From 76b3b6e20b2f7b66a5b7de16cee9a90423a225c0 Mon Sep 17 00:00:00 2001
From: Frank Sauerburger <frank@sauerburger.com>
Date: Fri, 9 Jul 2021 15:56:25 +0200
Subject: [PATCH 1/2] Test for kerning in PDF output with type 42 font

This commit adds an image comparison test to ensure that the text in a type 42
font has proper kerning in the PDF output.
---
 .../test_text/text_pdf_font42_kerning.pdf        | Bin 0 -> 5364 bytes
 lib/matplotlib/tests/test_text.py                |   7 +++++++
 2 files changed, 7 insertions(+)
 create mode 100644 lib/matplotlib/tests/baseline_images/test_text/text_pdf_font42_kerning.pdf

diff --git a/lib/matplotlib/tests/baseline_images/test_text/text_pdf_font42_kerning.pdf b/lib/matplotlib/tests/baseline_images/test_text/text_pdf_font42_kerning.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..a8ce9fca346c69d372dbbc0f2a0f4b60fdafa6da
GIT binary patch
literal 5364
zcma)Ac|25Y8&<u&)o8J&MNW#c&T59D!jOF#5ysG9Fk)uR48{^IrYK8#vy?4MB>R?7
zl&z3NiX<A9B}$|a>N|%PZ+(Bf^T(WXuHW<A&wXF_GiRRbk~YQbC?b_GFzK8gAiDyF
z1P}n-!40OS2Ea|pE)*7k0ucfLKjh1x0B|ibo9s?^0ru^KQD{!kK$RsEv^ng9=4>*X
z0$>&=>CkCx0KIr`N@lYuOd6mJJ;LCoOuC~vg$)qFb?`c1d<usR!1X-9H8sCqwZ32V
z01^N<cO^SfX)XYEak`a>gB!(>4d51n|4IkBv;Yknjm~BPNW@|WT3|L58VIN^J{wV-
zsANq#2h0H)fx#k_Q~)$aSqTe}zy!%maPW`v2^1FHi|GjR0Tuj{4~U{AWI%y`u%|<H
z2kQmkI_@ALc#0$43FOa^LUUoe0?0*D!L(UyCWY()<0QNbZ-Lby1cKGa%tza{8RD=q
zw{3!8y&W&L)t+E&Mm7GL{StBY+K{}sc-&0fgNBb2fl5tAN#pJY<9kJM?Y3GcxJR#1
zT2w`J&u>Oq9)kG{lN@B4q#BzOL*(j?n<#grCTN|=ixb_p=1QSH;VCVQo0{@~b*m#`
z_3ZV^SCM%CZR;}x_h#mZdnC6mo3Rg$G3rR)CDE-t8UB?hnY&^}R8cKAbeo$$<pR|B
z?^%D}0t?kZWg~IOAKPMqa|k1XT>cmbHS537gZkdEMYfOt27|^zEdM0yiKTDAp(mpI
zrjCt0yg)$-pFVJrBDK~ZkSY2r&DcS+tM0&ZneGn5IE4e)3j!hH8f%Jn|GiP?03NR|
znxk`8XeDuFr$e_~UB!;G7B+Yf(dTCGCvU|Hq|KM!2>R4HJMhB4iq*PrEOAy>z)Znp
zk8csWSjG%qBou1)AiViuIm(AI8sZg(-jv_^FfkFyzDGGDE=k#XOG=$yG+%VlSt|GB
z_$7JPy`6{a_S@it`Ch5A<mZ*y5|+q)cQ`XWr~Njxj3hTbZ0|ZPmn)lX%Q34Bkm7gj
zCJp3glgCE;Cf2;VR&=(Zy}s;r{bvRHqR6Z_(qE6~et6k}_m&)sly}jYJoePD!9aY2
zQhtuT(6_%8U2btbtIxhI5A?Hjcj;;9vL3c4q;(R-P=TR^nN9ZAsL96$D%jawI>yRR
zygu<xP3|;H7o-!6JoH-XpTgt9)U<atH(s+4@me1)c;8_nZ0|`O^Uc&t|6~kCrPbKm
zYac7mP^%%Q(wZ*XnG%TFH<Z-x`d=>Z=14f}%M?q;9dhfEq&XM`O-nr1{AZXLSJ^(P
z)S`=dyxtH-8t>8YiFzJ3xF;yLI_~m(X|zR$t?fwi!4}rMg_l}vY{x{86;Y#wBK^cL
z_DPJwl>uguX!FHb5wy#R>L!yG(XVciNS0ko^w%#;w>+C8jO)#F(H*vx*Un0h7-Tnb
zlyIX{_vb6yHg(QUu6bi>BVygB%<n54d*++tvo=|$N9y-f6`CrCo=v7{sdUk5Bb(cv
zBo;Mv8%-6AtABl>&Uoijf;}-5FP8DR!s3i)jNpq+A9MEB`F+jQ&A4y0b`?8sdPTHb
zq5Jms<y~x}$Cb(zI>uDX;{g|qSqii_V)41r{k2j;JjcPeXY(p0qlP{GI;;C4N0RHZ
zTLP|!D86pn`mt{)wdPDsB$w#V841XPza`^tk6=@qMk-WSU4HCEtg~`Us|`>$sB}Ho
z)p+gV4F9EKqvu-kR6*AI&{~J^nESU9CsifLjTUV2O16->EVmp_*AK4}@hyq<NQaf_
zmfEn}U55NWS^LF*E$lUZ^=Z#-61;r`54+K<%G}kL#Ys9kg<6-6^$dGEB%6STtslBn
zaVjE3qH<`*XNS7LI2+ehif^oXQZL6)4N9#a_?q2#>|Ax-x0>WGdQhbzvkvz5O8BGF
z#u5uQ!|u%R+S~(4H^K}qq((F6;6=IKPx}nL=3w3-s|nq@1D=zk*<A^<JIuV}ItNR_
z)Cb$v8^qUpNYvSdoqZ>ArcSmwu_Qd=<u0p`vU>ev=f7Urcgbdt{BbEa^7Ht05(ZhW
zuY1CeW|2w6oQ<I*HUVBFR{U--JgjTb&O^-Qkp8M^J9kf}@y2t}R<%)^E9bLb?6Hun
z{o;PNyUtqJ!cn2c#Z6XS(|@<1h8x+yxhiISV(n2{TO{kDqC=wh>Fy}K+N?wSTS{ec
zK0&F+EYEeMtl9j=^M+fm><P=h2LY`HE7ydk1@fsIXJ<w;ZPK==d+->>Re@CsuF>;0
zywcLQy&CtvJsB9UGkeUOP|tNKMpfoGi&9W<T(aH@Wc-ZvrGX?~iG|+`?&`HC=(3*T
zr{&&{>PjD82n}sO=WZHXUcKT-6LpgHUS)H&@~UIQulqmCdbLPjRk9Tf0X9yV;jKu7
zGI7I7enG6`j#Eco0E{$Jx5r$<=4a0-`1FrMtYZXtI+D68BI&G=Q9oa<@k~~9(4ZJ`
z*YIdMzf#w$P@&XEv4T4sO2m(bDW0ClI^i~CSw!R-72nCb7s~O7xZlZ4ZNit|z++rH
z(=6@XNe)D5%cIR{ShskAj#&{x3TbDE2y*or@pK8(Oyzu=FHtdHy5AOe=Dj~3Q9RDs
zi6JPM<Cr6>$N7(4K$o>2O0G@yxU$8Ir?r1VFaG0DDBXl1WS^vOEcDoB$HYYPz=fFy
zX^mom&L@w~1uwUU7j4*?$#7edYh2lC=UClJgX1;F{^|q{pVa;~IyzkIV$vKzN?EVk
zKIA+o=aF)(Dy71}W~c3Dg@)Al!DxfkO~;kCNv9?3yjqZyJ&^T!OrI)NP#j`=YiqpO
zEr*)bP9x7I+%V*oIbua_*cY^6#hz~8FTTvb)(L8r@V+pHOF8bsyS39@seYHY(k~fW
zOBuwcnnXQ|S^hrnYPZ_4q+1c2Z{830OS5&drajm7A)aP=mX~<v<zYXK@*k{=@6YXg
zShag^)o@z!+VIi0oi=wfj=rZK8|#iM$#=cd;(jH+_oa#_ZT`zAdgfl(=vdtR%~=j_
zs(|om7xP-a{e~8{NI><+j|FevIq#SD*w|Obkl50g6(qiX)6os(GG80UFLztMZqt<;
zw4<r<;@8ss=!y}C(>B=5l=Pn+@%Bz~`ph{cCUWoEKLa?t;oH=<Q~6U@q(6zGD_r?~
zItUdB{<`aw35qRgsLysL+x9IRWR`vS5Njx8FKZ}T9DD|4m65bpSKT1?e6k!>gq3M3
z+8pUDmnb(q@hFEz^pcU|l?Hol=BzJpstbDQWn0#%_POIuRT%AXeY?)qz^5}~YR)L-
z<WB=1$_L0MZw$9Q)i+#MeMvB=57k;5rRdNn8a9A}74+%HAU+4W?&>Td+!M6;SbDtt
zbmyL!T&$vW@tDZwS@m4j9-F}x0#$eC#AXJ`%E{<g=NO3a%nqzaIeac$^L}`>|9F{G
zfJEJ<RaJy(0x#cguKoo4<aE!u)~6ZGts=&&#;F3Cn$xV+K?D_5!DOeGj@<n+uhN9e
zC(;1pf|sGT0n>NIXUMB1!|r-=r$i&9U#a7I*Y!(q%N)log?oD<aUymu<DK!3JTXLX
z49^uCam%|(E%ih9499{u1*<ln&eil6$ltmRus2gt<Ez$dXokUa*o_;%@McuhG8wwO
zD>cQi*4=Y9yu&aP`K_j1`5!LwoAi!Go5+>Cdk%jt@>$(%_td+4Rd2osNBX}J=2jt+
zJ`6|t7ar!tYK)q33{xz4X553zr@1ROjD2o8v!d&^yyg4P4u=)C^<R;M@ie%}@}q~Q
ztrfg?s};Q1E~(q#z3TmjBU{e7^Y`!L=1%D}=EA0Q$j$2GOas0tckd?u{=#RWMuJ_}
z7(NJ_-dusOhQBSl1?osXdtQ@j8JR15f*G@4<|n+lpA@;w`M@0eyt1BwXhp8;_;8tT
zp)v0%QKy+GO4}4(nlRAK&asw|?YvpIM<_e4Q0{nqIRWQi*v@t4isH&&!?^LbH_gc9
zHK>*L)2B0Jd{1eb3+HV(Lbaopcbpp$*(@5PQlJ7`Cuo&!Dyd0YUAkRiZHi8cO^SYs
z)G~gNflActIok*3yXl&nP7%05+~YO2T!)q4@FLs~k9Uj6*`_f{H8lc*6;^NGzx;50
zeRK8&<UumlKPmCVt={{h#+RONEbfU@EsagcIxie5FY0W|yNDLaV>ndQV0tfQAiU0+
z8^7bq&ZoT**(Gyu2a!w=u-0&rv(1xkYSy~n6*)=$AZwe!7?v)aAmEhr&PB!6NxkcL
zy)?jo?`6NM;UQg%I7F}Fu3$u9f_a3GV^5c^>ch~&sGF{Qgw&`2a=M}AO@xr1qZEaJ
zX}~+MQ%MotQf(1tjpC{`@(2G$IB;^d^6EcWt=;lAy0#tJac+CBuIkRedv(`Zil;L)
z)(WY#23v8Ki;LIr8Iuq2&8M|R%6FOY<^&!#A5c5R;Z`l1<mrDh3vs-9lQU__ke?)k
zP)=CYySS#ZDR-U)$MJ22wV8RMY_^}hn$|CnCBO?N-EDBduFT#N!<SrWob^%o(Uu3?
ze)`7D4x#M{z{)n6<?@}OQQtP4`r5f{?s?lcNzsf)U4Ne27mnG`c^svJL;O6XLjvo+
z9MX}h|2mR89ua5U(Yz>U2{^CdTjt9r@aKBqjQ<uIg;D-jY?7%uy%rF7F(Zo0(84EV
zBq<nJRB&`{p1c$By2G&2RR+RVeVg5OzJg$KGSlB@tZKU4e>f-fPT$FY)O8*;Tt8k+
zJc$=)U}7u|n(8!E#s;abX;lfG^@>pOvfTR=bEi*M<z7L}d+YLLlHz<HI<Fwa_tw1s
z*t7Av{jF~eYdnHi93`4w%j(gd&W;n{h>h*l_7v_^Z~ThdXBo+O5NM%u!I(G7tpi3w
za#YWkK5{LJNwN6WAE~W8q&R=b%<ly8Za#j0A}aBvWx*q_l##t8VIxxD$ko%sd!A#N
zYfgwY?s~8#{aJYZyQbcmbp5ea%O(TH{||eRrvA|wO)`tJAg<w>MrLMsOL;uSjcn;<
zPNuP7a4kK&xi5=N@zA3=(*a1yn=qXyOh`TN)^h?yEtTyn3&05!7b>Wze1Y8>ppvJ6
zh`ASo;ZE^@WG@1002JvE{h%J+h|B<x3(6bRv!IH`Q&^5nDuYdD0?6<3-rUQ9y=W2W
z;h`arUx0#rw+$9kJq#ch#q$!m4!HsUp(p?jEeYyKGK~!dV}Y6-xo9$gQeB${s%xk`
z_>b}wb)jm$X9T5!038Op1&z8O!xuK<PYu72+>(@_f_|i=0WHb~R8gvs_rcJmP+eRh
za=xd-1{6`CG=((23kyJl^+Lfk7d-_<Bm#v56fsydfW)a{K+%szLrw-X1T+UIC=doi
zfnErN7?V9ff8dX9f+c~D3)NAB=HgC)7B(Vt4pW^#4*{!!0c!-iYzZKi^1P%saI2pQ
zFRAZ`V?zY6${>wU91;M2fw!Q00^VYOqCchkQciS`0CK^^fwstZX}@G|{sNcu_7}M1
z_3{f`VhsEOA&=xoAGGLRU>lGC+<@w|(0j<{AwX;{`cjKl)sneOY@lB#he>gUApjH#
zhWPmbNl->B1J1xt4BF_>2cZ4HFgUP>zhhVonDuYtusHA)_Zx--BmIGa8{-cQ0~Y-6
zxd;T<q2Dny8YKI-ao9iCK%fv{qkkKxtnz0rXxyI|0<v%ZJr{#Q{lNhmi}<4s1Qz;R
z`E9N;xcz^_&{)+!>P7#*{8W~%%Vv_nUk-D@o?LjlK}14!9Dz=U_Q%4#In|E>HhJ+c
Vgv?|wv>l^@QHDuNYn$l6{s+5*s~Z3S

literal 0
HcmV?d00001

diff --git a/lib/matplotlib/tests/test_text.py b/lib/matplotlib/tests/test_text.py
index abefe3c3ab04..8c575a7be1a6 100644
--- a/lib/matplotlib/tests/test_text.py
+++ b/lib/matplotlib/tests/test_text.py
@@ -741,3 +741,10 @@ def test_parse_math():
     ax.text(0, 0, r"$ \wrong{math} $", parse_math=True)
     with pytest.raises(ValueError, match='Unknown symbol'):
         fig.canvas.draw()
+
+
+@image_comparison(['text_pdf_font42_kerning.pdf'], style='mpl20')
+def test_pdf_font42_kerning():
+    plt.rcParams['pdf.fonttype'] = 42
+    plt.figure()
+    plt.figtext(0.1, 0.5, "ATAVATAVATAVATAVATA", size=30)

From 67aea087db15623c888e40b34d6b53b04872f51c Mon Sep 17 00:00:00 2001
From: Frank Sauerburger <frank@sauerburger.com>
Date: Fri, 9 Jul 2021 16:27:36 +0200
Subject: [PATCH 2/2] Emit Type 42 text with TJ and kerning in PDF

The commit applies the same kerning algorithm used for Type 3 fonts also to Type
42 fonts. A string is split into chunks with kerning between the chunks.
---
 lib/matplotlib/backends/backend_pdf.py | 45 +++++++++++++++++---------
 1 file changed, 30 insertions(+), 15 deletions(-)

diff --git a/lib/matplotlib/backends/backend_pdf.py b/lib/matplotlib/backends/backend_pdf.py
index 10063bd9a7b3..d4cde3155af4 100644
--- a/lib/matplotlib/backends/backend_pdf.py
+++ b/lib/matplotlib/backends/backend_pdf.py
@@ -2236,6 +2236,20 @@ def encode_string(self, s, fonttype):
             return s.encode('cp1252', 'replace')
         return s.encode('utf-16be', 'replace')
 
+    @staticmethod
+    def _font_supports_char(fonttype, char):
+        """
+        Returns True if the font is able to provided the char in a PDF
+
+        For a Type 3 font, this method returns True only for single-byte
+        chars.  For Type 42 fonts this method always returns True.
+        """
+        if fonttype == 3:
+            return ord(char) <= 255
+        if fonttype == 42:
+            return True
+        raise NotImplementedError()
+
     def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
         # docstring inherited
 
@@ -2270,26 +2284,27 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
             }
             self.file._annotations[-1][1].append(link_annotation)
 
-        # If fonttype != 3 emit the whole string at once without manual
-        # kerning.
-        if fonttype != 3:
+        # If fonttype is neither 3 nor 42, emit the whole string at once
+        # without manual kerning.
+        if fonttype not in [3, 42]:
             self.file.output(Op.begin_text,
                              self.file.fontName(prop), fontsize, Op.selectfont)
             self._setup_textpos(x, y, angle)
             self.file.output(self.encode_string(s, fonttype),
                              Op.show, Op.end_text)
 
-        # There is no way to access multibyte characters of Type 3 fonts, as
-        # they cannot have a CIDMap.  Therefore, in this case we break the
-        # string into chunks, where each chunk contains either a string of
-        # consecutive 1-byte characters or a single multibyte character.
-        # A sequence of 1-byte characters is broken into multiple chunks to
-        # adjust the kerning between adjacent chunks.  Each chunk is emitted
-        # with a separate command: 1-byte characters use the regular text show
-        # command (TJ) with appropriate kerning between chunks, whereas
-        # multibyte characters use the XObject command (Do).  (If using Type
-        # 42 fonts, all of this complication is avoided, but of course,
-        # subsetting those fonts is complex/hard to implement.)
+        # A sequence of characters is broken into multiple chunks. The chunking
+        # serves two purposes:
+        #   - For Type 3 fonts, there is no way to access multibyte characters,
+        #     as they cannot have a CIDMap.  Therefore, in this case we break
+        #     the string into chunks, where each chunk contains either a string
+        #     of consecutive 1-byte characters or a single multibyte character.
+        #   - A sequence of 1-byte characters is split into chunks to allow for
+        #     kerning adjustments between consecutive chunks.
+        #
+        # Each chunk is emitted with a separate command: 1-byte characters use
+        # the regular text show command (TJ) with appropriate kerning between
+        # chunks, whereas multibyte characters use the XObject command (Do).
         else:
             # List of (start_x, [prev_kern, char, char, ...]), w/o zero kerns.
             singlebyte_chunks = []
@@ -2298,7 +2313,7 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
             prev_was_multibyte = True
             for item in _text_helpers.layout(
                     s, font, kern_mode=KERNING_UNFITTED):
-                if ord(item.char) <= 255:
+                if self._font_supports_char(fonttype, item.char):
                     if prev_was_multibyte:
                         singlebyte_chunks.append((item.x, []))
                     if item.prev_kern: