Home | History | Annotate | Line # | Download | only in dist 
      1 /*	$NetBSD: cvt.c,v 1.4 2023/10/06 05:49:49 simonb Exp $	*/
      2 
      3 /*
      4  * Copyright (C) 1984-2023  Mark Nudelman
      5  *
      6  * You may distribute under the terms of either the GNU General Public
      7  * License or the Less License, as specified in the README file.
      8  *
      9  * For more information, see the README file.
     10  */
     11 
     12 /*
     13  * Routines to convert text in various ways.  Used by search.
     14  */
     15 
     16 #include "less.h"
     17 #include "charset.h"
     18 
     19 extern int utf_mode;
     20 
     21 /*
     22  * Get the length of a buffer needed to convert a string.
     23  */
     24 public int cvt_length(int len, int ops)
     25 {
     26 	if (utf_mode)
     27 		/*
     28 		 * Just copying a string in UTF-8 mode can cause it to grow
     29 		 * in length.
     30 		 * Four output bytes for one input byte is the worst case.
     31 		 */
     32 		len *= 4;
     33 	return (len + 1);
     34 }
     35 
     36 /*
     37  * Allocate a chpos array for use by cvt_text.
     38  */
     39 public int * cvt_alloc_chpos(int len)
     40 {
     41 	int i;
     42 	int *chpos = (int *) ecalloc(sizeof(int), len);
     43 	/* Initialize all entries to an invalid position. */
     44 	for (i = 0;  i < len;  i++)
     45 		chpos[i] = -1;
     46 	return (chpos);
     47 }
     48 
     49 /*
     50  * Convert text.  Perform the transformations specified by ops.
     51  * Returns converted text in odst.  The original offset of each
     52  * odst character (when it was in osrc) is returned in the chpos array.
     53  */
     54 public void cvt_text(char *odst, char *osrc, int *chpos, int *lenp, int ops)
     55 {
     56 	char *dst;
     57 	char *edst = odst;
     58 	char *src;
     59 	char *src_end;
     60 	LWCHAR ch;
     61 
     62 	if (lenp != NULL)
     63 		src_end = osrc + *lenp;
     64 	else
     65 		src_end = osrc + strlen(osrc);
     66 
     67 	for (src = osrc, dst = odst;  src < src_end;  )
     68 	{
     69 		int src_pos = (int) (src - osrc);
     70 		int dst_pos = (int) (dst - odst);
     71 		struct ansi_state *pansi;
     72 		ch = step_char(&src, +1, src_end);
     73 		if ((ops & CVT_BS) && ch == '\b' && dst > odst)
     74 		{
     75 			/* Delete backspace and preceding char. */
     76 			do {
     77 				dst--;
     78 			} while (dst > odst && utf_mode &&
     79 				!IS_ASCII_OCTET(*dst) && !IS_UTF8_LEAD(*dst));
     80 		} else if ((ops & CVT_ANSI) && (pansi = ansi_start(ch)) != NULL)
     81 		{
     82 			/* Skip to end of ANSI escape sequence. */
     83 			while (src < src_end)
     84 			{
     85 				if (ansi_step(pansi, ch) != ANSI_MID)
     86 					break;
     87 				ch = *src++;
     88 			}
     89 			ansi_done(pansi);
     90 		} else
     91 		{
     92 			/* Just copy the char to the destination buffer. */
     93 			if ((ops & CVT_TO_LC) && IS_UPPER(ch))
     94 				ch = TO_LOWER(ch);
     95 			put_wchar(&dst, ch);
     96 			/* Record the original position of the char. */
     97 			if (chpos != NULL)
     98 				chpos[dst_pos] = src_pos;
     99 		}
    100 		if (dst > edst)
    101 			edst = dst;
    102 	}
    103 	if ((ops & CVT_CRLF) && edst > odst && edst[-1] == '\r')
    104 		edst--;
    105 	*edst = '\0';
    106 	if (lenp != NULL)
    107 		*lenp = (int) (edst - odst);
    108 	/* FIXME: why was this here?  if (chpos != NULL) chpos[dst - odst] = src - osrc; */
    109 }
    110