lbbs/src/str_process.c

/* SPDX-License-Identifier: GPL-3.0-or-later */
/*
 * str_process
 *   - common string processing features with UTF-8 support
 *
 * Copyright (C) 2004-2025  Leaflet <leaflet@leafok.com>
 */

#include "common.h"
#include "log.h"
#include "str_process.h"
#include <ctype.h>
#include <stdio.h>
#include <string.h>

int str_length(const char *str, int skip_ctrl_seq)
{
        int i;
        char c;
        int ret = 0;

        for (i = 0; str[i] != '\0'; i++)
        {
                c = str[i];

                if (c == '\r' || c == '\7') // skip
                {
                        continue;
                }

                if (skip_ctrl_seq && c == '\033' && str[i + 1] == '[') // Skip control sequence
                {
                        for (i = i + 2; isdigit(str[i]) || str[i] == ';' || str[i] == '?'; i++)
                                ;

                        if (str[i] == 'm') // valid
                        {
                                // skip
                        }
                        else if (isalpha(str[i]))
                        {
                                // unsupported ANSI CSI command
                        }
                        else
                        {
                                i--;
                        }

                        continue;
                }

                // Process UTF-8 Chinese characters
                if (c & 0x80) // head of multi-byte character
                {
                        c = (c & 0x70) << 1;
                        while (c & 0x80)
                        {
                                i++;
                                c = (c & 0x7f) << 1;
                        }

                        ret += 2;
                }
                else
                {
                        ret++;
                }
        }

        return ret;
}

int split_line(const char *buffer, int max_display_len, int *p_eol, int *p_display_len, int skip_ctrl_seq)
{
        int i;
        *p_eol = 0;
        *p_display_len = 0;
        char c;

        for (i = 0; buffer[i] != '\0'; i++)
        {
                c = buffer[i];

                if (c == '\r' || c == '\7') // skip
                {
                        continue;
                }

                if (skip_ctrl_seq && c == '\033' && buffer[i + 1] == '[') // Skip control sequence
                {
                        i += 2;
                        while (buffer[i] != '\0' && buffer[i] != 'm')
                        {
                                i++;
                        }
                        continue;
                }

                if (c & 0x80) // head of multi-byte character
                {
                        if (*p_display_len + 2 > max_display_len)
                        {
                                break;
                        }

                        c = (c & 0x70) << 1;
                        while (c & 0x80)
                        {
                                i++;
                                c = (c & 0x7f) << 1;
                        }

                        (*p_display_len) += 2;
                }
                else
                {
                        if (*p_display_len + 1 > max_display_len)
                        {
                                break;
                        }
                        (*p_display_len)++;

                        // \n is regarded as 1 character wide in terminal editor, which is different from Web version
                        if (c == '\n')
                        {
                                i++;
                                *p_eol = 1;
                                break;
                        }
                }
        }

        return i;
}

long split_data_lines(const char *p_buf, int max_display_len, long *p_line_offsets, long line_offsets_count,
                                          int skip_ctrl_seq, int *p_line_widths)
{
        int line_cnt = 0;
        int len;
        int end_of_line = 0;
        int display_len = 0;

        p_line_offsets[line_cnt] = 0L;

        do
        {
                len = split_line(p_buf, max_display_len, &end_of_line, &display_len, skip_ctrl_seq);

                if (p_line_widths)
                {
                        p_line_widths[line_cnt] = display_len;
                }

                // Exceed max_line_cnt
                if (line_cnt + 1 >= line_offsets_count)
                {
                        // log_error("Line count %d reaches limit %d\n", line_cnt + 1, line_offsets_count);
                        return line_cnt;
                }

                p_line_offsets[line_cnt + 1] = p_line_offsets[line_cnt] + len;
                line_cnt++;
                p_buf += len;
        } while (len > 0);

        return line_cnt;
}

int str_filter(char *buffer, int skip_ctrl_seq)
{
        int i;
        int j;

        for (i = 0, j = 0; buffer[i] != '\0'; i++)
        {
                if (buffer[i] == '\r' || buffer[i] == '\7') // skip
                {
                        continue;
                }

                if (skip_ctrl_seq && buffer[i] == '\033' && buffer[i + 1] == '[') // Skip control sequence
                {
                        i += 2;
                        while (buffer[i] != '\0' && buffer[i] != 'm')
                        {
                                i++;
                        }
                        continue;
                }

                buffer[j] = buffer[i];
                j++;
        }

        buffer[j] = '\0';

        return j;
}
1	sysadm	1.23	/* SPDX-License-Identifier: GPL-3.0-or-later */
2			/*
3			* str_process
4			* - common string processing features with UTF-8 support
5			*
6	sysadm	1.24	* Copyright (C) 2004-2025 Leaflet <leaflet@leafok.com>
7	sysadm	1.23	*/
8	sysadm	1.1
9			#include "common.h"
10			#include "log.h"
11	sysadm	1.18	#include "str_process.h"
12	sysadm	1.22	#include <ctype.h>
13	sysadm	1.1	#include <stdio.h>
14			#include <string.h>
15
16	sysadm	1.20	int str_length(const char *str, int skip_ctrl_seq)
17			{
18			int i;
19			char c;
20			int ret = 0;
21
22			for (i = 0; str[i] != '\0'; i++)
23			{
24			c = str[i];
25
26			if (c == '\r' \|\| c == '\7') // skip
27			{
28			continue;
29			}
30
31			if (skip_ctrl_seq && c == '\033' && str[i + 1] == '[') // Skip control sequence
32			{
33	sysadm	1.22	for (i = i + 2; isdigit(str[i]) \|\| str[i] == ';' \|\| str[i] == '?'; i++)
34			;
35
36			if (str[i] == 'm') // valid
37			{
38			// skip
39			}
40			else if (isalpha(str[i]))
41			{
42			// unsupported ANSI CSI command
43			}
44			else
45	sysadm	1.20	{
46	sysadm	1.22	i--;
47	sysadm	1.20	}
48	sysadm	1.22
49	sysadm	1.20	continue;
50			}
51
52			// Process UTF-8 Chinese characters
53	sysadm	1.21	if (c & 0x80) // head of multi-byte character
54	sysadm	1.20	{
55	sysadm	1.21	c = (c & 0x70) << 1;
56			while (c & 0x80)
57	sysadm	1.20	{
58			i++;
59	sysadm	1.21	c = (c & 0x7f) << 1;
60	sysadm	1.20	}
61
62			ret += 2;
63			}
64			else
65			{
66			ret++;
67			}
68			}
69
70			return ret;
71			}
72
73	sysadm	1.17	int split_line(const char buffer, int max_display_len, int p_eol, int *p_display_len, int skip_ctrl_seq)
74	sysadm	1.1	{
75	sysadm	1.7	int i;
76	sysadm	1.1	*p_eol = 0;
77	sysadm	1.4	*p_display_len = 0;
78	sysadm	1.11	char c;
79	sysadm	1.1
80	sysadm	1.7	for (i = 0; buffer[i] != '\0'; i++)
81	sysadm	1.1	{
82	sysadm	1.11	c = buffer[i];
83	sysadm	1.1
84			if (c == '\r' \|\| c == '\7') // skip
85			{
86			continue;
87			}
88
89	sysadm	1.17	if (skip_ctrl_seq && c == '\033' && buffer[i + 1] == '[') // Skip control sequence
90	sysadm	1.1	{
91			i += 2;
92	sysadm	1.7	while (buffer[i] != '\0' && buffer[i] != 'm')
93	sysadm	1.1	{
94			i++;
95			}
96			continue;
97			}
98
99	sysadm	1.21	if (c & 0x80) // head of multi-byte character
100	sysadm	1.1	{
101	sysadm	1.6	if (*p_display_len + 2 > max_display_len)
102	sysadm	1.1	{
103			break;
104			}
105	sysadm	1.20
106	sysadm	1.21	c = (c & 0x70) << 1;
107			while (c & 0x80)
108	sysadm	1.20	{
109			i++;
110	sysadm	1.21	c = (c & 0x7f) << 1;
111	sysadm	1.20	}
112
113	sysadm	1.10	(*p_display_len) += 2;
114	sysadm	1.1	}
115			else
116			{
117	sysadm	1.6	if (*p_display_len + 1 > max_display_len)
118	sysadm	1.1	{
119			break;
120			}
121	sysadm	1.4	(*p_display_len)++;
122	sysadm	1.12
123	sysadm	1.14	// \n is regarded as 1 character wide in terminal editor, which is different from Web version
124	sysadm	1.12	if (c == '\n')
125			{
126			i++;
127			*p_eol = 1;
128			break;
129			}
130	sysadm	1.1	}
131			}
132
133			return i;
134			}
135
136	sysadm	1.19	long split_data_lines(const char p_buf, int max_display_len, long p_line_offsets, long line_offsets_count,
137			int skip_ctrl_seq, int *p_line_widths)
138	sysadm	1.1	{
139	sysadm	1.7	int line_cnt = 0;
140	sysadm	1.13	int len;
141	sysadm	1.1	int end_of_line = 0;
142	sysadm	1.4	int display_len = 0;
143
144	sysadm	1.1	p_line_offsets[line_cnt] = 0L;
145
146	sysadm	1.13	do
147	sysadm	1.1	{
148	sysadm	1.17	len = split_line(p_buf, max_display_len, &end_of_line, &display_len, skip_ctrl_seq);
149	sysadm	1.8
150	sysadm	1.19	if (p_line_widths)
151			{
152			p_line_widths[line_cnt] = display_len;
153			}
154
155	sysadm	1.8	// Exceed max_line_cnt
156	sysadm	1.9	if (line_cnt + 1 >= line_offsets_count)
157	sysadm	1.1	{
158	sysadm	1.16	// log_error("Line count %d reaches limit %d\n", line_cnt + 1, line_offsets_count);
159	sysadm	1.8	return line_cnt;
160	sysadm	1.1	}
161	sysadm	1.8
162			p_line_offsets[line_cnt + 1] = p_line_offsets[line_cnt] + len;
163			line_cnt++;
164			p_buf += len;
165	sysadm	1.25	} while (len > 0);
166	sysadm	1.1
167			return line_cnt;
168			}
169	sysadm	1.14
170	sysadm	1.17	int str_filter(char *buffer, int skip_ctrl_seq)
171	sysadm	1.14	{
172			int i;
173			int j;
174
175			for (i = 0, j = 0; buffer[i] != '\0'; i++)
176			{
177	sysadm	1.15	if (buffer[i] == '\r' \|\| buffer[i] == '\7') // skip
178	sysadm	1.14	{
179			continue;
180			}
181
182	sysadm	1.17	if (skip_ctrl_seq && buffer[i] == '\033' && buffer[i + 1] == '[') // Skip control sequence
183	sysadm	1.14	{
184			i += 2;
185			while (buffer[i] != '\0' && buffer[i] != 'm')
186			{
187			i++;
188			}
189			continue;
190			}
191
192			buffer[j] = buffer[i];
193			j++;
194			}
195
196			buffer[j] = '\0';
197
198			return j;
199			}
webmaster@leafok.com	ViewVC Help
Powered by ViewVC 1.3.0-beta1