Bill Jolitz @ Berkeley received this version from Lorindia Cherry

[unix-history] / usr / src / usr.bin / diction / style1 / style1.l

diff --git a/usr/src/usr.bin/diction/style1/style1.l b/usr/src/usr.bin/diction/style1/style1.l
index a994ceb..9fbc6de 100644 (file)
--- a/usr/src/usr.bin/diction/style1/style1.l
+++ b/usr/src/usr.bin/diction/style1/style1.l
@@ -2,10 +2,11 @@
 /* break out words, output cap + word(inverted) */
 
 #ifndef lint
 /* break out words, output cap + word(inverted) */
 
 #ifndef lint

-static char sccsid[] = "@(#)style1.l   4.1     (Berkeley)      82/11/06";
+static char sccsid[] = "@(#)style1.l   4.2     (Berkeley)      82/11/06";

 #endif not lint
 
 #include <stdio.h>
 #endif not lint
 
 #include <stdio.h>

+#include <ctype.h>

 #define OUT()  for(i=yyleng-1;i>=0; i--)putchar(yytext[i]); putchar('\n')
 #define OUT1(nam)      printf("%c:%s\n",nam,yytext)
 #define OUTN(string)   printf("%s\n",string)
 #define OUT()  for(i=yyleng-1;i>=0; i--)putchar(yytext[i]); putchar('\n')
 #define OUT1(nam)      printf("%c:%s\n",nam,yytext)
 #define OUTN(string)   printf("%s\n",string)


@@ -13,9 +14,11 @@ static char sccsid[] = "@(#)style1.l 4.1     (Berkeley)      82/11/06";
 #include "nhash.c"
 #include "dict.c"
 #include "ydict.c"
 #include "nhash.c"
 #include "dict.c"
 #include "ydict.c"

+#include "abbrev.c"

 char nt[]  = "D:n't";
 char qs[]  = "c:'s";
 char fin[]  = "E:.";
 char nt[]  = "D:n't";
 char qs[]  = "c:'s";
 char fin[]  = "E:.";

+int NOCAPS = 0;                /* if set all caps are turned to lower case */

 int i,j;
 int dot = 0;
 int first  = 1;
 int i,j;
 int dot = 0;
 int first  = 1;


@@ -23,39 +26,75 @@ int qflg,nflg;
 int cap  = 0;
 %}
 %p 3000
 int cap  = 0;
 %}
 %p 3000

-%a 2500
+%a 3300
+%o 4500

 
 L      [a-z]
 N      [0-9]
 C      [A-Z]
 
 L      [a-z]
 N      [0-9]
 C      [A-Z]

+A      [a-zA-Z]
+P      [a-zA-Z0-9]

 
 %%
 
 %%

-(St|Dr|Drs|Mr|Mrs|Ms)"."       {
+^[.!].+[\n]    {
+       if(dot){
+               OUTN(fin);
+               dot = 0;
+               first = 1;
+       }
+       printf(":%s",yytext);
+       }
+May    {
+               if(first == 0){
+                       OUT1(NOUN);
+               }
+               else {
+                       first = 0;
+                       yytext[0] = tolower(yytext[0]);
+                       cap = 1;
+                       goto wd;
+               }
+       }
+"U.S."         {

                OUT1(NOUN);
                }
 {C}{L}*'[s]    {
                pos(1);
                if(first==1)first=0;
                }
                OUT1(NOUN);
                }
 {C}{L}*'[s]    {
                pos(1);
                if(first==1)first=0;
                }

-{C}+['][s]*    {
+{C}+['][s]     {
+               if(NOCAPS)
+                       for(i=0;i<yyleng;i++)
+                               if(isupper(yytext[i]))yytext[i] = tolower(yytext[i]);

                OUT1(POS);
                }
                OUT1(POS);
                }

-(({C}+{L}*)|({C}*{L}+))+([-](({C}*{L}+)|({C}+{L}*))+)+ {
+{P}+([-]{P}+)+ {
+               if(NOCAPS)
+                       for(i=0;i<yyleng;i++)
+                               if(isupper(yytext[i]))yytext[i] = tolower(yytext[i]);

                OUT1(NOUN_ADJ);
                }
 {C}{C}+        {
                OUT1(NOUN_ADJ);
                }
 {C}{C}+        {

+               if(NOCAPS)
+                       for(i=0;i<yyleng;i++)
+                               yytext[i] = tolower(yytext[i]);

                if((i=input()) == 's'){
                        yytext[yyleng++] = 's';
                        yytext[yyleng] = '\0';
                if((i=input()) == 's'){
                        yytext[yyleng++] = 's';
                        yytext[yyleng] = '\0';

-                       OUT1(NOUN);
+                       OUT1(PNOUN);

                }
                else {
                        unput(i);
                }
                else {
                        unput(i);

-                       for(i=0;i<yyleng;i++)yytext[i]+= 'a' - 'A';
+                       if(!NOCAPS)
+                               for(i=0;i<yyleng;i++)yytext[i] = tolower(yytext[i]);

                                goto wd;
                }
                }
 [LD][']{C}{L}* {
                                goto wd;
                }
                }
 [LD][']{C}{L}* {

+               if(NOCAPS){
+                       yytext[0] = tolower(yytext[0]);
+                       yytext[2] = tolower(yytext[2]);
+               }

                OUT1(NOUN_ADJ);
                }
 {C}{L}*        {
                OUT1(NOUN_ADJ);
                }
 {C}{L}*        {


@@ -66,13 +105,10 @@ C  [A-Z]
                        cap = 0;
                        goto wd;
                }
                        cap = 0;
                        goto wd;
                }

-               yytext[0]+= 'a' - 'A';
+               yytext[0] = tolower(yytext[0]);

                goto wd;
        }
                goto wd;
        }

-({N}+[-]{N}+[-]*)+     {
-               OUT1(NOUN_ADJ);
-               }
-({N}+[-]*{L}+[-]*)+    {
+{N}":"{N}{N}   {

                OUT1(NOUN_ADJ);
                }
 ({N}*[,])*({N}+".")+[ \t\n]+{C}        {
                OUT1(NOUN_ADJ);
                }
 ({N}*[,])*({N}+".")+[ \t\n]+{C}        {


@@ -84,8 +120,23 @@ C   [A-Z]
                OUTN(fin);
                first = 1;
        }
                OUTN(fin);
                first = 1;
        }

-[ \t`][a-zA-Z0-9.]*("\/"[a-zA-Z0-9]+"."*)+[']* {
-               if(yytext[yyleng-1] == '.')dot=1;
+([hH]e"/"[sS]he)|([sS]he"/"[hH]e)              {
+       if(NOCAPS)
+               if(isupper(yytext[0]))yytext[0] = tolower(yytext[0]);
+       OUT1(PRONS);
+       }
+([hH]is"/"[hH]er)|([hH]er"/"[hH]is)    {
+       if(NOCAPS)
+               if(isupper(yytext[0]))yytext[0] = tolower(yytext[0]);
+       OUT1(POS);
+       }
+[ \t`]*[a-zA-Z0-9.]*("\/"[a-zA-Z0-9.]+)+[']*   {
+               if(yytext[yyleng-1] == '.'){
+                       if(ahead() == 0)dot=1;
+               }
+               if(NOCAPS)
+                       for(i=0;i<yyleng;i++)
+                               if(isupper(yytext[i]))yytext[i] = tolower(yytext[i]);

                OUT1(NOUN_ADJ);
                }
 {N}+([,]{N}+)*("."{N}+)*[']*[s]*       {
                OUT1(NOUN_ADJ);
                }
 {N}+([,]{N}+)*("."{N}+)*[']*[s]*       {


@@ -98,15 +149,14 @@ C  [A-Z]
        if(yytext[yyleng-1] == '.')dot=1;
        OUT1(NOUN_ADJ);
        }
        if(yytext[yyleng-1] == '.')dot=1;
        OUT1(NOUN_ADJ);
        }

-{L}+[-]*{N}+   {
-               OUT1(NOUN_ADJ);
-               }
-{C}+[-]*{N}+   {
+({A}*{N}+{A}*)+        {
+       if(input() == '.')
+               ahead();
+       if(NOCAPS)
+               for(i=0;i<yyleng;i++)
+                       if(isupper(yytext[i]))yytext[i]=tolower(yytext[i]);

        OUT1(NOUN_ADJ);
        }
        OUT1(NOUN_ADJ);
        }

-{N}+[-]+{C}+   {
-       OUT1(NOUN_ADJ);
-               }

 {N}+[%]                {
                OUT1(NOUN_ADJ);
                }
 {N}+[%]                {
                OUT1(NOUN_ADJ);
                }


@@ -149,24 +199,6 @@ C  [A-Z]
 "et al."       {
                OUT1(NOUN);
                }
 "et al."       {
                OUT1(NOUN);
                }

-[Nn][Oo][s]*"."        {
-               OUT1(NOUN_ADJ);
-               }
-[Ff]ig[s]*"."  {
-               OUT1(NOUN_ADJ);
-               }
-[Dd]ept[s]*"." {
-               OUT1(NOUN_ADJ);
-       }
-[Ee]q"."       {
-               OUT1(NOUN_ADJ);
-               }
-dB"."  {
-               OUT1(NOUN_ADJ);
-               }
-vs"."  {
-       OUT1(PREP);
-       }

 in"."[ \n]*{C} {
                unput(yytext[yyleng-1]);
                yytext[2] = '\0';
 in"."[ \n]*{C} {
                unput(yytext[yyleng-1]);
                yytext[2] = '\0';


@@ -174,24 +206,9 @@ in"."[ \n]*{C}     {
                OUTN(fin);
                first = 1;
                }
                OUTN(fin);
                first = 1;
                }

-(in|ft|yr|ckts|mi)"."  {
-               OUT1(NOUN_ADJ);
-               }

 Ph"."[ ]*[Dd]"."       {
                OUT1(ADJ);
                }
 Ph"."[ ]*[Dd]"."       {
                OUT1(ADJ);
                }

-[Jj]r"."       {
-       OUT1(ADJ);
-       }
-[Cc]h"."       {
-               OUT1(NOUN_ADJ);
-       }
-[Rr]ef[s]*"."  {
-               OUT1(NOUN_ADJ);
-       }
-Inc"." {
-               OUT1(ADJ);
-       }

 [A-Z]"."       {
                dot=1;
                OUT1(NOUN);
 [A-Z]"."       {
                dot=1;
                OUT1(NOUN);


@@ -205,6 +222,9 @@ can't               {
 won't          {
                OUT1('X');
                }
 won't          {
                OUT1('X');
                }

+ain't          {
+               OUT1('g');
+               }

 {L}+n't                {
                nflg=1;
                yytext[yyleng-3]='\0';
 {L}+n't                {
                nflg=1;
                yytext[yyleng-3]='\0';


@@ -212,13 +232,13 @@ won't             {
                goto wd;
                }
 [A-Z]{L}+n't   {
                goto wd;
                }
 [A-Z]{L}+n't   {

-               yytext[0]+= 'a' - 'A';
+               yytext[0] = tolower(yytext[0]);

                nflg=1;
                yytext[yyleng-3]='\0';
                yyleng -= 3;
                goto wd;
                }
                nflg=1;
                yytext[yyleng-3]='\0';
                yyleng -= 3;
                goto wd;
                }

-o'clock        {
+o'clock                {

                OUT1(ADV);
        }
 {L}+'[s]       {
                OUT1(ADV);
        }
 {L}+'[s]       {


@@ -247,7 +267,8 @@ wd:
        if((j = lookup(yytext,1,0)) != 0){
                first=0;
                if(cap){
        if((j = lookup(yytext,1,0)) != 0){
                first=0;
                if(cap){

-                       yytext[0] += 'A' - 'a';
+                       if(!NOCAPS)
+                               yytext[0] = toupper(yytext[0]);

                        cap = 0;
                        if(dot)OUTN(fin);
                }
                        cap = 0;
                        if(dot)OUTN(fin);
                }


@@ -272,7 +293,7 @@ wd:
                                        OUT1(NOUN);
                                        break;
                                }
                                        OUT1(NOUN);
                                        break;
                                }

-                                look(gy,yyleng-2,ADJ);
+                               look(gy,yyleng-2,ADJ);

                                break;
                        case 'r':       switch(yytext[yyleng-3]){
                                case 'a': look(ary,yyleng-3,ADJ);
                                break;
                        case 'r':       switch(yytext[yyleng-3]){
                                case 'a': look(ary,yyleng-3,ADJ);


@@ -289,10 +310,10 @@ wd:
                                break;
                        default: OUT();
                        }
                                break;
                        default: OUT();
                        }

-               }
+       }

                else {
                        if(cap){
                else {
                        if(cap){

-                               yytext[0] += 'A' - 'a';
+                               if(!NOCAPS)yytext[0] = toupper(yytext[0]);

                                cap = 0;
                                OUT1(NOUN_ADJ);
                        }
                                cap = 0;
                                OUT1(NOUN_ADJ);
                        }


@@ -336,6 +357,30 @@ wd:
        first = 1;
        OUT1(END);
        }
        first = 1;
        OUT1(END);
        }

+{A}{A}+"."     {
+               yytext[yyleng-1] = '\0';
+               if((j=abbrev(yytext,1,0)) != 0){
+                       if(isupper(yytext[0])){
+                               if(NOCAPS)yytext[0] = tolower(yytext[0]);
+                               if(first == 1)first=0;
+                       }
+                       yytext[yyleng-1] = '.';
+                       OUT1(j);
+               }
+               else {
+                       j = ahead();
+                       if(j == 0)
+                               yyleng--;
+                       for(i=0;i<yyleng;i++)
+                               if(isupper(yytext[i])){
+                                       yytext[i] = tolower(yytext[i]);
+                                       if(i == 0)cap = 1;
+                                       else cap = 0;
+                               }
+                       if(j == 0)goto wd;
+                       OUT1(NOUN_ADJ);
+               }
+       }

 "."    {
        first=1;
        OUT1(END);
 "."    {
        first=1;
        OUT1(END);


@@ -396,7 +441,7 @@ char cc;
 }
 pos(flg){
        int ii,j;
 }
 pos(flg){
        int ii,j;

-       if(flg==1)yytext[0] += 'a' - 'A';
+       if(flg == 1)yytext[0] = tolower(yytext[0]);

        for(ii=yyleng-1;yytext[ii] != '\''; ii--);
        yytext[ii] = '\0';
        if((j=lookup(yytext,1,0)) != 0){
        for(ii=yyleng-1;yytext[ii] != '\''; ii--);
        yytext[ii] = '\0';
        if((j=lookup(yytext,1,0)) != 0){


@@ -405,7 +450,7 @@ pos(flg){
                OUTN(qs);
        }
        else{
                OUTN(qs);
        }
        else{

-               if(flg==1)yytext[0] += 'A' - 'a';
+               if(flg==1 && !NOCAPS)yytext[0] = toupper(yytext[0]);

                yytext[ii] = '\'';
                OUT1(POS);
        }
                yytext[ii] = '\'';
                OUT1(POS);
        }


@@ -419,9 +464,11 @@ char       *argv[];
        register int rc=0;
        putchar(':'); putchar('\n');
        getd();
        register int rc=0;
        putchar(':'); putchar('\n');
        getd();

+       getab();

        ygetd();
        if(argc<=1) {
                yylex();
        ygetd();
        if(argc<=1) {
                yylex();

+               OUTN(fin);

        }else{
                while(argc>1) {
                        if(freopen(argv[1],"r",stdin)==NULL) {
        }else{
                while(argc>1) {
                        if(freopen(argv[1],"r",stdin)==NULL) {


@@ -430,9 +477,24 @@ char       *argv[];
                        }else{
                                filename=argv[1];
                                yylex();
                        }else{
                                filename=argv[1];
                                yylex();

+                               OUTN(fin);

                        }
                        argc--; argv++;
                }
        }
        return(rc);
 }
                        }
                        argc--; argv++;
                }
        }
        return(rc);
 }

+ahead(){
+       register int c;
+       if(isalnum((c=input()))){
+               yytext[yyleng++] = '.';
+               while(!isspace((c=input() )))
+                       yytext[yyleng++] = c;
+               yytext[yyleng] = '\0';
+               unput(c);
+               return(1);
+       }
+       unput(c);
+       unput('.');
+       return(0);
+}