From ea673addf53df27796d63a6539fb89b4ee9b5ef2 Mon Sep 17 00:00:00 2001
From: Pedro Paulo <pedro.pereira@g.globo>
Date: Mon, 13 Nov 2023 14:10:54 -0300
Subject: [PATCH] add clear image metadata option

---
 README.md                         |   2 ++
 remotecv/image.py                 |  20 +++++++++++++++++++-
 remotecv/worker.py                |   9 +++++++++
 tests/fixtures/with_metadata.tiff | Bin 0 -> 6062 bytes
 tests/test_images.py              |  17 +++++++++++++++++
 5 files changed, 47 insertions(+), 1 deletion(-)
 create mode 100644 tests/fixtures/with_metadata.tiff

diff --git a/README.md b/README.md
index 0145a42..ac39aeb 100644
--- a/README.md
+++ b/README.md
@@ -118,6 +118,8 @@ Options:
     --sentry-url TEXT             Sentry URL  [env var: SENTRY_URL]
     --metrics TEXT                Metrics client, should be the full name of a
                                   python module  [env var: METRICS_CLIENT]
+    --clear-image-metadata        Clears metadata when loading image
+                                  [env var: CLEAR_IMAGE_METADATA]
   Memcached store arguments:
     --memcached-hosts TEXT        Comma separated list of memcached hosts
                                   [env var: MEMCACHED_HOSTS]
diff --git a/remotecv/image.py b/remotecv/image.py
index 1677f78..5eae807 100644
--- a/remotecv/image.py
+++ b/remotecv/image.py
@@ -2,6 +2,8 @@
 
 from PIL import Image as PilImage
 
+from remotecv.utils import config
+
 PilImage.IGNORE_DECODING_ERRORS = True
 PilImage.MAXBLOCK = 2**25
 
@@ -23,4 +25,20 @@ def parse_image(self, image_buffer):
             img.load()
         except IOError:
             pass
-        return img
+
+        return self.clear_metadata(img)
+
+    def clear_metadata(self, image):
+        if (
+            not hasattr(config, "clear_image_metadata")
+            or not config.clear_image_metadata
+        ):
+            return image
+
+        if hasattr(image, "tag"):
+            image.tag.clear()
+
+        if hasattr(image, "tag_v2"):
+            image.tag_v2.clear()
+
+        return image
diff --git a/remotecv/worker.py b/remotecv/worker.py
index a1887d0..36f0dde 100644
--- a/remotecv/worker.py
+++ b/remotecv/worker.py
@@ -297,6 +297,14 @@ def import_modules():
     default="remotecv.metrics.logger_metrics",
     help="Metrics client, should be the full name of a python module",
 )
+@optgroup.option(
+    "--clear-image-metadata",
+    envvar="CLEAR_IMAGE_METADATA",
+    show_envvar=True,
+    is_flag=True,
+    default=False,
+    help="Clears metadata when loading image",
+)
 @optgroup.group("Memcached store arguments")
 @optgroup.option(
     "--memcached-hosts",
@@ -335,6 +343,7 @@ def main(**params):
     config.timeout = params["timeout"]
     config.worker_ttl = params["worker_ttl"]
     config.prune_dead_members = params["prune_dead_members"]
+    config.clear_image_metadata = params["clear_image_metadata"]
     config.server_port = params["server_port"]
     config.log_level = params["level"].upper()
     config.loader = import_module(params["loader"])
diff --git a/tests/fixtures/with_metadata.tiff b/tests/fixtures/with_metadata.tiff
new file mode 100644
index 0000000000000000000000000000000000000000..b85ccab889905b647db0df2e22d77ed2e62f77e7
GIT binary patch
literal 6062
zcmbVQ2S60p)*fttNUzcbF(_aI3-%bhNz|Z*7^B9nu|#9k7h{QF4<?EzQbh!@D|Uof
z5RJVCB{qa*S(fdyJALN7GrK;2lRwE{#<SPE%gp`mY2P_>hY#;SI1mKkL=bk(2s@m$
z$B*5cd1stAv1>{+#s9bCrzy_md*n}xH}fVqcYkw#4$i&b{4PJ^^UgZHZ{}{k1QGbo
zIv0}ci57ToI)3EK!TK+@YnRwAT}25{B&*0XSp<~V&z=bsrLC{G==BoAv82VI(Sjh-
z7RqEWQf8L12(*z`yi_wLi7^Sl0SEvResE@%z#xJ~0KN4q{N?p`(1@<6eg&dPnMi?R
zbm~TsK&hxR8gu|6&vN*+L7@f_NS(!~A%TAvnv%dIf=&P<$bb0j%ck!iwU%`!jV2VM
zR%g_kfJ37|?Q4a=@jPcy)V-o;Bg<LT4fRhdAM4N>_;*2+Reqy2BY{N%QUvq=6<U%c
zM8y42W*SEeNpYNzla<@Mckh5euXd3^;Sv5rhxg0KOa)+`R6b}>DI}2v2~C04Mm2G-
zEQ%<xXrjN?8Gn7_Us=<)7Qo+7k$RN|M~&8KCYi|gF-|RAF)Ewc6TSh?POf&IzK(8Q
zEpVimM%*1LC;^RdR%ojQa6vT1KU{x=e)~<;pvA3oJQz)sy+f;}%^ZAu1H65G{QSKG
zg1y~59b7%^z5H5w_&BtQ@ZrHAfXQlP`VBN<TJiq@*%r?>-BD{BG%v}~YT+cBgUM58
z1%`&(JGnT!d$e+P4GwAJ<>lq$=k4T-=uRPFzFz*Vk~Xd2<hVlrTA|Ts2#ucq+Yqzv
z$eK!4YfF+SLnAH<Bg#HQK}K+B<=n#F5%WDLI5Z?M0wpN0?WT<>a-!N1!EOA4!hJa~
z^FWI>gTIBA)9vr|7UV4c6SSO}tcYXb!6SrS6IXXPl}h=#=7p!bN5aSno!Z6o7`eEt
z2uMo4t$jSb-EQ2yj=4=Sq`Y5A;273oFpH?e{|XvVYgtotF+fdyqgrR&_+v^d7pL7B
zJ1~-YMz2xS#74wzSe^nD<nAayGTF8@4bi=Py!Y(QH6jIAObkbJEG>z=$TB$oKcG=3
zqlMbFE3HLy;#zr0e)cvWR{~h0PZOsf=6;vGWw!%iziP?qthDTw4$duFI3%yzV4*2N
z5;=iIXgSLSmZ8xIe;?$s!R4ruGwZFd0UkfCqFHKhUY4s9u{n9^{OJiE%?Z_$>z)KL
zFtUBk?Ypi7!E5w>t{$!}U0j{rR;*YllZw^1g20Ia10wT3L0iN6U1)1`;UZd{3V=fL
zcWy%L+4lXn(?%ZN_5GE?txqoGr!1Mm)IL04l;dG<7wYGO#bn{aMJyxIjKB-{C{9Q~
zK9L#bAGEe*)H_<^-AKVE1IL*_GP&6k?fsh{%2;!+B>V98g?Eo7-9EDVYSEgrdsdy;
zv%cg=mOa)pH|H7CW}?wpPQqvLGMi1vZ!*RG12o<$%O9a}yF5jk8v*o-rjMDC(EFDI
zKjy5PdFyC$N#^Ijom%+Yx$n>A&HrLV*u9GdSe3n89fuAcg3z+UR>zT~1>>Bups)Y_
zvO?P^^vy1;8xaLs;!Uhkar5G#P9ZIRnE%OF<9j53HLzgA$bU^}y<m**ym8@~o2E@4
z-#fs=&aG9`0DnKN&V(yiD2}1A-cq;=iy7@^h5oBiXEL&)>>dnGmJ{BPt|S-@8cZk<
z4F6g@wY57jeZu>@w=Vu>+Q7^oChlDR$+?`B8y1gSI&;+Y#9@w2h_E(+e*V5)W4kdd
zrlx$ENj+(36j{n<QX2{UX=DMZL&E4VGqf4ggJzKpc#Z)cLzJp~c-;lFh#;m<8n$rO
z=*2TX{MWRRDc?+-GNN1N#$}B9#o40;E>6unJYD<)16sN`Pnr6q2tsX5wIm`BkQAwV
z18t>r+3oKLrk{nU5trgvju%m^<p$lxbF?X8)R2e}PlOgMwE&tYx6ho(`?27s?*w%v
znCn0^0%K_D*v!?{xn1OYet|)DO`B9)zbf!7Y0_fpRM*!a^xJ~?r-_-POpF<EZFGyN
z!SgZ&NgVs^$>V0&*Vz+;`^SnFm0tB!vM6ZfW8l?btkG0I)+(RVW^GdPYI{eAR&MUW
zVc}R+9i1GkUe?nl>}NOx{;v?SMf870TEL)SFeeZiaV-=>Q4Gg%xZAEBV=!x4Hn&5E
zfJ8OaJOx%~X?VtKtDoE{2i`~-8_}oeT;%zHkWi!+gm(9I3k?a9TQX!70Sk&;zW!1e
zk>x=^Xqrch3%EnuRa5=izIhAeiay=DODsAKdo#|0BqN$RBaZa*s>eK*3x-i?wCBq%
z5lveJhqsH4>4Y}-aCc4oG(qMQiAAju7nl9-BgNAQjp0qR433D59%*dU+BbLT7~Ki`
z;s<x`;*+d}!p@MveuXqBs$Z~-nPIT^0~5xjwdnNdGk|Dj@9pCk-7zL4I1nE^erP{x
zY%@5niA(=!L})9C1BHgSLf6*Tm6rbE?&{HmXzJ3+RZdI+XcJlwSej!fYzC0)%tn)$
z<|U(r!6ppp5|u*5Jv_aKjTnvw@6tIMEhZZqA>WDWcjvgU&Y(w+n9L;3QFqKRXIB?j
zS9h$2vuDlYSgH24lA|Syi4`yv{}Uv#`>dddi=ro~Wh+)WySVj#f55urwJy$%Apw4i
z=Fh{Oi;TdU840Z;0%u|6o)I5Lg{`1C8caqML{DD74*lim=;-9+{NTYOWt|!y$9ibc
zQt~?1=MdB;vc9%Cz>C+4<}DoBwTs-aZhhPEHePP7tsLx0y+LLY8yvcZ;&3jYlQ~Jw
zCIOuw>#H+pacys3FKiE52l!z(q;Aw=rW=fw2DJ`xmGzppTH7YJ;K6yFQjLpvd;5lm
zN6el%YxKvXqB}&%p>ppbJ}BGyoe_(w>_ki~TaiT~<(Y?%A7Hx`9uYeFlTp3;#N(d~
zMiK=VM~=nr)P`uQ?C*{iGpo}Xx_9r{sZ*D@uCa^1UKA4*jdsi1yU(WDTODytnYgeX
zW2^v>W=v*`7jf~DMa^2+z1JbUd(YUcoZSsNm6^35$)k4ljT*cY4J6wLJtW8LJEH_O
z)wNOWquYdpb?6YiWlKs}SledJ?U$}zWfFK&6bu}T!pQGqq7mO=ura}<uouE`m6MAP
zYa8P0?->y6i{r*E$s8~y+KA|Cofbtlk&J6~`5#9`O|8fap6*^AZk_`N4qCZ#b=&Yr
zY@w$toM*yF1`*#CHNfLsEwVT<V4k2QdDI=_Pr{DK#m&jx%h}t{wP(LLS8pc{sK-?g
zvHReFCs|BG8!kqtRczg?byO?sa48=z-$4JM0sRM|wDu2a?(F)}l!?fvwF0FC76pl^
znkDOjYe2FSc!QZlXgOPr+K@o+W=)8I)-5B#z2f6zLPPv9gr7WrN(qbwH)wskkvkt*
zWm`Xtd!Ruifzj4f-FI%@q-}6$$EdEsZ6f{Kyyp-c5HW7>P0*KcHJ3nj1+>?IED<#o
zz+M4P!>SmNmC~pi1+%eNOj|cQqHm06e0z_+ox*xXN4E=Uhc>2wBSEibURgMWAQ-R_
z(isGj6lt2~EH%Kt1pfEpymh1ccInxnLugRDh^QX^F|kDZAYxXZZ>h3HmeQqS#RBT;
zOjE@Yc(GWhUC1_=LB{~8Y^Vl-M2Bqp?u)Z~mp!?>{qEV6qy=LJbZk30A)%#%W69+b
zEvHn`zYCz!8LM$Kwe=)+7+B_L@^_#g988~EbvbkX*a4x=j+>J<M?`mb3T;CKJNCY~
zu5aD05tjXf%?Cb`&W(oaBc!`=s`9~Xwc1k01IhAA0EP0|(d?}Wmkv)W-Jf{t^t^|q
zYd;?vhD~qBb^#t9EuX!7#Dj?wl)Rv!ES%mb*462iby&xhhq9NJ?ECiiiS^<3#0N2<
z1G>jKySw2*#)n%LjH}LexiohezrU|}Z@jsnpKzuxcc~j$(VKn3@c5dh7x))9uIxxp
z8F@H!WZ|~B;*4J9MWgN(&u+M%KCG*gS2H3a0Lh$r_Vk|Fi1slUbVdv+Wt|pV5#VdK
zCr!9^Y~_RE<XOYt3v5a_6GT{;Pp@G;a$lV3cW*6mVPbzFryuO=NM^=z2jciM1L1x=
zJVJ)$nfhO^<mBx5EMs$@+h-@7&+b)udhD&DkI$tID&IS~?t0p9WtpI<0^Uri>VVTR
zCZ)izq>++Cp10IgmS%6Ba=GA}gPX=HF7H}DXK3QksQI%#Dy=A5R=l>?y%m1Xmj>3Z
zilI}Zx%3`zq&;^aN;uaQZgM6AYpU72YJ9=23AfJ7xOr@BNqYY)86T9V5C8r6%)5nO
zmh7BRmbLIg&KgtYP0%QTY5-b;6`m%|CXG)0`YA}Y$qNRbOr3IL-@@|jc`wU;zH|2H
zf~;=~_ODrcA-UU~rJdPrPL=aptJZ{De(nr~ZN<Vi+~H^_NA3o*DtXE1ecL|0et5>^
z+;MmIO?|X?)`Of`#jA%r%AavLZG1(}j4RoTj&A-M^sg)mY><pVBOCFwUdtQn8P&b*
z-;FKHnuE9fvTM?{{P`D)mLADnaQbA@cgL5+-CPt!rTEo-@2y-FY}(RR$c%!+k#Mp-
zTm%yfq~>%%(t#ac9NPL}Sw`aZ-81fH&bz&H?&I9~6{%yZ4o)xK@o`1&muI)nEZi^`
z$X7t?fFX?y$Pgs@-Gv}>H%vR7It`)kADnk#_rz0~Q?H#}vom$lrVFbES0x42t&KEo
z^;acDa%quLP8T>4#1yrMGVKdB==G(CQVO?E-oNF;b7|u)ZJ%^w=d63X=ibYl`dh*0
z_x6sul>WiPBa`nQ`sz~NHwV&I-Y7lE=#{UY-p|k7*|kFu@<3_9`qMk+oJ*Zlo;~^6
zzUfyE&O1>sCogBtma?@`C0_-;SliK<8Yg7MLS8Hsb%Z1B#ADI$3myX~pFYh_S(5kD
z<kEf9&+Hm|a?3{-Q^)_Bo_H&3?8WVUFK6_vI+J*QSHEk8pI_Pk_3_+weLA<rcR;sR
zM2E;gPiG>?jY#`pPT|Js75U#*<j*{xHKAz7xZ<Mur;67mZX7#Nl{-)<8VE(bg#2zo
zUL5T23P(G^$xcw(@KWDcttdN@xogwJBN>V3^2U_!PbkYCawawYTK?ejyuRnsV~=j{
z^7#Cuqp5?gpIA1avrpp31Hbut^6WXEefrtRSu+x6PD=Q&chvdpb;p06b8Y{kibD%d
z?VI!Cx-lC!efbb77qe%_!0Exl!FW2a8@;E8Sl9!O#lfjqxL|BV1_pKQgOubil9u&O
z-`M+1-ms#dy52hRNzvAr;`I2^{Gq=Tjy}9=NM-r5#b34u>N!Y8fv@MK1`3M+9?D6~
zolEC_T0Z+^`ij#TUmZ!GoV0vkdfIG*<;i_`wZL>{1b6IxbAC@#4x;xI3%ZL3y9h_(
z;C%J#I<*?#qfI8=1GDzx-K*<Q<|k%u>XEuC>R86e6B%RA?wfizV`6^lrz_@nG%1T|
z?PGybX>?EVM2!N?gQ~k~wHk}Sk*u^OlSj-<h+4mV=&NUk0bbtw{l-HuEafhbhT<V~
zVShTWFSjS2-y6r}cVr5?!zt|2YSlGHb0aTcaZ>^Fh}V|^e?|Me@OaLGwF`#hE8dz#
zqt`6?SgS0?!v?di5_>?E=8@9y8wGfzXvS7Uz}HHK2Xnosy8}?E(_gQAbp>hes;F5m
zUQ2)zgQ&fIE!ln8?06xkhm_wPin_ru1E{rJ1HLL?qf%d^6iFkme+`TZO9~?3yC&Aj
z`l>2?f6{BOYa6dhY#l?_HRvB{xtI8?!iZ;k6p5#>cqir>^IWaCWH3KMgFIEMe+6!V
zrEEA`)St}iV@dDL?(V_w?g82HaG)z3YLsdkm^%5OO$3#K#DF&{8d<eg)R^%^nW_a6
z>jRB>tmnDsf>b3k3Yt=Ipd_JIK~-x-5>F@fI_n7~p5Cz40MBXRrG{6jMW_I79(iFT
YUo@EAGf2$sCuMeH(>kFD*<GOU-<z~V0{{R3

literal 0
HcmV?d00001

diff --git a/tests/test_images.py b/tests/test_images.py
index 25d62d9..1f3fbc4 100644
--- a/tests/test_images.py
+++ b/tests/test_images.py
@@ -2,6 +2,7 @@
 
 from preggy import expect
 
+from remotecv.utils import config
 from tests import create_image
 
 
@@ -20,3 +21,19 @@ def test_when_image_is_pallete(self):
         image = create_image("pallete.png")
         expect(image).Not.to_be_null()
         expect(image.size).to_equal((3317, 2083))
+
+    def test_should_clear_image_metadata(self):
+        config.clear_image_metadata = True
+        image = create_image("with_metadata.tiff")
+        expect(image).Not.to_be_null()
+        expect(image.tag).to_equal({})
+        expect(image.tag_v2).to_equal({})
+        expect(image.size).to_equal((41, 48))
+
+    def test_should_not_clear_image_metadata(self):
+        config.clear_image_metadata = False
+        image = create_image("with_metadata.tiff")
+        expect(image).Not.to_be_null()
+        expect(image.tag).Not.to_equal({})
+        expect(image.tag_v2).Not.to_equal({})
+        expect(image.size).to_equal((41, 48))