t6022: New tests checking for unnecessary updates of files
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 3fd9e0c70315727f835f39d9a0230ee9f24e7ea8..559bf574a88f97402f32e1bd4ab51f6f0c9d5026 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -23,7 +23,7 @@
 #endif
 
 static int diff_detect_rename_default;
-static int diff_rename_limit_default = 200;
+static int diff_rename_limit_default = 400;
 static int diff_suppress_blank_empty;
 int diff_use_color_default = -1;
 static const char *diff_word_regex_cfg;
@@ -628,7 +628,7 @@ struct diff_words_style {
        const char *newline;
 };
 
-struct diff_words_style diff_words_styles[] = {
+static struct diff_words_style diff_words_styles[] = {
        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
@@ -1050,8 +1050,16 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                        emit_line(ecbdata->opt, plain, reset, line, len);
                        fputs("~\n", ecbdata->opt->file);
                } else {
-                       /* don't print the prefix character */
-                       emit_line(ecbdata->opt, plain, reset, line+1, len-1);
+                       /*
+                        * Skip the prefix character, if any.  With
+                        * diff_suppress_blank_empty, there may be
+                        * none.
+                        */
+                       if (line[0] != '\n') {
+                             line++;
+                             len--;
+                       }
+                       emit_line(ecbdata->opt, plain, reset, line, len);
                }
                return;
        }
@@ -1242,7 +1250,7 @@ static void show_stats(struct diffstat_t *data, struct diff_options *options)
        uintmax_t max_change = 0, max_len = 0;
        int total_files = data->nr;
        int width, name_width;
-       const char *reset, *set, *add_c, *del_c;
+       const char *reset, *add_c, *del_c;
        const char *line_prefix = "";
        struct strbuf *msg = NULL;
 
@@ -1269,7 +1277,6 @@ static void show_stats(struct diffstat_t *data, struct diff_options *options)
 
        /* Find the longest filename and max number of changes */
        reset = diff_get_color_opt(options, DIFF_RESET);
-       set   = diff_get_color_opt(options, DIFF_PLAIN);
        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 
@@ -1539,8 +1546,36 @@ static void show_dirstat(struct diff_options *options)
                struct diff_filepair *p = q->queue[i];
                const char *name;
                unsigned long copied, added, damage;
+               int content_changed;
+
+               name = p->two->path ? p->two->path : p->one->path;
+
+               if (p->one->sha1_valid && p->two->sha1_valid)
+                       content_changed = hashcmp(p->one->sha1, p->two->sha1);
+               else
+                       content_changed = 1;
 
-               name = p->one->path ? p->one->path : p->two->path;
+               if (!content_changed) {
+                       /*
+                        * The SHA1 has not changed, so pre-/post-content is
+                        * identical. We can therefore skip looking at the
+                        * file contents altogether.
+                        */
+                       damage = 0;
+                       goto found_damage;
+               }
+
+               if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
+                       /*
+                        * In --dirstat-by-file mode, we don't really need to
+                        * look at the actual file contents at all.
+                        * The fact that the SHA1 changed is enough for us to
+                        * add this file to the list of results
+                        * (with each file contributing equal damage).
+                        */
+                       damage = 1;
+                       goto found_damage;
+               }
 
                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
                        diff_populate_filespec(p->one, 0);
@@ -1564,14 +1599,18 @@ static void show_dirstat(struct diff_options *options)
                /*
                 * Original minus copied is the removed material,
                 * added is the new material.  They are both damages
-                * made to the preimage. In --dirstat-by-file mode, count
-                * damaged files, not damaged lines. This is done by
-                * counting only a single damaged line per file.
+                * made to the preimage.
+                * If the resulting damage is zero, we know that
+                * diffcore_count_changes() considers the two entries to
+                * be identical, but since content_changed is true, we
+                * know that there must have been _some_ kind of change,
+                * so we force all entries to have damage > 0.
                 */
                damage = (p->one->size - copied) + added;
-               if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
+               if (!damage)
                        damage = 1;
 
+found_damage:
                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
                dir.files[dir.nr].name = name;
                dir.files[dir.nr].changed = damage;
@@ -3956,6 +3995,28 @@ static int is_summary_empty(const struct diff_queue_struct *q)
        return 1;
 }
 
+static const char rename_limit_warning[] =
+"inexact rename detection was skipped due to too many files.";
+
+static const char degrade_cc_to_c_warning[] =
+"only found copies from modified paths due to too many files.";
+
+static const char rename_limit_advice[] =
+"you may want to set your %s variable to at least "
+"%d and retry the command.";
+
+void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
+{
+       if (degraded_cc)
+               warning(degrade_cc_to_c_warning);
+       else if (needed)
+               warning(rename_limit_warning);
+       else
+               return;
+       if (0 < needed && needed < 32767)
+               warning(rename_limit_advice, varname, needed);
+}
+
 void diff_flush(struct diff_options *options)
 {
        struct diff_queue_struct *q = &diff_queued_diff;
@@ -4237,6 +4298,10 @@ void diffcore_std(struct diff_options *options)
 int diff_result_code(struct diff_options *opt, int status)
 {
        int result = 0;
+
+       diff_warn_rename_limit("diff.renamelimit",
+                              opt->needed_rename_limit,
+                              opt->degraded_cc_to_c);
        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
                return status;
@@ -4350,20 +4415,20 @@ void diff_change(struct diff_options *options,
                DIFF_OPT_SET(options, HAS_CHANGES);
 }
 
-void diff_unmerge(struct diff_options *options,
-                 const char *path,
-                 unsigned mode, const unsigned char *sha1)
+struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
 {
+       struct diff_filepair *pair;
        struct diff_filespec *one, *two;
 
        if (options->prefix &&
            strncmp(path, options->prefix, options->prefix_length))
-               return;
+               return NULL;
 
        one = alloc_filespec(path);
        two = alloc_filespec(path);
-       fill_filespec(one, sha1, mode);
-       diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
+       pair = diff_queue(&diff_queued_diff, one, two);
+       pair->is_unmerged = 1;
+       return pair;
 }
 
 static char *run_textconv(const char *pgm, struct diff_filespec *spec,