Latest version

3 files changed, 293 insertions, 207 deletions

diff --git a/.replit b/.replit
index 330a26e..6ba928f 100644
--- a/.replit
+++ b/.replit
@@ -1,2 +1,2 @@
 language = "cpp"

-run = "g++ src/*.cpp -o compress -std=c++1y; echo Compilation done; ./compress soc1.txt 89"
+run = "g++ src/*.cpp -o compress -std=c++1y; echo Compilation done; ./compress soc1.txt 30"
diff --git a/adjlist b/adjlist
index c3254d8..5c33973 100644
--- a/adjlist
+++ b/adjlist
@@ -1,200 +1,227 @@
-good	231605
-other	215885
-more	204207
-new	184735
-many	151831
-first	122599
-great	112203
-such	96282
-own	80354
-few	79347
-same	79274
-high	78209
-last	77135
-most	69223
-different	68446
-small	61453
-large	59995
-important	58807
-next	57079
-big	56641
-little	53993
-old	53835
-social	53217
-able	52787
-available	52513
-online	52116
-free	48563
-long	47269
-easy	47229
-local	45900
-much	44002
-several	41300
-full	41250
-real	41198
-sure	39843
-public	39216
-possible	38240
-least	37647
-bad	37080
-personal	35343
-low	35335
-late	33974
-young	33739
-hard	33348
-current	31799
-only	31724
-right	31587
-second	30897
-early	30367
-special	30135
-simple	30023
-major	27928
-human	27633
-short	26420
-strong	26274
-true	25912
-open	25816
-whole	25675
-less	25493
-financial	25155
-common	25046
-due	24836
-top	24722
-past	24478
-various	23942
-certain	23903
-recent	23392
-single	23271
-political	22713
-clear	22560
-specific	22495
-main	22323
-particular	22136
-happy	21700
-similar	21546
-natural	21203
-interesting	21086
-national	21039
-American	20879
-private	20479
-international	20327
-difficult	19955
-effective	19781
-unique	19645
-professional	19607
-perfect	19566
-economic	19529
-additional	19480
-key	19379
-mobile	19355
-original	18890
-nice	18845
-medical	18745
-third	18696
-entire	18664
-likely	18598
-necessary	18475
-global	18339
-general	18271
-popular	18118
-successful	18021
-beautiful	17655
-wrong	17654
-significant	17540
-legal	17391
-enough	17266
-final	17107
-healthy	16889
-white	16680
-ready	16642
-huge	16599
-interested	16547
-wide	16201
-former	15669
-safe	15532
-close	15450
-traditional	15384
-amazing	15252
-future	15195
-individual	15147
-physical	15039
-basic	14808
-complete	14715
-positive	14316
-black	14289
-federal	14283
-digital	14120
-deep	14066
-potential	14046
-useful	13966
-regular	13955
-hot	13834
-further	13751
-previous	13639
-serious	13514
-multiple	13493
-extra	13462
-excellent	13356
-poor	13322
-responsible	13308
-wonderful	13284
-quick	13281
-modern	13223
-daily	13007
-active	12973
-critical	12920
-favorite	12918
-annual	12761
-powerful	12745
-total	12661
-creative	12500
-appropriate	12418
-green	12219
-worth	12174
-normal	11877
-actual	11801
-fresh	11796
-fine	11705
-direct	11408
-present	11321
-cheap	11305
-military	11086
-rich	10952
-primary	10817
-relevant	10754
-essential	10702
-environmental	10660
-aware	10548
-fast	10518
-cool	10513
-corporate	10490
-red	10373
-technical	10346
-overall	10329
-light	10255
-live	10218
-independent	10022
-commercial	9957
-complex	9863
-average	9636
-cultural	9401
-dark	9233
-sexual	9229
-foreign	9166
-standard	8997
-educational	8949
-awesome	8937
-expensive	8934
-numerous	8871
-clean	8840
-proper	8724
-cold	8723
-academic	8657
-heavy	8588
-mental	8586
-initial	8566
-central	8561
-video	8554
-negative	8494
-exciting	8444
+Shy
+Bad
+Odd
+Real
+Hurt
+Rich
+Drab
+Dull
+Easy
+Sore
+Kind
+Lazy
+Blue
+Evil
+Long
+Tame
+Fair
+Busy
+Calm
+Fine
+Nice
+Ugly
+Open
+Vast
+Good
+Wild
+Poor
+Cute
+Dark
+Dead
+Zany
+Alert
+Alive
+Angry
+Scary
+Dizzy
+Shiny
+Itchy
+Silly
+Eager
+Awful
+Jolly
+Black
+Light
+Super
+Bored
+Lucky
+Brave
+Tense
+Misty
+Tasty
+Fancy
+Muddy
+Mushy
+Tired
+Tough
+Nasty
+Clean
+Frail
+Clear
+Nutty
+Funny
+Upset
+Crazy
+Weary
+Plain
+Cruel
+Witty
+Happy
+Wrong
+Proud
+Homely
+Quaint
+Hungry
+Amused
+Sleepy
+Smoggy
+Elated
+Joyous
+Better
+Stormy
+Stupid
+Bloody
+Lively
+Lonely
+Lovely
+Brainy
+Tender
+Bright
+Famous
+Modern
+Fierce
+Filthy
+Clever
+Cloudy
+Clumsy
+Gentle
+Gifted
+Creepy
+Wicked
+Grumpy
+Poised
+Putrid
+Defiant
+Annoyed
+Selfish
+Anxious
+Ashamed
+Jealous
+Smiling
+Average
+Jittery
+Elegant
+Strange
+Envious
+Excited
+Careful
+Foolish
+Fragile
+Naughty
+Nervous
+Ugliest
+Frantic
+Unusual
+Uptight
+Panicky
+Perfect
+Crowded
+Curious
+Worried
+Healthy
+Prickly
+Helpful
+Puzzled
+Zealous
+Homeless
+Adorable
+Horrible
+Relieved
+Distinct
+Annoying
+Doubtful
+Innocent
+Arrogant
+Splendid
+Spotless
+Blushing
+Talented
+Faithful
+Terrible
+Thankful
+Cautious
+Charming
+Troubled
+Cheerful
+Friendly
+Obedient
+Colorful
+Gleaming
+Confused
+Glorious
+Gorgeous
+Graceful
+Grieving
+Pleasant
+Handsome
+Powerful
+Precious
+Helpless
+Defeated
+Depressed
+Agreeable
+Different
+Difficult
+Repulsive
+Disgusted
+Important
+Disturbed
+Sparkling
+Beautiful
+Energetic
+Blue-eyed
+Expensive
+Exuberant
+Breakable
+Fantastic
+Unsightly
+Obnoxious
+Combative
+Concerned
+Glamorous
+Condemned
+Vivacious
+Wandering
+Grotesque
+Wide-eyed
+Worrisome
+Dangerous
+Hilarious
+Delightful
+Aggressive
+Determined
+Impossible
+Attractive
+Enchanting
+Bewildered
+Successful
+Motionless
+Thoughtful
+Mysterious
+Frightened
+Outrageous
+Victorious
+Courageous
+Adventurous
+Inexpensive
+Inquisitive
+Embarrassed
+Encouraging
+Magnificent
+Thoughtless
+Comfortable
+Outstanding
+Cooperative
+Enthusiastic
+Uninterested
+Old-fashioned
diff --git a/src/main.cpp b/src/main.cpp
index 3d5562a..7b1546e 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -14,13 +14,50 @@ struct weightstruct
 		return weightstruct(word+w.word,clause+w.clause,sentence+w.sentence);
 	}
 };
+map<string,int> wordlist;
+set<string> adjlist;
+void getlists()
+{
+  ifstream listfile("fullist");
+  for(int i=0;i<500;i++)
+  {
+    string a;int b;
+    listfile>>a>>b;
+    wordlist[a]=b;
+  }
+  ifstream listfileadj("adjlist");
+  for(int i=0;i<227;i++)
+  {
+    string a;
+    listfileadj>>a;
+    adjlist.insert(a);
+  }
+}
 
 weightstruct getweight(string s)
 {
 	if(s[0]=='$')
-		return weightstruct(200,100,100);
-	if(s=="one"||s=="two"||s=="three"||s=="four"||s=="five"||s=="six"||s=="seven"||s=="eight"||s=="nine")return weightstruct(100,50,50);
-	if(s=="example")return weightstruct(-300,-200,-200);
+		return weightstruct(200,50,10);
+	if(s=="one"||s=="two"||s=="three"||s=="four"||s=="five"||s=="six"||s=="seven"||s=="eight"||s=="nine")return weightstruct(100,25,10);
+	if(s=="example")return weightstruct(-300,-200,-100);
+  
+
+  
+  if(s=="spi")
+    return weightstruct(-100,-10,-200);
+  if(s=="ipf")
+    return weightstruct(-100,30,50);
+  if(s=="please") return weightstruct(-300,0,0);
+  if(adjlist.find(s)!=adjlist.end())
+  {
+    return weightstruct(-100,-10,-10);
+  }
+  if(wordlist.find(s)!=wordlist.end())
+  {
+    return weightstruct(-wordlist[s]/100000,-10,-5);
+  }
+  
+  /*
 	if(s=="a"||s=="is"||s=="are"||s=="to"||s=="so"||s=="as") return weightstruct(-200,0,0);
 	if(s=="the")
 		return weightstruct(-200,0,0);
@@ -38,6 +75,14 @@ void preprocessword(int stage,string& s,string& next,int pos)
 	{
 		if(s[0]=='$')
 			s=s.substr(1,s.length()-2);
+    if(s.length()>5)
+    {
+      string tmp;
+      for(char x:s)
+        if(x!='a'&&x!='e'&&x!='i'&&x!='o'&&x!='u')
+          tmp+=x;
+      s=tmp;
+    }
 	}
 	if(stage==0)
 	{
@@ -48,6 +93,19 @@ void preprocessword(int stage,string& s,string& next,int pos)
 	}
 	if(stage==1)
 	{
+    if(s=="sample"&&next=="input")
+      s="spi",next="";
+    if(s=="sample"&&next=="output")
+      s="spo",next="";
+    if(s=="input"&&next=="format")
+      s="ipf",next="";
+    if(s=="output"&&next=="format")
+      s="opf",next="";
+
+    if(s=="input")
+      s="ipt";
+    if(s=="output")
+      s="opt";
 		if(s=="and")
 			s=",";
 		if(s.size()>=2)
@@ -94,7 +152,7 @@ struct clause
 			}
 			else
 			{
-				if(s.at(j)!=' ')
+				if(s.at(j)!=' '&&s.at(j)!='\"'&&s.at(j)!='\"')
 					word+=(s.at(j));
 				//if(('a'<=s.at(j)&&s.at(j)<='z')||('A'<=s.at(j)&&s.at(j)<='Z')||('0'<=s.at(j)&&s.at(j)<='9')||(iseq&&s.at(j)!=' '))
 				//  word+=('A'<=s.at(j)&&s.at(j)<='Z')?(s.at(j)-'A'+'a'):(s.at(j));
@@ -156,7 +214,7 @@ struct sentence
 		for(int j=0;j<s.length();j++)
 		{
 			if(s.at(j)=='$') iseq=!iseq;
-			if(s.at(j)!=','|| iseq) cl+=s.at(j);
+			if((s.at(j)!=','&&s.at(j)!='('&&s.at(j)!=')'&&s.at(j)!=';'&&s.at(j)!=':')|| iseq) cl+=s.at(j);
 			else
 			{
 				clause tmp;
@@ -313,6 +371,7 @@ int compressrate;
 int targetsize;
 
 int main(int argc, char *argv[]) {
+  getlists();
 	string input="";
 
   string input_file = argv[1];
@@ -371,7 +430,7 @@ int main(int argc, char *argv[]) {
 	output.calcweight();
 	output.preprocess(-1);
 	string s = output.gettextstring(-100000);
-  //cout << s << '\n';
+  cout << s << '\n';
     
   //where can i find a library of words in categories can't find any
   // not really sure