V10/cmd/wwb/prose.c
/* NOTICE-NOT TO BE DISCLOSED OUTSIDE BELL SYS EXCEPT UNDER WRITTEN AGRMT */
/* Writer's Workbench version 2.3, April 7, 1981 */
#define NVAR 14
#define MAXLENG 79.0
#define READAB 0
#define NONF 1
#define PASS 8
#define NOM 9
#define AVW 10
#define EXPL 11
#define S_CPX 12
#include <stdio.h>
char *disclm ="BECAUSE YOUR TEXT IS SHORT \(< 2000 WORDS & < 100 SENTENCES\),\n\
THE FOLLOWING ANALYSIS MAY BE MISLEADING.\n\n";
char *note0 =
"NOTE: Your document is being compared against standards\n\
derived from 30 TMs, classified as good by Department Heads\n\
in the Research Area.\n";
char *note2 =
"NOTE: Your text is being compared against standards derived\n\
from 34 instructional texts produced by Department 45272.\n\n";
char *note1 =
"NOTE: Your text is being compared to standards for Craft\n\
derived from the research of E. Coke.\n\n";
char *note3 =
"NOTE: Your text is being compared to standards given in file\n\"%s.\"\n";
char *kincaid = "\nREADABILITY\n\
\n The Kincaid readability formula predicts that your text\n\
can be read by someone with %2.0f or more years of schooling,\n";
char *more_str =
"\n More Information on Sentence Structure\n\
\n Since passives, nominalizations, as well as expletives\n\
all require some form of the verb \"to be,\" an easy way to\n\
identify difficult paragraphs is to look for the overuse of\n\
forms of \"be.\" The `findbe' program makes this easy; it for-\n\
mats a text and underlines and capitalizes all forms of \"to\n\
be.\" To use it type:\n\n\
findbe filename\n\n\
You can also use the `style' program to print your sentences\n\
containing both a passive verb and a nominalization. Type:\n\n\
style -n filename\n";
char *read0 =
" Good TMs\n\
average close to 13th grade level, even though the audience\n\
has more education than that.\n";
char *read2 =
" Good train-\n\
ing materials average close to the 10th grade level, even\n\
though the audience has more education than that.\n";
char *read3 =
" Your good\n\
materials average close to grade %4.1f.\n";
char *read1 =
" Since Coke\n\
found that 62%% of Bell System craft have difficulty reading\n\
documents with readability scores above 12.0, the majority\n\
of the craft will probably not be able to read this.\n";
char *begmsg =
"\n Writing teachers also stress that no more than 75\n\
percent of the sentences in a text should begin with the\n\
subject of the sentence; these start with the subject\n\
%2.0f%% of the time. Try starting more of your sentences with\n\
prepositions, adverbs, or conjunctions. This change will\n\
have the added benefit of adding variety of sentence length\n\
and type.\n";
char *pasnom ="\n Passives and Nominalizations\n\
\n You have appropriately limited your use of passives and\n\
nominalizations \(nouns made from verbs, e.g. \"description\"\).\n";
char *badpas =
"\n This text contains a %s higher percentage of passive verbs\n\
(%4.1f%%) than is common in good documents of this type (%2.0f%%).\n";
char *badnom ="\n This text has a higher percentage of nominalizations\n\
(%4.1f%%) than is common in good documents of this type (%2.0f%%).\n";
char *badexp ="\n Expletives\n\
\n This text contains a higher percentage of expletives\n\
(%4.1f%%) than is common in good documents of this type (%2.0f%%).\n";
FILE *fp;
main(argc,argv)
int argc;
char *argv[];
{
char *rather="rather";
char *very="very";
char *too="too";
char *much="much";
char *null="";
char *qual;
char *note;
char *readmsg;
FILE *fpath;
int h,h0,fflag,sflag,ret;
char save[81], buf[100];
int i,aud,npara,p,j;
int numabst;
float kindex, avw,snonf, simple,complex, compound, compdx;
float ppron,padj, var[NVAR],array[NVAR][4],sc[NVAR];
float passive,explet;
int nomin,x;
int numsent, maxsent, qcount, icount,ml,lsum,mg,gsum;
long numwds,numnonf;
int maxindex,minindex,minsent;
float aindex,cindex,fgrad,findex,avl,fnumnonf,centlsum,centgsum;
int nsimple,ncomplex,ncompound,ncompdx,ntobe;
int npassive,pron,pos,adj,art,beg,begadv,begprep,begverb,begscon;
int begconj, begexp,noun;
float ptobe,paux,pinfin,padv,pconjc,pprepc,pnoun,pnomin,tot;
float pbegprep,pbegadv,pbegverb,pbegscon,pbegconj,pbegexp;
int tobe,aux,infin,prepc,conjc,adv;
char path[80];
fflag=0;
sflag=0;
aud=0;
note = note0;
readmsg = read0;
strcpy(path,LIB");
if(argc >= 2){
i=0;
if(*(argv[1]+i) =='f') {
fflag=1;
i++;
}
if(*(argv[1]+i) =='s') {
sflag=1;
i++;
}
switch (*(argv[1]+i)){
case 'm':
aud=0;
note = note0;
readmsg = read0;
break;
case 't':
aud=2;
note = note2;
readmsg = read2;
strds(LIB/train.st","training");
break;
case 'c':
aud=1;
note = note1;
readmsg = read1;
strds(LIB/crft.st","Craft");
break;
case 'x':
aud=3;
note = note3;
readmsg = read3;
if((fp=fopen(argv[2],"r"))==NULL){
fprintf(stderr,"Prose can't find your standards file \"%s,\" try specifying a more complete pathname.\n",argv[2]);
exit(1);
}
break;
}
}
if(aud==0)
strds(LIB/tm.st","TM");
findbeg:
if(fgets(save,81,stdin)==NULL){
fprintf(stderr,"The file named after the -f flag doesn't seem\
to contain a style table.\nOr else style cannot produce a table for your\
file.\n Try running style alone on your file.\n");
exit(1);
}
ret=sscanf(save,"%*s%f%*s%f%*s%f%*s%f (%f)",
&kindex,&aindex,&cindex,&fgrad,&findex);
if(ret !=5 ) goto findbeg;
fgets(buf,100,stdin);
scanf("%*s%*s%d%*s%*s%ld",&numsent,&numwds);
scanf("%*s%*s%*s%f%*s%*s%*s%f",&avw,&avl);
scanf("%*s%*s%d%*s%*s%d",&qcount,&icount);
scanf("%*s%*s%*s%d%f%%%*s%*s%f",&numnonf,&fnumnonf,&snonf);
scanf("%*s%*s (<%d) %f%% (%d) %*s%*s (>%d) %f%% (%d)",
&ml,¢lsum,&lsum,&mg,¢gsum,&gsum);
scanf("%*s%*s%d%*s%*s%*s%d;%*s%*s%d%*s%*s%*s%d\n",
&maxsent,&maxindex,&minsent,&minindex);
fgets(buf,100,stdin);
scanf("%*s%f%% (%d) %*s %f%% (%d)",&simple,&nsimple,&complex,&ncomplex);
scanf("%*s%f%% (%d) %*s%f%% (%d)\n",
&compound,&ncompound,&compdx,&ncompdx);
fgets(buf,100,stdin);
fgets(buf,100,stdin);
scanf("%*s%f%% (%d) %*s %f%% (%d)%*s%f%% (%d)",
&ptobe,&tobe,&paux,&aux,&pinfin,&infin);
scanf("%*s%*s %% %*s%*s%*s%f%% (%d)\n",&passive,&npassive);
fgets(buf,100,stdin);
scanf("%*s %f%% (%d)%*s%f%% (%d)%*s%f%% (%d)",
&pprepc,&prepc,&pconjc,&conjc,&padv,&adv);
scanf("%*s%f%% (%d)%*s %f%% (%d)%*s%f%% (%d)",
&pnoun,&noun, &padj,&adj,&ppron,&pron);
scanf("%*s%f %% (%d)\n",&pnomin,&nomin);
fgets(buf,100,stdin);
scanf("%*s%*s%*s (%d)%*s (%d)%*s (%d)%*s (%d)%*s (%d)%*s%f%%",
&noun,&pron,&pos,&adj,&art,&tot);
scanf("%*s%f%% (%d)%*s%f%% (%d) ",&pbegprep,&begprep,&pbegadv,&begadv);
scanf("%*s%f%% (%d) ",&pbegverb,&begverb);
scanf("%*s%f%% (%d)%*s%f%% (%d)",&pbegscon,&begscon,&pbegconj,&begconj);
scanf("%*s%f%% (%d)",&pbegexp,&begexp);
/*variables used by the prose program*/
var[READAB]=kindex;
var[NONF]=snonf;
var[2]=centlsum;
var[3]=centgsum;
var[4]=simple;
var[5]=complex;
var[6]=compound;
var[7]=compdx;
var[PASS]=passive;
var[NOM]=pnomin;
var[AVW]=avw;
var[EXPL]=pbegexp;
var[S_CPX]=simple-complex;
var[13]=compound+compdx;
/*read criterion values for variables from appropriate file*/
ret=0;
for(i=0;i <NVAR; i++)
for(j=0;j<4; j++){
ret+=fscanf(fp,"%f",&array[i][j]);
}
/* make sure values were read correctly*/
if(ret!=(NVAR*4)){
fprintf(stderr,"The standards file is not formatted\
properly.\nIt should have %d lines, each with 4 floating point numbers.\n",
NVAR);
exit(1);
}
/* Compare criterion values against STYLE produced scores*/
for(i=0; i< NVAR; i++){
sc[i]=0;
for(j=0; j < 4;j++){
if(var[i] > array[i][j]) sc[i]=j+1;
}
}
/* Read percent of abstract words */
if(numwds < 2000 && numsent <100)
printf(disclm);
if(sflag) goto shortans ;
printf(note,argv[2]);
/*READABILITY*/
printf(kincaid,kindex);
if(sc[READAB]==0){
printf("which is a low score for this type of document.");
if (aud==0)
printf(" If this is\nan instructional text\
(in paragraph form), run prose -t for\na more appropriate review.\n");
else if(aud==2)putchar('\n');
}
else if(sc[READAB] < 3)
printf("which is a good score for documents like this.\n");
else { /* ==3 or 4*/
if(sc[READAB] == 3)qual=rather;
else qual=very;
printf("which is %s high for this type of document.",qual);
if(aud != 1)printf(readmsg,(array[READAB][1]+array[READAB][2])/2.0);
if(sc[NONF] > 2)
prtfile(LIB/rd2.t");
if(aud == 1)printf(readmsg);
}
printf("\nVARIATION\n");
if(sc[S_CPX]==4 || sc[S_CPX]==0 || tot >MAXLENG)
prtfile(LIB/var.t");
if(sc[S_CPX]==0||sc[S_CPX]==4)
printf("\n In this text %2.0f%% of the sentences are\
simple, %2.0f%% are\ncomplex, giving a difference of %3.0f. This\
difference should\nrange from %2.0f to %2.0f for good documents of this\
type.\n",simple,complex, simple-complex,array[S_CPX][1],array[S_CPX][2]);
if(sc[S_CPX]==4){
if(sc[AVW] < 2)
prtfile(LIB/type.t");
else
prtfile(LIB/type2.t");
}
else if(sc[S_CPX]==0){
if(sc[READAB] < 3)
prtfile(LIB/var4.t");
else
prtfile(LIB/var6.t");
}
else printf("\n You have an appropriate distribution of sentence\
types.\n");
if(maxsent >= 50){
if(sc[S_CPX]==0 || sc[S_CPX]==4)
printf("\n Additionally, the longest sentence is\
%d words long.\n",maxsent);
else printf("\n The longest sentence,\
however, is %d words long.\n",maxsent);
prtfile(LIB/var5.t");
}
if(sc[AVW] > 2){
if(sc[S_CPX]==4 || sc[S_CPX]==0)
printf("\n Such changes would also help to\
reduce the average\n");
else printf("\n You should, however, consider\
shortening your average\n");
printf("sentence length. Your average is %2.0f words,\
which is ",var[AVW]);
if(sc[AVW]==3)printf("high.\n");
if(sc[AVW]==4) printf(" very\nhigh. ");
printf("A good average would be %2.0f to %2.0f words.\n",
array[AVW][1],array[AVW][2]);
}
if(tot >MAXLENG){
printf(begmsg,tot);
}
printf("\nSENTENCE STRUCTURE\n");
if(sc[PASS]<=2 && sc[NOM] <=2) {
printf(pasnom);
}
else {
printf("\n Passives\n");
if(sc[PASS] > 2){
if(sc[PASS] == 3)qual=null;
else qual=much;
printf(badpas,qual,var[PASS],(array[PASS][1]+array[PASS][2])/2.0);
prtfile(LIB/pass.t");
}
else printf("\n You have limited your passives appropriately.\n");
printf("\n Nominalizations\n");
if(sc[NOM] > 2){
printf(badnom,var[NOM],(array[NOM][1]+array[NOM][2])/2.0);
prtfile(LIB/nom.t");
}
else printf("\n You have appropriately limited your\
nominalizations\n\(nouns made from verbs, e.g., \"description\"\).\n");
}
if(sc[EXPL]==4){
printf(badexp,var[EXPL],(array[EXPL][1]+array[EXPL][2])/2.0);
prtfile(LIB/exp.t");
}
if(sc[PASS]>2 && sc[NOM]>2){
printf(more_str);
}
if(!fflag){
if(aud==0)
prtfile(LIB/popttm.t");
else if(aud==2)
prtfile(LIB/poptt.t");
}
prtfile(LIB/fur.t");
exit(0);
shortans :
if(aud==0) printf("Compared to TMs.\n\n");
else if(aud==1) printf("Evaluated for Craft.\n\n");
else if(aud==2) printf("Compared to training material.\n\n");
else if(aud==3)printf("Compared to file \"%s\"\n",argv[2]);
printf("Reading grade level--%2.0f: ",kindex);
if(sc[READAB]==1 || sc[READAB]==2) printf("Good\n");
else if(sc[READAB]==0) printf("Low\n");
else if(sc[READAB]==3) printf("High\n");
else if(sc[READAB]==4)printf("Very high\n");
printf("Variation--");
if(sc[S_CPX] == 4)
printf("Too many short, simple sentences--subordinate.\n");
else if(sc[S_CPX]==0) printf("Too many complex sentences.\n");
else printf("Good sentence type distribution.\n");
if(sc[AVW] > 2){
if(sc[AVW]==3)
qual = too;
else qual = very;
printf(" Sentences are %s long--avg\
length=%2.1f words\n",qual,var[AVW]);
printf(" --Good length\
= %2.1f to %2.1f\n",array[AVW][1],array[AVW][2]);
}
if( maxsent >50)
printf("Longest sentence is %d words---perhaps it is a list\n"
,maxsent);
if(tot >MAXLENG)
printf(" %2.0f%%= Too many sentences begin with\
the subject.\n",tot);
printf("Passives--%2.0f%%: ",passive);
if(sc[PASS]<=2)printf("Good\n");
else if(sc[PASS] ==3)printf("High\n");
else if(sc[PASS]==4)printf("Very high\n");
printf("Nominalizations--%2.0f%%: ",pnomin);
if(sc[NOM]<=2)printf("Good\n");
else if(sc[NOM]==3)printf("High\n");
else if(sc[NOM]==4)printf("Very high\n");
if(sc[EXPL]==4)printf("Expletive--%2.0f%%: Very high\n",pbegexp);
if (!fflag) printf("Don't forget the styl.tmp file.\n");
putchar('\n');
}
prtfile(s)
char *s;
{
FILE *ff;
char ch[512];
int num;
if((ff=fopen(s,"r"))==NULL){
fprintf(stderr,"Prose can't find the file %s\n",s);
exit(1);
}
while((num=fread(ch,sizeof(ch[0]),512,ff)))
fwrite(ch,sizeof(ch[0]),num,stdout);
fclose(ff);
}
strds(s1,s2)
char *s1, *s2;
{
if((fp=fopen(s1,"r"))==NULL){
fprintf(stderr,"Prose can't find the %s standards file %s\n",
s2,s1);
exit(1);
}
}