src/buf_text.h

   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 "buffer.h"
  11
  12 typedef enum {
  13         GIT_BOM_NONE = 0,
  14         GIT_BOM_UTF8 = 1,
  15         GIT_BOM_UTF16_LE = 2,
  16         GIT_BOM_UTF16_BE = 3,
  17         GIT_BOM_UTF32_LE = 4,
  18         GIT_BOM_UTF32_BE = 5
  19 } git_bom_t;
  20
  21 typedef struct {
  22         git_bom_t bom; /* BOM found at head of text */
  23         unsigned int nul, cr, lf, crlf; /* NUL, CR, LF and CRLF counts */
  24         unsigned int printable, nonprintable; /* These are just approximations! */
  25 } git_buf_text_stats;
  26
  27 /**
  28  * Append string to buffer, prefixing each character from `esc_chars` with
  29  * `esc_with` string.
  30  *
  31  * @param buf Buffer to append data to
  32  * @param string String to escape and append
  33  * @param esc_chars Characters to be escaped
  34  * @param esc_with String to insert in from of each found character
  35  * @return 0 on success, <0 on failure (probably allocation problem)
  36  */
  37 extern int git_buf_text_puts_escaped(
  38         git_buf *buf,
  39         const char *string,
  40         const char *esc_chars,
  41         const char *esc_with);
  42
  43 /**
  44  * Append string escaping characters that are regex special
  45  */
  46 GIT_INLINE(int) git_buf_text_puts_escape_regex(git_buf *buf, const char *string)
  47 {
  48         return git_buf_text_puts_escaped(buf, string, "^.[]$()|*+?{}\\", "\\");
  49 }
  50
  51 /**
  52  * Unescape all characters in a buffer in place
  53  *
  54  * I.e. remove backslashes
  55  */
  56 extern void git_buf_text_unescape(git_buf *buf);
  57
  58 /**
  59  * Fill buffer with the common prefix of a array of strings
  60  *
  61  * Buffer will be set to empty if there is no common prefix
  62  */
  63 extern int git_buf_text_common_prefix(git_buf *buf, const git_strarray *strs);
  64
  65 /**
  66  * Check quickly if buffer looks like it contains binary data
  67  *
  68  * @param buf Buffer to check
  69  * @return true if buffer looks like non-text data
  70  */
  71 extern bool git_buf_text_is_binary(const git_buf *buf);
  72
  73 /**
  74  * Check quickly if buffer contains a NUL byte
  75  *
  76  * @param buf Buffer to check
  77  * @return true if buffer contains a NUL byte
  78  */
  79 extern bool git_buf_text_contains_nul(const git_buf *buf);
  80
  81 /**
  82  * Check if a buffer begins with a UTF BOM
  83  *
  84  * @param bom Set to the type of BOM detected or GIT_BOM_NONE
  85  * @param buf Buffer in which to check the first bytes for a BOM
  86  * @param offset Offset into buffer to look for BOM
  87  * @return Number of bytes of BOM data (or 0 if no BOM found)
  88  */
  89 extern int git_buf_text_detect_bom(
  90         git_bom_t *bom, const git_buf *buf, size_t offset);
  91
  92 /**
  93  * Gather stats for a piece of text
  94  *
  95  * Fill the `stats` structure with counts of unreadable characters, carriage
  96  * returns, etc, so it can be used in heuristics.  This automatically skips
  97  * a trailing EOF (\032 character).  Also it will look for a BOM at the
  98  * start of the text and can be told to skip that as well.
  99  *
 100  * @param stats Structure to be filled in
 101  * @param buf Text to process
 102  * @param skip_bom Exclude leading BOM from stats if true
 103  * @return Does the buffer heuristically look like binary data
 104  */
 105 extern bool git_buf_text_gather_stats(
 106         git_buf_text_stats *stats, const git_buf *buf, bool skip_bom);
 107
 108 #endif