From 9c968456c1bc1e855d17d204be256c0df3a67afe Mon Sep 17 00:00:00 2001
From: Adrian Sampson <adrian@radbox.org>
Date: Fri, 13 Nov 2015 12:21:36 -0800
Subject: [PATCH] Fix #1666: malformed binary data in SoundCheck

---
 beets/mediafile.py                |   4 ++--
 docs/changelog.rst                |   2 ++
 test/rsrc/soundcheck-nonascii.m4a | Bin 0 -> 5862 bytes
 test/test_mediafile_edge.py       |  15 +++++++++++----
 4 files changed, 15 insertions(+), 6 deletions(-)
 create mode 100644 test/rsrc/soundcheck-nonascii.m4a

diff --git a/beets/mediafile.py b/beets/mediafile.py
index 64ab49ac2..5fe1fa308 100644
--- a/beets/mediafile.py
+++ b/beets/mediafile.py
@@ -215,9 +215,9 @@ def _sc_decode(soundcheck):
     # SoundCheck tags consist of 10 numbers, each represented by 8
     # characters of ASCII hex preceded by a space.
     try:
-        soundcheck = soundcheck.replace(' ', '').decode('hex')
+        soundcheck = soundcheck.replace(b' ', b'').decode('hex')
         soundcheck = struct.unpack(b'!iiiiiiiiii', soundcheck)
-    except (struct.error, TypeError, UnicodeEncodeError):
+    except (struct.error, TypeError):
         # SoundCheck isn't in the format we expect, so return default
         # values.
         return 0.0, 0.0
diff --git a/docs/changelog.rst b/docs/changelog.rst
index 27567f283..700a16d6d 100644
--- a/docs/changelog.rst
+++ b/docs/changelog.rst
@@ -56,6 +56,8 @@ Fixes:
 * :doc:`/plugins/duplicates`: Fix a crash when merging items. :bug:`1699`
 * :doc:`/plugins/smartplaylist`: More gracefully handle malformed queries and
   missing configuration.
+* Fix a crash with some files with unreadable iTunes SoundCheck metadata.
+  :bug:`1666`
 
 .. _Emby Server: http://emby.media
 
diff --git a/test/rsrc/soundcheck-nonascii.m4a b/test/rsrc/soundcheck-nonascii.m4a
new file mode 100644
index 0000000000000000000000000000000000000000..29f5de5310f799d5f17514d280f32d4cab82f9cd
GIT binary patch
literal 5862
zcmeGdX;hQP_5;GAvO_RHq+tgk7_yKAk$q7M$R;8x=F0}D34|ob(kfvQsZ>yl6}2KF
zV53yAEr=EcErQmfzTy)ROYsR65i6}r)mqz`FXUCxp5yCzzk0kUH}}q6=590dAqYYe
zDsrV!^dJ%f08)vRP8G>BByf!&h8rXq8LI&!maI<WLrG`kt(BdTcLCQJiXg2BqJ{mS
zN)2HB0!&#4zMy<*p^YHg1qvCu3TT!qR$(#q|BMNvK>%@|157w2j>b}B7N}t`sy$1>
z7omV9N%*g>YjG!L=#a)2%QUX?jI4AGuc2DK5>dJktW4z+^-hx-C%#%v<jVw54*}+A
zm?aaF)V$U<xq>SOe5za_$GWSQBTK*(V0IAxtV<g)g#y@j05pM|FNchdQmJ$R6bGnf
zARZwA-Yy40o*=qFqHS)V2Z$DEEJBbg6qpO(;kI7D2@u2$gy%s90JMWT{3O)@6dFO<
zH^43rPz?ZoOlJ%zbrRnK6lMd_6(AX)2S9Itb0&EgKq&yB9wJ;j19S)I1`zsq0-OyH
z%3vIg4$3sP(2foe`uhNcc2Ey}ohB&_P#9Me#}M!^U!q*W%K#JtsL`YgLCkL<h*ds<
zOxv2pSD=7yln5Zj;Dv(#4v!3#O5y0#(1akm>qTO@0xXHDbW{SUB_GASVdoHGmKa()
zsDk1W0JV8ur$`f!P?<u6c{{1HW!NrLEqECcX@*<?52L*b6>|Z#o@xWcSunJ{DpipF
zDgz*v34ns6kgb9SgvlKMi-9)m1pqs~NDi-{)f7Dr(FIbXijU?3`a*?N3I)#CnkN;(
zO3fxc;W@$%kF_)vl}=SarHuj<2B`V-9*7>0_^1>aP~$bJ!8$2Q$VwN;NvoM&K3-H$
zQf!upw<<v-5s;W(bT0<5jwS;DlqeCPuto%Q&I?XU;HCBjA~P@|*asvM<qBZBFSSxU
zsfc@x-SC>-q*+|CNS=lb=`VN8eT^OMH9JOS$wj<i87fN8BL!h+gtUMU@})oD5++Yb
z5uj{7ndQS}lYxp)=Avu?nMz|YP#-#r%@)98obwuiNv{f=AVYbpUJ;kYWwEFXCWB02
zp$sye&Y+S}Cdwr9xLg+M&7gS;*|5ofJ8lp>&aa5e7O<E!7K2S@p(ve9XR;||E}O+5
zGf*~zCSVDKK2#Qn`^jt6YpJOh9*yWlaZ$ie`h0Im6bS6lP&E@2qGsqc8WWSyeJCN2
zp|Yt|I>t~bAv8?I3=a=~f$<IxVFp8qcNilK9@V?;$H9Nw|C(ofR7^1ND}Fgous0Tw
z!lF+yG$xiFP^cj&gpQ?&t<<)F{d)|^!9gX?g-_B;4;m1Qc<^K6q~b~b>OqqT(y<Sc
zlPXmP-v;9I6~SjQRhE$@6{V*Fog1otI!RM)ut)E|52(%+p-g~XZQ%0;-gMwi2mTQq
z(6M}Tum2H)d{e~#tpoaQ67VrZE+YC`cKD+4<p;dI=XUz(b|8=NMRjySu}KYQ!6xC5
zU|Y(}p}Te!F*u6jwIe%so#a+^ercJwNc&2EzT4~*3zeBH<Y#<Qai#%LPY+R%1N~3e
zjV>e8p4pel6sc`~+-k99OWDXJ;^zA+bS9!BE6$QWRKggiVnUS(rl%#W_hMS+KlN{O
zJ8yloC36KgzrGcDf-mZf3F}uX4N3c(LT~hM_FuSIOv9ZEa=Pp;axa?Mfms9@I2I@k
zbUl1&cR5L@LCf~4vyWrv9_Tn+RXRF^Jj55(I*^Ku&+APq??a~F6KcJiIn5wD`NPf{
zTAo(B(9u1*KhqUv0?kZTos(G|@4vj)DAlZD-*2v!V}^M5+}RCPS08@q*>t`+M{kdn
zF^q~C%Zy5bO_LJC&Bw=?>1h`lh|$&W<vHIM;3>#r!NCI`%=j3FFAN4<#I595vqv&&
zCl-DhnnyY0{_~^jHUr+_`K--Pj*Oh0<<C?caV%SGIyBjK#}ASW1|AG<KXHA?)cRXa
zYkA>dUygo42}+14T`L@(Z?W3X?zfhc>IQHE5JQ*JJ|Ege(@OCy-(iD@q1#)gn~jz?
z?<nr>y**+PZHq4)!m^Gta128FRO}0WpI;JnJ70b@*09akve{}w&-t-~xx(u6$bFD-
zf@!hw435D;`+@~)veMEModU}2JcjqUB_Ak0S+M7t0<BLJ9ridZCce+YCO|WFjKOV0
zC~vjd{<JlLvxRolcC@<JF<?#ozaL!va{N-65}MV;8aU1ubDB#}9<MhioO^qpd#rm{
z;ZG0z^%i}0>i(hlOAY$atjmGm7&C2vFFj|oyM2Iq>(jOJfit~#H@e3BI6kk(<*7Xp
ziy0Gv_?1(lza#b-7lb|9Zfm`==z*)Q_d4GPW>sY)cFVDenv+URJjFALGTSa4^DAid
zAD!rNx%*Xh!$R7ctt-s=u}VuNRzWQ%yrzb=z~iG9U-bG%b1C)R{t}}9-0!1K+!?6Y
z-Dgxk$2=9AjMQI=407XnBI})%b&*DSzH8@iWE4fwR0P+iAdgX(Vj*e$Wj92M&E1@0
z&+pKw-Ezp+#JM)4iH#5b{6}Mt70+nNCALRLc?)GFr*JJ=70|Yh6Q(rSgnF)Mc3&PW
z%Ess0o$<^i-VFTeSZM3xT~A#*_u=Y4xu&Vf_+p1rkCJ(B$YdV3nji^2Z5)*Scz#o#
zz2T0~`2Cw^u(0q9l064^ie9pBB;}_&dizs+o!HG6)}_kLI8n~)zD`b7>`z&|cd1(j
zHk?`qLa9j|h5hjs$LN0@ar~w+&+h7yopTjNL1b@!?)R0nx|ACA5pp0oj3eRy$sTWe
zQta8R(<<BWtnnSWbLH-D@5uXw)dK0fHmp^g2t%cIV8@Mb8399W1kTng<^^XhdZIo%
z^RVl;rNJpUeBay<>;<Uh1fS6&b=?rfJ=(TjbgSH_lw!nwdcrSnxP0ZzvG$P;O@o}H
zaNi7sCwCLpmHlFtLv(S!6Hj#hsEeU58#}&7+@0yC?vEQG%rih~;owK9A6mO)P&udU
zLi4j9M}Al4QWbWqPp2$$w<qB`Sn7EaiuI*$k#J2zX1C5+D?)_I{eI=wmZtZ_pikrO
z;N{IW#(P^zyk~w~E$Z@|)`|t-L=bdzkGALK`H%FjzP--v+V8(dEM54;x_MQ$wHY_X
z-Ak~YX(ZEN?TfT)xY#s8<mE1>Ib-X&e!r~!yqR_T*Majp;x~SpKKq>qNrKaT%q@<l
zIIAJ-<~g_(8wXy|w=fFge_}fO6z85<e!uJW(#Deqc^OCi6Mr?oi5j{zJJ%4g7@Y7z
zquRs6-{;&|gKPD>t^M^=VT{=_>Q$XglO&sAulx<J3!0_ywqUPS9btCwl^vH|$2N4I
zjoY55wJhVi-o`s<VeyH|r0#esUh0ZH`dw60`x;B!t>lTpoX?t1JRX1Iq-sx0|HAnl
zw|j{;9W&L_2rn#9J}loAUJ|&Iw`R>m$SDh|zx9A$*Ud$Yz4k|A2j`^9LMtY)YmXX$
zP7K-JC#<)m>b(C&yjzOx!DQVEe*VO}4KW3EyLH#T-#<{*u-Rui5`PSPF>5VJ#b%xy
zd>h$2J9-ca>agyqOp19sZBO>w1G<hD7CIDST;tNiNxR5c#<iTVP17AtFOX!``DfRj
z@s=~*wLh73Y}uh%bAA{I=#8kT&%p*qB{(=MlZI#1pqHu-mj7_!S5?=Ao9(|Gvnr6~
fAF7H(JG6-K3%EZ{<PacLe+?U_72glk=~(^;R{g~!

literal 0
HcmV?d00001

diff --git a/test/test_mediafile_edge.py b/test/test_mediafile_edge.py
index 7a17fe86a..7886abdd2 100644
--- a/test/test_mediafile_edge.py
+++ b/test/test_mediafile_edge.py
@@ -89,6 +89,13 @@ class EdgeTest(unittest.TestCase):
             beets.mediafile._image_mime_type(jpg_data),
             'image/jpeg')
 
+    def test_soundcheck_non_ascii(self):
+        # Make sure we don't crash when the iTunes SoundCheck field contains
+        # non-ASCII binary data.
+        f = beets.mediafile.MediaFile(os.path.join(_common.RSRC,
+                                                   'soundcheck-nonascii.m4a'))
+        self.assertEqual(f.rg_track_gain, 0.0)
+
 
 class InvalidValueToleranceTest(unittest.TestCase):
 
@@ -269,19 +276,19 @@ class SoundCheckTest(unittest.TestCase):
         self.assertEqual(peak, 1.0)
 
     def test_decode_zero(self):
-        data = u' 80000000 80000000 00000000 00000000 00000000 00000000 ' \
-               u'00000000 00000000 00000000 00000000'
+        data = b' 80000000 80000000 00000000 00000000 00000000 00000000 ' \
+               b'00000000 00000000 00000000 00000000'
         gain, peak = beets.mediafile._sc_decode(data)
         self.assertEqual(gain, 0.0)
         self.assertEqual(peak, 0.0)
 
     def test_malformatted(self):
-        gain, peak = beets.mediafile._sc_decode(u'foo')
+        gain, peak = beets.mediafile._sc_decode(b'foo')
         self.assertEqual(gain, 0.0)
         self.assertEqual(peak, 0.0)
 
     def test_special_characters(self):
-        gain, peak = beets.mediafile._sc_decode(u'caf\xe9')
+        gain, peak = beets.mediafile._sc_decode(u'caf\xe9'.encode('utf8'))
         self.assertEqual(gain, 0.0)
         self.assertEqual(peak, 0.0)