Home | History | Annotate | Line # | Download | only in indent
pr_comment.c revision 1.165
      1 /*	$NetBSD: pr_comment.c,v 1.165 2023/06/14 14:11:28 rillig Exp $	*/
      2 
      3 /*-
      4  * SPDX-License-Identifier: BSD-4-Clause
      5  *
      6  * Copyright (c) 1985 Sun Microsystems, Inc.
      7  * Copyright (c) 1980, 1993
      8  *	The Regents of the University of California.  All rights reserved.
      9  * All rights reserved.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  * 3. All advertising materials mentioning features or use of this software
     20  *    must display the following acknowledgement:
     21  *	This product includes software developed by the University of
     22  *	California, Berkeley and its contributors.
     23  * 4. Neither the name of the University nor the names of its contributors
     24  *    may be used to endorse or promote products derived from this software
     25  *    without specific prior written permission.
     26  *
     27  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     29  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     30  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     31  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     32  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     33  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     34  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     35  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     36  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     37  * SUCH DAMAGE.
     38  */
     39 
     40 #include <sys/cdefs.h>
     41 __RCSID("$NetBSD: pr_comment.c,v 1.165 2023/06/14 14:11:28 rillig Exp $");
     42 
     43 #include <string.h>
     44 
     45 #include "indent.h"
     46 
     47 static void
     48 com_add_char(char ch)
     49 {
     50 	buf_add_char(&com, ch);
     51 }
     52 
     53 static void
     54 com_add_delim(void)
     55 {
     56 	if (opt.star_comment_cont)
     57 		buf_add_chars(&com, " * ", 3);
     58 }
     59 
     60 static bool
     61 fits_in_one_line(int max_line_length)
     62 {
     63 	for (const char *start = inp_p, *p = start; *p != '\n'; p++) {
     64 		if (p[0] == '*' && p[1] == '/') {
     65 			while (p - inp_p >= 2
     66 			    && ch_isblank(p[-1])
     67 			    && ch_isblank(p[-2]))
     68 				p--;
     69 			int ind = ind_add(ps.comment_ind + 3,
     70 			    start, (size_t)(p - start));
     71 			ind += p == start || ch_isblank(p[-1]) ? 2 : 3;
     72 			return ind <= max_line_length;
     73 		}
     74 	}
     75 	return false;
     76 }
     77 
     78 static void
     79 analyze_comment(bool *p_may_wrap, bool *p_delim, int *p_line_length)
     80 {
     81 	bool may_wrap = true;
     82 	bool delim = false;
     83 	int ind;
     84 	int line_length = opt.max_line_length;
     85 
     86 	if (inp_p - inp.s == 2 && !opt.format_col1_comments) {
     87 		may_wrap = false;
     88 		ind = 0;
     89 	} else {
     90 		if (inp_p[0] == '-' || inp_p[0] == '*' ||
     91 		    token.s[token.len - 1] == '/' ||
     92 		    (inp_p[0] == '\n' && !opt.format_block_comments))
     93 			may_wrap = false;
     94 		if (code.len == 0 && inp_p[strspn(inp_p, "*")] == '\n')
     95 			out.line_kind = lk_block_comment;
     96 
     97 		if (com.len > 0)
     98 			output_line();
     99 		if (lab.len == 0 && code.len == 0) {
    100 			ind = (ps.ind_level - opt.unindent_displace)
    101 			    * opt.indent_size;
    102 			if (ind <= 0)
    103 				ind = opt.format_col1_comments ? 0 : 1;
    104 			line_length = opt.block_comment_max_line_length;
    105 			if (may_wrap && inp_p[0] == '\n')
    106 				delim = true;
    107 			if (may_wrap && opt.comment_delimiter_on_blank_line)
    108 				delim = true;
    109 		} else {
    110 			int target_ind = code.len > 0
    111 			    ? ind_add(compute_code_indent(), code.s, code.len)
    112 			    : ind_add(compute_label_indent(), lab.s, lab.len);
    113 
    114 			ind = ps.line_has_decl || ps.ind_level == 0
    115 			    ? opt.decl_comment_column - 1
    116 			    : opt.comment_column - 1;
    117 			if (ind <= target_ind)
    118 				ind = next_tab(target_ind);
    119 			if (ind + 25 > line_length)
    120 				line_length = ind + 25;
    121 		}
    122 	}
    123 
    124 	if (!may_wrap) {
    125 		/* Find out how much indentation there was originally, because
    126 		 * that much will have to be ignored by output_line. */
    127 		size_t len = (size_t)(inp_p - 2 - inp.s);
    128 		ps.comment_shift = -ind_add(0, inp.s, len);
    129 	} else {
    130 		ps.comment_shift = 0;
    131 		if (!(inp_p[0] == '\t' && !ch_isblank(inp_p[1])))
    132 			while (ch_isblank(inp_p[0]))
    133 				inp_p++;
    134 	}
    135 
    136 	ps.comment_ind = ind;
    137 	*p_may_wrap = may_wrap;
    138 	*p_delim = delim;
    139 	*p_line_length = line_length;
    140 }
    141 
    142 static void
    143 copy_comment_start(bool may_wrap, bool *delim, int line_length)
    144 {
    145 	ps.comment_in_first_line = true;
    146 	com_add_char('/');
    147 	com_add_char(token.s[token.len - 1]);	/* either '*' or '/' */
    148 
    149 	if (may_wrap) {
    150 		if (!ch_isblank(inp_p[0]))
    151 			com_add_char(' ');
    152 
    153 		if (*delim && fits_in_one_line(line_length))
    154 			*delim = false;
    155 		if (*delim) {
    156 			output_line();
    157 			com_add_delim();
    158 		}
    159 	}
    160 }
    161 
    162 static void
    163 copy_comment_wrap_text(int line_length, ssize_t *last_blank)
    164 {
    165 	int now_len = ind_add(ps.comment_ind, com.s, com.len);
    166 	for (;;) {
    167 		char ch = inp_next();
    168 		if (ch_isblank(ch))
    169 			*last_blank = (ssize_t)com.len;
    170 		com_add_char(ch);
    171 		now_len++;
    172 		if (memchr("*\n\r\b\t", inp_p[0], 6) != NULL)
    173 			break;
    174 		if (now_len >= line_length && *last_blank != -1)
    175 			break;
    176 	}
    177 
    178 	if (now_len <= line_length)
    179 		return;
    180 	if (ch_isspace(com.s[com.len - 1]))
    181 		return;
    182 
    183 	if (*last_blank == -1) {
    184 		/* only a single word in this line */
    185 		output_line();
    186 		com_add_delim();
    187 		return;
    188 	}
    189 
    190 	const char *last_word_s = com.s + *last_blank + 1;
    191 	size_t last_word_len = com.len - (size_t)(*last_blank + 1);
    192 	com.len = (size_t)*last_blank;
    193 	output_line();
    194 	com_add_delim();
    195 
    196 	/* Assume that output_line and com_add_delim don't invalidate the
    197 	 * "unused" part of the buffer beyond com.s + com.len. */
    198 	memmove(com.s + com.len, last_word_s, last_word_len);
    199 	com.len += last_word_len;
    200 	*last_blank = -1;
    201 }
    202 
    203 static bool
    204 copy_comment_wrap_newline(ssize_t *last_blank, bool seen_newline)
    205 {
    206 	*last_blank = -1;
    207 	if (seen_newline) {
    208 		if (com.len == 0)
    209 			com_add_char(' ');	/* force empty output line */
    210 		if (com.len > 3) {
    211 			output_line();
    212 			com_add_delim();
    213 		}
    214 		output_line();
    215 		com_add_delim();
    216 	} else {
    217 		if (!(com.len > 0 && ch_isblank(com.s[com.len - 1])))
    218 			com_add_char(' ');
    219 		*last_blank = (int)com.len - 1;
    220 	}
    221 	++line_no;
    222 
    223 	/* flush any blanks and/or tabs at start of next line */
    224 	inp_skip();		/* '\n' */
    225 	while (ch_isblank(inp_p[0]))
    226 		inp_p++;
    227 	if (inp_p[0] == '*' && inp_p[1] == '/')
    228 		return false;
    229 	if (inp_p[0] == '*') {
    230 		inp_p++;
    231 		while (ch_isblank(inp_p[0]))
    232 			inp_p++;
    233 	}
    234 
    235 	return true;
    236 }
    237 
    238 static void
    239 copy_comment_wrap_finish(int line_length, bool delim)
    240 {
    241 	if (delim) {
    242 		if (com.len > 3)
    243 			output_line();
    244 		else
    245 			buf_clear(&com);
    246 		com_add_char(' ');
    247 	} else {
    248 		size_t len = com.len;
    249 		while (ch_isblank(com.s[len - 1]))
    250 			len--;
    251 		int end_ind = ind_add(ps.comment_ind, com.s, len);
    252 		if (end_ind + 3 > line_length)
    253 			output_line();
    254 	}
    255 
    256 	while (com.len >= 2
    257 	    && ch_isblank(com.s[com.len - 1])
    258 	    && ch_isblank(com.s[com.len - 2]))
    259 		com.len--;
    260 	buf_terminate(&com);
    261 
    262 	inp_p += 2;
    263 	if (com.len > 0 && ch_isblank(com.s[com.len - 1]))
    264 		buf_add_chars(&com, "*/", 2);
    265 	else
    266 		buf_add_chars(&com, " */", 3);
    267 }
    268 
    269 /*
    270  * Copy characters from 'inp' to 'com'. Try to keep comments from going over
    271  * the maximum line length. To do that, remember where the last blank, tab, or
    272  * newline was. When a line is filled, print up to the last blank and continue
    273  * copying.
    274  */
    275 static void
    276 copy_comment_wrap(int line_length, bool delim)
    277 {
    278 	ssize_t last_blank = -1;	/* index of the last blank in 'com' */
    279 	bool seen_newline = false;
    280 
    281 	for (;;) {
    282 		if (inp_p[0] == '\n') {
    283 			if (had_eof)
    284 				goto unterminated_comment;
    285 			if (!copy_comment_wrap_newline(&last_blank,
    286 				seen_newline))
    287 				break;
    288 			seen_newline = true;
    289 		} else if (inp_p[0] == '*' && inp_p[1] == '/')
    290 			break;
    291 		else {
    292 			copy_comment_wrap_text(line_length, &last_blank);
    293 			seen_newline = false;
    294 		}
    295 	}
    296 
    297 	copy_comment_wrap_finish(line_length, delim);
    298 	return;
    299 
    300 unterminated_comment:
    301 	diag(1, "Unterminated comment");
    302 	output_line();
    303 }
    304 
    305 static void
    306 copy_comment_nowrap(void)
    307 {
    308 	char kind = token.s[token.len - 1];
    309 
    310 	for (;;) {
    311 		if (inp_p[0] == '\n') {
    312 			if (kind == '/')
    313 				return;
    314 
    315 			if (had_eof) {
    316 				diag(1, "Unterminated comment");
    317 				output_line();
    318 				return;
    319 			}
    320 
    321 			if (com.len == 0)
    322 				com_add_char(' ');	/* force output of an
    323 							 * empty line */
    324 			output_line();
    325 			++line_no;
    326 			inp_skip();
    327 			continue;
    328 		}
    329 
    330 		com_add_char(*inp_p++);
    331 		if (com.len >= 2
    332 		    && com.s[com.len - 2] == '*'
    333 		    && com.s[com.len - 1] == '/'
    334 		    && kind == '*')
    335 			return;
    336 	}
    337 }
    338 
    339 /*
    340  * Scan, reformat and output a single comment, which is either a block comment
    341  * starting with '/' '*' or an end-of-line comment starting with '//'.
    342  */
    343 void
    344 process_comment(void)
    345 {
    346 	bool may_wrap, delim;
    347 	int line_length;
    348 
    349 	analyze_comment(&may_wrap, &delim, &line_length);
    350 	copy_comment_start(may_wrap, &delim, line_length);
    351 	if (may_wrap)
    352 		copy_comment_wrap(line_length, delim);
    353 	else
    354 		copy_comment_nowrap();
    355 }
    356