author | rubidium |
Wed, 23 Apr 2008 20:56:08 +0000 | |
changeset 10314 | 9cfcdd5b5ddb |
parent 9207 | 2b390a312384 |
child 10465 | 0c68afe3d725 |
permissions | -rw-r--r-- |
2186 | 1 |
/* $Id$ */ |
2 |
||
8844 | 3 |
/** @file string_func.h Functions related to low-level strings. */ |
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
4 |
|
8710
52015340050c
(svn r11777) -Codechange: split the string header and make do not include it when it's not necessary.
rubidium
parents:
8609
diff
changeset
|
5 |
#ifndef STRING_FUNC_H |
52015340050c
(svn r11777) -Codechange: split the string header and make do not include it when it's not necessary.
rubidium
parents:
8609
diff
changeset
|
6 |
#define STRING_FUNC_H |
1317
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
7 |
|
8609
8c0c3e9dd6a0
(svn r11674) -Codechange: refactor some functions out of macros.h into more logical locations.
rubidium
parents:
8006
diff
changeset
|
8 |
#include "core/bitmath_func.hpp" |
8710
52015340050c
(svn r11777) -Codechange: split the string header and make do not include it when it's not necessary.
rubidium
parents:
8609
diff
changeset
|
9 |
#include "string_type.h" |
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
10 |
|
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
11 |
/** |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
12 |
* usage ttd_strlcpy(dst, src, lengthof(dst)); |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
13 |
* @param dst destination buffer |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
14 |
* @param src string to copy/concatenate |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
15 |
* @param size size of the destination buffer |
1317
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
16 |
*/ |
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
17 |
void ttd_strlcat(char *dst, const char *src, size_t size); |
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
18 |
void ttd_strlcpy(char *dst, const char *src, size_t size); |
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
19 |
|
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
20 |
/** |
1317
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
21 |
* usage: strecpy(dst, src, lastof(dst)); |
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
22 |
* @param dst destination buffer |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
23 |
* @param src string to copy |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
24 |
* @param last pointer to the last element in the dst array |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
25 |
* if NULL no boundary check is performed |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
26 |
* @return a pointer to the terminating \0 in the destination buffer |
1317
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
27 |
*/ |
7814
fe643468ad64
(svn r10673) -Cleanup: some assorted style cleanups. Primarily type* var -> type *var.
rubidium
parents:
6916
diff
changeset
|
28 |
char *strecat(char *dst, const char *src, const char *last); |
fe643468ad64
(svn r10673) -Cleanup: some assorted style cleanups. Primarily type* var -> type *var.
rubidium
parents:
6916
diff
changeset
|
29 |
char *strecpy(char *dst, const char *src, const char *last); |
1317
f382f1b439c7
(svn r1821) Move generic string handling functions to string.[ch] and introduce stre{cpy,cat}, see string.h for their semantics
tron
parents:
diff
changeset
|
30 |
|
7814
fe643468ad64
(svn r10673) -Cleanup: some assorted style cleanups. Primarily type* var -> type *var.
rubidium
parents:
6916
diff
changeset
|
31 |
char *CDECL str_fmt(const char *str, ...); |
2234 | 32 |
|
2775
d3ed38a97250
(svn r3322) - Fix: Network window crash when it receives invalid information for example from the integrated nightly, so validate the network-input when it is received
Darkvater
parents:
2436
diff
changeset
|
33 |
/** Scans the string for valid characters and if it finds invalid ones, |
d3ed38a97250
(svn r3322) - Fix: Network window crash when it receives invalid information for example from the integrated nightly, so validate the network-input when it is received
Darkvater
parents:
2436
diff
changeset
|
34 |
* replaces them with a question mark '?' */ |
d3ed38a97250
(svn r3322) - Fix: Network window crash when it receives invalid information for example from the integrated nightly, so validate the network-input when it is received
Darkvater
parents:
2436
diff
changeset
|
35 |
void str_validate(char *str); |
d3ed38a97250
(svn r3322) - Fix: Network window crash when it receives invalid information for example from the integrated nightly, so validate the network-input when it is received
Darkvater
parents:
2436
diff
changeset
|
36 |
|
5101
797a070e5b22
(svn r7172) -Fix [r6931]: The console showed '?' characters instead of colours. Now strip all
Darkvater
parents:
4300
diff
changeset
|
37 |
/** Scans the string for colour codes and strips them */ |
797a070e5b22
(svn r7172) -Fix [r6931]: The console showed '?' characters instead of colours. Now strip all
Darkvater
parents:
4300
diff
changeset
|
38 |
void str_strip_colours(char *str); |
797a070e5b22
(svn r7172) -Fix [r6931]: The console showed '?' characters instead of colours. Now strip all
Darkvater
parents:
4300
diff
changeset
|
39 |
|
5164
a4fb0ede4ce5
(svn r7271) -Codechange: Revert the strtolower part of r7199 as it can actually become broken due to
Darkvater
parents:
5108
diff
changeset
|
40 |
/** Convert the given string to lowercase, only works with ASCII! */ |
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
41 |
void strtolower(char *str); |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
42 |
|
5317
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
43 |
|
8006
947dbf950dc7
(svn r11025) -Fix: safeguard StrEmpty from calls with NULL.
rubidium
parents:
7814
diff
changeset
|
44 |
static inline bool StrEmpty(const char *s) { return s == NULL || s[0] == '\0'; } |
5889
272800d151ac
(svn r8097) Replace strlen() {==,!=,>} 0 by the more concise {,!}StrEmpty(). Additionally the test takes O(1) instead of O(n) now
tron
parents:
5885
diff
changeset
|
45 |
|
272800d151ac
(svn r8097) Replace strlen() {==,!=,>} 0 by the more concise {,!}StrEmpty(). Additionally the test takes O(1) instead of O(n) now
tron
parents:
5885
diff
changeset
|
46 |
|
5317
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
47 |
/** Get the length of a string, within a limited buffer */ |
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
48 |
static inline int ttd_strnlen(const char *str, int maxlen) |
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
49 |
{ |
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
50 |
const char *t; |
9191
ae14770c829a
(svn r12368) -Codechange: use explicit body for loops and conditions and remove -Wno-empty-body from the configure script
smatz
parents:
8844
diff
changeset
|
51 |
for (t = str; *t != '\0' && t - str < maxlen; t++) {} |
5317
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
52 |
return t - str; |
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
53 |
} |
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
54 |
|
5885
262234e81333
(svn r8093) -Codechange: Add a function to get a string representation of an MD5SUM and use it.
Darkvater
parents:
5726
diff
changeset
|
55 |
/** Convert the md5sum number to a 'hexadecimal' string, return next pos in buffer */ |
262234e81333
(svn r8093) -Codechange: Add a function to get a string representation of an MD5SUM and use it.
Darkvater
parents:
5726
diff
changeset
|
56 |
char *md5sumToString(char *buf, const char *last, const uint8 md5sum[16]); |
5317
e235a3a573e3
(svn r7475) -Fix (r7348): sanity check NewGRF action 8 strings for null terminator
peter1138
parents:
5164
diff
changeset
|
57 |
|
4300
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
58 |
/** |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
59 |
* Only allow certain keys. You can define the filter to be used. This makes |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
60 |
* sure no invalid keys can get into an editbox, like BELL. |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
61 |
* @param key character to be checked |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
62 |
* @param afilter the filter to use |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
63 |
* @return true or false depending if the character is printable/valid or not |
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
64 |
*/ |
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
65 |
bool IsValidChar(WChar key, CharSetFilter afilter); |
4300
687a17c9c557
(svn r5946) -Add: merged the TGP branch to mainline. TGP adds:
truelight
parents:
4299
diff
changeset
|
66 |
|
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
67 |
size_t Utf8Decode(WChar *c, const char *s); |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
68 |
size_t Utf8Encode(char *buf, WChar c); |
6541
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
69 |
size_t Utf8TrimString(char *s, size_t maxlen); |
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
70 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
71 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
72 |
static inline WChar Utf8Consume(const char **s) |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
73 |
{ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
74 |
WChar c; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
75 |
*s += Utf8Decode(&c, *s); |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
76 |
return c; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
77 |
} |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
78 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
79 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
80 |
/** Return the length of a UTF-8 encoded character. |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
81 |
* @param c Unicode character. |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
82 |
* @return Length of UTF-8 encoding for character. |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
83 |
*/ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
84 |
static inline size_t Utf8CharLen(WChar c) |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
85 |
{ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
86 |
if (c < 0x80) return 1; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
87 |
if (c < 0x800) return 2; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
88 |
if (c < 0x10000) return 3; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
89 |
if (c < 0x110000) return 4; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
90 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
91 |
/* Invalid valid, we encode as a '?' */ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
92 |
return 1; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
93 |
} |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
94 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
95 |
|
6541
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
96 |
/** |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
97 |
* Return the length of an UTF-8 encoded value based on a single char. This |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
98 |
* char should be the first byte of the UTF-8 encoding. If not, or encoding |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
99 |
* is invalid, return value is 0 |
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
100 |
* @param c char to query length of |
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
101 |
* @return requested size |
6541
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
102 |
*/ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
103 |
static inline size_t Utf8EncodedCharLen(char c) |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
104 |
{ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
105 |
if (GB(c, 3, 5) == 0x1E) return 4; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
106 |
if (GB(c, 4, 4) == 0x0E) return 3; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
107 |
if (GB(c, 5, 3) == 0x06) return 2; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
108 |
if (GB(c, 7, 1) == 0x00) return 1; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
109 |
|
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
110 |
/* Invalid UTF8 start encoding */ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
111 |
return 0; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
112 |
} |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
113 |
|
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
114 |
|
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
115 |
/* Check if the given character is part of a UTF8 sequence */ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
116 |
static inline bool IsUtf8Part(char c) |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
117 |
{ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
118 |
return GB(c, 6, 2) == 2; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
119 |
} |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
120 |
|
6537
dbf4fa55395f
(svn r9003) -Codechange: Introduce a function Utf8PrevCharLen that finds the starting character of an UTF-8 sequence from a given position and returns the length to the first UTF-8 encoding byte of that sequence.
Darkvater
parents:
5889
diff
changeset
|
121 |
/** |
6540
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
122 |
* Retrieve the previous UNICODE character in an UTF-8 encoded string. |
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
123 |
* @param s char pointer pointing to (the first char of) the next character |
6916
e87d54a598ea
(svn r9556) -Documentation: doxygen and comment-style changes. 'R', 'S'.. The end of the preliminary work is near
belugas
parents:
6574
diff
changeset
|
124 |
* @return a pointer in 's' to the previous UNICODE character's first byte |
6540
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
125 |
* @note The function should not be used to determine the length of the previous |
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
126 |
* encoded char because it might be an invalid/corrupt start-sequence |
6537
dbf4fa55395f
(svn r9003) -Codechange: Introduce a function Utf8PrevCharLen that finds the starting character of an UTF-8 sequence from a given position and returns the length to the first UTF-8 encoding byte of that sequence.
Darkvater
parents:
5889
diff
changeset
|
127 |
*/ |
6540
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
128 |
static inline char *Utf8PrevChar(const char *s) |
6537
dbf4fa55395f
(svn r9003) -Codechange: Introduce a function Utf8PrevCharLen that finds the starting character of an UTF-8 sequence from a given position and returns the length to the first UTF-8 encoding byte of that sequence.
Darkvater
parents:
5889
diff
changeset
|
129 |
{ |
6540
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
130 |
const char *ret = s; |
9191
ae14770c829a
(svn r12368) -Codechange: use explicit body for loops and conditions and remove -Wno-empty-body from the configure script
smatz
parents:
8844
diff
changeset
|
131 |
while (IsUtf8Part(*--ret)) {} |
6540
4db000b2cc41
(svn r9011) -Codechange (r9003): Rework Utf8PrevChar so that it returns a pointer to the previous UTF8 character's first byte instead of a byte-length offset
Darkvater
parents:
6537
diff
changeset
|
132 |
return (char*)ret; |
6537
dbf4fa55395f
(svn r9003) -Codechange: Introduce a function Utf8PrevCharLen that finds the starting character of an UTF-8 sequence from a given position and returns the length to the first UTF-8 encoding byte of that sequence.
Darkvater
parents:
5889
diff
changeset
|
133 |
} |
dbf4fa55395f
(svn r9003) -Codechange: Introduce a function Utf8PrevCharLen that finds the starting character of an UTF-8 sequence from a given position and returns the length to the first UTF-8 encoding byte of that sequence.
Darkvater
parents:
5889
diff
changeset
|
134 |
|
5108
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
135 |
|
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
136 |
static inline bool IsPrintable(WChar c) |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
137 |
{ |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
138 |
if (c < 0x20) return false; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
139 |
if (c < 0xE000) return true; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
140 |
if (c < 0xE200) return false; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
141 |
return true; |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
142 |
} |
dc67d70b5a45
(svn r7182) -Feature: Merge utf8 branch. This brings us support for Unicode/UTF-8 and the option for fonts rendered by FreeType. Language changes to come.
peter1138
parents:
5101
diff
changeset
|
143 |
|
6541
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
144 |
/** |
9207
2b390a312384
(svn r12385) -Fix [FS#ln]: Non-breaking spaces should not be broken.
peter1138
parents:
9191
diff
changeset
|
145 |
* Check whether UNICODE character is whitespace or not, i.e. whether |
2b390a312384
(svn r12385) -Fix [FS#ln]: Non-breaking spaces should not be broken.
peter1138
parents:
9191
diff
changeset
|
146 |
* this is a potential line-break character. |
6541
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
147 |
* @param c UNICODE character to check |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
148 |
* @return a boolean value whether 'c' is a whitespace character or not |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
149 |
* @see http://www.fileformat.info/info/unicode/category/Zs/list.htm |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
150 |
*/ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
151 |
static inline bool IsWhitespace(WChar c) |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
152 |
{ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
153 |
return |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
154 |
c == 0x0020 /* SPACE */ || |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
155 |
c == 0x3000 /* IDEOGRAPHIC SPACE */ |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
156 |
; |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
157 |
} |
ff575414f0d2
(svn r9012) -Fix/Feature (UTF8): When cutting strings into multiple lines also take into consideration whitespace characters of more than 1 byte length (eg IDEOGRAPHIC SPACE, IsWhitespace() function). When trimming such strings, account for multiple-byte long sequences so use *Utf8PrevChar(v) = '\0'.
Darkvater
parents:
6540
diff
changeset
|
158 |
|
8710
52015340050c
(svn r11777) -Codechange: split the string header and make do not include it when it's not necessary.
rubidium
parents:
8609
diff
changeset
|
159 |
#endif /* STRING_FUNC_H */ |