From d1dea2e22ec3c501f534aeba34bf2e11efc40631 Mon Sep 17 00:00:00 2001
From: voje <voje@iskratel.si>
Date: Mon, 11 Mar 2019 09:26:49 +0100
Subject: [PATCH] fixed the weird bug (defined a list instead of dict... should
 have gone to sleep yesterday)

---
 .gitignore                                     |   2 +-
 data/kres_srl_ikt                              |   1 +
 src/pkg/corpusparser/corpusparser/Parser.py    |  17 ++++++++---------
 .../__pycache__/Parser.cpython-35.pyc          | Bin 5389 -> 0 bytes
 .../__pycache__/Sentence.cpython-35.pyc        | Bin 448 -> 0 bytes
 .../__pycache__/__init__.cpython-35.pyc        | Bin 249 -> 0 bytes
 6 files changed, 10 insertions(+), 10 deletions(-)
 create mode 120000 data/kres_srl_ikt
 delete mode 100644 src/pkg/corpusparser/corpusparser/__pycache__/Parser.cpython-35.pyc
 delete mode 100644 src/pkg/corpusparser/corpusparser/__pycache__/Sentence.cpython-35.pyc
 delete mode 100644 src/pkg/corpusparser/corpusparser/__pycache__/__init__.cpython-35.pyc

diff --git a/.gitignore b/.gitignore
index c57e05a..43fb148 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,3 @@
 data/samples/
-*/__pycache__/
 *egg-info/
+*.pyc
diff --git a/data/kres_srl_ikt b/data/kres_srl_ikt
new file mode 120000
index 0000000..465d987
--- /dev/null
+++ b/data/kres_srl_ikt
@@ -0,0 +1 @@
+/home/voje/work_data/final_json
\ No newline at end of file
diff --git a/src/pkg/corpusparser/corpusparser/Parser.py b/src/pkg/corpusparser/corpusparser/Parser.py
index fb8b408..1c5dd6b 100644
--- a/src/pkg/corpusparser/corpusparser/Parser.py
+++ b/src/pkg/corpusparser/corpusparser/Parser.py
@@ -75,9 +75,9 @@ class Parser():
         if len(sent_srl_links) == 0:
             print("HI")
             return []
-        print(sent_srl_dict)
+        print(sent_srl_links)
         # find the correspointing json file with srl links
-        return []
+        return sent_srl_links
 
     def parse(self):
         if self.corpus == "kres":
@@ -112,7 +112,7 @@ class Parser():
         else:
             divs = root.findall(".//div")
 
-        res_dict = []  # TODO: try making an iterator instead
+        res_dict = {}
 
         # parse divs
         for div in divs:
@@ -150,7 +150,7 @@ class Parser():
                             sentence_text += el.text
                             sentence_tokens += [{
                                 "word": False,
-                                "tid": el_id,
+                                "tid": int(el_id),
                                 "text": el.text,
                             }]
                         elif el.tag in self.S_TAGS:
@@ -166,16 +166,15 @@ class Parser():
                         raise KeyError("duplicated id: {}".format(sentence_id))
                     jos_links = self.parse_jos_links(s)
                     srl_links = srl_dict.get(sentence_id) if self.corpus == "kres" else None
-                    srl_links_fixed = self.parse_srl_links(s, srl_links)
-                    print(srl_links)
+                    srl_links_parsed = self.parse_srl_links(s, srl_links)
                     res_dict[sentence_id] = {
                         "sid": sentence_id,
                         "text": sentence_text,
                         "tokens": sentence_tokens,
-                        "jos_links": "BBBB",
-                        "srl_links": "AAAAA",
+                        "jos_links": jos_links,
+                        "srl_links": srl_links_parsed
                     }
-
+                    print("------------------------------------------------- END")
                     print(res_dict[sentence_id])
                     print("------------------------------------------------- END")
                     return  # TODO dev break
diff --git a/src/pkg/corpusparser/corpusparser/__pycache__/Parser.cpython-35.pyc b/src/pkg/corpusparser/corpusparser/__pycache__/Parser.cpython-35.pyc
deleted file mode 100644
index e438b679e512adfece77441901df5aaacf36d388..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 5389
zcmb7IOOqSb6+XAS)q0F%&D%46VA&WCGO>mt5RBuvFg6epG9ie{Pz_E;R<{OeS}o1(
z7RX|>!6C4tvPfl-tg=fLe;~Uovdk~&ZC3mRZ<6nvmNX;9EYdT5bsy*4d*0{V-Yk_0
zU;g#WkE8QM|E5!qiT>+&BNvT_e?<XNSCOMg@z|iipsq=dnT|~gEOIP?ML$DMM*0>7
zHg&V)WJ$g4Va}%C0o|s@201zMOsZG2-{BCdsw7wU!`Kg-KKj;sjrfy2bTWRdeZR>d
z)5k#Pb-d9r8lRXb(Bv3QCMd%JAPZZ2mh?f@^kAkcOI}XqZSwNu6{L^JCGv{W2U#WZ
z&PX2sW%8D#Um$OVyjAG~cnzelC-!>{9r?P+661$|`MrlX`V*RC(4^?lApHu(7=lWJ
zwjV13@vfRqnX@T_!kQpig0w${I6*QgEK)37VVf_oQ@WWjd64$OTb3q$VHm6)7;+}q
zBamL$ORPifM_Pf%@3?WyDC&%^-_-|$E)-E|_H=&`RiaONgTSji^ea19->LL;Wrq=W
zYKi(RQO!iv9i!i@<~8%3Eb*G#>II&!ld=pV9ZV;3*urhK13xMJx)BWgH?{8R#7>V&
z?0xrxyKmPM``+ZYF27YXu_8Zc=`tUXZHK%;^f%HQ7(aP=x37C0zZvgFy1CmwJlIWl
z>x*LM2dT1Z{YOd8b=zS(cHJfH5v`*kqo~%@rYfpUqiE#uTU4X+1Y-)_WZUD1e`k7r
z5d)~RBaXtVz92)<3{+I-H3(m<6NX<G1(<u<2A*R_kHWYytePo%>q58N=|yhP4i6)j
zMVPFmvyJf4WEOI)8hQ;IB-sesa{WNWkv5w=UtZYh93~?^l2loh%+cnAw*{~niy1%`
z@QNvaiw+n_Sp-jGgJSS!5c)9aLn3{HVi>o{Gcd{nqOnOsEf#Ol_WcJf8e4P>x6Dv%
z(y>LIsSS6;ba|(O_yKHpvYeIubEgpUJadAuBnWMjjr~;|cyx#MaYnsb+#BU;ySpMo
zodxxym3xhFC$6;Gp;wXVn%I6~wREqWs9It*T7xh#;+|uAe*e4w;L#OeR_!FqSdAb^
z%)k$m>^r|q4cBakabh0$agvGpK|8KyL?xO<on(Ro-H(zy`-7a9Sp2h~CcOpN%Xm{G
zglbk{c(z(mrn-XBi^k~u(-kF_G1C?s<IEs$U@!}Y6SawA?n8J`od?NFpo9W>s$R|P
zH8}?G!=`QFt>cZJMFUmBf*)+tSS*@tfF^q1{)0hBHt<b-5s)J2RAx$F!$j3cjM|Hd
z8ZL1Cq;NZEM{%>)?cez!pTmddX|y}XDhG?Wp&?%vZS8i@Yc_)D4mwlZC#o_>H=DQ6
zeH)z#-7hiS9C~42MB9^SygTtn80?pbts(4En3^(VrN*IR;R%^YU&WU2x$F4RFL1-*
z=fvOi4={Wg?}Qg%hy={kORd-+#*P_8@d92l@%RO1c_RxAWiF@%3z&zTrJfvs^LW_`
zqQ9V-^H-C;B5H}Vg5}?1)fCmk6t>})kkvRt!zMf4{Rbb>*q(+kn`0Mz7G{F~F<eu;
z^%!XdsO+jKL{1zNViXT$Pew{*DD<ZOnW3ZC;Gdb)KMNE8d|UVDl*nwaF%vB1i4rd_
z8sFC>`4MC)n(7Q?L;R?Z@<Bfb%+tC;AyE(?bQxelX*c<-i2>H;3~(;qKF$78Be7E>
zMc9WyJcA}N8)4&#x|RCFB__tw?BtqLg;19oNA4tE=$CkD2aS^vQuXsZQRN8}2MEz%
zO~5$7>uoeY!aIu`7ZEo~Y73F0pf0E_V|4EOxtW>GT;O0|gs-Pzo-^5{sTrif9?6TB
zzc-ng_4nUh$Yh-TQ_*uK6QTjZ`c1>+8A)F=s73FXWU<-$J!Y8QoU_=nv5>`PU~}xT
zZUv7ujYxnH08>_T2@Jemmo<r#3T9UaBr+O9@~CR<ImS=#ipJ4-2USU?uaN-s73}Ji
zW^8!vW_+4M{VLWM*wU=Gcko6<Gz5LE;<Z#^G1|`<N~8Vh7|n_kuJ+KGQ`#*W-kj(S
zs(MWD=ymZQs1BJo<^H{+OSB&@3mhh~WC^bbgY~MB7AMGOTytqIlALTi_O;j6$(cTv
zR_?HivJRD)E@5?AjO1ps^U)EqVx;#08Wqs6_BLUxIb&2jZKSh&#>UOABpf~guWdvq
zsbEixb08rO<$x5`P!uBDHi`gd#^e$iF%Gj_EeLPtxd1`kf$CV|**6zB=*3y;<T$Ez
z5V<4=;GBJk;aBf}$RQNaxUVSsW{sg|aEOpoAOt{M4y0B=e2kZ<Q{+nqB4HZ;5C|We
z>SPJ4C0=@Lay*So)H%b$4z3!q2@V_QBqU=8@G`3_SJg4*v6`o_Oz3x(`KH3~sP7@5
z0uBFV9Ot3%e;Q+iUyy=3OIaX_H2l`!8Q?7Q%>4&{HpU<X=TCu7BzX;zJVV1z4PFMu
zN{ZndV;o$58paMV0GOxoiqNu3!%#8kYHEVNRpT`}UZW0LFttvHHt9baJoCTDKab7U
z>3CgKic1&AqIHAExO$PKY+hfH4L4Bwp|{GHG7lgd6mK#vc=XW00UP9<l@3liEAw0P
zlI>0D*y7c2dkVbGLzc=p!KxycWzt%6m5%<QPS`ms>?rdzJN>^K-g!#-IFao2l(naj
zet}769;ZZ4^X9@?TXGgo2R9|*9i-rY-@8a1kY1!?7!-62`@)9!F9-(m$T{+!5tt3;
z2mgG3gf`BzQZXFk=+hSWm5CA&`oShxO{=6aL<tyNa$pLOra|K^iZ4o^zYid1K;GC0
z<W)p1m&v<A-m~OgC5R8y0N^%bNRMyB3M211K7+q6)c23-YX7LF_5o3^zK}+vw8Z2P
zILg-$7;BxV7bZ$OhJHB84C2;}o1^l_5h73+{p89=zq#_@g)0f_@Mdp-D4k`zOa-gA
za8*IyM)~2jKTV9bH?q+0OB?AL79H!ep7uUS%((68U7pAUez)6j%x>gquJUl5XRxpr
z`frVLpO0(uKU(nyxCk^GvF}yd-mS{#V_d7-o@2+oLqEiAV0N){tT*r^`B{W_GI#kI
zy>jFKeJXG6-PaW+>M2e2ne<|5g^jK+Rfp!BBe!*D>9cnHi5m@Ct@bdndVO4OERkv^
z=rz13v9#at65I2eJ<m@J?Sq@aLv3LL$NcaY?<O|B*I+3vZn)}7EKU;;T;o75^EoU%
zsh#x-_f{FkZuNAx5huBy`;X)%rSS!dG^c0%8k*{MS|I!weJ95f;-u=}78dv}q?u<L
z)9(mItM6D3QB^xxSintmKy<GcI~H^sIXS+439^Z?W3_;uD4_JE6FFAI8%p-(PfxCt
z>19sl=}^?>7=GYn{D481W@|8wH<)5hQBHZLN6<C&y~Tuoi-GV)U!WlrYGq>+McS&m
zV5}NzMp>O#8^)G_c~vyZhHaKq*#rc?fVRywhE>b>E1|%%aY@Vr_OiML{0)4$NM~()
zy;wC$rlmH_qS^q&RW2>CZ~C`vo<X5eQlpL2S$CGp_o_wyG{UChx=F!xyFG6JnY&5R
zb&m#(AYIXC!9%JuzP_eaEobahg<Ra|+uShMrz9aYfz+=2ae*%@5!*x2D3*)aVyT$J
zYZvpyoaXgvJ}I95mZQ0jbFzJW84TJF#entCFs0w(c`JZxi)9M((@srtw^M%Za1fC?
MgyqR!dX}yK0+*{<x&QzG

diff --git a/src/pkg/corpusparser/corpusparser/__pycache__/Sentence.cpython-35.pyc b/src/pkg/corpusparser/corpusparser/__pycache__/Sentence.cpython-35.pyc
deleted file mode 100644
index c0ae9f75396887f2b01bd1afd2afe7d6e563d333..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 448
zcmZWku}Z{15S`uJK{)ON8%40V(ln;C5YYn#t#U=If-r<+k?38r?rsXNRX@gG+SZQ{
zteoAD=y3z{c4lYZo0)^zZ1R2e@o)zKU*v-Y6fMa0A;E14P%?P;_{JflfI}Ju1r-P5
z6fMa0F@XTc7!WVxj!Nys&H^X{;*D&WirR>}5IYLjPGOJ0bki6xQb{A8Y>s-;9I>h^
zgLo31&+nHv%ZS;rR8?(0LAlL6{G7#7HBX{2u~tPapC99*QL@!CS6Zm;q?bOGuQu%E
z-9_2~u+9F^(N`092V$MC1)`lGuA8!5xp<2B(&nqKqQ?C~f2*mH`(sX++>e+tpC?iG
T{_FJzIf4x@x)GP2ML+NhUEyBP

diff --git a/src/pkg/corpusparser/corpusparser/__pycache__/__init__.cpython-35.pyc b/src/pkg/corpusparser/corpusparser/__pycache__/__init__.cpython-35.pyc
deleted file mode 100644
index a365c8e295dbed9a0472980368c09666df7d3ef2..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 249
zcmWgR<>jikZ;c6LU|@I*#Bjg_WH|tFu?mn#0U}0*90rD5MusRx21XE{i6MuHA(xpU
ziW$gfW(d|~c?nd>pvicPEg-R|IJF4K;s{R7D@n~uPW97di4soEFDfW4E&xmFLDUq1
ztcemuk%8$cVg|~sWGG?*QefhjnSMc0epYI7iGFcWvVK8!x<0Z-R8D++W?p7Ve7s&k
X<t+}I-29Z%oK!nTpi#vjD+L$<aB4ti