From 6be2d372f2bd26e11aad22c01ae67850a5813228 Mon Sep 17 00:00:00 2001
From: retoor <retoor@molodetz.nl>
Date: Sun, 1 Dec 2024 19:40:01 +0100
Subject: [PATCH] A few optimizations.

---
 isspam            | Bin 33312 -> 33312 bytes
 retoor_c/isspam.c |  41 ++++++++++++++++-------------------------
 2 files changed, 16 insertions(+), 25 deletions(-)

diff --git a/isspam b/isspam
index 37029aad7d45889dd466433f9cc526bd57ede09c..4e85c07d49ad11fa871713ef503b6014d23bf852 100755
GIT binary patch
delta 3180
zcmZ`*eN<G(6@Tw7t4l!NDk2<pg_y-4mB^~71W@@hv-s9>PPKj_#3&#hv9X#qRb0ti
z+Vk?dYCBDvlMfFyCh=6$)4C18G)00a)|`4$wWje4s<aMKi=}>~M&15qc2`zUvVZKI
zH}~G(@7_Bzcc#gwH2IX@+2q&-lO*w55Ev#`D<htWZ~x)tl$6(h;!dkMUPq4Z8uqKE
zz)vj`?Mlx$`eDKe*&gtZs2@c?oDk=2mEn0X8GkJPQZPu2VyBHXwyD0|m?FCMg~^g6
zLso_5Jji6XUT>2&Jl|RDjN!blHZ`SKlHBG3J25A=tF7gd@9c(YY%c)aMxSP83B9Ja
zUMjj3Q)SyQlx1B6mnjui<cw+BII2Kxt&rV%r|dhs@eDCHBLw5$>{?$isDnZ_@#oGh
zv8~@pFtic9BYA0g!Low#g5?DjhlrW4yAZ2C(P@{YRz(EFu7jXAHAtV!dSdSBe}K%t
zNG3)XE5*Fq=oBgbBAg<;MT$boj&DP%&E?V&iKkJ?K7fnIxU2ay*V^e{vqsB1=(Q{@
zk!A_fsp<f!bdk<;rI-Gzdw_2VxZOsF;LApADCLjg$oXCnQg&28M)P&T#bf+a^Jmx7
zHrE)riJo`m4k?ZTZJ}wCQxlG3)n(Eu4T7)(S$4cf7fqh+{XU;(JG=O5_OWIz!^vj=
z!)+W9!4zTP5yW=|h;E%Dr0kdkiNvF}#rpto8;6BcAFr5c9S0*GG5m*eKh0ms6Y75o
zdqW#yq**YD(p9tCZt9#;EbpePr;PT#3N<ku4G@}f+pS+!*xMIGkeZRUi7WL(3Naj&
z7&4P<^bVV5u!q-L;fDUix8euNz77%gzIqQ1X8xs2q4q_o%eeZp`#D125z?=U(!03+
z<@-YYc1S%ZN<F-b=1rZbjOe1ZQ+F$M7pOgZ*IHuaIiqTH_8h!%bdp7t#_-abj^+#=
ziuD67bnEnqe6@7zlVPbsCWnvPS=F~!R5r#*%o?1(YpU-xOsFY7{7W=tnp4R+PfMn)
ztL7YfM|xCEs-6m)VM(^g1vBn%e1{()U|(Pvy}{s>x76m%&ZH|Qs&ht!l(;E&pm@(9
z^xi$lzG}yR)v<R977~AUc^n-*eY|%sto>!OtIbGk!jOG(HW)mZihx)HM#MNIvVCc=
z$nQN^$W`jP#Ka-u`xX9;eCr`oN$2e394itUR%ScM2W%Ec;C-ee<X*?kHUCQIc=#+9
zzA+8fnHjWWy3>2=3r^R^-Z&Es<^?x)6VrmTO!UlAVG>11EP$P*!zL7DD@FtT>?<4-
zK(h&<Fdb!BVtjwfzad*7qrzMi4)aGG!^`7*llz;?(0)){JB|pP-Vn})1w%s&8?1GZ
zBEgS8hiCZQWdkMB1gOBVu7I7!kqVB)I5`#xVTtu^lW~~SF+mJ#&IBn*Dp8xKm-udN
zSk6p9p*ax0qrgV*%}7#eKBpsdioE;Nc^-!KkC5#2-@(nnS>6pEMRrUqnmq4}#4JN^
zJ>+U*KSz99^DAL4sg0|6XF_*8e>LO?o5RF7!iHe~pzx^l3W2X%OSxnlIRWh7yeZ)K
z+q_Zu=rF*aBJeFC{QDMh=Gm4<@PEr9_}q}M!<8pF{>TRj#P<t^TH8Yehe2=xHe2(5
z4SOWuT_70xnBx%!&jW^C92pER3x@YU1Fk61C$|K{sD~Lk!SK9bm>FW&hZ1Q<kbQI|
z<QOT@)yw!cyu3yQVTK8UVLoR-jrQFU_{Ro6ltBmZhYS2QX(EGrXd0SvnUz0+?-%$S
z{^Iyh0>`*yH&638_;qN`AdTMOhD4aOg-!Mdlas=P-==!zJ7LW>p*bKlMdGG!`~&#W
zkl77f*BVaQRq+n!Vm;zJ$j%9QnVsVEdgih)Qn8STAblnI)~cAa&`KLv2T@masTl)Z
zs+WK<(x-^N$tJ139q8C4t{&pgTr!CFI`|dYHRB@gez8|e&G>Ftn_KUY*%$EEa-X9)
zd1UB98277-tqE8#wvBGj8zUFdeR)pr_^^U^6gCSUZqs9H;qM!q054uDp%c*fXQCD#
z1((frg53TOG1{>dc9{2i(FGBPTSv(d-)+PP6MtqQ?_8KiuJ%Z9#F&p3YsP;H=aYZq
zOX1}cdb{k-o$s`*KiPW#?p2|Ok?!Sq?g#nyBNzer%s1rw5)eJP)wcELdK2KKrffwP
z2n^3I)z@@4Qn%kka7uu)=IaXgv;bPgfokFQ0BDOy@T&;WiU`o0ZwRRW6c4D1<9rM+
z`m>pj(<L)`<$o~qMen(ByvB+3sCDH1<%UDA`%7M!{6B@%%cIm8T<!XwQ11+>XGN*w
zxcYd=+Z$4+Myao!g!*ZCGX*13Wacjo8Qn#dA`u+rMvnmlmF73wcd!dw$dhg7KDfM$
zPPXzkA9gM6eVy)~HIJ6f``-T9sv<9>m4(T4bm7>*oWhZo$wzS{Kf<PJL2t2P*X_Z>
zbUtMI^<Z!Z<d={)AZyU8W~0-xJArK_;|Iw%sjMZ?meS#h(oCQAB+`6su(cAqG=|=r
zJ6GA$Lrb+%S);FLxz-&x=IyX|(0;AZdKX`H^qR+MskFq>O3%o+xp4%!^&pB=+Th7h
zT4Fe2oF|Sy-LpVGKkygN8dWBNobrmtl<q0?wdz)x(8V?HDoxX<vU<5(NuOPvB&X6B
zSHCMu^zqv5a#rB?wMiD`dKNvl_Ls^{DX^k$>`>WC8#YghC9?pG9|;`TTy6b7*`vs}

delta 3279
zcmZ`+Yfuzd7VaJx9R+QSA`=zyF}OrU8I_0`CSj0?Jusw9$wpJIiW|X4L{kR0)ItaC
z;YX)sL%hkBm93;KV~my5R*Ym_)^!mPbV()AHNK)Q8k4LSUnIKGC`#{lZqLY=+U3Wb
z+xI-a@0@$?>1JElwk>Si?U4JlPL#x7YyALuv2D<hb@A<Q94vCp==`?rZ*!Yo|Kk0&
zt@WGZ#wXcIMzehbkIU8dwS(3V3HeHXJxkTQRP9<uql^FtocQ61-<qbX7r2ttnkFSY
z2ntodplbG4CrOeF`BFp{K|1+{`>@v5IAB!Zaw_4&2yvvql}LQtO-fpxq#8k&Q#GDX
zQjOeZ-uNJ5AJ9WBl~pixSf=F@l(aA{rGQZaaAR0=fwI@*51W<K_)>5R=<PuxLrIeY
zN@J(*c0q;Ys2spMzcEdp(s)`mq-le8tsd%E_HK{^+F==vt52#%6+-Fld{?JwYQT~W
z`~*DsS+#X;P`}KRpb!k9O(boi>hU?0O{z)zRW2vLaf4P=53MC*?mXXbeDi$seG3k$
zhIgPF#i>pHRFR}c8<@MBc=aXI?AYDSZcQ9rR1R)x?Hn2P<7crB=to3N--Ia&*$duO
z(A!xg<<tw1%HDYrN*F#@S;}W1RIz?PSeyAkVGFZ{$|KnNtYGqJm;tn<e-CKgH>^D?
zqwseZd+@Gc9nf0@UN*P^XRyElZUnCrT@#Bp8x)K6?}N43{n$+RFgc4=xIO(&T<8g^
zvOl{=44e-?S#6YhK_nrvoa$xVofnFueLjKg^-lh5jNd-21Nvb>W;!*LX8-pD3ZmIh
z2`Q%@fJEXkedrfXt-`63DrVUCz=#;b?_Kf>)|OI?O;|gF{q%H(1xmM=-@Cv*om3!C
zVRt7D4P{_gHBtva1ob-s?V62ew+oU%Jw1Vx+98{&r~Z(M?jK4Tt;G@4`J-pO1zMDS
zGCDhY)+TH_wo*eI-d8R4pcwT<QlIXgCiEqi{-=Z<_k*N=ty`$GE%m7w_4ao5`Q-7o
z9qsJi<Q=x|f3a;-wpVcd4(#=KS4^SG&(juepD0xd)z?V%jU~GFd=i?yd7M@4JrVgf
zq?3FcYRx$%yi=4n-lZB#uyI`pzl9AYjeVT$vbCLOC$m@Db6vcQv%Dt<2kYpD`Et5f
z?-cigZr46;npfXb_Ex!4t{N!DG69~P(5ENa)8~-vpO9>M-7RJKJ>P7#Hv4*C_R*7L
zLLOMx&XnCvdU6()d@f86i~s{Z04r5LB=XwSTV&#eg<P(zOnx7gMrP^s4A~EIgnA^;
zIn56017<o7(1IU5iwHEwS}gbOCq&yUa*csc2R*3q8|_~{!AhpOLW`~tTqhq7r(ARO
zMb(HCqc8JAVW`=+2uWq*4PQq|K1UGPv*u+NZUsWGAxqFqy@(UVb7>@k3K>IlPK0J(
ziV-57?~p$xCI(=Y%YhpK=kXRSrTMs6ZnSr2NRmkKdGLtNm{zQ#cb<tj_M5N^QY7Eu
zco)SwgAEKM>K3hurcMJkRNdZ|A*Dzdo>Tqd+m-YAP!Pf_P-|2F*_+koq}YBw!*=FO
z59M3#_994Leg@bSoNxY#SY*HHz)OS%p=!)@;KE}bjsmA@@6BIPL)DZ)hbzb{u33&4
z&4*O|Fn<UAV^x|ADbKScx&0wP*N*}q6nqq+6$l4mRPC*^5bnnjLMKoB1w!sagf}7x
zYX!oN(-sp4?|Vxi-0vNef~pfhsQ{Qj03oGj77(aOKDugoMsqlK8PD{h%*=`)I0eG=
zhX`%A1;Vaa1iTV}V8W)_`}r7Aqz^@rOkVg4fOP-}SqMjANmcp_Nq*^+#l*pP-4h5e
zJc`f)gm(qP9hX43ixENZz7FbloL@AB#~-G!A_DM&0N5x1sKY}>j#^_@CL_yW#AJ#v
zSrFlXmt^Loh-Qe;_=U!gxzeQ6>;^tHKsDaybWCf+w?-#_MLe5)iIA6hDeY^SJl)Fk
zn2>k}mt`zc$Hk|k`Or-@bQ08lnAbTy!@A?dNz<eckASwpAt~WH3~2Xtbn;c2476`N
zi;S;ea7B7Gk-M8ztySjhz%}4m!P-1(|I4txsk~@^55?reSz7Ke+0SxwU7@`ZMF}a6
z0xMt?I~q_@%-d`wKH5GtZK_i657qEH+{e{BKdO2&j$s?jJ2Xr|`i#*)9sIh8-*NCe
z4qw_A-x^VJdI^0zB_4zGOK?_=5N&A9a==r9kC?AI{towYC9MG8O<IWJ6=B2g{tf{*
zMFY-A3hI}l`4$?97u`((twrXKA!4y-v18Twj%0W#Y1POAQ(`^E*^VzeE6!6^9OQyH
zFaNJVpPj&Yddeu!a=1kT1*|}KyCMg03kPN$fdBs-z$U~Acq%B)A$YNoPmg3*yj1z#
zPp=IvA4N4z?uS}OU%_90g@OitrowoOE%gng5LNFZwfnA6Pq5TSW7K~n^_P~nVyUS^
zquy1dehS`v%W;I5Zg{Vb6bv@U7|kK0zKDUNqc__ubiqL+%G00_^X&*<bcdD~=5~x@
z`OiGJCr`1lVfka@XRlnfsNS2O7B?ve<7z+pDdW0nIxym#KbU3-<W|T!$R!wm|A9Pn
z(=_w?NYb3H`rrG<^peZi_2N<X8U0J+#<0@j!3n+u71T<5Nm3<yP*h}de$NVnh4Oy(
zR?uUYFaw*J986_5gZcL5cv=B>S6pe_W5PLKIDZEkb!=ntSlj-1wy-!kA&@}-7Gwdv
zrT&ZJmle68e%ibReQh(Ru(uaC$_?zrB_G(5rZU^o`SLNgaA}HM$JQ?WKz@adT=u%W
tt^Si`DRH(^H~XyO54JM7{<kZJ_m@|*ja3u-G(3aA_-m;DYt>@=e*wKm1QY-O

diff --git a/retoor_c/isspam.c b/retoor_c/isspam.c
index f91d387..2907e89 100644
--- a/retoor_c/isspam.c
+++ b/retoor_c/isspam.c
@@ -174,22 +174,14 @@ bool is_fully_capitalized_word(char *word) {
     return true;
 }
 
-sl *get_capitalized_words(char *content) {
+sl *get_capitalized_words(sl *all_words) {
     sl *capitalized_words = sln();
-    sl *sentences = get_sentences(content);
-    for (uint j = 0; j < sentences->count; j++) {
-        char *sentence = sentences->strings[j];
-        sl *all_words = get_words(sentence);
-
-        // Always skip the first word since sentences start with
         for (uint i = 0; i < all_words->count; i++) {
             if (is_fully_capitalized_word(all_words->strings[i])) {
                 rstring_list_add(capitalized_words, all_words->strings[i]);
             }
         }
-        slf(all_words);
-    }
-    slf(sentences);
+    
     return capitalized_words;
 }
 
@@ -251,16 +243,13 @@ bool containswordi(sl *words, char *word) {
     return false;
 }
 
-sl *get_forbidden_words(char *content) {
-    sl *words = get_words(content);
+sl *get_forbidden_words(sl *words) {
     sl *found = sln();
     for (int j = 0; forbidden_words[j] != NULL; j++) {
         if (containswordi(words, forbidden_words[j])) {
             rstring_list_add(found, forbidden_words[j]);
         }
     }
-
-    slf(words);
     return found;
 }
 unsigned int total = 0;
@@ -278,8 +267,18 @@ void analyze(FILE *f) {
 
     free(clean_data);
 
+      sl *words = get_words(data);
+
+    // All words
+    printf("Words: %llu\n", words->count);
+    if(show_words)
+    sld(words);
+    sbuf = slds(words);
+    stra(all, sbuf);
+    free(sbuf);
+
     // All capitalized words
-    sl *capitalized_words = get_capitalized_words(data);
+    sl *capitalized_words = get_capitalized_words(words);
     ulonglong capitalized_words_count = capitalized_words->count;
     printf("Capitalized words: %llu\n", capitalized_words_count);
     if(show_capitalized)
@@ -298,15 +297,7 @@ void analyze(FILE *f) {
     stra(all, sbuf);
     free(sbuf);
 
-    sl *words = get_words(data);
-
-    // All words
-    printf("Words: %llu\n", words->count);
-    if(show_words)
-    sld(words);
-    sbuf = slds(words);
-    stra(all, sbuf);
-    free(sbuf);
+  
 
     // Numbers
     sl *numbers = get_numbers(data);
@@ -318,7 +309,7 @@ void analyze(FILE *f) {
     free(sbuf);
 
     // Forbidden words
-    sl *fw = get_forbidden_words(data);
+    sl *fw = get_forbidden_words(words);
     printf("Forbidden words: %llu\n", fw->count);
     if(show_forbidden_words)
     sld(fw);