diff --git a/include/git2/diff.h b/include/git2/diff.h index 71a8b72bf..c989ba4ee 100644 --- a/include/git2/diff.h +++ b/include/git2/diff.h @@ -391,7 +391,7 @@ typedef enum { */ GIT_DIFF_LINE_FILE_HDR = 'F', GIT_DIFF_LINE_HUNK_HDR = 'H', - GIT_DIFF_LINE_BINARY = 'B' + GIT_DIFF_LINE_BINARY = 'B' /**< For "Binary files x and y differ" */ } git_diff_line_t; /** @@ -942,6 +942,28 @@ GIT_EXTERN(int) git_diff_patch_get_line_in_hunk( size_t hunk_idx, size_t line_of_hunk); +/** + * Look up size of patch diff data in bytes + * + * This returns the raw size of the patch data. This only includes the + * actual data from the lines of the diff, not the file or hunk headers. + * + * If you pass `include_context` as true (non-zero), this will be the size + * of all of the diff output; if you pass it as false (zero), this will + * only include the actual changed lines (as if `context_lines` was 0). + * + * @param patch A git_diff_patch representing changes to one file + * @param include_context Include context lines in size if non-zero + * @param include_hunk_headers Include hunk header lines if non-zero + * @param include_file_headers Include file header lines if non-zero + * @return The number of bytes of data + */ +GIT_EXTERN(size_t) git_diff_patch_size( + git_diff_patch *patch, + int include_context, + int include_hunk_headers, + int include_file_headers); + /** * Serialize the patch to text via callback. * diff --git a/src/diff.h b/src/diff.h index d1bec00c6..bec7e27d7 100644 --- a/src/diff.h +++ b/src/diff.h @@ -82,6 +82,13 @@ extern const char *git_diff_delta__path(const git_diff_delta *delta); extern bool git_diff_delta__should_skip( const git_diff_options *opts, const git_diff_delta *delta); +extern int git_diff_delta__format_file_header( + git_buf *out, + const git_diff_delta *delta, + const char *oldpfx, + const char *newpfx, + int oid_strlen); + extern int git_diff__oid_for_file( git_repository *, const char *, uint16_t, git_off_t, git_oid *); diff --git a/src/diff_patch.c b/src/diff_patch.c index 69bb08198..cc45b6ddb 100644 --- a/src/diff_patch.c +++ b/src/diff_patch.c @@ -42,7 +42,7 @@ struct git_diff_patch { git_array_t(diff_patch_hunk) hunks; git_array_t(diff_patch_line) lines; size_t oldno, newno; - size_t content_size; + size_t content_size, context_size, header_size; git_pool flattened; }; @@ -810,6 +810,39 @@ notfound: return diff_error_outofrange(thing); } +size_t git_diff_patch_size( + git_diff_patch *patch, + int include_context, + int include_hunk_headers, + int include_file_headers) +{ + size_t out; + + assert(patch); + + out = patch->content_size; + + if (!include_context) + out -= patch->context_size; + + if (include_hunk_headers) + out += patch->header_size; + + if (include_file_headers) { + git_buf file_header = GIT_BUF_INIT; + + if (git_diff_delta__format_file_header( + &file_header, patch->delta, NULL, NULL, 0) < 0) + giterr_clear(); + else + out += git_buf_len(&file_header); + + git_buf_free(&file_header); + } + + return out; +} + git_diff_list *git_diff_patch__diff(git_diff_patch *patch) { return patch->diff; @@ -904,6 +937,8 @@ static int diff_patch_hunk_cb( hunk->header[header_len] = '\0'; hunk->header_len = header_len; + patch->header_size += header_len; + hunk->line_start = git_array_size(patch->lines); hunk->line_count = 0; @@ -924,6 +959,7 @@ static int diff_patch_line_cb( git_diff_patch *patch = payload; diff_patch_hunk *hunk; diff_patch_line *line; + const char *content_end = content + content_len; GIT_UNUSED(delta); GIT_UNUSED(range); @@ -938,34 +974,43 @@ static int diff_patch_line_cb( line->len = content_len; line->origin = line_origin; - patch->content_size += content_len; - /* do some bookkeeping so we can provide old/new line numbers */ - for (line->lines = 0; content_len > 0; --content_len) { + line->lines = 0; + while (content < content_end) if (*content++ == '\n') ++line->lines; - } + + patch->content_size += content_len; switch (line_origin) { case GIT_DIFF_LINE_ADDITION: + patch->content_size += 1; case GIT_DIFF_LINE_DEL_EOFNL: line->oldno = -1; line->newno = patch->newno; patch->newno += line->lines; break; case GIT_DIFF_LINE_DELETION: + patch->content_size += 1; case GIT_DIFF_LINE_ADD_EOFNL: line->oldno = patch->oldno; line->newno = -1; patch->oldno += line->lines; break; - default: + case GIT_DIFF_LINE_CONTEXT: + patch->content_size += 1; + patch->context_size += 1; + case GIT_DIFF_LINE_CONTEXT_EOFNL: + patch->context_size += content_len; line->oldno = patch->oldno; line->newno = patch->newno; patch->oldno += line->lines; patch->newno += line->lines; break; + default: + assert(false); + break; } hunk->line_count++; diff --git a/src/diff_print.c b/src/diff_print.c index 0de548813..4ddd72443 100644 --- a/src/diff_print.c +++ b/src/diff_print.c @@ -98,12 +98,12 @@ static int diff_print_one_compact( if (delta->old_file.path != delta->new_file.path && strcomp(delta->old_file.path,delta->new_file.path) != 0) - git_buf_printf(out, "%c\t%s%c -> %s%c\n", code, + git_buf_printf(out, "%c\t%s%c %s%c\n", code, delta->old_file.path, old_suffix, delta->new_file.path, new_suffix); else if (delta->old_file.mode != delta->new_file.mode && delta->old_file.mode != 0 && delta->new_file.mode != 0) - git_buf_printf(out, "%c\t%s%c (%o -> %o)\n", code, - delta->old_file.path, new_suffix, delta->old_file.mode, delta->new_file.mode); + git_buf_printf(out, "%c\t%s%c %s%c\n", code, + delta->old_file.path, old_suffix, delta->new_file.path, new_suffix); else if (old_suffix != ' ') git_buf_printf(out, "%c\t%s%c\n", code, delta->old_file.path, old_suffix); else @@ -198,13 +198,13 @@ int git_diff_print_raw( return error; } -static int diff_print_oid_range(diff_print_info *pi, const git_diff_delta *delta) +static int diff_print_oid_range( + git_buf *out, const git_diff_delta *delta, int oid_strlen) { - git_buf *out = pi->buf; char start_oid[GIT_OID_HEXSZ+1], end_oid[GIT_OID_HEXSZ+1]; - git_oid_tostr(start_oid, pi->oid_strlen, &delta->old_file.oid); - git_oid_tostr(end_oid, pi->oid_strlen, &delta->new_file.oid); + git_oid_tostr(start_oid, oid_strlen, &delta->old_file.oid); + git_oid_tostr(end_oid, oid_strlen, &delta->new_file.oid); /* TODO: Match git diff more closely */ if (delta->old_file.mode == delta->new_file.mode) { @@ -228,14 +228,65 @@ static int diff_print_oid_range(diff_print_info *pi, const git_diff_delta *delta return 0; } +static int diff_delta_format_with_paths( + git_buf *out, + const git_diff_delta *delta, + const char *oldpfx, + const char *newpfx, + const char *template) +{ + const char *oldpath = delta->old_file.path; + const char *newpath = delta->new_file.path; + + if (git_oid_iszero(&delta->old_file.oid)) { + oldpfx = ""; + oldpath = "/dev/null"; + } + if (git_oid_iszero(&delta->new_file.oid)) { + newpfx = ""; + newpath = "/dev/null"; + } + + return git_buf_printf(out, template, oldpfx, oldpath, newpfx, newpath); +} + +int git_diff_delta__format_file_header( + git_buf *out, + const git_diff_delta *delta, + const char *oldpfx, + const char *newpfx, + int oid_strlen) +{ + if (!oldpfx) + oldpfx = DIFF_OLD_PREFIX_DEFAULT; + if (!newpfx) + newpfx = DIFF_NEW_PREFIX_DEFAULT; + if (!oid_strlen) + oid_strlen = GIT_ABBREV_DEFAULT + 1; + + git_buf_clear(out); + + git_buf_printf(out, "diff --git %s%s %s%s\n", + oldpfx, delta->old_file.path, newpfx, delta->new_file.path); + + if (diff_print_oid_range(out, delta, oid_strlen) < 0) + return -1; + + if ((delta->flags & GIT_DIFF_FLAG_BINARY) == 0) + diff_delta_format_with_paths( + out, delta, oldpfx, newpfx, "--- %s%s\n+++ %s%s\n"); + + return git_buf_oom(out) ? -1 : 0; +} + static int diff_print_patch_file( const git_diff_delta *delta, float progress, void *data) { diff_print_info *pi = data; - const char *oldpfx = pi->diff ? pi->diff->opts.old_prefix : NULL; - const char *oldpath = delta->old_file.path; - const char *newpfx = pi->diff ? pi->diff->opts.new_prefix : NULL; - const char *newpath = delta->new_file.path; + const char *oldpfx = + pi->diff ? pi->diff->opts.old_prefix : DIFF_OLD_PREFIX_DEFAULT; + const char *newpfx = + pi->diff ? pi->diff->opts.new_prefix : DIFF_NEW_PREFIX_DEFAULT; uint32_t opts_flags = pi->diff ? pi->diff->opts.flags : GIT_DIFF_NORMAL; GIT_UNUSED(progress); @@ -247,33 +298,8 @@ static int diff_print_patch_file( (opts_flags & GIT_DIFF_INCLUDE_UNTRACKED_CONTENT) == 0)) return 0; - if (!oldpfx) - oldpfx = DIFF_OLD_PREFIX_DEFAULT; - if (!newpfx) - newpfx = DIFF_NEW_PREFIX_DEFAULT; - - git_buf_clear(pi->buf); - git_buf_printf(pi->buf, "diff --git %s%s %s%s\n", - oldpfx, delta->old_file.path, newpfx, delta->new_file.path); - - if (diff_print_oid_range(pi, delta) < 0) - return -1; - - if (git_oid_iszero(&delta->old_file.oid)) { - oldpfx = ""; - oldpath = "/dev/null"; - } - if (git_oid_iszero(&delta->new_file.oid)) { - newpfx = ""; - newpath = "/dev/null"; - } - - if ((delta->flags & GIT_DIFF_FLAG_BINARY) == 0) { - git_buf_printf(pi->buf, "--- %s%s\n", oldpfx, oldpath); - git_buf_printf(pi->buf, "+++ %s%s\n", newpfx, newpath); - } - - if (git_buf_oom(pi->buf)) + if (git_diff_delta__format_file_header( + pi->buf, delta, oldpfx, newpfx, pi->oid_strlen) < 0) return -1; if (pi->print_cb(delta, NULL, GIT_DIFF_LINE_FILE_HDR, @@ -284,10 +310,10 @@ static int diff_print_patch_file( return 0; git_buf_clear(pi->buf); - git_buf_printf( - pi->buf, "Binary files %s%s and %s%s differ\n", - oldpfx, oldpath, newpfx, newpath); - if (git_buf_oom(pi->buf)) + + if (diff_delta_format_with_paths( + pi->buf, delta, oldpfx, newpfx, + "Binary files %s%s and %s%s differ\n") < 0) return -1; if (pi->print_cb(delta, NULL, GIT_DIFF_LINE_BINARY, diff --git a/tests-clar/diff/patch.c b/tests-clar/diff/patch.c index 3f14a0de7..6a33fa990 100644 --- a/tests-clar/diff/patch.c +++ b/tests-clar/diff/patch.c @@ -128,6 +128,11 @@ void test_diff_patch__to_string(void) cl_assert_equal_s(expected, text); + cl_assert_equal_sz(31, git_diff_patch_size(patch, 0, 0, 0)); + cl_assert_equal_sz(31, git_diff_patch_size(patch, 1, 0, 0)); + cl_assert_equal_sz(31 + 16, git_diff_patch_size(patch, 1, 1, 0)); + cl_assert_equal_sz(strlen(expected), git_diff_patch_size(patch, 1, 1, 1)); + git__free(text); git_diff_patch_free(patch); git_diff_list_free(diff); @@ -408,7 +413,7 @@ void test_diff_patch__hunks_have_correct_line_numbers(void) static void check_single_patch_stats( git_repository *repo, size_t hunks, - size_t adds, size_t dels, size_t ctxt, + size_t adds, size_t dels, size_t ctxt, size_t *sizes, const char *expected) { git_diff_list *diff; @@ -437,6 +442,18 @@ static void check_single_patch_stats( cl_git_pass(git_diff_patch_to_str(&text, patch)); cl_assert_equal_s(expected, text); git__free(text); + + cl_assert_equal_sz( + strlen(expected), git_diff_patch_size(patch, 1, 1, 1)); + } + + if (sizes) { + if (sizes[0]) + cl_assert_equal_sz(sizes[0], git_diff_patch_size(patch, 0, 0, 0)); + if (sizes[1]) + cl_assert_equal_sz(sizes[1], git_diff_patch_size(patch, 1, 0, 0)); + if (sizes[2]) + cl_assert_equal_sz(sizes[2], git_diff_patch_size(patch, 1, 1, 0)); } /* walk lines in hunk with basic sanity checks */ @@ -481,6 +498,23 @@ void test_diff_patch__line_counts_with_eofnl(void) git_buf content = GIT_BUF_INIT; const char *end; git_index *index; + const char *expected = + /* below is pasted output of 'git diff' with fn context removed */ + "diff --git a/songof7cities.txt b/songof7cities.txt\n" + "index 378a7d9..3d0154e 100644\n" + "--- a/songof7cities.txt\n" + "+++ b/songof7cities.txt\n" + "@@ -42,7 +42,7 @@ With peoples undefeated of the dark, enduring blood.\n" + " \n" + " To the sound of trumpets shall their seed restore my Cities\n" + " Wealthy and well-weaponed, that once more may I behold\n" + "-All the world go softly when it walks before my Cities,\n" + "+#All the world go softly when it walks before my Cities,\n" + " And the horses and the chariots fleeing from them as of old!\n" + " \n" + " -- Rudyard Kipling\n" + "\\ No newline at end of file\n"; + size_t expected_sizes[3] = { 115, 119 + 115 + 114, 119 + 115 + 114 + 71 }; g_repo = cl_git_sandbox_init("renames"); @@ -495,14 +529,14 @@ void test_diff_patch__line_counts_with_eofnl(void) git_buf_consume(&content, end); cl_git_rewritefile("renames/songof7cities.txt", content.ptr); - check_single_patch_stats(g_repo, 1, 0, 1, 3, NULL); + check_single_patch_stats(g_repo, 1, 0, 1, 3, NULL, NULL); /* remove trailing whitespace */ git_buf_rtrim(&content); cl_git_rewritefile("renames/songof7cities.txt", content.ptr); - check_single_patch_stats(g_repo, 2, 1, 2, 6, NULL); + check_single_patch_stats(g_repo, 2, 1, 2, 6, NULL, NULL); /* add trailing whitespace */ @@ -514,7 +548,7 @@ void test_diff_patch__line_counts_with_eofnl(void) cl_git_pass(git_buf_putc(&content, '\n')); cl_git_rewritefile("renames/songof7cities.txt", content.ptr); - check_single_patch_stats(g_repo, 1, 1, 1, 3, NULL); + check_single_patch_stats(g_repo, 1, 1, 1, 3, NULL, NULL); /* no trailing whitespace as context line */ @@ -537,22 +571,7 @@ void test_diff_patch__line_counts_with_eofnl(void) cl_git_rewritefile("renames/songof7cities.txt", content.ptr); check_single_patch_stats( - g_repo, 1, 1, 1, 6, - /* below is pasted output of 'git diff' with fn context removed */ - "diff --git a/songof7cities.txt b/songof7cities.txt\n" - "index 378a7d9..3d0154e 100644\n" - "--- a/songof7cities.txt\n" - "+++ b/songof7cities.txt\n" - "@@ -42,7 +42,7 @@ With peoples undefeated of the dark, enduring blood.\n" - " \n" - " To the sound of trumpets shall their seed restore my Cities\n" - " Wealthy and well-weaponed, that once more may I behold\n" - "-All the world go softly when it walks before my Cities,\n" - "+#All the world go softly when it walks before my Cities,\n" - " And the horses and the chariots fleeing from them as of old!\n" - " \n" - " -- Rudyard Kipling\n" - "\\ No newline at end of file\n"); + g_repo, 1, 1, 1, 6, expected_sizes, expected); git_buf_free(&content); git_config_free(cfg);