From 142fc90cc56207fa0cff4fe3d83d1d925a92b3da Mon Sep 17 00:00:00 2001
From: David Huggins-Daines <dhd@ecolingui.ca>
Date: Sat, 19 Aug 2023 09:29:21 -0600
Subject: [PATCH] Support for marked content section IDs (#961)

---
 README.md                   |   9 ++++++
 pdfplumber/page.py          |  56 ++++++++++++++++++++++++++++++++++--
 tests/pdfs/mcid_example.pdf | Bin 0 -> 24694 bytes
 tests/test_convert.py       |   7 +++--
 tests/test_mcids.py         |  55 +++++++++++++++++++++++++++++++++++
 5 files changed, 122 insertions(+), 5 deletions(-)
 create mode 100644 tests/pdfs/mcid_example.pdf
 create mode 100644 tests/test_mcids.py

diff --git a/README.md b/README.md
index 19134d6..9bd2a0d 100644
--- a/README.md
+++ b/README.md
@@ -158,6 +158,8 @@ Each object is represented as a simple Python `dict`, with the following propert
 |`bottom`| Distance of bottom of the character from top of page.|
 |`doctop`| Distance of top of character from top of document.|
 |`matrix`| The "current transformation matrix" for this character. (See below for details.)|
+|`mcid`| The marked content section ID for this character if any (otherwise None)|
+|`tag`| The marked content section tag for this character if any (otherwise None)|
 |`ncs`|TKTK|
 |`stroking_pattern`|TKTK|
 |`non_stroking_pattern`|TKTK|
@@ -191,6 +193,8 @@ my_char_rotation = my_char_ctm.skew_x
 |`linewidth`| Thickness of line.|
 |`stroking_color`|The color of the line. See [docs/colors.md](docs/colors.md) for details.|
 |`non_stroking_color`|The non-stroking color specified for the line’s path. See [docs/colors.md](docs/colors.md) for details.|
+|`mcid`| The marked content section ID for this line if any (otherwise None)|
+|`tag`| The marked content section tag for this line if any (otherwise None)|
 |`object_type`| "line"|
 
 #### `rect` properties
@@ -210,6 +214,8 @@ my_char_rotation = my_char_ctm.skew_x
 |`linewidth`| Thickness of line.|
 |`stroking_color`|The color of the rectangle's outline. See [docs/colors.md](docs/colors.md) for details.|
 |`non_stroking_color`|The rectangle’s fill color. See [docs/colors.md](docs/colors.md) for details.|
+|`mcid`| The marked content section ID for this rect if any (otherwise None)|
+|`tag`| The marked content section tag for this rect if any (otherwise None)|
 |`object_type`| "rect"|
 
 #### `curve` properties
@@ -231,6 +237,8 @@ my_char_rotation = my_char_ctm.skew_x
 |`fill`| Whether the shape defined by the curve's path is filled.|
 |`stroking_color`|The color of the curve's outline. See [docs/colors.md](docs/colors.md) for details.|
 |`non_stroking_color`|The curve’s fill color. See [docs/colors.md](docs/colors.md) for details.|
+|`mcid`| The marked content section ID for this curve if any (otherwise None)|
+|`tag`| The marked content section tag for this curve if any (otherwise None)|
 |`object_type`| "curve"|
 
 #### Derived properties
@@ -531,6 +539,7 @@ Many thanks to the following users who've contributed ideas, features, and fixes
 - [Shannon Shen](https://github.com/lolipopshock)
 - [Matsumoto Toshi](https://github.com/toshi1127)
 - [John West](https://github.com/jwestwsj)
+- [David Huggins-Daines](https://github.com/dhdaines)
 - [Jeremy B. Merrill](https://github.com/jeremybmerrill)
 
 ## Contributing
diff --git a/pdfplumber/page.py b/pdfplumber/page.py
index 521002f..c86a363 100644
--- a/pdfplumber/page.py
+++ b/pdfplumber/page.py
@@ -22,7 +22,7 @@ from pdfminer.layout import (
     LTPage,
     LTTextContainer,
 )
-from pdfminer.pdfinterp import PDFPageInterpreter
+from pdfminer.pdfinterp import PDFPageInterpreter, PDFStackT
 from pdfminer.pdfpage import PDFPage
 from pdfminer.psparser import PSLiteral
 
@@ -62,6 +62,8 @@ ALL_ATTRS = set(
         "stream",
         "stroke",
         "stroking_color",
+        "mcid",
+        "tag",
     ]
 )
 
@@ -115,6 +117,56 @@ def normalize_color(
     return separate_pattern(tuplefied)
 
 
+class PDFPageAggregatorWithMarkedContent(PDFPageAggregator):
+    """Extract layout from a specific page, adding marked-content IDs to
+    objects where found."""
+
+    cur_mcid: Optional[int] = None
+    cur_tag: Optional[str] = None
+
+    def begin_tag(self, tag: PSLiteral, props: Optional[PDFStackT] = None) -> None:
+        """Handle beginning of tag, setting current MCID if any."""
+        self.cur_tag = decode_text(tag.name)
+        if isinstance(props, dict) and "MCID" in props:
+            self.cur_mcid = props["MCID"]
+        else:
+            self.cur_mcid = None
+
+    def end_tag(self) -> None:
+        """Handle beginning of tag, clearing current MCID."""
+        self.cur_tag = None
+        self.cur_mcid = None
+
+    def tag_cur_item(self) -> None:
+        """Add current MCID to what we hope to be the most recent object created
+        by pdfminer.six."""
+        # This is somewhat hacky and would not be necessary if
+        # pdfminer.six supported MCIDs.  In reading the code it's
+        # clear that the `render_*` methods methods will only ever
+        # create one object, but that is far from being guaranteed.
+        # Even if pdfminer.six's API would just return the objects it
+        # creates, we wouldn't have to do this.
+        cur_obj = self.cur_item._objs[-1]
+        cur_obj.mcid = self.cur_mcid  # type: ignore
+        cur_obj.tag = self.cur_tag  # type: ignore
+
+    def render_char(self, *args, **kwargs) -> float:  # type: ignore
+        """Hook for rendering characters, adding the `mcid` attribute."""
+        adv = super().render_char(*args, **kwargs)
+        self.tag_cur_item()
+        return adv
+
+    def render_image(self, *args, **kwargs) -> None:  # type: ignore
+        """Hook for rendering images, adding the `mcid` attribute."""
+        super().render_image(*args, **kwargs)
+        self.tag_cur_item()
+
+    def paint_path(self, *args, **kwargs) -> None:  # type: ignore
+        """Hook for rendering lines and curves, adding the `mcid` attribute."""
+        super().paint_path(*args, **kwargs)
+        self.tag_cur_item()
+
+
 class Page(Container):
     cached_properties: List[str] = Container.cached_properties + ["_layout"]
     is_original: bool = True
@@ -174,7 +226,7 @@ class Page(Container):
     def layout(self) -> LTPage:
         if hasattr(self, "_layout"):
             return self._layout
-        device = PDFPageAggregator(
+        device = PDFPageAggregatorWithMarkedContent(
             self.pdf.rsrcmgr,
             pageno=self.page_number,
             laparams=self.pdf.laparams,
diff --git a/tests/pdfs/mcid_example.pdf b/tests/pdfs/mcid_example.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..2a90e125c8490a11d9df4655d968b4e3d56ce8b9
GIT binary patch
literal 24694
zcmeIa1z42Z7B_AYA}t^wF@S*d%rL~zozfxQLw7d{3eqVZ(p?H7EhQ!0(p}Oi`3-pP
zIqEt0-tV6ObD!t`e9s*nXTN*)T6^ua*1PuW`t3E8vO*#ZAVyXc%BGB_nx^+nDJWn7
z5MZTeio(Uk^aN&UWN!>$hWE%biI|w%!)%#E%ysQyLNI+R0~ikvik-bJOxFU%IkE8R
zxakBC)8pvrN+qT+;ba5;+sY_B6shW7zka`72wOj2Ow@&^=mR%OVb<yd%(qkXNpTd!
zp@yG=Y<G(}dB(V=r!~&qsq$sACrKNT(^2XB5*3WtL5x~#^;K3|Z3<2y%B|Wd#mPm*
ztRo&?XRG^71iDQn9FkMV6_p#x>(VBs{x|(0>t?p@4R?{=)_t2|<}n^;#&_;%w!Xt1
zdE4`GR$PNjQ-NPdQpu3<zW#YBr?yi}_{H4C;nvcKSU1Ht%5Oe_nQva<M61<2=Fhah
z6P5Nv>Y>gca5Jn!f`6$Y=17<Ga{H2Eb|LjPoi(-mZQKQ?*0?lj%40F?L6pZ&Uu7D4
zQnE-d2^Wakj(D^ry*97($vbxc=z>AbY<K70dlZ}@)%_PSTv0nQg`&-pB*XGC&^sS=
zil58920f}-cup?3gsIc%i6g9v<MlxN$jLgw@DQsu&ZQ&XAbGJjA@*L=c@yb|=&>)l
zzAu#@xMjLI)m@<4#oF``YUs!D)t3H<b#nd&=~~I@bKXJSky|!l-$)i0Wc?;Pdy<!6
zjBG^i#fau{!``ybZjff%*ZF<Jqo<dRBt@i>%cD`CQ1lj#EimJXcXFESszxsn+M&;w
zA><sz0Smm0^pW{upwU1Ju#~WxzZttRD-d%7S3gxsWE}f-AA65XmNk@AT!opKll9Be
zbiSC@SO#4(ACZ4V$!BMido*&YZFn2c4Rsgrg^P5=m5g6S%x#mugN2To7CcXH#-*j~
zVi6M0yw}v5VbV+(DJO#+lM|8Fi{&uH<+X%Gi_Qb3er<LpuX#^|c8Pw^?m^EUtHS-v
zmPG+7%V#K^RYc{Ce3TZcl%|1M9FeSiFOeoJUKZ(OT98o%n7u7`dE-)KgHDTJizTOE
z-VGjr@KG8Nkwi9A($BvvJ>Fx-5;P-PzE#1<((xe6=3|^g;-{B~1IJi=v8`R~p%~AO
z9FB(+?;i_MswqAbyhM%jK#MaEzsu<ETh=yV)}|Qv<P-WNIj0L+oMkw4R!On0VNruy
z9c|rD8Hbs)Mb|DTZ8Yw&BWu8|oWqTZ$&k6F^clCf!nDM0q1i6BPW(@K57apl?lr0F
zTHA*?i*+4)tC9K-O$^L+-NtJY54_utYGl+@juHOJ_zO(lAAiFgvvGMi{Ggq3=lr;F
znm;#${g`O40*Fz${~6i(=;kvS35{wUZh6-S1fC%w8VvQY?4}AXJF)w&Wgr{h2*mf8
zboMV)Cb7#dI`1WZ?$o(paq;^^q*=CI?lu={Jtry$6M5QXtl`@+`hf81{zq3P@Gdg7
zZDwN|=&cfvASM;t<uzpE#NfPSkusruRVC%ar@-~fyGH`064M8aLji1^N^`?{xl8rY
zm8x^t?+WN(L$}c9>K<L*l;Ar=L4jEs{1zzSt*h9Aaurgp8XzDG<m&6|PB81ALJyej
zpTwTuViO1e0)l{$A0yEs)Xe_jP~1^MxcOK#LwhMDRABh=eNGz#RnKO!?#}yT(MhBJ
zxV(H4fN1mf=#SBg;?l^WE%c#=#PS+%_y8|IrqaV!b)Oism6X}ge0>k2>5XQ<k*`sA
z{bDB(5F=mZ=T|#jO9*vDws!2f7!NmEa;x7;3Qv>{@$-*<f|)EPsNc|>C!h}^ThPAy
zd@l92iLRN`s^JTxyR)7=iTDK7N2E-r&oGA+UM%pifL)D`IeEEHo(hy<37{y8Ha7Hb
zR#735WvTal@<qmWUBXx-zu=n7r_f)`UprQ(!@9gZSDwC;s-t$4KrZmvu=!~SLa_uS
zcY<hc9qWN!L+Q5H<YhYMoQE;*>sQ(RWVX)FF?U*{Qp9}kYaXH3J!AhepB;vHoa%?y
zKonqCBasxD-`>dtxMadOkdv3-7*zKUc{J)U*1#*>(FPL0o5~!mTz4O76upd*GX__<
zwNWh4*hbaxGd{&9nWK~gheTzzVzlRrW%|W@r3?*k?*W~%LtmZ+X*`<{VXmB+rF4pn
zT+oho*@*zzqJ7%c+(jG=PnpSSZa;lP#uoOVckNVc+<{DiNScnoaAq5OEn$IY6M3w?
z)~5~BF{_c7%q0`j<XM*Zi7S-Rux622q+X(^$&ATA>^AZ{!P<jJ?Z}#!y!y%`;YkZt
z?T+9op?Ov3XFaVK2jNI1S-b|2r_pia4cZMJq4PYox5JAgvABGmRoaPu+elGJz1(Ry
ziWJD1qZ6BY`C-B}E#dN2zV`7mZK+-H%SMgp_q~<dMm+4K3TQcu@>WW<$HF|ebwz}N
z7H_w!znLuwu1&6qmZ*nflb3IeuOA&89UOk$aon038iMl|t<EkiZ>iNZoqF%kxg;g^
zkZGz{<>`8!jdHQ3f@-On2CO1Z64g~xPM6;^nTk!ca1II>@<x5m8dr>on*3&z3XwB$
zNFV{)zgA#*pZlgNG75M~@wJ{P5`{m%Dl)Py<&?K%+f7x0_SwBhnL!z4kMFwb-WwCp
zeC2&6pZ_?~yVlJenbZRJ+}l06p&Gv+&p?cf+>z6?B*(Z+AmJ+%GtT^;P7qKtY?}Q2
zi@l^F1CII|4Yyt_l<Vmc4qM(vv^E=Om9i635S!G?vS<kVIyrA<RzAAnX>@>IDP|6%
z?{9q40cEgdWTp`^Sy^}m)Z}J%S*PJ9KxHo@1L4$srrzTpOj;PHqOPZXNM!qZl=i9S
zc?bbBM(2>8S;E&h%A9U^=l02Vci&bCFrh?xk;R5MA5Jbty%A{7?Y(>{E4!p-u(C6@
zs=K38NKMV1HWP>X1W|_J32K11R$}6?DfN8|70bDURDXqgJw&B#$?E5Cc`%<#c~Uya
z$Yj^(_3)&8R&stJhDmcgLjQ$L?W?aa!Pau=-DreBAKdrjCIlgg5p4kS-DVPP#0PC}
z7RAXci9F-($#y@9R`?<Vy|3Dx=E~PR<#>2P1ygDX2^;mr<;9MoMzpCadJ~LwVB6=1
zSIw?!jv24vNkyll$SEMlul!UmT-Gf2RyUeb+#qwtO4H|!0t8}3!hn}lSY5K57JBM@
z-#9Tr={IlQo1AS!6-beZR>KPOmT3KwWu}>fg{sq(R#*`CR`N+3Wtv-QHKgFZadBww
zV^dr4z$do^1ERm0>aR-}C!m%Y3cPvW%w3l)-qiSY-?hQ^?ehZa3uRh!^9GuH^Mw7E
zzA*KV*o_Ws^vCp$s=NAr=V?q|v-SOO^Ww`(X1Cv|uI@P;1|+#1p0LtBH^3iE%uCzg
zVovLc0Qua0Y(@+54t=-ni8DySzYPFXRm>=L7xvENwt6$u5^EgcH6Y<dy#54SpiO0%
z$ZL~xkCS5|&W;W#C?^U?ONvGhoV4m8>NZ6m^Pa<Jft2MCtKwNIwDr!ozqbcs7iPwx
z*UsE;O4ANPe!FncH(~6e^7^%2NjzQcnz}Co4PLbWXn2yaF$fmg^tSASEoak%zL339
z&bQ{e@)VAT((?0ZdSN;XoCX4-DqoCCN@|}m8e<)4@rO#rGpw|mh^GZY-})0`IlZ82
z(VE7i3O;Rjh|bR~*6Y8U<Y8&VVrj8yf`z^n7d>OWr8Pa^>dwS9vFsLcyNGh;P^MTU
zdCTM`DPd5D94)21&>Q6PL7!K??dr+V$Z<LvxD-J56&y^F0Rt00UuheHc%J#32xtVd
zACZzT1uRoZY{0<ye$b~769Gx8#eB-(C~A{Rfo?Eb)czy+<c6bBE&z48EONBT;F|7&
z0uYo$tqR7z|CkI$&NboH?Qp3zEzZD`#PBsBWMcoL4M?<P>{!L*zJR@K(Saz{ZO^LF
zS6J#kUI3pDXycD8oA_+YZ0qnHI}y?n`V(ksp@KcB`x-=Yz*~7TI$O2i-n*mA@8%<L
zlrNgWxxA8`i>*>H`Abe}<$_!FY-gC${E|9>_F1uXGZ6P&IY~%Zs~N-vktY=6c<)#*
z#QL_U{Dk?PBrL;V>o3R^1hcLbsQ5SiZ>4KPjb6kATYK+di>$DE2ZO-&W5^gl<-}ZZ
zksR2qW@Ibo%{%EM;EpAPCF*p?hx+!+Q$*>Vs-y%6&sRT#1erf~-I>=jE-DD^d=Oe}
z5Sn58l{fT)0DUps(C;}Uc5J9u#}VyoeNBTbY8I8Cpk`t1j-Jm438b_JOU#kMXSY$a
z<VVlcnzCcm)f8LR85VhK^ES&x0@pSpp6c$M-78+gb9c|o46lvG+G?z>@$<S_e_ul=
zM8d7}hCLliW|h2|3&GBb1MEFmx-;grNnl{GmRG4({?z;kp-~4vqu>2Spc4B1#jwYI
zd{F1k=TS|pHR6l1>gWQ{l}_D|_EcHc(v||o)(m8++*n9y5T-ZPXJqCK6MhqMC}hy;
z!-q2DW~JA@2M-DF2+|aE(3BS!-x(TS7iFDLRX;71oO0hW$|mFRa(aYbp^a#TBSW%>
z_rSbtsOH`p#&inl=JWYb9`{}RbE?sbZN{T}pB)%>B-b}8De3Ns-_2EfYP>9!Wm`P0
z*T09L<;7@~sDkA0fT@Bs<p3z~p&oAd$g!e@G{zIYgh=RmJFs?DV-&I$L4@ikrSH=|
zDm(pQ3V}s~LPX1o7I#E30N>rePqEvsKK3GHdAz*4syjcKa|uIR`EnyS{8VF|Fw*VB
zX`@cx>K$^tdUioeeSPUcc>aj`MxkJ`nbo^m$kOh6)(tIH?E%}nso$b>c%+BWn#I0k
zvV0R3uB3CGGEg?TSEl8qW%<151!fZoFNjA6aSXM~&^pcprTr^}1ybVDxE+|Vwr<ij
zlu!E*O;4X062%%7;EVnFc{Hgb5@Jy$<HT9Mwnfq8Al;N=1kvoKb&<<Pd7Vu0^yaV?
z+xRH`#wS%*?Dd3DEN+TP9oYw%ETA|RVXb<Rp4ZOjiE_Z(pr!(X8p9kGpD9#3RF4)^
zX8Ml%hOywL7X;Xb$g%ujK@r&A?Vtr%hMLF_^C$W9`%jqDeMdTZorD<}D-CRIP*m@a
z&8%*yAbx@9Z@Ok0vHP7x?h)6!pC43QY>dd9cQLSqkk-s*-F3E`0$!$goh%f4d7f7q
z8G4<1oTMWjW)cMM#Ga7pJxA+bZxNxeDvecM$S-(6zyQN5t}lp+D)?x+O=n-J67Q{1
zqQVb*iBYgEqx_*A#f0uf!AuG5KHYR_BU9ma4Q}$=#-t~=#*!rMFP`@ufs(LAyg-to
z>K@9*UHxq5;RLRuC<>Aqg%ypaRQv@WyHs_DQHoV&U*J*|gq3(3AT*%3(zT%uF;~J8
zhWU0UO1(@xVwAnU`9^E<2ac;5a_SSSkg;mjm=CEu4@u9(si^YuT1mC?(2uF^uFQMy
zaxO|*%U;aUc)B+HK|4-!r=KO$-!1gWV>fkjwQQb?qXeU6jZywt^*CFmZraciJ}MPO
z%E&r3*NRJ>A}zz=iKK&?PxrF~O%wmSid4oYAMYqo&0td@$9^=s)qaPWDoo1SJJtK{
zJ{5o}Rv!mPleUt}CI0Y6r@PHPYPznI6j%i(<?QF8@Hl94risdcv7gx^qw+^eoSA1x
z%bn5_sOUrE1R5vK3ULFfpYFk=8}CoUZ+bx(nb|f>)%!H_rD;s3V#GV!1P2OLbrmsM
z;qsQMw~wr<6Rh@6q|im{zR1(S|5MJR4Ma|;_2jq8IXXdmz`f{u-}1u%+I>s0YO)1i
z9=5V7-P(uB)v3*Bv}~WNW@9X4ZEmCRetu!4#+VrX$ny9j+7@tjIp&@pkN!K8+?6KP
z)XX}i6f6~c8W2GPT4Hi#lAUC}aa3{1V_0-QjwH>j@3V)daofAWIy1LIr+X~%B}x02
z?=reX&o+IA>8bk*#;tx`*2o`Om`m@{vZ80u?O`=PYhmeiE(PDr3Y4l8A*mp7!!kG|
z$(inroeVghU9KCMH(BCjeYTU#`RQ`LAuh%KPQ2r^*JWsOuGo0@D^ayPu8lpXBC{ID
zxvhFmFI)5J3o(1($I0VfimHcbi)@+ct*014EFoou9}0HbldL9Aa~OFyW7p0M%EZ-(
zi)%CX)GW`c)~zlicC!+S(%j+@j`o?)=0Bu+2@(QYUeYRQ*(A3YXz6M^50yxlepy`2
zeIZaQm#}<bCMi=g=}c7(W$R)bE518E^Wew<TCHvL*qb2QvU6<Ci&8-qmagKiL!*?r
z*er^pNpJ6@BABm=Xuk?c`rx>4)@yBjArpUS6u(^LZQILZIq#{xbha=BHX4bidOYiE
zbI?naZ=Lc1DMg*qAant;u;i_*Zu%L7=`-~aHy4fTY1+sG*K=3RB34q;!f%<kzt*B_
zlQZ{Vz%&PtOm8!q3m%5Yy98p56m+z+&@-A2uo};7+L-5ER-Yg(Pa<Im*i1#oyx(T}
z+)DpRW^zo4TgyvzN!2^`HSzda*Qp_*z50cKdEB&=*ThLjPxm=3>sCX>iXZa1+-n7~
zoEWndD~cFKlm-{xOcT5K?Tyol3~yeEnwr=9Hc}Zc^74_#y0}H+v77*3n+Cj{aR*${
zSsU*j-#&Z(aMA3WpAzrWiae{Q1!nDz6M<l>*RFHqu{Vw=E>glpDi<-;gJvUIoNrJv
z$?1eFMCc+c`M(UJDM5ju)0M>IM7ZIw`>nK9;IPMNNF~ebpn0a`$jxxc-?2pK_QQ9^
z{qRhq`CMm*_cqH!G46*j$SxrQBRSo-LWnEwx$vINg**Zxd5$x}l32KeqE528?9dfw
z5gZjM<-EGQkI)Q_gEzO9BQO-Raa9C|^JHkPNh-z8vu-WhS>;~leMB%wUwVL2KpC@$
z``DVSlALKh|5ZVcTR95{wv=cwy59>~ATf^boWEfEir6ED0^&UEHM4comxs>o1`ztP
z=16~MaBRpYeU@(w%Yj4pxQ){&VvFz`n|)2I(tKuv3VI((i}<-UQp~3}EF2EXvg4sp
zBV=3GHGey%oHYm_KVWfS30{s}?i5iFkQJl$N!?Zfx@2OKYLWPbp@s@6tXgCH_r>;V
z52#Z(`>a~iGu+W9OSO_Kb#rFtz%gu2wFVOl%{p^<hiie7gR%tr$krH?-o<{5Fx}5;
zLTYR*(%~SMZqeV=LGjQv@aR+FQeq$q67nN&Y|xQFeyPwGjt@q{xZCM)X8{ouqeAa_
zLy83Xss2P$yEQF#a4<4W{M$5!42lMblq-EAd>!8D>3$_J{kwyN%_PJ@SrjYI^oiJx
zei$g+JEb~bOXHtBR!Pt;$i*a<d^C@w64X5?3nmIleW&x79iKQjt55AwEGoKgONn`3
zDo#E!o`BdL{qFZW23%O5#F#9$IH+IUG|Y<&2qod`QRKiw_sbDQJ>_%8dEL8X%;ibh
zWyGG`%qO_|%1(TtX-9=nX<h(`)7V_9wv)O8`a(@_$MI0Z`Z0m1mL4Qlgib#<UW_AY
zbitOG%ejOqr_2zUQ^6-%Q@dG`RN}>J&qukhOZynjy@K9bCLY94^;M@!u1M>6aN#Gu
zK~}ss)?SdGtDvDnknV4vM}<pV=ua)GF`3NRFOg<&rw5Y!G!l3h>t+vNj{k+PSmL8L
zvIqlttPGT6VLN0t>F0xyYVv;Bd|%{fL(b#|ABf#+6dz0*0la%i4U+fKexp`5sKzb8
zErCP)q?$05IEVKIUX`pBW~Jl?N=A%a+J`RrE-Y|}&7JVKNg1+qL0~c}GEC}ZZNvG_
z;~3!M^0wU84{O`zHE)(;#)8u*pDS^6u|xVnc;~Oh0@J4+3<>g80@t-rpt`;<W0g#+
zNhw|9qGC+?B{JeF@FUl%pS4=<vj?3B7DTGHlNlyR3w^RBntdfdKvyy)j>rFM6=o-y
zy4>S9)A+8hGmZ13`=*|KKsJ6)u(L!vox(dFJYk~TGCU*@_8ME`3?Sd=mB|1>AWggv
zyF=rH6D*v`H~lgAdp3Oj5o?Nc2PB3gIC~ETNpA%*4J<C!At+}9@?EnA^xLGE$H+25
z=a$3Xc6=(KCD8Wi;NWOeh6%Mh2aVyjS=NvmC}0_R1dcr|B}vJbFtqf24aJSMyfw?-
zNv<wRy^hxP=rehyY*^tL{~$iBj=5BL)c%gDRg*;V0KJcSt$;)pWbIBmyZFO|dsS0S
z5>iYp9%<{|c+9vO90_IA>GV^B*yA8P3u930lk(iboH?4voyX&$MH*jwfC&{(gatpx
z@7bC1d^U|#uKA=YM0RGtksMdZz83P%(_mP^Fn%B<S0ikQt!RaEv?xkhk3fH&qLB*7
z&pI2Jc`rRZg^<-x@@ZGtTSL>RSj|f_Zo;SDO4h3<VvL`_z92`MeBSpT);`4)Oq)}?
z+F-2doeB%Yk%r1cXSwd#4<*(c4VQA;x!}c_1d%5j?qsMin5m#sv~7uCXbQPJpuWeZ
zFZO^vYp-S;iR{S@a-e`R=EFN}miU14k|cA4yIa`16V6uriBo1eba8x@g<{jb$9Nf5
zpK1quM+~{DcWTgNaFYrA)uOCtCo^ZHoAc%7VKuinY03$5_v`NOpx0E;IbrmVZ%<{-
zaALM{-sNLF4-eLq@sZ%T`!1!iGsSzGM&lCX%<Joals&b4VOP{~*gUZcEjq1zew6GS
zTI_*gzxCWVaVX^=87)yNk=urqTKkC6xSh0NdweYTeMLp`>Sfn(#YsEC=dVvT2)5$<
znRf{e2xd;F=Ym<!gMbR0k+E^HE-*t!LrW3n9iLoM>uIlH*Z{YgTlBl6uRJd2;yl}r
zS4hgaJ6yKzO|(i8BTd?j+l<&;OzE?{*WXxtU%jQKpHOSI$+$V#__@6r1ufYq^<h<;
za>49o<eB$||3+rXy@CGOA>ZtKJ(gVa&QreBp3D3;50e>JB3O5CPe1b>XQ<$`tY12$
zsK~!x@niz*aF5q!Z?J3PK9V%)jQkX>I@eN3%a@U{4#Lmpm1yo>=vDk%{UHqj<;fqP
z89d}m^>8z@a2wAzry~qF9f~bML2#%EwpyzG#_sjdE52uF-?`JmZ9DOx+XdR5kbSUH
zWieS~c`?{nfU{YaUY_Kibf5DZaWbvC=&49T(d6)^saNIs#^KQyJH#)uH_!ICW}zFi
zlrpb7DO@@>k^?#D&p-M0d){6-IY6uOL@((>%b3no@3rBWe75OITujmRA_3@4DBUn<
zuU*{K=R)h0nyoPGeAI5OQnZr!)TBoXx?J+)<Y;qG>b85A*Ir+Gv&BZ#x|@gX$F;2}
zL1ZzUSUx_hBv$MS-q#YGqc_o!8t0uULu`<+kFvpvZ`-Ita5Ag|B|wzcvd>Y`ogHLJ
zuyaV>9A4n0KIpJFh{|!40JTOjKR}E<x@0Q|m^1i`MEZ4t{3?6S!p_3{M<$*9k4(Cv
zi#3c%#LCiM2xg~mYhrD01qboqm(scxFeU;1s~6o96Fr!%uDywsr2@>>#1Q2Q<ukGa
zK;RAdNC5#WXLSY^2s40z8ORI(f!WvqP<D_eleoRExrsi%rI9%d07PNpx6_AN+F!M<
zruwZZsB0|-Gchu@zal|lQm}_vC<EBxW3D*EF+~^{aGm#;M}a5te@6nKf5`tE2nl9m
z{x=Y$W_Xw73jof`BQAu_RlQpr>5ka?;G|E@!d~*@NPHjEB2ezw%#O#9j#Fv!)g`!f
zcP;W+PU>f8P!ucgkr0t&V&mb}s-ZhS6Q_+@R1w=eq@8j%M8w+KsAy@J$dwymA-Afl
zV6UESQd+sWiRn^$t3dHSkNbqH_EA1a8|Y&jo0ipZON4jXKhB8twtta6wYz%N#tMU%
z_fgl4ZvCW8r4M)|_@5fUEI0FLEUYKrqurvsICwh=4myieW$LPl>Ytl%?xu(lo>5p>
z!#Pf5Hw|cmqG?chJp3S`fmkQwe%D=G1EDSpYI7kt$HiD_b$=Mm<)Zo7H`BsUfG^23
zi53Iy*+pO<)rYxnh^#Iz=KsQ3n1O$&^>?>mQgG0-zw!}9TL;*6uYj%{?8-X-+&fHc
z?d%1Ob#1R~{Y3Y-HVAIsrzQsW#&+s#a1UVx0Ra%W&oHw?0aqRaW&r|NfIt8<GxPQJ
zRr?>;n%~{w4-W-^uRK^Dg-Ov$$<pKsYzBa@eBviBVFvxNoPR*oSLpR$Vbwq2MNrnC
z@gf&D9R=9?Ej72Tnv;%n@dVbFUoktHcuXGOA`kBMwGCj+R3wx&HtuH~r2>ibyUX#F
zDTi~A+sTQrb71JoI^N7&VOOs0UU`hvo!AFm?v6|4m?uiY(v@9Kg>ajBjWN8mIi8<5
zot|8+cvn{r|1<6C9?C~PO#A|M<tE8Qq2`70z`;TJjcvvo@oH?3gz7eHO@uIRTi06j
z9TriCOz$E*>%I8~Iy_aGYf<H~r+ovAa^o!=Cpoz-Rrjvt2J3V4&#M3tH`{#QlY2S5
z<>_=bqN-n*5^|;m&ZE1<@-H*ZJRwOZf`i2?yVF<}%bn1*0V06!xn_f7ljgS@2!_x0
zhx)|0N+0@)+{_Y%xj^uOcxv`n00*Zxk6W4rPugS<!%ZuDcDyAD`_}sVnZfV&5z}eL
z&D+~wURse1Rp&X)k__RMe55+8`--*XL%@0~$dmSnu{M(KayqRyt1hI0&Kmzvn`-T)
z+p5P3cq)Gz@hnVErZFTxmis0WuIYoX&yg->KOB?<e|3*ps^2febh>}W6z_uR$@dbB
zpb1&G2-ttReaM#3h_P*PjK?KOQIF8Ndj6;|sQSabZ3au|i%RWXbsah`F`msP%&!?c
z4BMI-XlAbxT!;BE$)DnH_;?<ZOo-=OpUGJeoD+eWNGKODB9bOCy~k-HPXZ%FnLe7&
z9)-o7A=Gq?-#y*izNIh*&dSeS9I<1vooLe^biw(O`)!98c^Un}H(zM)@<h%m)uVyF
zhGL|5rJj=Q{-L&<uU49IqAinZZO6z&S(=TDPsK&n!<eD-f&MR)$?<C~ftJ=!8EQNm
z9QKyGY8|B)3wGc3N6ZlnWaCOZNZX7S8m9B7ypBuG8ZaHCeZGttAbeNd^iiEG#PzJ3
zFuVSJhDt%Hqhbus2s<}J#L5H86YMOgnHfRnYD_bpT)c?(#oe!X?NSuuMfj`I!S<n{
zLOvgE;I%2R3$GfKD-bm+p>+vu<v5-Qt<tAk8I%Eo9l*Ks)?tW(l<Bx;k9vGV>f-Sl
zU)^3c2}}y~F`;Y$hDA#oaLB<CX!{#R^g<%m%%)2i)^EC^vMFWE&Th7k_i89u*s;4)
zwn%2u8?oGP&@+^mXNrb=&cKgIZ4#X0+mW4a@<CJ?k&nW%Mm(#4NY<p-pB~58)ge97
z4~Gn>GIZ7KIVtaQJ;vqCH_IR&Go9Ssaam)O@{h>toxwS+yJOqxayZZl^?aGm2ywp)
zrm+d6)R_7jMw;NKH#xFw6OddN&2u?p+{2M=Y<-AkV6ZQ2_IP5(g6h6cIuo}+Bp1f%
zBSA@|?jxEax9Ab^F|_;y%A#Hh$76p%FD2<#jf5t4o=>Nt&3wA4GqNv;(V?gqXjQtz
z4jIY5ON!br9NtsRFw*UcYXs605JsxR8;JMB8fetEh*uVxnQunEXw1B6geH7ORzLVI
zB+sli5kH&1<K$h}>{n)j78e)~tMVg7$=hUd@>1!5@L?f@yoaipJs&w#y`(*o%KPpN
z1X0-}M}BT|G)-h-=$Q6Jrz60m=`9z|6;SD;ugv$iK7AcDgc=nS?pyy78!ykWJ*p#7
zuKlY9Bw3R{St_SA4vq{(xRl;uC=7^JRBjr6tuinopX3b~w`GN-+!$He{%GFkIEe7B
z;VA<~x9Srm*21h}XfM^6y7^5tCkwrIU1XIWQMxJbpevD`A><Id(N6(Qt)?(xJI(ZW
zMY>9R8VF(%fQ0)fW<|hx^Y@vbWp7G+5hxt}JmyS_2Ei2E$DGwDdo~r3B`ev*chg9t
z()s}dW<npLe)?Tsh(f_#WxDx=3z&$KEtH{%_+3_XwSkdMVb`o(MskE+NOKw(WMaJd
zW;n#(ID!0m^qQRwgMNEyS%)waE2oci$~(cRi7CS|HYgW;hXhdN^J@)wf{Kky1&7Ky
z%(;L9Nnsafj0*9oTI0$5<QKYT$<}7QlwsA8QR;_Iw205cm|pWS>K<vN2X7$kdOP+{
z(GePpzeAX0n{n(6tkkbFZ?1&&E}NTceH0n+v<=;mc453fs5==TOxaRa&Om6bXVvuV
zq1(f!EGfWV%_1k^JuLLf7KT9g+X~{t?gDqLnHj0=U|ADbiy`JFP=_+}nvE9^-}s3m
ze9}1|=D+i4AY=zq4XF~RCoJBWFrf{NNc=@qGdHUmAepsgU6<<+^%)V^DZZo7VSR!?
zgUaY3U!21WcBZ~b1wU)ZZRS2!KDAB)H%vD~drNgwA&g+_Y|KMe&G;%4rM*5>0Lx*P
zKVCb=>%elK2&q<yVEnLqi<a7=+JQ1!=3#=lsk~9Wm0|r08e)8d1oH9BAH*b_7$d$Z
zf$M2oo_4yUy?(p|$UTzXti|rYEvXvgmz4kPH9l;_PG{61njn!fn~r_UDZ-K**WPM@
z|F)wORkreusaCeq&aKis>F6X#y!8}Udgn^KQt}3n<83hdyg1qML1gD~WFCXIg|d=o
z5}iA_Hag~EzlVa%`~Ej=7I<YuG0CaeceDzJ`wO{eZ__E_5;VXHZ7Vx8-fE@?7F5v^
zizvZTJ6B4+wj3_3JI^24ERyIpSYdeXmpg=-SA=TaTeWO%LUg(<^9-TGXR!vwPsv&2
zr64T_Jo|^@riibUs`=6tH#r6HrMS@(cR1fVDmT&aY^R|jD|uW&#{BGgULKB!x7fuI
zu|n<Xkoj(8`YGYe0ct2><EENc`c3}cHAUA3@7@(b{Gr(r^qv)KF~11(qty~oKRMzr
zxz6^^p5Dgw=%)@UJwnrWQjrVvf$}st_NvYTdo!i_t%`H0g4$^J_UbS0>Q+H3BY3F?
zpCf0M^blJwtl2w0Ptk?!I@Z+`@TDXec7q?ddKd~ShmXX(zD-!%L;gnXqy0_uO<SDK
z9Yv_b;%JZYW6`ZRTO!)l6Wa;x5bi}4<&YCwF71$Pf>h0Pq2OmlastRoxI!IoRGH&-
zrnvHhcw=6Nosm&nca*J;NQ=Z67uXx7+MyUhVl{A~l|^HAMGm<PBDZ;~qee{wp{+`#
zn3cJ=c~j23SBq{K+7?kf^g7ucr=CZWfEI2G9`t>&2=+L#938XQwy<3k<o@P><~+Ga
zOg5N?wqcC=je4v9Gt_yfYeD0J`ly!B-RzjH7)i^p{EK8gHSc{(Z};u$2eI~5kT#jl
z3fx#*fav1EB0l)QL<g$*Oc3EQ?u^SJ3B$+PTSP{iOdpLKGIw>q<>g4YaMzhGWI%%g
zctx^p4(&$_7)Q!6+jRrOQ=GaO>I$nvb20k2R6?|mKW9-)&nMSgG<~Iz3I00sP`gq@
ziL02$b^T*C1uSa*b$SYla-+F7*XYWj!NZ|=t`o|{RM75w_j1BxuI&@>UN8N^;_Chh
zQL&6w?~Yaq<8m3ApZSMr-*eWlrMiQ@46OTKN+x_OLb<JJjE3Kqn^`L5bVi6)M%saw
zzi4!aOC%VVJu&Jw={;*M+9Y<7`4-uee0j1ivpvGJdP+L_UNqRtW!1>cR8r(7wLXvG
z%tH7`fd}td_bfL(#zvWuL5^dSfyxTrJ%D6!`IEOR{3U^3yy+upxHp+iFj3y<m|uqR
zms&JKtZqSE9;)W#C2~uNfVpus?eI)jdUQ>kxDWU1l24|jc3PQ{y1ZP@YeUx@&&_I_
zz)psCM}4AOb=5~pLOt9S4>4*E4*eHpcs;DUD!6IP+2m^WkjGJt9eM_GED)k;2yQ5#
zpdpwZd^SS~MocH0WmbDvF(t-SBWES#6M&c4Wm)mb!x61G@DvgiN&A$x1iYYV5ctV-
zVYe+WX}7V>M)5P5Xh2DS%PW79T$bqe#|!5NM_Zw4+VKG=Q`AVFz3(kYn8zzD-09~L
zY&RZwojG1AU|V|4pO0vPJRL4Qb{O0$nk}atjUKI6j!Z|9ume?rB)V=Y1TE)k@1G{<
zgm6yhfNc`-lwvIq8y+`%sHI~K44G!cKE>Ju#{_BTu7N%(c85qj`r?gaG??yFe?B9I
zRB`?~<<V*7SX{eR4`>FVK%&l931g2*OXCo^UapqPZDGI*%#5lPP#igzfL;k~vS7_+
z=0j7ao%|q^J<DK|Ld2E%xD|s8Vx3FeB&R#qYW(!2xb?-xUQa!H;C98kRwu4=I1)Bq
zZ}s3v%~pF?H%OOm%6C0-T4J+7)mTDHtS<6Qco##351%87X!#?~c2yu1!4Sb}<&9O-
zDd<qFRrl6mTvv^B4DQh?v!kK4Ngk><m6rEXRhfJga`0kQCgdfChTeO__$P0<OjYE^
zHd#g6s<)?%9>_`QnsO<+?EBw-7t%a^ymGvGy5+H(vNc(CPbOKM%7&P+6?jO+TS6+~
z8OBxby76FZ>)VCCdNR}_cCO7Q)V+<GbXSLVvAkVVE#tk>lF@@(2Fokpp|xFAXcY}}
z*5{SO)oh*jt{pg2I*(Tl2_Wv`N6q*$8pqQbNw3QUpGK2=Xwr&$zSsX8_ccl%F;9o=
zvY6D1{bW#cXT!^Nc|T=ShO6Y1t-YAl%4<8hYdp!#iR<`Iw|fC8sr&q8&86XV=x6VZ
z(JU{|b+1dEfl>{!<vTGhn<2potWKdC6Yo8ix!bk3!baxaDHeyy*RJKLQKw2|4%?dY
zFc7~i=Mln9`e<fSnEOPJV+ut{X`X<<PM}EHuZ8Ss0m8V;KueF-A}<yX;kUWma5g3N
z!>moFaZ%pI%g3LynRN?KIU<!4LR?l|5LhDvi0?w<@Q7C~{X`p&zYyHk&)N%(W^?GO
z?3TCW;6%<=BDA@fPC;%w3t#;DWM=L`O0_+e^<_Lq>jy24`)?aoS##dKTjc_(@2^38
z-XPr*_TE&xj~6P*<7G?#^|-yf#DIWQj&u-3xbJ=LON9qQY}@zrJ$znblxKP{ZItU%
zY`t}mpr)FVv<cth9GD!lpT^?rH9;7f8nVAXUQKvV@S=l?QNd9z^>Fv0RZYXR<Jw?L
zt4vaa{*!g7a{r+<!XvJ!u!QuXRa;c|I;oVvICo1%my+QahG)ZX9L#FJcy4%Zm~50N
zCp0K@$`&8M3<6o9tzW5VV%90w-tOJV<};mL%tKg?%GJocNrWp<l!fE6P8#t*?oyk5
zey5f$(kj`L>DwXmsDB{O$9_tt&nFnO`3v=w^%5bErkfiRw9a;JFrMFdxUIEXiQr@*
zHPlYAJNbOmOWD{)+XzG`r4}q686#7=EJb)PW%Xdkci<a)RH!^pve8zbJ5hJ5&C-Q?
z(2}+swK*H8E5E8nHY(8Dt`XN;!dH5HeVNxTbl>Mek%`Hdn&nXt+OGO=1<6HBIC%4$
zioyH+<2L1?>UOu{>G-UPhjWS)(vST_C)fg+jZ7iY^d2-l>kQunS_;I65vMbVu!7>_
z(0GDBAuFlo^aL2#thy^l_r-1=$UVw-^$25De11w-Qh&>h7O3`|0e7YFqwS}fgATVB
z5mDp8_O3ys<M-Kc$9Qqa74odVCZ=m9)x2hlG-LS6p5XvW+I%)LAlKbT<s-wrH>rL3
zZEuA*p>dU$w&ir{JjSf3(@Bx(Zh>;XSxB*|N8{1bl0|a)l3;ZTcNShzHPd_d*I3zg
zTw(MFtj1}+viqBdNXj1Bebm;A&wZDo6?rz&m!WD(I=+NQgHPT;HH$~?3WQKfW@f%j
zDH_p5EVeT{NNiu9w{^8u`4%4fjk>*l$0s}~5cjRZ%NI5MD+TtN>u*2McSmtfu+*tv
zo+qEuZaJ?rJRKS|Ej*CiTpnlC7-j9ZS5Q($L0a^`Bv_=u_%gu~X40)}1{sShV@0jh
zN|~$~s~K?rL?7FQ=#vwY=Qfx<)MSdAgSLh@Oib=8BFbCQ-@noy6RltGo;0DgNLyXL
zvdfaDu|%cS$WqV^tUhXpW;l7k(B$Xowt2RF<fUXyPKWnptV!zI(ExE0ftz;Px&td=
zpXbWBYX0yg{wR+P^U&C%#uSlklfvrJ9O>*mu93X1-IFrJsVAM}M)kO-g`@f{Ik@Tt
zE9H8XxIHSZ`OBm!N4qw>I}?ffLq<kSGooJSO?2b+dxIf<Bwi<EmbSZ2_IaGWZLg_&
zD0%Tcmj}J-E;|QuN3rW_xKB?rorc#QxZgb-ynI@-crkuI=E8xqGQyF{rF9^JDc5fw
zaxaptVQz=txrNtXedq@DOZKmeOX1tuy`mfZYtv5(W4=h3ZVXzcDT{PT7W%}?95Zl9
zPLEgm^h#DS<$HD<hqiGHR^0n&Z<PhjYOp=wx>?^;z2ni{#oeg6@L-qw%&!q;)>Bsn
zCf}vRom%x~D(}MIFrF&)O<5!Tmr{i|P;c!F9(i47Uw2nVB#3LXv%{tybAM1!Sn{?7
zO4mT#XlC!NU>CkrrLeozxN2ys+d%?mGt&67#tgBgVyO;cFO^V|e93C$QU_9>x!Ig~
z9ZFQD(AiTd?&^=e1*8-|dh~cD=ZH(YxEDF{8`jGufMf{`1moSyI24AD2D8d9mHF$e
zV+8V3`J6RUS+-vS-}I`*rk*4VtiR!NkWU@>kdn_Mw_+cg@1dH?JcGBfjFNPB-2u*8
z=l!OTLk@Qo@+1N3tD9bpOz%a~67*6JLte0_J7lTI@#Jf0VEdOjA-1%Ht3a^#s0G5U
zb7S8hci7<0*}j!qm%#~?Y%0v`6j=ApZO}{SCDJA_-`d=7wR}~9*z{cI2|ZRZzP0et
zk`6uN0B+_x2|A|sB7A~?Ku-|X9+J1_#i)f|#)d7b9{ngu0JS$r&$m%e!9+iI<$&6I
z_QeVYQZw#+z=4HD>M%h6J)-_yUbK)$9LP&2n<|e<kjp#XwMe%1$sh~slYbz4JhdY8
z$a_c3Ih<G84(shhz0@MH(hsE+gI~xj_gh=|fF!0Ek=Bz`JGR8(8J@e8L?PyY$cNr7
zFRB^vf}CD7Zs;X);mtRw<Vb<Ys5s8<j_nockmH#iI5}FTWjF|nI}pQ6^dI+;^=xi&
zsl*HL>zbq+2c#Am)4zS*K<8ia$-4pAHIH#T@{&XF?%TfE&~NHADmbOGW{C(sdH|90
z*@*Z}l}Ch+!sue?t07@JnYRc+?Jbc%jjdP%@n3cNBZmbhWeDAO5+Sc>FDet(yTcT=
z@;L5U1p~Hpn5ob{(yaw%1ls;Dla$%7$P@8zFIxo*i{WkTQ@2Sxttq1jf7Q)IH~zs!
zKUce5f{rpXiathDDtZ(LjW+jHyK)srjNDR_?QoT<DU6yhhVlM-Iu0qu^ZT!Fw#=zW
zb<-wgKH=9z6~4dI@0{}vf^Ei>e?;$Nip<Yhey5z14EZKz?zh}6>O5R>%$HJ_`sp06
z=sI|3bg4X^_97VLv0z2_<{-Coifd<Va0|?*f88*0q`>r*kH`Rx4Q(i4mTNS~b?yiH
zt4ZLgwSNNltBdygM++NgXyMO2DJHJE>z5)j!I!8!_o6+zfK1?SdA|CGs(nK$|Ij@;
z<UV@S*V8#p4hdutujoJBr3wc*f+i}km<0S<hUNhM9{`=|iI4b)?*a4QZcq{SwTZaI
zWfA$@NnrRaQh4l3ux|2+wciQ`C3-m4VcuCD_LhjOO3;7FHNZZI3>jDpnx2E~O}Abo
z<fnzFj(=RL3Ex@tOl-%K<1IRV9WaJwUdOoBUOFl!w&22vzhSh43bd$v-x#pOk;l4|
zxA!WbNV{D!mIsS4L4al*hi=DGe~UeV?TOthp=?`va0a;QJ)P_;dOc6POU3SLv3MPQ
zb^mGc6sZo0^|a#*L(S}h%wThvW6-JeJh6g`AyW&v5`%?=ML^I)N{+^oP~qYiFriPR
z1~~>DJ<2B>{VN808P@hC@!Ri$5MQHG<9{iAjDS;Y@Ivkr&2vI!wB0py+Hei=)5f<}
zWw8On%>hx&eeK(ys7KM6%Wi%O3=&l)8Or?}wfiA*gSR5sC^J||mDcs9Vc`O4b473q
zRqAI#-B-8v#TTJUa#;NvhN_Z<X^Lpn7mT)^c+;=a9>A()1seKWl!cM<tKm;(_Y4S9
zBY86mPN<4rE5kCQXk$AsT3x|Kj{|3#m9QSR_6y_3(3tUX)eTos7knvw0_pF57qo#b
zm+8qRvCvFM-)6Sh;bDt6PR(>gt`tnRMvb4$9tfk^OHEohfHZ++Hz$B6D3d|;vy}}u
zIliF>?Y+7CfOa~tSE`<8!}+$y`06Q#^{o3`Po`Cb&qJ2oEycqrKkrijdhpCF)(fmB
z`Im$(#XFK)sHX@$sHZ7hBcnZT0bk8}EO=U1dxlcf-7J?Yo0b(<NzX0co>!2%em%kH
zJ5Qx<y=W+D^)x;Hyj`g~#r-vYp5RbrBPEHrJf$Txfv50Xe}I(s&1z&QujNUxn(+yD
z&=)U{^SCVLHn#I6*Sb~SZ*tWa+^23tZk1k!<u1~-A;qW+CG|^5hL?SzMx&xKZ=8S}
zqs(m>A?TH&NK&{Yr#pG`2dxyXY^2jpXWKTx2f{%lV_rfWjq9O0m-immK0A3Jo6oD}
zus<PJC%P8HE>0LBE$Kcso7XgPv7E8|A$3`rxAiq~9!K-WDH%F3mq%DWZ^_<`r;Y&I
znW5Y1NH1z-q|4n>n%gYh9oRMxpS<O*qZ=ujhE12djC5<JS1T@tR*^2clC-RJCnsrK
zN)vr?aixBGx&6hD0%PyUO=RQr{k+Gx;n@8hTDL?Z{}&HNo028mc+?VhHHN8z1FhsE
z*BKtEOK#xx(4Vd;l^-hYCu!=M%F8*wK&i`(Y(@3FSs%2uu(hPLOeiw>9I<;+7CJI-
z`8aVFy4aFQ(x)XE9V5&6wShU=j-;s72Z|d_MX{EEihDwZPILLpvARy8Ub$YCY2^TR
z;_4w(uiz!Zn>V6}wUwA0dSm$Z;ldjfq*5;(VSPEd$v1}fE=d_XMbRZsY0aaiH+_{N
zs-AjhDdi_1VUq4xOqbrqlgf*vB}~YS7O&%=;d`M@<@yTvgs|<oaGf)G@(N=F=@YE5
z$Fu>hFqB0tIf+k?=>sBTeZ#RGuqQE@0;qu`*)0(OPofb_;_O1-W1ACUF(2bsdR`OQ
zQYF5qeRSdR<RsL!>|_3P1F9n*y>3S--+KR7P~AVuFj?8yp}*szaOmv!;>_OxygyL0
zKX6^aD_Hj@Tvyl9?%%+=%y2*s$_D+9VO=)pU&FfB@Z8UU9pn%3e+TSX{vE8_FKOum
z#6+B_m6?TAy()SBLP$W2M*I<;M@|{i4L7myfKpUd>r+>*ve#8&`_MyFFD@aNL(vmn
z(PB$Qmd`co!H&RSr|f4H*6V1Ippn}65v5~gL4DN)vvpX!Zq5_7C1~K(*%YMurIoa+
z`RsFHCTXAbu3#&kjxWXQ)#cj2XK|G}!VN6;bv+9DDG4{zWXlW7rZ<;i>yj7fHB5NU
zM8gqjIdV+~4-tw$Y&w;g2tCCk-p~6k59eM;e3ABG?IwMFk%Q?^Uo$_W@+NY|!(X~x
zbU!39^=RQ92gOz|(i%Z}=)xTj^6V8+TW4kQsq~(O+b7tGzSG4G1cx!+LFbF{moKu;
zXm})xzLkb*?9F>cMe#PF+IEq47k^vrA0BU5Z~Ar-^fJHN&hybMCGxwEH2a92)Au>m
z3Q$t4ga+QU|CN8S{9%gUJ?URS!T-#m{t^iWGyj1EU!lG%tltq}FdXZLzpt+U-TMkX
zhL1t{2U^Vb1Kkyxe2pH%`TQ9<zP?9(LXMfQ-0o*b{GrlN1O&Lo#o?$i7{Gj`mY;`S
zRUlp+uz=S?UxDHP<||q_e_2~AeFd1kIupEvnn@AnY_Iv_35Fk_>1*}M>KehAWOZ$?
zPDOyPCt;F@*;zT*>ci{+%-^dVrC<gox>si{u3&mr5E#G)1#7}jOxQZ;+kYpuy8_%r
zZLJ)v;nFC;5qtPuZGFX5--Ss~0e<Yl(L^66FDk$!4uDg^3$&R8;UdH5v;$lN<3Aj+
zVE*-ph!o6T*Fe`^_je~yk$Uov3UCM%$O@<Zt%#G0*V$TE-wb9C(1RJ7SaOqZRo9aP
zObocmpRz~+rK|;E#wKE}wlD=(X+?ckGkvH5IWG?imota6g|!8I7J##bxuqS4GdH;|
zyzmpo0dHS*Gm-P40Jv-o3_0Y5M1C6szv3o0wzs$DU}AD|a$<C1X0)<3Vgf;-P$nRl
z2@Ga{k6^HKv9#B9X0WuQxMK00$3LPX{7ehril?rgm4iJuIr(pltqlwvbgxAIfw``U
ziN!Vh>o1us?U+mqu7!fL{B1l3T-)|MATFk#uCL}~x|-yNN&f>ueiy>hj`3O}jQUm<
zS2Hky7@2<%fCKJw1`hf#TORm%9b1@;p&{Jb05(SEYjOVg_Wz4Ga1;HtIL^A(aB+bD
z5C=YvLl9p34?j($Xk}&2^VHS^{_6l-OuzQSWdU&g-U$<e|9Aio7zk!&0J1Xxp^6{~
z2mCJ!lmQ6g00Ms?|331ctmM10;a&pwDyHAv>PqAPZ@m7ukpI8g@qZTipGE#l8u*`U
z{g)K`x7q&BB7a+T|0TlzS>(T@*uTy8e-`=Ms{1by{?8)+CB^=2w*Rxp-&Wm!iSU0G
z`7bH<Z?pY>$0C1Z#o$#;+~iJIIj_G+yh47c!I6O%&$z(TQt)J=xvoCU0-mU25-_*Y
zH~TS@_-mr}dp=Ou9A*JeyTh}Cx|RmIwgzw(fbVHeCRroERrV2{qT~l$&jxQv!gGp0
z<tHIOBq;x{#P}g&dY#bx&&2pH&=2WH$UpE6cp~%vYW^S6uzzFz>+I)`T7{>Ve^2ZF
z1wYv<eSR-E`az$P>foQX2>L;b{{a`~U%33Bi~oQN<QFc#VnaV$=%1~K<rgkLKs5h>
zELVy6AFK=d*G~J-WMTV-%fI%yf5wIV7cT$WbN(3@=r3G;O_%@QX&3HhKP%>k!p8qV
zy>LJKi4FMgc`qDN_?gY$b6?Oe{tJg@{?5Gwcaoo_`Fjow`o;Iaf6s$Kzc?Sfa`E>?
z{qq3zz2Ff5WdyRJF#T5N%EAZ&f>D^Rt7h4tjBvCE!pzRd4q#yhGqSM&SfR{}(4Q9m
zuYE$q#K^%Gb`_6)D~bFmzF$ds9pA5h3G=EAxQ<EikL!>JfZOW2^LlRpT;G@Q@2(*%
z0DQ^U-(B+sfUd(G;M%q10pRP14FEylh~4!l@b{UnwJZO#;UGU~{GY4phw$*P@1=jY
zrCl%OYD@br@K37zL6X0?y@agv9j-RnD`hfYQ^Q-o@0OD4*SH+uyK=tk=DTWsn&O)9
zH$d=*ITUSSFnKGhE8f>A-_<n$a?KHiN#4pFCZ&4~+Wod)Gs#}XOZXzb8|J4OuBm>T
zLE6E>P8|gZPzQV$?)#emrm0at|H9P2ng2&?1Ae!*<`34phA)3qz!kO$KV0xDo{7Tr
zJD>^opx+TqxbHAYni$xre}783x)ii>fWxAoALfAk8u|s{$fUWI5nQ?Oxqd@XQJ7?`
zVV3;*S9R;^-~E+IMO6)8>0oaD{U0El`S(Em5A*zP_b0lRMvo0`83g%h;DYMfnu%K)
zTEQPiu9`4|-{8}$C45Ik;bqKMx51Ah{;;q7hJ<m62#bgSfh=r5AR8D6WQD>%A@EN)
zSj`58Tl*jQ-EZR{g5Srm!pB~ZhmV86=~-ESd&B8j;iodL=;1?v%y7P7VfZIAoY(b~
z@P%I?YLA6Dt}rzqI}i$jfFRd!8Z{6|P4k1e&bBZ^xF12-QQ(|@djVKkSm21aA>ex-
z2rf9>%dTDk%isFo?W@4>a~}wFb<6+M2Zq<@!)^9+ACMjTD=ioVyn^F@9tQ%k!rkuY
zJ}~Q*5B$;x`ICHL5cuj*_UCaR=Bpt1Yajcc`yecT?qj(MeZSDMK>ysw%KGO%wyQHQ
zztFM+|DuoWPwNFiuXdcDX~95v{QXlO`>!?vgF*0<5WkLtTjbY1W+3#}d4X`<|H==@
z2LD;+UufZdf8qy(FZWmb0@+yqWJ@3$JL|9GKoGX8F!$4X;eD*X$_#=)S>Six&*R`u
z`>U?t8UX%@mIZ$2{E3#G_0PIuzdHBv(;Q$R`=4mRK<J-!4+33nu|Ly-!O&l227_6k
zzsd(6c6C_gXIi*Qf3*>qnVIDW`Rr|V;Xj*Udvzd1!Ne85$*{xs7b|<fb<Bi63tR;<
z00chS)q{b!5Pa+4f(kJU2(pO?vq0f1U=;?!l`qIDAk56qFAQd8g|NYK^nW|#Dp12W
zP(fpuzL}ka1%L@6Bmn<`PgW?DA8rn)0J9K0c7a&g;GV+D&nhA;c(uXV*~8CET!$Vo
Pd=L<Yl2TYk1m*t#02!{8

literal 0
HcmV?d00001

diff --git a/tests/test_convert.py b/tests/test_convert.py
index 0fd03e1..400bdaa 100644
--- a/tests/test_convert.py
+++ b/tests/test_convert.py
@@ -70,7 +70,7 @@ class Test(unittest.TestCase):
         assert c.split("\r\n")[9] == (
             "char,1,45.83,58.826,656.82,674.82,117.18,117.18,135.18,12.996,"
             '18.0,12.996,,,,,,TimesNewRomanPSMT,,,"(1, 0, 0, 1, 45.83, 660.69)"'
-            ',DeviceRGB,"(0, 0, 0)",,,18.0,,,,,,Y,,1,'
+            ',,DeviceRGB,"(0, 0, 0)",,,18.0,,,,,,,Y,,1,'
         )
 
         io = StringIO()
@@ -125,7 +125,7 @@ class Test(unittest.TestCase):
         assert res.decode("utf-8").split("\r\n")[9] == (
             "char,1,45.83,58.826,656.82,674.82,117.18,117.18,135.18,12.996,"
             '18.0,12.996,,,,,,TimesNewRomanPSMT,,,"(1, 0, 0, 1, 45.83, 660.69)"'
-            ',DeviceRGB,"(0, 0, 0)",,,18.0,,,,,,Y,,1,'
+            ',,DeviceRGB,"(0, 0, 0)",,,18.0,,,,,,,Y,,1,'
         )
 
     def test_cli_csv_exclude(self):
@@ -141,6 +141,7 @@ class Test(unittest.TestCase):
                 "3",
                 "--exclude-attrs",
                 "matrix",
+                "mcid",
                 "ncs",
                 "non_stroking_pattern",
                 "stroking_pattern",
@@ -150,7 +151,7 @@ class Test(unittest.TestCase):
         assert res.decode("utf-8").split("\r\n")[9] == (
             "char,1,45.83,58.826,656.82,674.82,117.18,117.18,135.18,12.996,"
             "18.0,12.996,,,,,,TimesNewRomanPSMT,"
-            ',,"(0, 0, 0)",,18.0,,,,,Y,,1,'
+            ',,"(0, 0, 0)",,18.0,,,,,,Y,,1,'
         )
 
     def test_cli_csv_include(self):
diff --git a/tests/test_mcids.py b/tests/test_mcids.py
new file mode 100644
index 0000000..006454e
--- /dev/null
+++ b/tests/test_mcids.py
@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+
+import os
+import unittest
+
+import pdfplumber
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+
+
+class TestMCIDs(unittest.TestCase):
+    """Test MCID extraction."""
+
+    def test_mcids(self):
+        path = os.path.join(HERE, "pdfs/mcid_example.pdf")
+
+        pdf = pdfplumber.open(path)
+        page = pdf.pages[0]
+        # Check text of MCIDS
+        mcids = []
+        for c in page.chars:
+            if "mcid" in c:
+                while len(mcids) <= c["mcid"]:
+                    mcids.append("")
+                if not mcids[c["mcid"]]:
+                    mcids[c["mcid"]] = c["tag"] + ": "
+                mcids[c["mcid"]] += c["text"]
+        assert mcids == [
+            "Standard: Test of figures",
+            "",
+            "P: 1 ligne",
+            "P: 2 ligne",
+            "P: 3 ligne",
+            "P: 4 ligne",
+            "P: 0",
+            "P: 2",
+            "P: 4",
+            "P: 6",
+            "P: 8",
+            "P: 10",
+            "P: 12",
+            "P: Figure 1: Chart",
+            "",
+            "P: 1 colonne",
+            "P: 2 colonne",
+            "P: 3 colonne",
+        ]
+        # Check line and curve MCIDs
+        line_mcids = set(x["mcid"] for x in page.lines)
+        curve_mcids = set(x["mcid"] for x in page.curves)
+        assert all(x["tag"] == "Figure" for x in page.lines)
+        assert all(x["tag"] == "Figure" for x in page.curves)
+        assert line_mcids & {1, 14}
+        assert curve_mcids & {1, 14}
+        # No rects to test unfortunately!