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_sys_hashsig_h__ 8 #define INCLUDE_sys_hashsig_h__ 9 10 #include "git2/common.h" 11 12 GIT_BEGIN_DECL 13 14 /** 15 * Similarity signature of arbitrary text content based on line hashes 16 */ 17 typedef struct git_hashsig git_hashsig; 18 19 /** 20 * Options for hashsig computation 21 * 22 * The options GIT_HASHSIG_NORMAL, GIT_HASHSIG_IGNORE_WHITESPACE, 23 * GIT_HASHSIG_SMART_WHITESPACE are exclusive and should not be combined. 24 */ 25 typedef enum { 26 /** 27 * Use all data 28 */ 29 GIT_HASHSIG_NORMAL = 0, 30 31 /** 32 * Ignore whitespace 33 */ 34 GIT_HASHSIG_IGNORE_WHITESPACE = (1 << 0), 35 36 /** 37 * Ignore \r and all space after \n 38 */ 39 GIT_HASHSIG_SMART_WHITESPACE = (1 << 1), 40 41 /** 42 * Allow hashing of small files 43 */ 44 GIT_HASHSIG_ALLOW_SMALL_FILES = (1 << 2) 45 } git_hashsig_option_t; 46 47 /** 48 * Compute a similarity signature for a text buffer 49 * 50 * If you have passed the option GIT_HASHSIG_IGNORE_WHITESPACE, then the 51 * whitespace will be removed from the buffer while it is being processed, 52 * modifying the buffer in place. Sorry about that! 53 * 54 * @param out The computed similarity signature. 55 * @param buf The input buffer. 56 * @param buflen The input buffer size. 57 * @param opts The signature computation options (see above). 58 * @return 0 on success, GIT_EBUFS if the buffer doesn't contain enough data to 59 * compute a valid signature (unless GIT_HASHSIG_ALLOW_SMALL_FILES is set), or 60 * error code. 61 */ 62 GIT_EXTERN(int) git_hashsig_create( 63 git_hashsig **out, 64 const char *buf, 65 size_t buflen, 66 git_hashsig_option_t opts); 67 68 /** 69 * Compute a similarity signature for a text file 70 * 71 * This walks through the file, only loading a maximum of 4K of file data at 72 * a time. Otherwise, it acts just like `git_hashsig_create`. 73 * 74 * @param out The computed similarity signature. 75 * @param path The path to the input file. 76 * @param opts The signature computation options (see above). 77 * @return 0 on success, GIT_EBUFS if the buffer doesn't contain enough data to 78 * compute a valid signature (unless GIT_HASHSIG_ALLOW_SMALL_FILES is set), or 79 * error code. 80 */ 81 GIT_EXTERN(int) git_hashsig_create_fromfile( 82 git_hashsig **out, 83 const char *path, 84 git_hashsig_option_t opts); 85 86 /** 87 * Release memory for a content similarity signature 88 * 89 * @param sig The similarity signature to free. 90 */ 91 GIT_EXTERN(void) git_hashsig_free(git_hashsig *sig); 92 93 /** 94 * Measure similarity score between two similarity signatures 95 * 96 * @param a The first similarity signature to compare. 97 * @param b The second similarity signature to compare. 98 * @return [0 to 100] on success as the similarity score, or error code. 99 */ 100 GIT_EXTERN(int) git_hashsig_compare( 101 const git_hashsig *a, 102 const git_hashsig *b); 103 104 GIT_END_DECL 105 106 #endif 107
