[util] Improve scientific notation handling in js_to_json (closes #14789)

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 92b22e639b30af5e93cce69fa8c057be2cb1e0c0..2fe9cf585db817e1d86831c71ef28502b4a16ee5 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -39,6 +39,7 @@ from .compat import (
     compat_HTMLParser,
     compat_basestring,
     compat_chr,
+    compat_ctypes_WINFUNCTYPE,
     compat_etree_fromstring,
     compat_expanduser,
     compat_html_entities,
@@ -159,6 +160,8 @@ DATE_FORMATS = (
     '%Y-%m-%dT%H:%M',
     '%b %d %Y at %H:%M',
     '%b %d %Y at %H:%M:%S',
+    '%B %d %Y at %H:%M',
+    '%B %d %Y at %H:%M:%S',
 )
 
 DATE_FORMATS_DAY_FIRST = list(DATE_FORMATS)
@@ -1328,24 +1331,24 @@ def _windows_write_string(s, out):
     if fileno not in WIN_OUTPUT_IDS:
         return False
 
-    GetStdHandle = ctypes.WINFUNCTYPE(
+    GetStdHandle = compat_ctypes_WINFUNCTYPE(
         ctypes.wintypes.HANDLE, ctypes.wintypes.DWORD)(
-        (b'GetStdHandle', ctypes.windll.kernel32))
+        ('GetStdHandle', ctypes.windll.kernel32))
     h = GetStdHandle(WIN_OUTPUT_IDS[fileno])
 
-    WriteConsoleW = ctypes.WINFUNCTYPE(
+    WriteConsoleW = compat_ctypes_WINFUNCTYPE(
         ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE, ctypes.wintypes.LPWSTR,
         ctypes.wintypes.DWORD, ctypes.POINTER(ctypes.wintypes.DWORD),
-        ctypes.wintypes.LPVOID)((b'WriteConsoleW', ctypes.windll.kernel32))
+        ctypes.wintypes.LPVOID)(('WriteConsoleW', ctypes.windll.kernel32))
     written = ctypes.wintypes.DWORD(0)
 
-    GetFileType = ctypes.WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)((b'GetFileType', ctypes.windll.kernel32))
+    GetFileType = compat_ctypes_WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)(('GetFileType', ctypes.windll.kernel32))
     FILE_TYPE_CHAR = 0x0002
     FILE_TYPE_REMOTE = 0x8000
-    GetConsoleMode = ctypes.WINFUNCTYPE(
+    GetConsoleMode = compat_ctypes_WINFUNCTYPE(
         ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE,
         ctypes.POINTER(ctypes.wintypes.DWORD))(
-        (b'GetConsoleMode', ctypes.windll.kernel32))
+        ('GetConsoleMode', ctypes.windll.kernel32))
     INVALID_HANDLE_VALUE = ctypes.wintypes.DWORD(-1).value
 
     def not_a_console(handle):
@@ -1835,10 +1838,20 @@ def parse_duration(s):
         days, hours, mins, secs, ms = m.groups()
     else:
         m = re.match(
-            r'''(?ix)(?:P?T)?
+            r'''(?ix)(?:P?
+                (?:
+                    [0-9]+\s*y(?:ears?)?\s*
+                )?
+                (?:
+                    [0-9]+\s*m(?:onths?)?\s*
+                )?
+                (?:
+                    [0-9]+\s*w(?:eeks?)?\s*
+                )?
                 (?:
                     (?P<days>[0-9]+)\s*d(?:ays?)?\s*
                 )?
+                T)?
                 (?:
                     (?P<hours>[0-9]+)\s*h(?:ours?)?\s*
                 )?
@@ -1933,7 +1946,7 @@ class PagedList(object):
 
 
 class OnDemandPagedList(PagedList):
-    def __init__(self, pagefunc, pagesize, use_cache=False):
+    def __init__(self, pagefunc, pagesize, use_cache=True):
         self._pagefunc = pagefunc
         self._pagesize = pagesize
         self._use_cache = use_cache
@@ -2254,7 +2267,7 @@ def js_to_json(code):
         "(?:[^"\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^"\\]*"|
         '(?:[^'\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^'\\]*'|
         {comment}|,(?={skip}[\]}}])|
-        [a-zA-Z_][.a-zA-Z_0-9]*|
+        (?:(?<![0-9])[eE]|[a-df-zA-DF-Z_])[.a-zA-Z_0-9]*|
         \b(?:0[xX][0-9a-fA-F]+|0+[0-7]+)(?:{skip}:)?|
         [0-9]+(?={skip}:)
         '''.format(comment=COMMENT_RE, skip=SKIP_RE), fix_kv, code)
@@ -2340,6 +2353,7 @@ def mimetype2ext(mt):
         'ttml+xml': 'ttml',
         'x-flv': 'flv',
         'x-mp4-fragmented': 'mp4',
+        'x-ms-sami': 'sami',
         'x-ms-wmv': 'wmv',
         'mpegurl': 'm3u8',
         'x-mpegurl': 'm3u8',
@@ -2362,7 +2376,7 @@ def parse_codecs(codecs_str):
     vcodec, acodec = None, None
     for full_codec in splited_codecs:
         codec = full_codec.split('.')[0]
-        if codec in ('avc1', 'avc2', 'avc3', 'avc4', 'vp9', 'vp8', 'hev1', 'hev2', 'h263', 'h264', 'mp4v'):
+        if codec in ('avc1', 'avc2', 'avc3', 'avc4', 'vp9', 'vp8', 'hev1', 'hev2', 'h263', 'h264', 'mp4v', 'hvc1'):
             if not vcodec:
                 vcodec = full_codec
         elif codec in ('mp4a', 'opus', 'vorbis', 'mp3', 'aac', 'ac-3', 'ec-3', 'eac3', 'dtsc', 'dtse', 'dtsh', 'dtsl'):