diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/diff_tform.c | 64 | ||||
| -rw-r--r-- | src/hashsig.c | 5 | ||||
| -rw-r--r-- | src/hashsig.h | 6 |
3 files changed, 71 insertions, 4 deletions
diff --git a/src/diff_tform.c b/src/diff_tform.c index 3939230b7..61cf5b392 100644 --- a/src/diff_tform.c +++ b/src/diff_tform.c @@ -169,6 +169,60 @@ int git_diff_merge( return error; } +#define FIND_SIMILAR_HASHSIG(NAME,OPT) \ + static int find_similar__hashsig_for_file ## NAME( \ + void **out, const git_diff_file *f, const char *path, void *p) { \ + GIT_UNUSED(f); GIT_UNUSED(p); \ + return git_hashsig_create_fromfile((git_hashsig **)out, path, OPT); \ + } \ + static int find_similar__hashsig_for_buf ## NAME( \ + void **out, const git_diff_file *f, const char *buf, size_t len, void *p) { \ + GIT_UNUSED(f); GIT_UNUSED(p); \ + return git_hashsig_create((git_hashsig **)out, buf, len, OPT); \ + } + +FIND_SIMILAR_HASHSIG(_default, GIT_HASHSIG_SMART_WHITESPACE); +FIND_SIMILAR_HASHSIG(_ignore_whitespace, GIT_HASHSIG_IGNORE_WHITESPACE); +FIND_SIMILAR_HASHSIG(_include_whitespace, GIT_HASHSIG_NORMAL); + +static void find_similar__hashsig_free(void *sig, void *payload) +{ + GIT_UNUSED(payload); + git_hashsig_free(sig); +} + +static int find_similar__calc_similarity( + int *score, void *siga, void *sigb, void *payload) +{ + GIT_UNUSED(payload); + *score = git_hashsig_compare(siga, sigb); + return 0; +} + +static git_diff_similarity_metric find_similar__internal_metrics[3] = { + { + find_similar__hashsig_for_file_default, + find_similar__hashsig_for_buf_default, + find_similar__hashsig_free, + find_similar__calc_similarity, + NULL + }, + { + find_similar__hashsig_for_file_ignore_whitespace, + find_similar__hashsig_for_buf_ignore_whitespace, + find_similar__hashsig_free, + find_similar__calc_similarity, + NULL + }, + { + find_similar__hashsig_for_file_include_whitespace, + find_similar__hashsig_for_buf_include_whitespace, + find_similar__hashsig_free, + find_similar__calc_similarity, + NULL + } +}; + #define DEFAULT_THRESHOLD 50 #define DEFAULT_BREAK_REWRITE_THRESHOLD 60 #define DEFAULT_TARGET_LIMIT 200 @@ -237,6 +291,16 @@ static int normalize_find_opts( opts->target_limit = limit; } + /* for now, always assign the same internal metric */ + if (!opts->metric) { + if (opts->flags & GIT_DIFF_FIND_IGNORE_WHITESPACE) + opts->metric = &find_similar__internal_metrics[1]; + else if (opts->flags & GIT_DIFF_FIND_DONT_IGNORE_WHITESPACE) + opts->metric = &find_similar__internal_metrics[2]; + else + opts->metric = &find_similar__internal_metrics[0]; + } + return 0; } diff --git a/src/hashsig.c b/src/hashsig.c index a9cd8fa53..60649fd11 100644 --- a/src/hashsig.c +++ b/src/hashsig.c @@ -266,7 +266,8 @@ static git_hashsig *hashsig_alloc(git_hashsig_option_t opts) int git_hashsig_create( git_hashsig **out, - const git_buf *buf, + const char *buf, + size_t buflen, git_hashsig_option_t opts) { int error; @@ -274,7 +275,7 @@ int git_hashsig_create( git_hashsig *sig = hashsig_alloc(opts); GITERR_CHECK_ALLOC(sig); - error = hashsig_add_hashes(sig, buf->ptr, buf->size, &prog); + error = hashsig_add_hashes(sig, buf, buflen, &prog); if (!error) error = hashsig_finalize_hashes(sig); diff --git a/src/hashsig.h b/src/hashsig.h index 70b47f5f3..8c920cbf1 100644 --- a/src/hashsig.h +++ b/src/hashsig.h @@ -7,7 +7,7 @@ #ifndef INCLUDE_hashsig_h__ #define INCLUDE_hashsig_h__ -#include "buffer.h" +#include "common.h" /** * Similarity signature of line hashes for a buffer @@ -32,11 +32,13 @@ typedef enum { * * @param out The array of hashed runs representing the file content * @param buf The contents of the file to hash + * @param buflen The length of the data at `buf` * @param generate_pairwise_hashes Should pairwise runs be hashed */ extern int git_hashsig_create( git_hashsig **out, - const git_buf *buf, + const char *buf, + size_t buflen, git_hashsig_option_t opts); /** |