r.status, http_client.NOT_FOUND)
@@ -200,7 +200,7 @@ class TestHTTP(BaseTest):
             r = conn.getresponse()
             self.ae(r.status, http_client.MOVED_PERMANENTLY)
             self.ae(r.getheader('Location'), '/somewhere-else')
-            self.ae('', r.read())
+            self.ae(b'', r.read())
 
             server.change_handler(lambda data:data.path[0] + data.read().decode('ascii'))
             conn = server.connect(timeout=base_timeout * 5)
@@ -277,7 +277,10 @@ class TestHTTP(BaseTest):
             for i in range(10):
                 conn._HTTPConnection__state = http_client._CS_IDLE
                 conn.request('GET', '/%d'%i)
-                responses.append(conn.response_class(conn.sock, strict=conn.strict, method=conn._method))
+                if ispy3:
+                    responses.append(conn.response_class(conn.sock, method=conn._method))
+                else:
+                    responses.append(conn.response_class(conn.sock, strict=conn.strict, method=conn._method))
             for i in range(10):
                 r = responses[i]
                 r.begin()
@@ -288,7 +291,7 @@ class TestHTTP(BaseTest):
             server.loop.opts.timeout = 10  # ensure socket is not closed because of timeout
             conn.request('GET', '/close', headers={'Connection':'close'})
             r = conn.getresponse()
-            self.ae(r.status, 200), self.ae(r.read(), 'close')
+            self.ae(r.status, 200), self.ae(r.read(), b'close')
             server.loop.wakeup()
             num = 10
             while num and server.loop.num_active_connections != 0:
@@ -302,8 +305,8 @@ class TestHTTP(BaseTest):
             conn = server.connect(timeout=1)
             conn.request('GET', '/something')
             r = conn.getresponse()
-            self.ae(r.status, 200), self.ae(r.read(), 'something')
-            self.assertIn('Request Timeout', eintr_retry_call(conn.sock.recv, 500))
+            self.ae(r.status, 200), self.ae(r.read(), b'something')
+            self.assertIn(b'Request Timeout', eintr_retry_call(conn.sock.recv, 500))
     # }}}
 
     def test_http_response(self):  # {{{

src/calibre/srv/utils.py

@@ -284,13 +284,13 @@ def get_translator_for_lang(cache, bcp_47_code):
 
 def encode_path(*components):
     'Encode the path specified as a list of path components using URL encoding'
-    return '/' + '/'.join(urlquote(x.encode('utf-8'), '').decode('ascii') for x in components)
+    return '/' + '/'.join(urlquote(x.encode('utf-8'), '') for x in components)
 
 
 class Cookie(SimpleCookie):
 
     def _BaseCookie__set(self, key, real_value, coded_value):
-        if not isinstance(key, bytes):
+        if not ispy3 and not isinstance(key, bytes):
             key = key.encode('ascii')  # Python 2.x cannot handle unicode keys
         return SimpleCookie._BaseCookie__set(self, key, real_value, coded_value)
 

src/calibre/srv/web_socket.py

@@ -295,7 +295,7 @@ class WebSocketConnection(HTTPConnection):
         if self.method != 'GET':
             return self.simple_response(http_client.BAD_REQUEST, 'Invalid WebSocket method: %s' % self.method)
 
-        response = HANDSHAKE_STR % as_base64_unicode(sha1(key + GUID_STR).digest())
+        response = HANDSHAKE_STR % as_base64_unicode(sha1((key + GUID_STR).encode('utf-8')).digest())
         self.optimize_for_sending_packet()
         self.socket.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
         self.set_state(WRITE, self.upgrade_connection_to_ws, ReadOnlyFileBuffer(response.encode('ascii')), inheaders)

src/calibre/utils/imghdr.py

@@ -7,6 +7,7 @@ from __future__ import (unicode_literals, division, absolute_import,
 from struct import unpack, error
 import os
 from calibre.utils.speedups import ReadOnlyFileBuffer
+from calibre.constants import ispy3
 from polyglot.builtins import string_or_bytes
 
 """ Recognize image file formats and sizes based on their first few bytes."""
@@ -125,16 +126,23 @@ def jpeg_dimensions(stream):
             raise ValueError('Truncated JPEG data')
         return ans
 
-    x = b''
+    if ispy3:
+        def read_byte():
+            return read(1)[0]
+    else:
+        def read_byte():
+            return ord(read(1)[0])
+
+    x = None
     while True:
         # Find next marker
-        while x != b'\xff':
-            x = read(1)
+        while x != 0xff:
+            x = read_byte()
         # Soak up padding
-        marker = b'\xff'
-        while marker == b'\xff':
-            marker = read(1)
-        q = ord(marker[0])  # [0] needed for memoryview
+        marker = 0xff
+        while marker == 0xff:
+            marker = read_byte()
+        q = marker
         if 0xc0 <= q <= 0xcf and q != 0xc4 and q != 0xcc:
             # SOFn marker
             stream.seek(3, os.SEEK_CUR)

src/calibre/utils/speedups.py

@@ -100,6 +100,8 @@ def svg_path_to_painter_path(d):
     while data.tell() < end:
         last_cmd = cmd
         cmd = data.read(1) if repeated_command is None else repeated_command
+        if isinstance(cmd, memoryview):
+            cmd = cmd.tobytes()
         repeated_command = None
 
         if cmd == b' ':
@@ -176,7 +178,7 @@ def svg_path_to_painter_path(d):
                 x1, y1 = x, y
             x, y = parse_floats(2, x, y)
             path.quadTo(x1, y1, x, y)
-        elif cmd[0] in b'-.' or b'0' <= cmd[0] <= b'9':
+        elif cmd[0:1] in b'-.0123456789':
             # A new number begins
             # In this case, multiple parameters tuples are specified for the last command
             # We rewind to reparse data correctly