From f9f7fae76a52dc52efb101ad61d92dd61a35b32b Mon Sep 17 00:00:00 2001
From: voje <kristjan.voje@gmail.com>
Date: Wed, 13 Feb 2019 16:49:45 +0100
Subject: [PATCH] added formatter for mate-tools parse_full.sh

---
 tools/main.py                                   |   6 +++++-
 .../parser/__pycache__/__init__.cpython-37.pyc  | Bin 0 -> 129 bytes
 tools/parser/__pycache__/msdmap.cpython-37.pyc  | Bin 0 -> 9043 bytes
 tools/parser/__pycache__/parser.cpython-37.pyc  | Bin 0 -> 3198 bytes
 tools/parser/parser.py                          |  16 +++++++++++++++-
 5 files changed, 20 insertions(+), 2 deletions(-)
 create mode 100644 tools/parser/__pycache__/__init__.cpython-37.pyc
 create mode 100644 tools/parser/__pycache__/msdmap.cpython-37.pyc
 create mode 100644 tools/parser/__pycache__/parser.cpython-37.pyc

diff --git a/tools/main.py b/tools/main.py
index 5fb8fc8..eeb21e6 100644
--- a/tools/main.py
+++ b/tools/main.py
@@ -17,7 +17,11 @@ if __name__ == "__main__":
     # kres_file = "../data/kres_example/F0019343.xml.parsed.xml"
     kres_dir = "../data/kres_example/"
     for kres_file in os.listdir(kres_dir):
+        out_file = ""
         res_dict = parser.parse_tei(join(kres_dir, kres_file))
         for _, sentence in res_dict.items():
-            parser.to_conll09(sentence)
+            out_file += parser.to_conll_2009_full(sentence)
+        with open(join(kres_dir, kres_file + ".tsv"), "wb+") as fp:
+            fp.write(out_file.encode("utf-8"))
+            fp.close()
     print("end parsing kres")
diff --git a/tools/parser/__pycache__/__init__.cpython-37.pyc b/tools/parser/__pycache__/__init__.cpython-37.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..8b49a1ebc35c5467384b158794c4ef7f27d2ab9d
GIT binary patch
literal 129
zcmZ?b<>g`kf<FhsVnFm`5CH>>K!yVl7qb9~6oz01O-8?!3`HPe1o2BtKRK(cM7Ovo
zN4F#~Jv}ooUB4thKc`r~AhD=8wMaicJ~J<~BtBlRpz;=nO>TZlX-=vg$c$njW&i*J
CV;vs=

literal 0
HcmV?d00001

diff --git a/tools/parser/__pycache__/msdmap.cpython-37.pyc b/tools/parser/__pycache__/msdmap.cpython-37.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..82d62dc9145da39670dd9193fe6bd0c936376dab
GIT binary patch
literal 9043
zcmd^FTbJ9^6_$LFHJ(YJgj=AOK)D&13E@)Ol7@r~q!btefdoW?(Rgc&#<Jub*(T#o
z3MJuAy4t>Ub+r%Vv44Um@+Zz~AG-41C!V^$ePl}(Gb=x!p5<dnXMbm({q24BxoDo-
zv!`kC=Rb$&?0=uJtbgNX`l<89QU0dSadE;Ti|ot%y;Sk6z@|#CqK;6NYE<u5y~>3j
z2bERp+^t?Us0G#~>zjH|y=q^wZR`94XDo6~S*=Eg+nIeTe07w+>7!f(mS>aYRfyZY
zY<X3*%fH+<Is99v2D$v}&;qy9)Xpm@_5DG|W+*=R{fNKE`I|n(cWB4-jT~TUfts{P
zOWmbyt8LL9+DkXl&GaL>h4#^{bQ|4HchH@*pYEc&>BsaF`YHX44$wVxFa4bEqx<Or
zdXOHX7ClUl(4(|WD|C<^qeJvKJwZ>>Q}hdZntn;YqF>W9beMiaN9b8PO2_CqI!>$f
zJiS0C=(qGD{f=Ium+2K+qgUx9y+*IoDSCt6q|<bU-lDh3qj%^mouhZDP4AIU>(n8l
zfHtU0?^8&d)T4;{6jMS2I!}sH8q$cy^Z{L<Et=5pX`3$6ALv8+i2g_)(<Qo0pU|iD
z8GTM)(3kWTeck<tSzx}I%PVWGnqEx(ev%ITC=PqtKBa4G$tcz~s+?}*cauolCv@#}
zpw_j0nx8HVR7k;C#82vlW3(A`hT%BS_A#z)aBUhTV{N~#ontg+n7WBQgXSiF*bi8V
zryD0#60;t)C{-huf395}X05)Y8*8I}pnO*SifJ>TDCuf@RX2~(AW6evn8d1@wL1<b
z+#9vPw|en-5^s)T?oq;49B<V|4SxLHl_-dz`en~O>8oMbiMYS7wQ8Qz52zmu)$J_h
zExi~I110tV{UW0=LXw{HH?Ljw$G)~-^_GsUud84j`bNduEVRTzaWERIYgfO!62ytZ
z=+W-$K^oEsR=m}6y@s#+^|Tji+t>ApKM2z~vpLSgZaCC7>Dmj4>Z{wh4ZzleG@zJa
z>*{`fG!Js3V8~t8cBCHU$|=b01Q9zGsE)3W75i8H2z_dFnFk@Zt7~s;4Fa`K)aVNN
z>H#zlgDHuU@2<q$g2Jt?ZUw2fxAf9#uo1@LklRR81*==GI><jeLFYQE9)r{&6py1Y
z#uc#e(zS2HF?1q3<N)0`7KeSGt*r6$maARzHziy}tFEWLh=bt{uG9{(S2*TiiLM>@
z(?IPPkb07T<gt1RS1?QcV)pE5TxIr@B&9PYNz;*Xn;^K{reU`vInPX~ZNrR+P!I*z
zzTGtDs7KKJUM_TVAR!zM5X4HYpyFmOTJ8vAz`!sfga#*BkHVg>9Ekmr3vCZ{{j|gw
z#vP+z`4n~SYA-ZIi9#U?D8(~`Wpo*$OcxUbeQXfAIu1$ncukJ1jcHy8I1{7;zEeh@
zu9F&dO}y!Pk_PLhciKsVP7;&0AJ@*wK&9+fgHT7I9OkHFD1_?gaD}X|R#6zLc^Glq
z1q<fL_8RC%h#{F8TMXU0LE~gIjI%K?OGjaph&{T7ah&nVxoJI6)1JKm-RN1a2q@w0
z#I#QR7FSt-?Sha%D<QH#+XW#kln`0K?TpB~i0CLGvmu!y5mJ(r*`9VP`~-&i8B#C3
z`T^&Yc1B6&f(0g%BSCWASp5d3J6w2+>!B}k04^WG<7ZJA7i$Mujy-UMpXD8$dB9D>
zxI2n`rJe^!2sKFc5+3{zCN6jxbvYt)!7d1KK?#w$U>AhApoGX=urngJ3@#`kGZ*B@
z-Cd9)UxbB5r_9(qGpl7hdjtiD*R{SM@_g1eA)m4K3aV#<bN%w%j<1NF5vhkEB^f3g
zDT_RZ$H_QbFN7wYO7d;SNuM!hEf6cosaf_tiB?`i@e~U2%<H&1g`${!OOoP|l4RzQ
zlB9T|B$;_)UUJS-RYDjey++TUz(Q^J42@{r+e1M=iFty`CmBPH(Id&X7xCJJ=T2X}
z!4SO|`@9;8(;ntE>EdIs_z>!x<B;u46IV0xdB#rK%&%%eNc;x^JZGR1PN;<xY&g)3
zet=oNr(Oj>xD{8pQPMG;l$MIFI0MOCTwL)Mp5>BDe9IGiNm5)<lFVFDk`y~jl9?;!
zCFfmHLKs(wNIy<^ZNQwO-#KWO8=OHj;zf<&r;r(L1Qz4phVe4F3DOvUBczZ~LK-q|
zgcNd0$Si&^Fn2dKw;K*GwXiE$H*=diC-J<6LIxQjU-T1M1b}Q^O9x40vdBW;PdlS1
z<b*F(&WRwV*+f`IDv#l2q`!kJ3AwYl$`~$bGV;P}Nj~FsM|okhB%iUlll(k~C0xc~
z38$C9!2J9-15W3^j0vOJYsNn3$Pha!WA?9MHrn-^bTnWBwLMaY@zCUAuMzv$#PI%N
zqTa-F*yYvNd5c75raX<OrhQ#u51iur9oOm=-hQCcth2MB?Bl@>ka<4Vr&4ok^c>sZ
z`nffF4tN64%u$Pc9S3F)fz1%~>P9+D2JxmZ5ushPhvSv{GpKY{gRTmKX=LowfWrBK
zEPoZ@3OE&@K!o!e7Grs;m1jVqKr*OM*$g>U&ECb#Lo%i9Gti<w_)M_5&loF6Y%HXz
z;JKOFY#A)|g%uAcZni8<ft4;7Ry-Uh!Ng~U{52#k;3krN7otVY_|@)O&@~Bt7BM0k
z7N!aqs@-@h_#Rp)nhA!Ib?oaf$7nmrXBh2tl0MIEh3VcnQ#TTo@Vcz0LNxajiaWT_
zOCcsg-lfNUbYLib8kPIG;JKrR>l`A9Ho~*lP>5s}nk7j|03}J&m2#UfSCu4<LraqL
zL0&?bLGL-(g|Hi>%X!$iJnQl<WIDAJ*!0|F67rnGnD@|;0!9Uqan-L-F^3r>q_E5_
zY_RTPA#YDQT2ExfHRM&US+5H4Zm#7+03I*hWD{<X1zR4Qi`Xy8X~=Y~Rt!$W${^u%
z{w=GN@RM#zCR*@E7s4p?)t0(%*3T@sSlyTxIddS;?JcNaBML6aY0rhCX)+E59O=Aa
zMK@)|EkhQr=R>9uGnvsMbux5K2<(pCi;l%8a*P@7yunoA`y82@tV}G+IdL{d!;#_u
z-zer-Ik)1a6?!LtLhm^Ec}s1a_At~HdZz|>Ag;_)(0M$|C9^)c+$ZDSoNICKQYR^R
zS8xUI;>?V*gx}9y^urS+1&z|IAn)Z{J_@M~)YvzVWnT=>>iR~=2R_<9WcJ0zrz)K_
zknf;}cs4?@io%Q@&rxBY_io1K`?+Fn0C{zl^!?b>J_wi@LfYj-Vdgv`D^C?JpE+@e
zat4&(EvAVv@VPi*MSXxGmozb=N|KT$N|ITcC`n41C`o2%VqS92%5pEVG~x15oG&pa
z0hz}fyO3i)9vWlHV=x^=yrh~NPGah#GgHN>*>HfvW(d(EqjoQZO9&HqJh>w5HbXN^
zJP1l8ID9C{=VoTsI*Z3yFkHa3G%do?nJqj!g+e4V-b<3gdr30my(B5Tmn1XZ=OyQP
zFCh%?Do30{2TN_gi)W^7?aDy~(sIrm%;Q6|Bu%rKSgHz?s|i}j0n+~OXU{~^wC^np
z!vP<rr9ED&4JGUUV61wz;npBg+rSId3|8454zZDug`rgEgQV$qI-}I2Aj!vtVX`Yy
z{B{v~a!3|L3;c8^W#YgR#5izQEy6|#V%XRXGVjt7)VNeTLlq2pJt#aBlGpJ4Q1J=i
z6LZPy53n(plpcMEXCI+hMUnLg<_|}k!DcYQHayYx<}O}pY$PfOyRk|DFF49`$9(WA
zPWm0H@=@Ll`#g;qC&_;1FzU^il38p#P(HsXK==tXUlr^L@oItdyr8vxzBqfbM$XuX
ziySt`p~uTGa~>lL6zOc1k{0k*rI+B;=RNcWD(A60Rr1Rg9L;1q<RaXitw0|DdIJS=
zM}GdUy131ZG+Tgx4>U4$d9rag0@i`up%8lZhA4^NOAyNeS@iC#Mq;-F$YOVAfW%1&
zkj2T4fH~e&326A1*GbpDEiiBn7L+vQ8L6TdIq|gl5UB0PU7qUsh7ZqLHt(#1Xyad&
z`i$}4KXkCOIUX*jDq0@;-7YTz4-S(gN)Ph7gkNGD?8`gPmBE&F+ikppX}7D4OYgL~
ztXAxu{#>{6fBl)<WZo37Od0EdGJFs;xIWI`^ieJ@{B+yewD1c1Vs*Q^ZSS|X@rv8s
zuAZ{ayJxJS&9A=gkIie-%G%_%Gd!fa2k;)|K(cXw54D5N`zx)Ay3CE~=4n3~1<&(n
zSFM`*6pyMMeveu|KMK^Au5t{B*G=&`>TfCpNt@qp-_3qaG3G3LiT_-Ca(iY>Nw5Q3
z@Tv26X46}IbK&94rUMsi+cknsw%AnLuJCo$T+i&Q;C21b-j>(){5!p`%5#39KC{2J
zb{(T<Qe9qNp4`0?69+iLN5M*~sy<;sUFUT%FO9zB>sDR(@D_u(%_o#WH{t!qS6rdK
zMsWoN21J1x89&@0Xt@WpV9^6w?qph~f0J7Z^kl*v<O_U2(UhJPDf@H#8~!hWZT>%j
z{npjW#mY9Dx8M50`po*w-l*We8ZNtB%kKV!C04CVHit!JySh=~x8}~;bqu=e{I#w_
z#q2*3a06zLV$yrL=vv#>xw|exlhv!L8uWfHny*&>QaNAa)+*oI%j&Nz+p^VP@Ke#1
zmB-}Rh%fmZKH~^cl#}v(^KcZVL+o3Qtbh}!cI?T684LZC+8ex1N8VXEjKkn4J;E1L
z3@iH+lue_P<%N4)^M^wvdN`Q>2A!f<Vsxq4>LbvK3~t#svlo`Qob1~{`&z4kl!X(q
zc3U^w?S4X|2-i#P_IZ90WNOqG(1kt;3+f-Z$^${J&<O?VVQh+-g1!Ht^`Ui_>$pv+
zfhxS~Xt%xkDc;g6&pA1WC;k4UabmDc%l*k(%MnTSPsa0Vj?d5AY*(uO&iCriD0b7W
Z-NUByN2CJ?!@W@d1m{^!z)rPk{|~4V?JxiU

literal 0
HcmV?d00001

diff --git a/tools/parser/__pycache__/parser.cpython-37.pyc b/tools/parser/__pycache__/parser.cpython-37.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..732b7528f05f6067bf33be15eb24a2c33884bba2
GIT binary patch
literal 3198
zcmbtWL312O74GionVlW&u69?l9VNyHg;cUdEQu5lV~mT$G880{iIqZTh}l|i&u-S5
zomusag1nx&Aah9-pum-)cB?qRjWfT3;y71Mf;$J7tKfURvSdd<QN_;o>+bj7bieod
zy|3r(YSm--{r#%?@cna){goz9e*%MVp(S@f6jMB8L+*1<V-X5pptnM6D18Z@>~r5%
zHmIW<Rq6}Bq+C^g#C%tIY7Tu_Ra6zdr)p{*{hSh8tW`h4>a4{JJ4kg9Y+~fRm#AT9
zRPdh^yj$=sUyQn4PHOlqpjFXspd~+mjf&-bz;cnY3^pE!L-EV?+{&y8pR!zLaw3$N
z^4w0X`%+m-X0TE@KbORrNChi1dum~3Pn^u2lt>%;PM_r^RnDA=J7qhLPL-GQ%<1#I
z6xGxmjVJI3nx#X&|EH|9!4CP|Kjv=cPCQixzYP9wF)OR;lpE<Uaxe2{(#f2vO}Uz1
zV)-2OH27szKi0dKSF*|(y{eLV4OY$X{{zy>lx?uPf6MDx{eCsA?Kp>Af0EVDyw%7W
zs<FgWb4NhqDrUA~KNHxuk=5|-{F^MVW%Ftwt!Iq`P8_q!;*@7xdFHLiaL!G6)HI{>
zQx5wt%<S*s%(2G&?h|84!&vf{r<NS@(eKm?*sJl3$ItJ)lr?DAp7q>1O~m|3R#Pt7
zxBF!G$u?6jP8PCqHkVa#Ud^nT3G5|wJzZ2ULDHP^1O8JEKKwbhUFg@+b6MqpN0hA>
ze<L%IHLtPVKX0?$-=lr9&9E+epWSA67;@(PlyR0(UVd)V7y91V-b;IzuZ-tEO!kLi
zlzemLM?YS<`=ynF9Z^D3+IHi;C>`6&kY9bwmkTjazI^BQ%`M;B+WOwuSzTLGgOA4&
zJyLr0=%4tRS)yr)f~Q)d_w6tk4m*W4Oq6eRqRwM})t6Bm1=@yq>^{s_&41ibd!umB
z?WBQP9;mC!5A&Avtz@7InFjl5;iU0S5G92j4x*i;b)k^)D2NKFgN`a36?9`26ha4h
zXYZb-D)+5B@4Z_%L1aX7CwPz)rQRS?oiHq{evlS+f-TcR4$?puR@&(o&UX8Qx4yHb
z8{~sEh_LSs!zrAt<GvT`VJ9uzcY+5uwT^W$H|pplXdAl_xG+wF)<WU-24OJjq#yZ}
z{$7wI?QX0`dr9HyAZe>XH!VbO<jZ?Ws{Im<;|d1wCXQ2IBC?5Zf%>x7Mg%w%M`n=t
zGQq&FCP9?Kv!G4E@#jycvnYum^z9%d)wvVdK>dN~Re1LJwYBd3kJHPE4ll#&{vhhF
zrEwf4YsM9Q+^vot6y=$7X)qv&N<IQ%T;iz-S9rXMzAhxMpt)$CRYP0ki_}OD^9n{p
zJudJ#;JV;k6ctei-^17wo`n@Lt_ji+;wzTR^*Z)CK}brM5}HBCeGD=-0Wt!Sc#4`n
z;7aT;eJK^nqIx0h-2Xo6MJ^GdY~)r-=qZ(rDhE=c%1cY^fCEd3>jD?#u?A4?S&g2B
z>XN9g*SBXi<ZX^!q`f(B?ZT3A<3=ZXHC?9MS~l#}SrJ$&80iW$G-10hJH5RKN2=pt
z%h4}_RahZ%xb)pyCX@6sslG(yB1p@o!={4_`+Ai4Wy+jcmPzXLmr0Y(b4J!^^a@A^
zEm;I%h-d}Trr5q9Tts$77>$#tlKFJ7CaQ0sauwqE2zO7)4iG_Q<rZ+<#T_(@;}g1j
zlnvx{b}YTVdkHhdF}E`@K?Ie%#ORhm+{?;45P2!H)Z9d5He{7023kF#j#-E1CnHsx
zSjRJH_i>Y``XiEKj}4sG1+0EXr*T#X_BOFD*-5+pry1nW$Y@r|N@~H}4$YaZK-k4;
z4QOrcUc&iX>Ku(zWX%D-88Qo)dVaG@a6>=p3I4Xw62d>lp&&L3-Zp2{623KxQN+EV
zlYZvt3}S>M9Vkt2n<eG=<unFzL}A!|V}1SW?XBDIu3ih{ZYNBxuL5Ca`cxOXNrW~x
zz-too$Uw0C;MT@1U#78&$FDxj%m2gQJEHB6ERe@MJINMe(vrqKO0YiScejGC2Cmo=
z`&8^oxXbmM5FPtlh}Cs{`RV*LQ6GEP!a<UD<KgJ~uOK%5I&A#PfARP9jsM3Y{SC6_
z-<YJYLGbHSTQn6$!GEZ~YI?6_YYK>8!Cls-B5o&PtY0RfydMvu!X9bd7%jJ`oj@*7
zT0hWVqt_%}8y2N_FKrXl+vYyQ<pzw_^G2aJ((1EG(Kr*25p)w;kqOsvfVp+lxImQy
zKj$2Sunthx!}CQvwc|(dOF#gi>P@_P5<fc46I8v9i2|PPzmBTFw-j)5ITsmF02OW#
zmQIlIn4?mr0(HQ5Yp4mX92#eo>ZMJ6nv4b;9g}XJzJkRy?STBCn6;#Si&$?Hxj|%u
z$n$VkG@p0Bw-<&Js3mm}x`%4Gb;##BeAC|suT}A7h>sTY1)wj0<5y>w+A77rGcsqW
zKM#Ta0+AIWbki7MG|@2&J!6TniO{uhuMK0h7Y5fU5fZ96g4YU6aJcKao9?{3{x45;
B>bC#@

literal 0
HcmV?d00001

diff --git a/tools/parser/parser.py b/tools/parser/parser.py
index dacc8c5..2409089 100644
--- a/tools/parser/parser.py
+++ b/tools/parser/parser.py
@@ -90,6 +90,7 @@ def parse_tei(filepath):
                             parse_links(s) if guess_corpus == "KRES" else None
                         )
                     }
+    fp.close()
     return res_dict
 
 
@@ -108,7 +109,7 @@ def parse_links(s_el):
     return res_links
 
 
-def to_conll09(sentence_entry):
+def to_conll_2009_SRL(sentence_entry):
 
     def fillpred(pos, feat):
         if False:
@@ -154,3 +155,16 @@ def to_conll09(sentence_entry):
     out_str += "\n"
     print(out_str)
     return out_str
+
+
+def to_conll_2009_full(sentence_entry):
+    out_str = ""
+    for token in sentence_entry["tokens"]:
+        t_id = token[1]
+        #            1   3
+        out_str += "{}\t{}\n".format(
+            t_id,  # id
+            token[2],  # form
+        )
+    out_str += "\n"
+    return out_str
\ No newline at end of file