![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to indent.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [CSRG BSD Unix] / 43BSDReno / pgrm / indent|
Return to indent.c CVS log | Up to [CSRG BSD Unix] / 43BSDReno / pgrm / indent |
1.1 root 1: /*
2: * Copyright (c) 1985 Sun Microsystems, Inc.
3: * Copyright (c) 1980 The Regents of the University of California.
4: * Copyright (c) 1976 Board of Trustees of the University of Illinois.
5: * All rights reserved.
6: *
7: * Redistribution and use in source and binary forms are permitted
8: * provided that: (1) source distributions retain this entire copyright
9: * notice and comment, and (2) distributions including binaries display
10: * the following acknowledgement: ``This product includes software
11: * developed by the University of California, Berkeley and its contributors''
12: * in the documentation or other materials provided with the distribution
13: * and in all advertising materials mentioning features or use of this
14: * software. Neither the name of the University nor the names of its
15: * contributors may be used to endorse or promote products derived
16: * from this software without specific prior written permission.
17: * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
18: * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED
19: * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
20: */
21:
22: #ifndef lint
23: char copyright[] =
24: "@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\
25: @(#) Copyright (c) 1980 The Regents of the University of California.\n\
26: @(#) Copyright (c) 1976 Board of Trustees of the University of Illinois.\n\
27: All rights reserved.\n";
28: #endif /* not lint */
29:
30: #ifndef lint
31: static char sccsid[] = "@(#)indent.c 5.15 (Berkeley) 6/1/90";
32: #endif /* not lint */
33:
34: #include <sys/param.h>
35: #include "indent_globs.h"
36: #include "indent_codes.h"
37: #include <ctype.h>
38:
39: char *in_name = "Standard Input"; /* will always point to name of input
40: * file */
41: char *out_name = "Standard Output"; /* will always point to name
42: * of output file */
43: char bakfile[MAXPATHLEN] = "";
44:
45: main(argc, argv)
46: int argc;
47: char **argv;
48: {
49:
50: extern int found_err; /* flag set in diag() on error */
51: int dec_ind; /* current indentation for declarations */
52: int di_stack[20]; /* a stack of structure indentation levels */
53: int flushed_nl; /* used when buffering up comments to remember
54: * that a newline was passed over */
55: int force_nl; /* when true, code must be broken */
56: int hd_type; /* used to store type of stmt for if (...),
57: * for (...), etc */
58: register int i; /* local loop counter */
59: int scase; /* set to true when we see a case, so we will
60: * know what to do with the following colon */
61: int sp_sw; /* when true, we are in the expressin of
62: * if(...), while(...), etc. */
63: int squest; /* when this is positive, we have seen a ?
64: * without the matching : in a <c>?<s>:<s>
65: * construct */
66: register char *t_ptr; /* used for copying tokens */
67: int type_code; /* the type of token, returned by lexi */
68:
69: int last_else = 0; /* true iff last keyword was an else */
70:
71:
72: /*-----------------------------------------------*\
73: | INITIALIZATION |
74: \*-----------------------------------------------*/
75:
76:
77: ps.p_stack[0] = stmt; /* this is the parser's stack */
78: ps.last_nl = true; /* this is true if the last thing scanned was
79: * a newline */
80: ps.last_token = semicolon;
81: combuf = (char *) malloc(bufsize);
82: labbuf = (char *) malloc(bufsize);
83: codebuf = (char *) malloc(bufsize);
84: tokenbuf = (char *) malloc(bufsize);
85: l_com = combuf + bufsize - 5;
86: l_lab = labbuf + bufsize - 5;
87: l_code = codebuf + bufsize - 5;
88: l_token = tokenbuf + bufsize - 5;
89: combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
90: * comment buffers */
91: combuf[1] = codebuf[1] = labbuf[1] = '\0';
92: ps.else_if = 1; /* Default else-if special processing to on */
93: s_lab = e_lab = labbuf + 1;
94: s_code = e_code = codebuf + 1;
95: s_com = e_com = combuf + 1;
96: s_token = e_token = tokenbuf + 1;
97:
98: in_buffer = (char *) malloc(10);
99: in_buffer_limit = in_buffer + 8;
100: buf_ptr = buf_end = in_buffer;
101: line_no = 1;
102: had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
103: sp_sw = force_nl = false;
104: ps.in_or_st = false;
105: ps.bl_line = true;
106: dec_ind = 0;
107: di_stack[ps.dec_nest = 0] = 0;
108: ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
109:
110:
111: scase = ps.pcase = false;
112: squest = 0;
113: sc_end = 0;
114: bp_save = 0;
115: be_save = 0;
116:
117: output = 0;
118:
119:
120:
121: /*--------------------------------------------------*\
122: | COMMAND LINE SCAN |
123: \*--------------------------------------------------*/
124:
125: #ifdef undef
126: max_col = 78; /* -l78 */
127: lineup_to_parens = 1; /* -lp */
128: ps.ljust_decl = 0; /* -ndj */
129: ps.com_ind = 33; /* -c33 */
130: star_comment_cont = 1; /* -sc */
131: ps.ind_size = 8; /* -i8 */
132: verbose = 0;
133: ps.decl_indent = 16; /* -di16 */
134: ps.indent_parameters = 1; /* -ip */
135: ps.decl_com_ind = 0; /* if this is not set to some positive value
136: * by an arg, we will set this equal to
137: * ps.com_ind */
138: btype_2 = 1; /* -br */
139: cuddle_else = 1; /* -ce */
140: ps.unindent_displace = 0; /* -d0 */
141: ps.case_indent = 0; /* -cli0 */
142: format_col1_comments = 1; /* -fc1 */
143: procnames_start_line = 1; /* -psl */
144: proc_calls_space = 0; /* -npcs */
145: comment_delimiter_on_blankline = 1; /* -cdb */
146: ps.leave_comma = 1; /* -nbc */
147: #endif
148:
149: for (i = 1; i < argc; ++i)
150: if (strcmp(argv[i], "-npro") == 0)
151: break;
152: set_defaults();
153: if (i >= argc)
154: set_profile();
155:
156: for (i = 1; i < argc; ++i) {
157:
158: /*
159: * look thru args (if any) for changes to defaults
160: */
161: if (argv[i][0] != '-') {/* no flag on parameter */
162: if (input == 0) { /* we must have the input file */
163: in_name = argv[i]; /* remember name of input file */
164: input = fopen(in_name, "r");
165: if (input == 0) /* check for open error */
166: err(in_name);
167: continue;
168: }
169: else if (output == 0) { /* we have the output file */
170: out_name = argv[i]; /* remember name of output file */
171: if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite
172: * the file */
173: fprintf(stderr, "indent: input and output files must be different\n");
174: exit(1);
175: }
176: output = fopen(out_name, "w");
177: if (output == 0) /* check for create error */
178: err(out_name);
179: continue;
180: }
181: fprintf(stderr, "indent: unknown parameter: %s\n", argv[i]);
182: exit(1);
183: }
184: else
185: set_option(argv[i]);
186: } /* end of for */
187: if (input == 0) {
188: fprintf(stderr, "indent: usage: indent file [ outfile ] [ options ]\n");
189: exit(1);
190: }
191: if (output == 0)
192: if (troff)
193: output = stdout;
194: else {
195: out_name = in_name;
196: bakcopy();
197: }
198: if (ps.com_ind <= 1)
199: ps.com_ind = 2; /* dont put normal comments before column 2 */
200: if (troff) {
201: if (bodyf.font[0] == 0)
202: parsefont(&bodyf, "R");
203: if (scomf.font[0] == 0)
204: parsefont(&scomf, "I");
205: if (blkcomf.font[0] == 0)
206: blkcomf = scomf, blkcomf.size += 2;
207: if (boxcomf.font[0] == 0)
208: boxcomf = blkcomf;
209: if (stringf.font[0] == 0)
210: parsefont(&stringf, "L");
211: if (keywordf.font[0] == 0)
212: parsefont(&keywordf, "B");
213: writefdef(&bodyf, 'B');
214: writefdef(&scomf, 'C');
215: writefdef(&blkcomf, 'L');
216: writefdef(&boxcomf, 'X');
217: writefdef(&stringf, 'S');
218: writefdef(&keywordf, 'K');
219: }
220: if (block_comment_max_col <= 0)
221: block_comment_max_col = max_col;
222: if (ps.decl_com_ind <= 0) /* if not specified by user, set this */
223: ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
224: if (continuation_indent == 0)
225: continuation_indent = ps.ind_size;
226: fill_buffer(); /* get first batch of stuff into input buffer */
227:
228: parse(semicolon);
229: {
230: register char *p = buf_ptr;
231: register col = 1;
232:
233: while (1) {
234: if (*p == ' ')
235: col++;
236: else if (*p == '\t')
237: col = ((col - 1) & ~7) + 9;
238: else
239: break;
240: p++;
241: }
242: if (col > ps.ind_size)
243: ps.ind_level = ps.i_l_follow = col / ps.ind_size;
244: }
245: if (troff) {
246: register char *p = in_name,
247: *beg = in_name;
248:
249: while (*p)
250: if (*p++ == '/')
251: beg = p;
252: fprintf(output, ".Fn \"%s\"\n", beg);
253: }
254: /*
255: * START OF MAIN LOOP
256: */
257:
258: while (1) { /* this is the main loop. it will go until we
259: * reach eof */
260: int is_procname;
261:
262: type_code = lexi(); /* lexi reads one token. The actual
263: * characters read are stored in "token". lexi
264: * returns a code indicating the type of token */
265: is_procname = ps.procname[0];
266:
267: /*
268: * The following code moves everything following an if (), while (),
269: * else, etc. up to the start of the following stmt to a buffer. This
270: * allows proper handling of both kinds of brace placement.
271: */
272:
273: flushed_nl = false;
274: while (ps.search_brace) { /* if we scanned an if(), while(),
275: * etc., we might need to copy stuff
276: * into a buffer we must loop, copying
277: * stuff into save_com, until we find
278: * the start of the stmt which follows
279: * the if, or whatever */
280: switch (type_code) {
281: case newline:
282: ++line_no;
283: flushed_nl = true;
284: case form_feed:
285: break; /* form feeds and newlines found here will be
286: * ignored */
287:
288: case lbrace: /* this is a brace that starts the compound
289: * stmt */
290: if (sc_end == 0) { /* ignore buffering if a comment wasnt
291: * stored up */
292: ps.search_brace = false;
293: goto check_type;
294: }
295: if (btype_2) {
296: save_com[0] = '{'; /* we either want to put the brace
297: * right after the if */
298: goto sw_buffer; /* go to common code to get out of
299: * this loop */
300: }
301: case comment: /* we have a comment, so we must copy it into
302: * the buffer */
303: if (!flushed_nl || sc_end != 0) {
304: if (sc_end == 0) { /* if this is the first comment, we
305: * must set up the buffer */
306: save_com[0] = save_com[1] = ' ';
307: sc_end = &(save_com[2]);
308: }
309: else {
310: *sc_end++ = '\n'; /* add newline between
311: * comments */
312: *sc_end++ = ' ';
313: --line_no;
314: }
315: *sc_end++ = '/'; /* copy in start of comment */
316: *sc_end++ = '*';
317:
318: for (;;) { /* loop until we get to the end of the comment */
319: *sc_end = *buf_ptr++;
320: if (buf_ptr >= buf_end)
321: fill_buffer();
322:
323: if (*sc_end++ == '*' && *buf_ptr == '/')
324: break; /* we are at end of comment */
325:
326: if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer
327: * overflow */
328: diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
329: fflush(output);
330: exit(1);
331: }
332: }
333: *sc_end++ = '/'; /* add ending slash */
334: if (++buf_ptr >= buf_end) /* get past / in buffer */
335: fill_buffer();
336: break;
337: }
338: default: /* it is the start of a normal statment */
339: if (flushed_nl) /* if we flushed a newline, make sure it is
340: * put back */
341: force_nl = true;
342: if (type_code == sp_paren && *token == 'i'
343: && last_else && ps.else_if
344: || type_code == sp_nparen && *token == 'e'
345: && e_code != s_code && e_code[-1] == '}')
346: force_nl = false;
347:
348: if (sc_end == 0) { /* ignore buffering if comment wasnt
349: * saved up */
350: ps.search_brace = false;
351: goto check_type;
352: }
353: if (force_nl) { /* if we should insert a nl here, put it into
354: * the buffer */
355: force_nl = false;
356: --line_no; /* this will be re-increased when the nl is
357: * read from the buffer */
358: *sc_end++ = '\n';
359: *sc_end++ = ' ';
360: if (verbose && !flushed_nl) /* print error msg if the line
361: * was not already broken */
362: diag(0, "Line broken");
363: flushed_nl = false;
364: }
365: for (t_ptr = token; *t_ptr; ++t_ptr)
366: *sc_end++ = *t_ptr; /* copy token into temp buffer */
367: ps.procname[0] = 0;
368:
369: sw_buffer:
370: ps.search_brace = false; /* stop looking for start of
371: * stmt */
372: bp_save = buf_ptr; /* save current input buffer */
373: be_save = buf_end;
374: buf_ptr = save_com; /* fix so that subsequent calls to
375: * lexi will take tokens out of
376: * save_com */
377: *sc_end++ = ' ';/* add trailing blank, just in case */
378: buf_end = sc_end;
379: sc_end = 0;
380: break;
381: } /* end of switch */
382: if (type_code != 0) /* we must make this check, just in case there
383: * was an unexpected EOF */
384: type_code = lexi(); /* read another token */
385: /* if (ps.search_brace) ps.procname[0] = 0; */
386: if ((is_procname = ps.procname[0]) && flushed_nl
387: && !procnames_start_line && ps.in_decl
388: && type_code == ident)
389: flushed_nl = 0;
390: } /* end of while (search_brace) */
391: last_else = 0;
392: check_type:
393: if (type_code == 0) { /* we got eof */
394: if (s_lab != e_lab || s_code != e_code
395: || s_com != e_com) /* must dump end of line */
396: dump_line();
397: if (ps.tos > 1) /* check for balanced braces */
398: diag(1, "Stuff missing from end of file.");
399:
400: if (verbose) {
401: printf("There were %d output lines and %d comments\n",
402: ps.out_lines, ps.out_coms);
403: printf("(Lines with comments)/(Lines with code): %6.3f\n",
404: (1.0 * ps.com_lines) / code_lines);
405: }
406: fflush(output);
407: exit(found_err);
408: }
409: if (
410: (type_code != comment) &&
411: (type_code != newline) &&
412: (type_code != preesc) &&
413: (type_code != form_feed)) {
414: if (force_nl &&
415: (type_code != semicolon) &&
416: (type_code != lbrace || !btype_2)) {
417: /* we should force a broken line here */
418: if (verbose && !flushed_nl)
419: diag(0, "Line broken");
420: flushed_nl = false;
421: dump_line();
422: ps.want_blank = false; /* dont insert blank at line start */
423: force_nl = false;
424: }
425: ps.in_stmt = true; /* turn on flag which causes an extra level of
426: * indentation. this is turned off by a ; or
427: * '}' */
428: if (s_com != e_com) { /* the turkey has embedded a comment
429: * in a line. fix it */
430: *e_code++ = ' ';
431: for (t_ptr = s_com; *t_ptr; ++t_ptr) {
432: CHECK_SIZE_CODE;
433: *e_code++ = *t_ptr;
434: }
435: *e_code++ = ' ';
436: *e_code = '\0'; /* null terminate code sect */
437: ps.want_blank = false;
438: e_com = s_com;
439: }
440: }
441: else if (type_code != comment) /* preserve force_nl thru a comment */
442: force_nl = false; /* cancel forced newline after newline, form
443: * feed, etc */
444:
445:
446:
447: /*-----------------------------------------------------*\
448: | do switch on type of token scanned |
449: \*-----------------------------------------------------*/
450: CHECK_SIZE_CODE;
451: switch (type_code) { /* now, decide what to do with the token */
452:
453: case form_feed: /* found a form feed in line */
454: ps.use_ff = true; /* a form feed is treated much like a newline */
455: dump_line();
456: ps.want_blank = false;
457: break;
458:
459: case newline:
460: if (ps.last_token != comma || ps.p_l_follow > 0
461: || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
462: dump_line();
463: ps.want_blank = false;
464: }
465: ++line_no; /* keep track of input line number */
466: break;
467:
468: case lparen: /* got a '(' or '[' */
469: ++ps.p_l_follow; /* count parens to make Healy happy */
470: if (ps.want_blank && *token != '[' &&
471: (ps.last_token != ident || proc_calls_space
472: || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
473: *e_code++ = ' ';
474: if (ps.in_decl && !ps.block_init)
475: if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
476: ps.dumped_decl_indent = 1;
477: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
478: e_code += strlen(e_code);
479: }
480: else {
481: while ((e_code - s_code) < dec_ind) {
482: CHECK_SIZE_CODE;
483: *e_code++ = ' ';
484: }
485: *e_code++ = token[0];
486: }
487: else
488: *e_code++ = token[0];
489: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
490: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
491: && ps.paren_indents[0] < 2 * ps.ind_size)
492: ps.paren_indents[0] = 2 * ps.ind_size;
493: ps.want_blank = false;
494: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
495: /*
496: * this is a kluge to make sure that declarations will be
497: * aligned right if proc decl has an explicit type on it, i.e.
498: * "int a(x) {..."
499: */
500: parse(semicolon); /* I said this was a kluge... */
501: ps.in_or_st = false; /* turn off flag for structure decl or
502: * initialization */
503: }
504: if (ps.sizeof_keyword)
505: ps.sizeof_mask |= 1 << ps.p_l_follow;
506: break;
507:
508: case rparen: /* got a ')' or ']' */
509: rparen_count--;
510: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
511: ps.last_u_d = true;
512: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
513: }
514: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
515: if (--ps.p_l_follow < 0) {
516: ps.p_l_follow = 0;
517: diag(0, "Extra %c", *token);
518: }
519: if (e_code == s_code) /* if the paren starts the line */
520: ps.paren_level = ps.p_l_follow; /* then indent it */
521:
522: *e_code++ = token[0];
523: ps.want_blank = true;
524:
525: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
526: * (...), or some such */
527: sp_sw = false;
528: force_nl = true;/* must force newline after if */
529: ps.last_u_d = true; /* inform lexi that a following
530: * operator is unary */
531: ps.in_stmt = false; /* dont use stmt continuation
532: * indentation */
533:
534: parse(hd_type); /* let parser worry about if, or whatever */
535: }
536: ps.search_brace = btype_2; /* this should insure that constructs
537: * such as main(){...} and int[]{...}
538: * have their braces put in the right
539: * place */
540: break;
541:
542: case unary_op: /* this could be any unary operation */
543: if (ps.want_blank)
544: *e_code++ = ' ';
545:
546: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
547: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
548: ps.dumped_decl_indent = 1;
549: e_code += strlen(e_code);
550: }
551: else {
552: char *res = token;
553:
554: if (ps.in_decl && !ps.block_init) { /* if this is a unary op
555: * in a declaration, we
556: * should indent this
557: * token */
558: for (i = 0; token[i]; ++i); /* find length of token */
559: while ((e_code - s_code) < (dec_ind - i)) {
560: CHECK_SIZE_CODE;
561: *e_code++ = ' '; /* pad it */
562: }
563: }
564: if (troff && token[0] == '-' && token[1] == '>')
565: res = "\\(->";
566: for (t_ptr = res; *t_ptr; ++t_ptr) {
567: CHECK_SIZE_CODE;
568: *e_code++ = *t_ptr;
569: }
570: }
571: ps.want_blank = false;
572: break;
573:
574: case binary_op: /* any binary operation */
575: if (ps.want_blank)
576: *e_code++ = ' ';
577: {
578: char *res = token;
579:
580: if (troff)
581: switch (token[0]) {
582: case '<':
583: if (token[1] == '=')
584: res = "\\(<=";
585: break;
586: case '>':
587: if (token[1] == '=')
588: res = "\\(>=";
589: break;
590: case '!':
591: if (token[1] == '=')
592: res = "\\(!=";
593: break;
594: case '|':
595: if (token[1] == '|')
596: res = "\\(br\\(br";
597: else if (token[1] == 0)
598: res = "\\(br";
599: break;
600: }
601: for (t_ptr = res; *t_ptr; ++t_ptr) {
602: CHECK_SIZE_CODE;
603: *e_code++ = *t_ptr; /* move the operator */
604: }
605: }
606: ps.want_blank = true;
607: break;
608:
609: case postop: /* got a trailing ++ or -- */
610: *e_code++ = token[0];
611: *e_code++ = token[1];
612: ps.want_blank = true;
613: break;
614:
615: case question: /* got a ? */
616: squest++; /* this will be used when a later colon
617: * appears so we can distinguish the
618: * <c>?<n>:<n> construct */
619: if (ps.want_blank)
620: *e_code++ = ' ';
621: *e_code++ = '?';
622: ps.want_blank = true;
623: break;
624:
625: case casestmt: /* got word 'case' or 'default' */
626: scase = true; /* so we can process the later colon properly */
627: goto copy_id;
628:
629: case colon: /* got a ':' */
630: if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
631: --squest;
632: if (ps.want_blank)
633: *e_code++ = ' ';
634: *e_code++ = ':';
635: ps.want_blank = true;
636: break;
637: }
638: if (ps.in_decl) {
639: *e_code++ = ':';
640: ps.want_blank = false;
641: break;
642: }
643: ps.in_stmt = false; /* seeing a label does not imply we are in a
644: * stmt */
645: for (t_ptr = s_code; *t_ptr; ++t_ptr)
646: *e_lab++ = *t_ptr; /* turn everything so far into a label */
647: e_code = s_code;
648: *e_lab++ = ':';
649: *e_lab++ = ' ';
650: *e_lab = '\0';
651:
652: force_nl = ps.pcase = scase; /* ps.pcase will be used by
653: * dump_line to decide how to
654: * indent the label. force_nl
655: * will force a case n: to be
656: * on a line by itself */
657: scase = false;
658: ps.want_blank = false;
659: break;
660:
661: case semicolon: /* got a ';' */
662: ps.in_or_st = false;/* we are not in an initialization or
663: * structure declaration */
664: scase = false; /* these will only need resetting in a error */
665: squest = 0;
666: if (ps.last_token == rparen && rparen_count == 0)
667: ps.in_parameter_declaration = 0;
668: ps.cast_mask = 0;
669: ps.sizeof_mask = 0;
670: ps.block_init = 0;
671: ps.block_init_level = 0;
672: ps.just_saw_decl--;
673:
674: if (ps.in_decl && s_code == e_code && !ps.block_init)
675: while ((e_code - s_code) < (dec_ind - 1)) {
676: CHECK_SIZE_CODE;
677: *e_code++ = ' ';
678: }
679:
680: ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
681: * structure declaration, we
682: * arent any more */
683:
684: if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
685:
686: /*
687: * This should be true iff there were unbalanced parens in the
688: * stmt. It is a bit complicated, because the semicolon might
689: * be in a for stmt
690: */
691: diag(1, "Unbalanced parens");
692: ps.p_l_follow = 0;
693: if (sp_sw) { /* this is a check for a if, while, etc. with
694: * unbalanced parens */
695: sp_sw = false;
696: parse(hd_type); /* dont lose the if, or whatever */
697: }
698: }
699: *e_code++ = ';';
700: ps.want_blank = true;
701: ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
702: * middle of a stmt */
703:
704: if (!sp_sw) { /* if not if for (;;) */
705: parse(semicolon); /* let parser know about end of stmt */
706: force_nl = true;/* force newline after a end of stmt */
707: }
708: break;
709:
710: case lbrace: /* got a '{' */
711: ps.in_stmt = false; /* dont indent the {} */
712: if (!ps.block_init)
713: force_nl = true;/* force other stuff on same line as '{' onto
714: * new line */
715: else if (ps.block_init_level <= 0)
716: ps.block_init_level = 1;
717: else
718: ps.block_init_level++;
719:
720: if (s_code != e_code && !ps.block_init) {
721: if (!btype_2) {
722: dump_line();
723: ps.want_blank = false;
724: }
725: else if (ps.in_parameter_declaration && !ps.in_or_st) {
726: ps.i_l_follow = 0;
727: dump_line();
728: ps.want_blank = false;
729: }
730: }
731: if (ps.in_parameter_declaration)
732: prefix_blankline_requested = 0;
733:
734: if (ps.p_l_follow > 0) { /* check for preceeding unbalanced
735: * parens */
736: diag(1, "Unbalanced parens");
737: ps.p_l_follow = 0;
738: if (sp_sw) { /* check for unclosed if, for, etc. */
739: sp_sw = false;
740: parse(hd_type);
741: ps.ind_level = ps.i_l_follow;
742: }
743: }
744: if (s_code == e_code)
745: ps.ind_stmt = false; /* dont put extra indentation on line
746: * with '{' */
747: if (ps.in_decl && ps.in_or_st) { /* this is either a structure
748: * declaration or an init */
749: di_stack[ps.dec_nest++] = dec_ind;
750: /* ? dec_ind = 0; */
751: }
752: else {
753: ps.decl_on_line = false; /* we cant be in the middle of
754: * a declaration, so dont do
755: * special indentation of
756: * comments */
757: if (blanklines_after_declarations_at_proctop
758: && ps.in_parameter_declaration)
759: postfix_blankline_requested = 1;
760: ps.in_parameter_declaration = 0;
761: }
762: dec_ind = 0;
763: parse(lbrace); /* let parser know about this */
764: if (ps.want_blank) /* put a blank before '{' if '{' is not at
765: * start of line */
766: *e_code++ = ' ';
767: ps.want_blank = false;
768: *e_code++ = '{';
769: ps.just_saw_decl = 0;
770: break;
771:
772: case rbrace: /* got a '}' */
773: if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
774: * omitted in
775: * declarations */
776: parse(semicolon);
777: if (ps.p_l_follow) {/* check for unclosed if, for, else. */
778: diag(1, "Unbalanced parens");
779: ps.p_l_follow = 0;
780: sp_sw = false;
781: }
782: ps.just_saw_decl = 0;
783: ps.block_init_level--;
784: if (s_code != e_code && !ps.block_init) { /* '}' must be first on
785: * line */
786: if (verbose)
787: diag(0, "Line broken");
788: dump_line();
789: }
790: *e_code++ = '}';
791: ps.want_blank = true;
792: ps.in_stmt = ps.ind_stmt = false;
793: if (ps.dec_nest > 0) { /* we are in multi-level structure
794: * declaration */
795: dec_ind = di_stack[--ps.dec_nest];
796: if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
797: ps.just_saw_decl = 2;
798: ps.in_decl = true;
799: }
800: prefix_blankline_requested = 0;
801: parse(rbrace); /* let parser know about this */
802: ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
803: && ps.il[ps.tos] >= ps.ind_level;
804: if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
805: postfix_blankline_requested = 1;
806: break;
807:
808: case swstmt: /* got keyword "switch" */
809: sp_sw = true;
810: hd_type = swstmt; /* keep this for when we have seen the
811: * expression */
812: goto copy_id; /* go move the token into buffer */
813:
814: case sp_paren: /* token is if, while, for */
815: sp_sw = true; /* the interesting stuff is done after the
816: * expression is scanned */
817: hd_type = (*token == 'i' ? ifstmt :
818: (*token == 'w' ? whilestmt : forstmt));
819:
820: /*
821: * remember the type of header for later use by parser
822: */
823: goto copy_id; /* copy the token into line */
824:
825: case sp_nparen: /* got else, do */
826: ps.in_stmt = false;
827: if (*token == 'e') {
828: if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
829: if (verbose)
830: diag(0, "Line broken");
831: dump_line();/* make sure this starts a line */
832: ps.want_blank = false;
833: }
834: force_nl = true;/* also, following stuff must go onto new line */
835: last_else = 1;
836: parse(elselit);
837: }
838: else {
839: if (e_code != s_code) { /* make sure this starts a line */
840: if (verbose)
841: diag(0, "Line broken");
842: dump_line();
843: ps.want_blank = false;
844: }
845: force_nl = true;/* also, following stuff must go onto new line */
846: last_else = 0;
847: parse(dolit);
848: }
849: goto copy_id; /* move the token into line */
850:
851: case decl: /* we have a declaration type (int, register,
852: * etc.) */
853: parse(decl); /* let parser worry about indentation */
854: if (ps.last_token == rparen && ps.tos <= 1) {
855: ps.in_parameter_declaration = 1;
856: if (s_code != e_code) {
857: dump_line();
858: ps.want_blank = 0;
859: }
860: }
861: if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
862: ps.ind_level = ps.i_l_follow = 1;
863: ps.ind_stmt = 0;
864: }
865: ps.in_or_st = true; /* this might be a structure or initialization
866: * declaration */
867: ps.in_decl = ps.decl_on_line = true;
868: if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
869: ps.just_saw_decl = 2;
870: prefix_blankline_requested = 0;
871: for (i = 0; token[i++];); /* get length of token */
872:
873: /*
874: * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
875: * : i);
876: */
877: dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
878: goto copy_id;
879:
880: case ident: /* got an identifier or constant */
881: if (ps.in_decl) { /* if we are in a declaration, we must indent
882: * identifier */
883: if (ps.want_blank)
884: *e_code++ = ' ';
885: ps.want_blank = false;
886: if (is_procname == 0 || !procnames_start_line) {
887: if (!ps.block_init)
888: if (troff && !ps.dumped_decl_indent) {
889: sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7);
890: ps.dumped_decl_indent = 1;
891: e_code += strlen(e_code);
892: }
893: else
894: while ((e_code - s_code) < dec_ind) {
895: CHECK_SIZE_CODE;
896: *e_code++ = ' ';
897: }
898: }
899: else {
900: if (dec_ind && s_code != e_code)
901: dump_line();
902: dec_ind = 0;
903: ps.want_blank = false;
904: }
905: }
906: else if (sp_sw && ps.p_l_follow == 0) {
907: sp_sw = false;
908: force_nl = true;
909: ps.last_u_d = true;
910: ps.in_stmt = false;
911: parse(hd_type);
912: }
913: copy_id:
914: if (ps.want_blank)
915: *e_code++ = ' ';
916: if (troff && ps.its_a_keyword) {
917: e_code = chfont(&bodyf, &keywordf, e_code);
918: for (t_ptr = token; *t_ptr; ++t_ptr) {
919: CHECK_SIZE_CODE;
920: *e_code++ = keywordf.allcaps && islower(*t_ptr)
921: ? toupper(*t_ptr) : *t_ptr;
922: }
923: e_code = chfont(&keywordf, &bodyf, e_code);
924: }
925: else
926: for (t_ptr = token; *t_ptr; ++t_ptr) {
927: CHECK_SIZE_CODE;
928: *e_code++ = *t_ptr;
929: }
930: ps.want_blank = true;
931: break;
932:
933: case period: /* treat a period kind of like a binary
934: * operation */
935: *e_code++ = '.'; /* move the period into line */
936: ps.want_blank = false; /* dont put a blank after a period */
937: break;
938:
939: case comma:
940: ps.want_blank = (s_code != e_code); /* only put blank after comma
941: * if comma does not start the
942: * line */
943: if (ps.in_decl && is_procname == 0 && !ps.block_init)
944: while ((e_code - s_code) < (dec_ind - 1)) {
945: CHECK_SIZE_CODE;
946: *e_code++ = ' ';
947: }
948:
949: *e_code++ = ',';
950: if (ps.p_l_follow == 0) {
951: if (ps.block_init_level <= 0)
952: ps.block_init = 0;
953: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
954: force_nl = true;
955: }
956: break;
957:
958: case preesc: /* got the character '#' */
959: if ((s_com != e_com) ||
960: (s_lab != e_lab) ||
961: (s_code != e_code))
962: dump_line();
963: *e_lab++ = '#'; /* move whole line to 'label' buffer */
964: {
965: int in_comment = 0;
966: int com_start = 0;
967: char quote = 0;
968: int com_end = 0;
969:
970: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
971: buf_ptr++;
972: if (buf_ptr >= buf_end)
973: fill_buffer();
974: }
975: while (*buf_ptr != '\n' || in_comment) {
976: CHECK_SIZE_LAB;
977: *e_lab = *buf_ptr++;
978: if (buf_ptr >= buf_end)
979: fill_buffer();
980: switch (*e_lab++) {
981: case BACKSLASH:
982: if (troff)
983: *e_lab++ = BACKSLASH;
984: if (!in_comment) {
985: *e_lab++ = *buf_ptr++;
986: if (buf_ptr >= buf_end)
987: fill_buffer();
988: }
989: break;
990: case '/':
991: if (*buf_ptr == '*' && !in_comment && !quote) {
992: in_comment = 1;
993: *e_lab++ = *buf_ptr++;
994: com_start = e_lab - s_lab - 2;
995: }
996: break;
997: case '"':
998: if (quote == '"')
999: quote = 0;
1000: break;
1001: case '\'':
1002: if (quote == '\'')
1003: quote = 0;
1004: break;
1005: case '*':
1006: if (*buf_ptr == '/' && in_comment) {
1007: in_comment = 0;
1008: *e_lab++ = *buf_ptr++;
1009: com_end = e_lab - s_lab;
1010: }
1011: break;
1012: }
1013: }
1014:
1015: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1016: e_lab--;
1017: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1018: * preprocessor line */
1019: if (sc_end == 0) /* if this is the first comment, we
1020: * must set up the buffer */
1021: sc_end = &(save_com[0]);
1022: else {
1023: *sc_end++ = '\n'; /* add newline between
1024: * comments */
1025: *sc_end++ = ' ';
1026: --line_no;
1027: }
1028: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1029: sc_end += com_end - com_start;
1030: if (sc_end >= &save_com[sc_size])
1031: abort();
1032: e_lab = s_lab + com_start;
1033: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1034: e_lab--;
1035: bp_save = buf_ptr; /* save current input buffer */
1036: be_save = buf_end;
1037: buf_ptr = save_com; /* fix so that subsequent calls to
1038: * lexi will take tokens out of
1039: * save_com */
1040: *sc_end++ = ' '; /* add trailing blank, just in case */
1041: buf_end = sc_end;
1042: sc_end = 0;
1043: }
1044: *e_lab = '\0'; /* null terminate line */
1045: ps.pcase = false;
1046: }
1047:
1048: if (strncmp(s_lab, "#if", 3) == 0) {
1049: if (blanklines_around_conditional_compilation) {
1050: register c;
1051: prefix_blankline_requested++;
1052: while ((c = getc(input)) == '\n');
1053: ungetc(c, input);
1054: }
1055: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1056: match_state[ifdef_level].tos = -1;
1057: state_stack[ifdef_level++] = ps;
1058: }
1059: else
1060: diag(1, "#if stack overflow");
1061: }
1062: else if (strncmp(s_lab, "#else", 5) == 0)
1063: if (ifdef_level <= 0)
1064: diag(1, "Unmatched #else");
1065: else {
1066: match_state[ifdef_level - 1] = ps;
1067: ps = state_stack[ifdef_level - 1];
1068: }
1069: else if (strncmp(s_lab, "#endif", 6) == 0) {
1070: if (ifdef_level <= 0)
1071: diag(1, "Unmatched #endif");
1072: else {
1073: ifdef_level--;
1074:
1075: #ifdef undef
1076: /*
1077: * This match needs to be more intelligent before the
1078: * message is useful
1079: */
1080: if (match_state[ifdef_level].tos >= 0
1081: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1082: diag(0, "Syntactically inconsistant #ifdef alternatives.");
1083: #endif
1084: }
1085: if (blanklines_around_conditional_compilation) {
1086: postfix_blankline_requested++;
1087: n_real_blanklines = 0;
1088: }
1089: }
1090: break; /* subsequent processing of the newline
1091: * character will cause the line to be printed */
1092:
1093: case comment: /* we have gotten a /* this is a biggie */
1094: if (flushed_nl) { /* we should force a broken line here */
1095: flushed_nl = false;
1096: dump_line();
1097: ps.want_blank = false; /* dont insert blank at line start */
1098: force_nl = false;
1099: }
1100: pr_comment();
1101: break;
1102: } /* end of big switch stmt */
1103:
1104: *e_code = '\0'; /* make sure code section is null terminated */
1105: if (type_code != comment && type_code != newline && type_code != preesc)
1106: ps.last_token = type_code;
1107: } /* end of main while (1) loop */
1108: }
1109:
1110: /*
1111: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1112: * backup file will be ".Bfile" then make the backup file the input and
1113: * original input file the output
1114: */
1115: bakcopy()
1116: {
1117: int n,
1118: bakchn;
1119: char buff[8 * 1024];
1120: register char *p;
1121:
1122: /* construct file name .Bfile */
1123: for (p = in_name; *p; p++); /* skip to end of string */
1124: while (p > in_name && *p != '/') /* find last '/' */
1125: p--;
1126: if (*p == '/')
1127: p++;
1128: sprintf(bakfile, "%s.BAK", p);
1129:
1130: /* copy in_name to backup file */
1131: bakchn = creat(bakfile, 0600);
1132: if (bakchn < 0)
1133: err(bakfile);
1134: while (n = read(fileno(input), buff, sizeof buff))
1135: if (write(bakchn, buff, n) != n)
1136: err(bakfile);
1137: if (n < 0)
1138: err(in_name);
1139: close(bakchn);
1140: fclose(input);
1141:
1142: /* re-open backup file as the input file */
1143: input = fopen(bakfile, "r");
1144: if (input == 0)
1145: err(bakfile);
1146: /* now the original input file will be the output */
1147: output = fopen(in_name, "w");
1148: if (output == 0) {
1149: unlink(bakfile);
1150: err(in_name);
1151: }
1152: }
1153:
1154: err(msg)
1155: char *msg;
1156: {
1157: extern int errno;
1158: char *strerror();
1159:
1160: (void)fprintf(stderr, "indent: %s: %s\n", msg, strerror(errno));
1161: exit(1);
1162: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.