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>
 */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

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

int UTF8_fixed_width = 1;

int str_length(const char *str, int skip_ctrl_seq)
{
        int str_len;
        char input_str[5];
        wchar_t wcs[2];
        int wc_len;
        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((int)str[i]) || str[i] == ';' || str[i] == '?'; i++)
                                ;

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

                        continue;
                }

                // Process UTF-8 Chinese characters
                if (c & 0x80) // head of multi-byte character
                {
                        str_len = 0;
                        c = (char)(c & 0xf0);
                        while (c & 0x80)
                        {
                                input_str[str_len] = str[i + str_len];
                                str_len++;
                                c = (c & 0x7f) << 1;
                        }
                        input_str[str_len] = '\0';

                        if (mbstowcs(wcs, input_str, 1) == (size_t)-1)
                        {
#ifdef _DEBUG
                                log_error("mbstowcs(%s) error\n", input_str);
#endif
                                wc_len = (UTF8_fixed_width ? 2 : 1); // Fallback
                        }
                        else
                        {
                                wc_len = (UTF8_fixed_width ? 2 : wcwidth(wcs[0]));
                        }

                        i += (str_len - 1);
                        ret += wc_len;
                }
                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;
        int str_len;
        char input_str[5];
        wchar_t wcs[2];
        int wc_len;

        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
                {
                        str_len = 0;
                        c = (char)(c & 0xf0);
                        while (c & 0x80)
                        {
                                input_str[str_len] = buffer[i + str_len];
                                str_len++;
                                c = (c & 0x7f) << 1;
                        }
                        input_str[str_len] = '\0';

                        if (mbstowcs(wcs, input_str, 1) == (size_t)-1)
                        {
#ifdef _DEBUG
                                log_error("mbstowcs(%s) error\n", input_str);
#endif
                                wc_len = (UTF8_fixed_width ? 2 : 1); // Fallback
                        }
                        else
                        {
                                wc_len = (UTF8_fixed_width ? 2 : wcwidth(wcs[0]));
                        }

                        if (*p_display_len + wc_len > max_display_len)
                        {
                                break;
                        }

                        i += (str_len - 1);
                        (*p_display_len) += wc_len;
                }
                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 (p_buf[0] != '\0' || end_of_line);

        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	/* SPDX-License-Identifier: GPL-3.0-or-later */
2	/*
3	* str_process
4	* - common string processing features with UTF-8 support
5	*
6	* Copyright (C) 2004-2025 Leaflet <leaflet@leafok.com>
7	*/
8
9	#ifdef HAVE_CONFIG_H
10	#include "config.h"
11	#endif
12
13	#include "common.h"
14	#include "log.h"
15	#include "str_process.h"
16	#include <ctype.h>
17	#include <stdio.h>
18	#include <stdlib.h>
19	#include <string.h>
20	#include <wchar.h>
21
22	int UTF8_fixed_width = 1;
23
24	int str_length(const char *str, int skip_ctrl_seq)
25	{
26	int str_len;
27	char input_str[5];
28	wchar_t wcs[2];
29	int wc_len;
30	int i;
31	char c;
32	int ret = 0;
33
34	for (i = 0; str[i] != '\0'; i++)
35	{
36	c = str[i];
37
38	if (c == '\r' \|\| c == '\7') // skip
39	{
40	continue;
41	}
42
43	if (skip_ctrl_seq && c == '\033' && str[i + 1] == '[') // Skip control sequence
44	{
45	for (i = i + 2; isdigit((int)str[i]) \|\| str[i] == ';' \|\| str[i] == '?'; i++)
46	;
47
48	if (str[i] == 'm') // valid
49	{
50	// skip
51	}
52	else if (isalpha((int)str[i]))
53	{
54	// unsupported ANSI CSI command
55	}
56	else
57	{
58	i--;
59	}
60
61	continue;
62	}
63
64	// Process UTF-8 Chinese characters
65	if (c & 0x80) // head of multi-byte character
66	{
67	str_len = 0;
68	c = (char)(c & 0xf0);
69	while (c & 0x80)
70	{
71	input_str[str_len] = str[i + str_len];
72	str_len++;
73	c = (c & 0x7f) << 1;
74	}
75	input_str[str_len] = '\0';
76
77	if (mbstowcs(wcs, input_str, 1) == (size_t)-1)
78	{
79	#ifdef _DEBUG
80	log_error("mbstowcs(%s) error\n", input_str);
81	#endif
82	wc_len = (UTF8_fixed_width ? 2 : 1); // Fallback
83	}
84	else
85	{
86	wc_len = (UTF8_fixed_width ? 2 : wcwidth(wcs[0]));
87	}
88
89	i += (str_len - 1);
90	ret += wc_len;
91	}
92	else
93	{
94	ret++;
95	}
96	}
97
98	return ret;
99	}
100
101	int split_line(const char buffer, int max_display_len, int p_eol, int *p_display_len, int skip_ctrl_seq)
102	{
103	int i;
104	*p_eol = 0;
105	*p_display_len = 0;
106	char c;
107	int str_len;
108	char input_str[5];
109	wchar_t wcs[2];
110	int wc_len;
111
112	for (i = 0; buffer[i] != '\0'; i++)
113	{
114	c = buffer[i];
115
116	if (c == '\r' \|\| c == '\7') // skip
117	{
118	continue;
119	}
120
121	if (skip_ctrl_seq && c == '\033' && buffer[i + 1] == '[') // Skip control sequence
122	{
123	i += 2;
124	while (buffer[i] != '\0' && buffer[i] != 'm')
125	{
126	i++;
127	}
128	continue;
129	}
130
131	if (c & 0x80) // head of multi-byte character
132	{
133	str_len = 0;
134	c = (char)(c & 0xf0);
135	while (c & 0x80)
136	{
137	input_str[str_len] = buffer[i + str_len];
138	str_len++;
139	c = (c & 0x7f) << 1;
140	}
141	input_str[str_len] = '\0';
142
143	if (mbstowcs(wcs, input_str, 1) == (size_t)-1)
144	{
145	#ifdef _DEBUG
146	log_error("mbstowcs(%s) error\n", input_str);
147	#endif
148	wc_len = (UTF8_fixed_width ? 2 : 1); // Fallback
149	}
150	else
151	{
152	wc_len = (UTF8_fixed_width ? 2 : wcwidth(wcs[0]));
153	}
154
155	if (*p_display_len + wc_len > max_display_len)
156	{
157	break;
158	}
159
160	i += (str_len - 1);
161	(*p_display_len) += wc_len;
162	}
163	else
164	{
165	if (*p_display_len + 1 > max_display_len)
166	{
167	break;
168	}
169	(*p_display_len)++;
170
171	// \n is regarded as 1 character wide in terminal editor, which is different from Web version
172	if (c == '\n')
173	{
174	i++;
175	*p_eol = 1;
176	break;
177	}
178	}
179	}
180
181	return i;
182	}
183
184	long split_data_lines(const char p_buf, int max_display_len, long p_line_offsets, long line_offsets_count,
185	int skip_ctrl_seq, int *p_line_widths)
186	{
187	int line_cnt = 0;
188	int len;
189	int end_of_line = 0;
190	int display_len = 0;
191
192	p_line_offsets[line_cnt] = 0L;
193
194	do
195	{
196	len = split_line(p_buf, max_display_len, &end_of_line, &display_len, skip_ctrl_seq);
197
198	if (p_line_widths)
199	{
200	p_line_widths[line_cnt] = display_len;
201	}
202
203	// Exceed max_line_cnt
204	if (line_cnt + 1 >= line_offsets_count)
205	{
206	// log_error("Line count %d reaches limit %d\n", line_cnt + 1, line_offsets_count);
207	return line_cnt;
208	}
209
210	p_line_offsets[line_cnt + 1] = p_line_offsets[line_cnt] + len;
211	line_cnt++;
212	p_buf += len;
213	} while (p_buf[0] != '\0' \|\| end_of_line);
214
215	return line_cnt;
216	}
217
218	int str_filter(char *buffer, int skip_ctrl_seq)
219	{
220	int i;
221	int j;
222
223	for (i = 0, j = 0; buffer[i] != '\0'; i++)
224	{
225	if (buffer[i] == '\r' \|\| buffer[i] == '\7') // skip
226	{
227	continue;
228	}
229
230	if (skip_ctrl_seq && buffer[i] == '\033' && buffer[i + 1] == '[') // Skip control sequence
231	{
232	i += 2;
233	while (buffer[i] != '\0' && buffer[i] != 'm')
234	{
235	i++;
236	}
237	continue;
238	}
239
240	buffer[j] = buffer[i];
241	j++;
242	}
243
244	buffer[j] = '\0';
245
246	return j;
247	}
webmaster@leafok.com	ViewVC Help
Powered by ViewVC 1.3.0-beta1