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
46 #define NULLSTR ((char *)0)
48 #define MISSINGBLANK "unlit: Program line next to comment"
49 #define EMPTYSCRIPT "unlit: No definitions in file (perhaps you forgot the '>'s?)"
50 #define USAGE "usage: unlit [-q] [-n] [-c] file1 file2\n"
51 #define CANNOTOPEN "unlit: cannot open \"%s\"\n"
52 #define DISTINCTNAMES "unlit: input and output filenames must differ\n"
53 #define MISSINGENDCODE "unlit: missing \\end{code}\n"
55 #define BEGINCODE "\\begin{code}"
56 #define LENBEGINCODE 12
57 #define ENDCODE "\\end{code}"
60 /* According to Will Partain, the inventor of pseudocode, this gone now. */
61 #define MISSINGENDPSEUDOCODE "unlit: missing \\end{pseudocode}\n"
62 #define BEGINPSEUDOCODE "\\begin{pseudocode}"
63 #define LENBEGINPSEUDOCODE 18
64 #define ENDPSEUDOCODE "\\end{pseudocode}"
65 #define LENENDPSEUDOCODE 16
68 typedef enum { START, BLANK, TEXT, DEFN, BEGIN, /*PSEUDO,*/ END, HASH, SHEBANG } line;
69 #define isWhitespace(c) (c==' ' || c=='\t')
70 #define isLineTerm(c) (c=='\n' || c==EOF)
72 static int noisy = 1; /* 0 => keep quiet about errors, 1 => report errors */
73 static int errors = 0; /* count the number of errors reported */
74 static int crunchnl = 0; /* don't print \n for removed lines */
75 static int leavecpp = 1; /* leave preprocessor lines */
76 static int ignore_shebang = 1; /* Leave out shebang (#!) lines */
78 static char* prefix_str = NULL; /* Prefix output with a string */
80 /* complain(file,line,what)
82 * print error message `what' for `file' at `line'. The error is suppressed
83 * if noisy is not set.
86 complain(file, lin, what)
92 fprintf(stderr, "%s ", file);
93 fprintf(stderr,"line %d: %s\n",lin,what);
100 /* As getc, but does TAB expansion */
105 static int spleft = 0;
106 static int linepos = 0;
117 else if (c == '\n' || c == '\f') {
120 } else if (c == '\t') {
121 spleft = TABPOS - linepos % TABPOS;
132 /* readline(istream, ostream)
134 * Read a line from the input stream `istream', and return a value
135 * indicating whether that line was:
136 * BLANK (whitespace only),
137 * DEFN (first character is DEFNCHAR),
138 * TEXT (a line of text)
139 * BEGIN (a \begin{code} line)
140 * PSEUDO (a \begin{pseodocode} line)
141 * HASH (a preprocessor line)
142 * or END (indicating an EOF).
143 * Lines of type DEFN are copied to the output stream `ostream'
144 * (without the leading DEFNCHAR). BLANK and TEXT lines are
145 * replaced by empty (i.e. blank lines) in the output stream, so
146 * that error messages refering to line numbers in the output file
147 * can also be used to locate the corresponding line in the input
151 line readline(istream,ostream)
152 FILE *istream, *ostream; {
163 if ( ignore_shebang ) {
166 while (c=egetc(istream), !isLineTerm(c)) ;
174 while (c=egetc(istream), !isLineTerm(c))
182 /* putc(' ',ostream);*/
183 while (c=egetc(istream), !isLineTerm(c))
192 while (isWhitespace(c))
199 while (c=egetc(istream), !isLineTerm(c))
200 if (i < sizeof buf - 1)
202 while(i > 0 && isspace(buf[i-1]))
205 if (strcmp(buf, BEGINCODE) == 0)
208 else if (strcmp(buf, BEGINPSEUDOCODE) == 0)
216 /* unlit(file,istream,ostream)
218 * Copy the file named `file', accessed using the input stream `istream'
219 * to the output stream `ostream', removing any comments and checking
220 * for bad use of literate script features:
221 * - there should be at least one BLANK line between a DEFN and TEXT
222 * - there should be at least one DEFN line in a script.
225 unlit(file, istream, ostream)
229 line last, this=START;
235 this = readline(istream, ostream);
239 if (last==DEFN && this==TEXT)
240 complain(file, linesread-1, MISSINGBLANK);
241 if (last==TEXT && this==DEFN)
242 complain(file, linesread, MISSINGBLANK);
244 /* start of code, copy to end */
247 if (fgets(lineb, sizeof lineb, istream) == NULL) {
248 complain(file, linesread, MISSINGENDCODE);
252 if (strncmp(lineb,ENDCODE,LENENDCODE) == 0) {
256 fputs(lineb, ostream);
261 if (this == PSEUDO) {
264 if (fgets(lineb, sizeof lineb, istream) == NULL) {
265 complain(file, linesread, MISSINGENDPSEUDOCODE);
270 if (strncmp(lineb,ENDPSEUDOCODE,LENENDPSEUDOCODE) == 0) {
279 complain(file,linesread,EMPTYSCRIPT);
284 * Main program. Processes command line arguments, looking for leading:
285 * -q quiet mode - do not complain about bad literate script files
286 * -n noisy mode - complain about bad literate script files.
287 * -r remove cpp droppings in output.
288 * Expects two additional arguments, a file name for the input and a file
289 * name for the output file. These two names must normally be distinct.
290 * An exception is made for the special name "-" which can be used in either
291 * position to specify the standard input or the standard output respectively.
297 FILE *istream, *ostream;
300 for (argc--, argv++; argc > 0; argc--, argv++)
301 if (strcmp(*argv,"-n")==0)
303 else if (strcmp(*argv,"-q")==0)
305 else if (strcmp(*argv,"-c")==0)
307 else if (strcmp(*argv,"-h")==0) {
312 prefix_str = (char*)malloc(sizeof(char)*(1+strlen(*argv)));
314 strcpy(prefix_str, *argv);
316 } else if (strcmp(*argv,"-#")==0)
322 fprintf(stderr, USAGE);
326 if (strcmp(argv[0],argv[1])==0 && strcmp(argv[0],"-")!=0) {
327 fprintf(stderr, DISTINCTNAMES);
332 if (strcmp(argv[0], "-")==0) {
337 if ((istream=fopen(argv[0], "r")) == NULL) {
338 fprintf(stderr, CANNOTOPEN, argv[0]);
342 if (strcmp(argv[1], "-")==0)
345 if ((ostream=fopen(argv[1], "w")) == NULL) {
346 fprintf(stderr, CANNOTOPEN, argv[1]);
350 /* Prefix the output with line pragmas */
352 fprintf(ostream, "#line 1 \"%s\"\n{-# LINE 1 \"%s\" #-}\n", prefix_str, prefix_str);
355 unlit(file, istream, ostream);
357 if (istream != stdin) fclose(istream);
358 if (ostream != stdout) fclose(ostream);
360 exit(errors==0 ? 0 : 1);