Add png->mkv muxing script and example file

The file is from wikipedia. Only 10 times larger than the gif (I never
said this was a good idea). Only mpv can play this file.

Hail Satan.

Author: wm4

matroska/png/example.mkv

@@ -0,0 +1,166 @@
+#!/usr/bin/env python3
+
+#
+# This file is part of mpv.
+#
+# mpv is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Lesser General Public
+# License as published by the Free Software Foundation; either
+# version 2.1 of the License, or (at your option) any later version.
+#
+# mpv is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public
+# License along with mpv.  If not, see <http://www.gnu.org/licenses/>.
+#
+
+import binascii
+import io
+import os
+import sys
+
+class Elem:
+
+    def __init__(self, ebml_id, sub = None):
+        self.ebml_id = ebml_id
+        self.sub = sub
+
+files = []
+outfile = None
+fps = 25
+for i in sys.argv[1:]:
+    if i.startswith("--o="):
+        assert outfile is None
+        outfile = open(i[4:], "wb")
+    elif i.startswith("--fps="):
+        fps = float(i[6:])
+    elif i.startswith("-"):
+        print("Unknown option '%s'" % i)
+        print("Usage: %s --fps=25 --o=outfile.mkv infile1.png infile2.png ..." %
+              sys.argv[0])
+        exit(1)
+    else:
+        files.append(open(i, "rb").read())
+
+if outfile is None or not files:
+    print("no")
+    exit(1)
+
+global duration
+duration = 0
+
+global clusters
+clusters = []
+
+for f in files:
+    # Making each image its own cluster is slightly wasteful, but simpler.
+    data = io.BytesIO()
+    data.write((0x80 | 1).to_bytes(1, "big")) # Track Number
+    data.write((0).to_bytes(2, "big")) # block Timestamp (relative to cluster's)
+    data.write((0).to_bytes(1, "big")) # Block Header flags
+    data.write(f)
+
+    block_duration = 1.0 / fps
+
+    clusters.append(
+        Elem('1f43b675', [ # Cluster
+            Elem('e7', int(duration * 1e9 / 1000000)), # Timestamp
+            Elem('a0', [ # BlockGroup
+                Elem('a1', data.getvalue()), # Block
+                Elem('9b', int(block_duration  * 1e9 / 1000000)), # BlockDuration
+            ])
+        ])
+    )
+
+    duration += block_duration
+
+def write_elem(dst, elem):
+    dst.write(binascii.unhexlify(elem.ebml_id))
+
+    content = io.BytesIO()
+
+    val = elem.sub
+    if type(val) == int:
+        # EBML/Matroska does have signed integers. But whether a value is signed
+        # or unsigned (i.e. meaning of the MSB) depends on the element ID.
+        # The Python type is ambiguous for this purpose. Fortunately, we use
+        # only unsigned elements.
+        assert(val >= 0)
+        blen = max((val.bit_length() + 7) // 8, 1)
+        assert(blen <= 8)
+        #debug print(val, val.to_bytes(blen, "big"))
+        content.write(val.to_bytes(blen, "big"))
+    elif type(val) == float:
+        # Why the fuck do integers have to_bytes, but floats don't?
+        import struct
+        content.write(struct.pack(">f", val))
+    elif type(val) == str:
+        content.write(val.encode("utf-8"))
+    elif type(val) == bytes:
+        content.write(val)
+    elif type(val) == list:
+        for sub in val:
+            write_elem(content, sub)
+    else:
+        assert False, "Unknown type %s" % type(val)
+
+    buf = content.getvalue()
+
+    # element length
+    e_len = len(buf)
+    num_bytes = -1
+    for i in range(1, 9):
+        if e_len < 2 ** (i * 8 - i) - 1:
+            num_bytes = i
+            break
+    assert(num_bytes >= 0 and num_bytes <= 8)
+    mask = 2 ** (8 - num_bytes)
+    len_bytes = e_len.to_bytes(num_bytes, "big")
+    dst.write((len_bytes[0] | mask).to_bytes(1, "big"))
+    dst.write(len_bytes[1:])
+
+    dst.write(buf)
+
+header = Elem('1a45dfa3', [ # EMBL
+    Elem('4286', 1), # EBMLVersion
+    Elem('42f7', 1), # EBMLReadVersion
+    Elem('42f2', 4), # EBMLMaxIDLength
+    Elem('42f3', 8), # EBMLMaxSizeLength
+    Elem('4282', "matroska"), # DocType
+    Elem('4287', 2), # DocTypeVersion
+    Elem('4285', 2), # DocTypeReadVersion
+])
+
+segment = Elem('18538067', [ # Segment
+    Elem('1549a966', [ # Info
+        Elem('4d80', "shitty python script"), # MuxingApp
+        Elem('5741', "sorry I don't know"), # WritingApp
+        Elem('2ad7b1', 1000000), # TimestampScale
+        # The stuff the Matroska devs smoked, was it xiph-laced with lead?
+        Elem('4489', duration * 1e9 / 1000000), # Duration
+    ]),
+    Elem('1654ae6b', [ # Tracks
+        Elem('ae', [ # TrackEntry
+            Elem('d7', 1), # TrackNumber
+            Elem('73c5', 1), # TrackUID
+            Elem('83', 1), # TrackType
+            Elem('23e383', int((1.0 / fps) * 1e9)), # DefaultDuration
+            Elem('86', "V_PNG"), # CodecID
+            # Note: elements like video dimension are probably required normally,
+            # but I can't be arsed.
+            Elem('e0', []), # Video
+        ]),
+    ]),
+    # Note: you could avoid buffering all clusters in memory by picking smaller
+    # clusters, and using "unknown size" for the segment EBML size, or reserve
+    # the maximum for it.
+] + clusters)
+
+write_elem(outfile, header)
+write_elem(outfile, segment)
+outfile.close()
+
+exit(1)