1 /*
2 * Copyright (C) the libgit2 contributors. All rights reserved.
3 *
4 * This file is part of libgit2, distributed under the GNU GPL v2 with
5 * a Linking Exception. For full terms see the included COPYING file.
6 */
7 #ifndef INCLUDE_buf_text_h__
8 #define INCLUDE_buf_text_h__
9
10 #include "common.h"
11
12 #include "buffer.h"
13
14 typedef enum {
15 GIT_BOM_NONE = 0,
16 GIT_BOM_UTF8 = 1,
17 GIT_BOM_UTF16_LE = 2,
18 GIT_BOM_UTF16_BE = 3,
19 GIT_BOM_UTF32_LE = 4,
20 GIT_BOM_UTF32_BE = 5
21 } git_bom_t;
22
23 typedef struct {
24 git_bom_t bom; /* BOM found at head of text */
25 unsigned int nul, cr, lf, crlf; /* NUL, CR, LF and CRLF counts */
26 unsigned int printable, nonprintable; /* These are just approximations! */
27 } git_buf_text_stats;
28
29 /**
30 * Append string to buffer, prefixing each character from `esc_chars` with
31 * `esc_with` string.
32 *
33 * @param buf Buffer to append data to
34 * @param string String to escape and append
35 * @param esc_chars Characters to be escaped
36 * @param esc_with String to insert in from of each found character
37 * @return 0 on success, <0 on failure (probably allocation problem)
38 */
39 extern int git_buf_text_puts_escaped(
40 git_buf *buf,
41 const char *string,
42 const char *esc_chars,
43 const char *esc_with);
44
45 /**
46 * Append string escaping characters that are regex special
47 */
git_buf_text_puts_escape_regex(git_buf * buf,const char * string)48 GIT_INLINE(int) git_buf_text_puts_escape_regex(git_buf *buf, const char *string)
49 {
50 return git_buf_text_puts_escaped(buf, string, "^.[]$()|*+?{}\\", "\\");
51 }
52
53 /**
54 * Unescape all characters in a buffer in place
55 *
56 * I.e. remove backslashes
57 */
58 extern void git_buf_text_unescape(git_buf *buf);
59
60 /**
61 * Replace all \r\n with \n.
62 *
63 * @return 0 on success, -1 on memory error
64 */
65 extern int git_buf_text_crlf_to_lf(git_buf *tgt, const git_buf *src);
66
67 /**
68 * Replace all \n with \r\n. Does not modify existing \r\n.
69 *
70 * @return 0 on success, -1 on memory error
71 */
72 extern int git_buf_text_lf_to_crlf(git_buf *tgt, const git_buf *src);
73
74 /**
75 * Fill buffer with the common prefix of a array of strings
76 *
77 * Buffer will be set to empty if there is no common prefix
78 */
79 extern int git_buf_text_common_prefix(git_buf *buf, const git_strarray *strs);
80
81 /**
82 * Check quickly if buffer looks like it contains binary data
83 *
84 * @param buf Buffer to check
85 * @return true if buffer looks like non-text data
86 */
87 extern bool git_buf_text_is_binary(const git_buf *buf);
88
89 /**
90 * Check quickly if buffer contains a NUL byte
91 *
92 * @param buf Buffer to check
93 * @return true if buffer contains a NUL byte
94 */
95 extern bool git_buf_text_contains_nul(const git_buf *buf);
96
97 /**
98 * Check if a buffer begins with a UTF BOM
99 *
100 * @param bom Set to the type of BOM detected or GIT_BOM_NONE
101 * @param buf Buffer in which to check the first bytes for a BOM
102 * @return Number of bytes of BOM data (or 0 if no BOM found)
103 */
104 extern int git_buf_text_detect_bom(git_bom_t *bom, const git_buf *buf);
105
106 /**
107 * Gather stats for a piece of text
108 *
109 * Fill the `stats` structure with counts of unreadable characters, carriage
110 * returns, etc, so it can be used in heuristics. This automatically skips
111 * a trailing EOF (\032 character). Also it will look for a BOM at the
112 * start of the text and can be told to skip that as well.
113 *
114 * @param stats Structure to be filled in
115 * @param buf Text to process
116 * @param skip_bom Exclude leading BOM from stats if true
117 * @return Does the buffer heuristically look like binary data
118 */
119 extern bool git_buf_text_gather_stats(
120 git_buf_text_stats *stats, const git_buf *buf, bool skip_bom);
121
122 #endif
123