1 /* unlit.c Wed Dec 5 17:16:24 GMT 1990
3 * Literate script filter. In contrast with the format used by most
4 * programming languages, a literate script is a program in which
5 * comments are given the leading role, whilst program text must be
6 * explicitly flagged as such by placing a `>' character in the first
7 * column on each line. It is hoped that this style of programming will
8 * encourage the writing of accurate and clearly documented programs
9 * in which the writer may include motivating arguments, examples
12 * Unlit is a filter that can be used to strip all of the comment lines
13 * out of a literate script file. The command format for unlit is:
14 * unlit [-n] [-q] ifile ofile
15 * where ifile and ofile are the names of the input (literate script) and
16 * output (raw program) files respectively. Either of these names may
17 * be `-' representing the standard input or the standard output resp.
18 * A number of rules are used in an attempt to guard against the most
19 * common errors that are made when writing literate scripts:
20 * 1) Empty script files are not permitted. A file in which no lines
21 * begin with `>' usually indicates a file in which the programmer
22 * has forgotten about the literate script convention.
23 * 2) A line containing part of program definition (i.e. preceeded by `>')
24 * cannot be used immediately before or after a comment line unless
25 * the comment line is blank. This error usually indicates that
26 * the `>' character has been omitted from a line in a section of
27 * program spread over a number of lines.
28 * Using the -q (quiet) flag suppresses the signalling of these error
29 * conditions. The default behaviour can be selected explicitly using
30 * the -n (noisy) option so that any potential errors in the script file
33 * The original idea for the use of literate scripts is due to Richard
34 * Bird of the programming Research Group, Oxford and was initially
35 * adopted for use in the implementation of the functional programming
36 * language Orwell used for teaching in Oxford. This idea has subsequently
37 * been borrowed in a number of other language implementations.
39 * Modified to understand \begin{code} ... \end{code} used in Glasgow. -- LA
40 * And \begin{pseudocode} ... \end{pseudocode}. -- LA
47 #define NULLSTR ((char *)0)
49 #define MISSINGBLANK "unlit: Program line next to comment"
50 #define EMPTYSCRIPT "unlit: No definitions in file (perhaps you forgot the '>'s?)"
51 #define USAGE "usage: unlit [-q] [-n] [-c] [-#] [-P] [-h label] file1 file2\n"
52 #define CANNOTOPEN "unlit: cannot open \"%s\"\n"
53 #define CANNOTWRITE "unlit: error writing \"%s\"\n"
54 #define CANNOTWRITESTDOUT "unlit: error writing standard output\n"
55 #define DISTINCTNAMES "unlit: input and output filenames must differ\n"
56 #define MISSINGENDCODE "unlit: missing \\end{code}\n"
58 #define BEGINCODE "\\begin{code}"
59 #define LENBEGINCODE 12
60 #define ENDCODE "\\end{code}"
63 /* According to Will Partain, the inventor of pseudocode, this gone now. */
64 #define MISSINGENDPSEUDOCODE "unlit: missing \\end{pseudocode}\n"
65 #define BEGINPSEUDOCODE "\\begin{pseudocode}"
66 #define LENBEGINPSEUDOCODE 18
67 #define ENDPSEUDOCODE "\\end{pseudocode}"
68 #define LENENDPSEUDOCODE 16
71 typedef enum { START, BLANK, TEXT, DEFN, BEGIN, /*PSEUDO,*/ END, HASH, SHEBANG } line;
72 #define isWhitespace(c) (c==' ' || c=='\t')
73 #define isLineTerm(c) (c=='\n' || c==EOF)
75 static int noisy = 1; /* 0 => keep quiet about errors, 1 => report errors */
76 static int errors = 0; /* count the number of errors reported */
77 static int crunchnl = 0; /* don't print \n for removed lines */
78 static int leavecpp = 1; /* leave preprocessor lines */
79 static int ignore_shebang = 1; /* Leave out shebang (#!) lines */
80 static int no_line_pragma = 0; /* Leave out initial line pragma */
82 static char* prefix_str = NULL; /* Prefix output with a string */
84 static char *ofilename = NULL;
86 /* complain(file,line,what)
88 * print error message `what' for `file' at `line'. The error is suppressed
89 * if noisy is not set.
92 complain(file, lin, what)
98 fprintf(stderr, "%s ", file);
99 fprintf(stderr,"line %d: %s\n",lin,what);
106 if (!strcmp(ofilename,"-")) {
107 fprintf(stderr, CANNOTWRITESTDOUT);
109 fprintf(stderr, CANNOTWRITE, ofilename);
117 if (putc(c,ostream) == EOF) {
124 /* As getc, but does TAB expansion */
129 static int spleft = 0;
130 static int linepos = 0;
141 else if (c == '\n' || c == '\f') {
144 } else if (c == '\t') {
145 spleft = TABPOS - linepos % TABPOS;
156 /* readline(istream, ostream)
158 * Read a line from the input stream `istream', and return a value
159 * indicating whether that line was:
160 * BLANK (whitespace only),
161 * DEFN (first character is DEFNCHAR),
162 * TEXT (a line of text)
163 * BEGIN (a \begin{code} line)
164 * PSEUDO (a \begin{pseodocode} line)
165 * HASH (a preprocessor line)
166 * or END (indicating an EOF).
167 * Lines of type DEFN are copied to the output stream `ostream'
168 * (without the leading DEFNCHAR). BLANK and TEXT lines are
169 * replaced by empty (i.e. blank lines) in the output stream, so
170 * that error messages refering to line numbers in the output file
171 * can also be used to locate the corresponding line in the input
175 line readline(istream,ostream)
176 FILE *istream, *ostream; {
187 if ( ignore_shebang ) {
190 while (c=egetc(istream), !isLineTerm(c)) ;
198 while (c=egetc(istream), !isLineTerm(c))
200 myputc('\n',ostream);
206 /* myputc(' ',ostream);*/
207 while (c=egetc(istream), !isLineTerm(c))
209 myputc('\n',ostream);
214 myputc('\n',ostream);
216 while (isWhitespace(c))
223 while (c=egetc(istream), !isLineTerm(c))
224 if (i < sizeof buf - 1)
226 while(i > 0 && isspace(buf[i-1]))
229 if (strcmp(buf, BEGINCODE) == 0)
232 else if (strcmp(buf, BEGINPSEUDOCODE) == 0)
240 /* unlit(file,istream,ostream)
242 * Copy the file named `file', accessed using the input stream `istream'
243 * to the output stream `ostream', removing any comments and checking
244 * for bad use of literate script features:
245 * - there should be at least one BLANK line between a DEFN and TEXT
246 * - there should be at least one DEFN line in a script.
249 unlit(file, istream, ostream)
253 line last, this=START;
259 this = readline(istream, ostream);
263 if (last==DEFN && this==TEXT)
264 complain(file, linesread-1, MISSINGBLANK);
265 if (last==TEXT && this==DEFN)
266 complain(file, linesread, MISSINGBLANK);
268 /* start of code, copy to end */
271 if (fgets(lineb, sizeof lineb, istream) == NULL) {
272 complain(file, linesread, MISSINGENDCODE);
276 if (strncmp(lineb,ENDCODE,LENENDCODE) == 0) {
277 myputc('\n', ostream);
280 fputs(lineb, ostream);
285 if (this == PSEUDO) {
288 if (fgets(lineb, sizeof lineb, istream) == NULL) {
289 complain(file, linesread, MISSINGENDPSEUDOCODE);
293 myputc('\n', ostream);
294 if (strncmp(lineb,ENDPSEUDOCODE,LENENDPSEUDOCODE) == 0) {
303 complain(file,linesread,EMPTYSCRIPT);
308 * Main program. Processes command line arguments, looking for leading:
309 * -q quiet mode - do not complain about bad literate script files
310 * -n noisy mode - complain about bad literate script files.
311 * -r remove cpp droppings in output.
312 * -P don't output any CPP line pragmas.
313 * Expects two additional arguments, a file name for the input and a file
314 * name for the output file. These two names must normally be distinct.
315 * An exception is made for the special name "-" which can be used in either
316 * position to specify the standard input or the standard output respectively.
322 FILE *istream, *ostream;
325 for (argc--, argv++; argc > 0; argc--, argv++)
326 if (strcmp(*argv,"-n")==0)
328 else if (strcmp(*argv,"-q")==0)
330 else if (strcmp(*argv,"-c")==0)
332 else if (strcmp(*argv,"-P")==0)
334 else if (strcmp(*argv,"-h")==0) {
339 prefix_str = (char*)malloc(sizeof(char)*(1+strlen(*argv)));
341 strcpy(prefix_str, *argv);
343 } else if (strcmp(*argv,"-#")==0)
349 fprintf(stderr, USAGE);
353 if (strcmp(argv[0],argv[1])==0 && strcmp(argv[0],"-")!=0) {
354 fprintf(stderr, DISTINCTNAMES);
359 if (strcmp(argv[0], "-")==0) {
364 if ((istream=fopen(argv[0], "r")) == NULL) {
365 fprintf(stderr, CANNOTOPEN, argv[0]);
370 if (strcmp(argv[1], "-")==0)
373 if ((ostream=fopen(argv[1], "w")) == NULL) {
374 fprintf(stderr, CANNOTOPEN, argv[1]);
378 /* Prefix the output with line pragmas */
379 if (!no_line_pragma && prefix_str) {
380 /* Both GHC and CPP understand the #line pragma.
381 * We used to throw in both a #line and a {-# LINE #-} pragma
382 * here, but CPP doesn't understand {-# LINE #-} so it thought
383 * the line numbers were off by one. We could put the {-# LINE
384 * #-} before the #line, but there's no point since GHC
385 * understands #line anyhow. --SDM 8/2003
387 fprintf(ostream, "#line 1 \"%s\"\n", prefix_str);
390 unlit(file, istream, ostream);
392 if (istream != stdin) fclose(istream);
393 if (ostream != stdout) {
394 if (fclose(ostream) == EOF) {
399 exit(errors==0 ? 0 : 1);