survey: checkpoint add blob pretty print

Signed-off-by: Jeff Hostetler <jeffhostetler@github.com>

builtin/survey.c

@@ -358,8 +358,9 @@ struct large_item {
 
 	/*
 	 * For blobs and trees the name field is the pathname of the
-	 * file or directory.  Root trees will have a zero-length
-	 * name.  The name field is not currenly used for commits.
+	 * file or directory (as reported by the treewalk). Root trees
+	 * are reported with a zero-length name, but we'll fix them up.
+	 * The name field is not currenly used for commits.
 	 */
 	struct strbuf *name;
 
@@ -387,13 +388,15 @@ struct large_item_vec {
 	const struct large_item_vec_labels *labels_json;
 	const struct large_item_vec_labels *labels_pretty;
 	uint64_t nr_items;
+	enum object_type type;
 	struct large_item items[FLEX_ARRAY]; /* nr_items */
 };
 
 static struct large_item_vec *alloc_large_item_vec(
 	const struct large_item_vec_labels *labels_json,
 	const struct large_item_vec_labels *labels_pretty,
-	uint64_t nr_items)
+	uint64_t nr_items,
+	enum object_type type)
 {
 	struct large_item_vec *vec;
 	size_t flex_len = nr_items * sizeof(struct large_item);
@@ -406,6 +409,7 @@ static struct large_item_vec *alloc_large_item_vec(
 	vec->labels_json = labels_json;
 	vec->labels_pretty = labels_pretty;
 	vec->nr_items = nr_items;
+	vec->type = type;
 
 	for (k = 0; k < nr_items; k++) {
 		struct strbuf *p = xcalloc(1, sizeof(struct strbuf));
@@ -470,6 +474,15 @@ static void maybe_insert_large_item(struct large_item_vec *vec,
 		strbuf_reset(pbuf_temp);
 		if (name && *name)
 			strbuf_addstr(pbuf_temp, name);
+		else if (vec->type == OBJ_TREE) {
+			/*
+			 * NEEDSWORK: Would it be better to wait and create
+			 * a name of the form "<name_rev>^{tree}" after the
+			 * treewalk is finished?
+			 */
+			strbuf_addf(pbuf_temp, "%s^{tree}",
+				    oid_to_hex(containing_commit_oid));
+		}
 
 		/* push items[k..] down one and insert data for this item here */
 
@@ -654,7 +667,8 @@ static void alloc_commit_by_parents(void)
 	if (survey_opts.show_largest_commits_by_nr_parents)
 		survey_stats.commits.vec_largest_by_nr_parents =
 			alloc_large_item_vec(&json, &pretty,
-					     survey_opts.show_largest_commits_by_nr_parents);
+					     survey_opts.show_largest_commits_by_nr_parents,
+					     OBJ_COMMIT);
 }
 
 static void alloc_commit_by_size(void) {
@@ -670,7 +684,8 @@ static void alloc_commit_by_size(void) {
 	if (survey_opts.show_largest_commits_by_size_bytes)
 		survey_stats.commits.vec_largest_by_size_bytes =
 			alloc_large_item_vec(&json, &pretty,
-					     survey_opts.show_largest_commits_by_size_bytes);
+					     survey_opts.show_largest_commits_by_size_bytes,
+					     OBJ_COMMIT);
 };
 
 static void alloc_tree_by_entries(void)
@@ -687,7 +702,8 @@ static void alloc_tree_by_entries(void)
 	if (survey_opts.show_largest_trees_by_nr_entries)
 		survey_stats.trees.vec_largest_by_nr_entries =
 			alloc_large_item_vec(&json, &pretty,
-					     survey_opts.show_largest_trees_by_nr_entries);
+					     survey_opts.show_largest_trees_by_nr_entries,
+					     OBJ_TREE);
 }
 
 static void alloc_tree_by_size(void)
@@ -704,7 +720,8 @@ static void alloc_tree_by_size(void)
 	if (survey_opts.show_largest_trees_by_size_bytes)
 		survey_stats.trees.vec_largest_by_size_bytes =
 			alloc_large_item_vec(&json, &pretty,
-					     survey_opts.show_largest_trees_by_size_bytes);
+					     survey_opts.show_largest_trees_by_size_bytes,
+					     OBJ_TREE);
 }
 
 static void alloc_blob_by_size(void)
@@ -721,7 +738,8 @@ static void alloc_blob_by_size(void)
 	if (survey_opts.show_largest_blobs_by_size_bytes)
 		survey_stats.blobs.vec_largest_by_size_bytes =
 			alloc_large_item_vec(&json, &pretty,
-					     survey_opts.show_largest_blobs_by_size_bytes);
+					     survey_opts.show_largest_blobs_by_size_bytes,
+					     OBJ_BLOB);
 }
 
 static void do_load_refs(struct ref_array *ref_array)
@@ -1788,6 +1806,7 @@ static void fmt_pbin_row(struct strbuf *buf,
 
 static void fmt_large_item_hdr(struct strbuf *buf,
 			       int indent,
+			       int name_length,
 			       int name_rev_length,
 			       const char *item_hdr_label)
 {
@@ -1797,13 +1816,16 @@ static void fmt_large_item_hdr(struct strbuf *buf,
 		strbuf_addchars(buf, ' ', indent);
 
 	strbuf_addf(buf, "%-*s | %14s", column0, "OID", item_hdr_label);
+	if (name_length)
+		strbuf_addf(buf, " | %-*s", name_length, "Name");
 	strbuf_addf(buf, " | %-*s", name_rev_length, "Name Rev");
 
 	strbuf_addch(buf, '\n');
 }
 
 static void fmt_large_item_hr(struct strbuf *buf,
 			      int indent,
+			      int name_length,
 			      int name_rev_length)
 {
 	int column0 = the_hash_algo->hexsz;
@@ -1814,6 +1836,10 @@ static void fmt_large_item_hr(struct strbuf *buf,
 	strbuf_addchars(buf, '-', column0);
 	strbuf_addstr(buf, "-+-");
 	strbuf_addchars(buf, '-', 14);
+	if (name_length) {
+		strbuf_addstr(buf, "-+-");
+		strbuf_addchars(buf, '-', name_length);
+	}
 	strbuf_addstr(buf, "-+-");
 	strbuf_addchars(buf, '-', name_rev_length);
 
@@ -1822,6 +1848,8 @@ static void fmt_large_item_hr(struct strbuf *buf,
 
 static void fmt_large_item_row(struct strbuf *buf,
 			       int indent,
+			       int name_length,
+			       int name_rev_length,
 			       struct large_item *pitem)
 {
 	int column0 = the_hash_algo->hexsz;
@@ -1832,7 +1860,10 @@ static void fmt_large_item_row(struct strbuf *buf,
 	strbuf_addf(buf, "%-*s | %14"PRIu64,
 		    column0, oid_to_hex(&pitem->oid),
 		    pitem->size);
-	strbuf_addf(buf, " | %-s", pitem->name_rev->buf);
+	if (name_length)
+		strbuf_addf(buf, " | %-*s", name_length,
+			    (pitem->name ? pitem->name->buf: ""));
+	strbuf_addf(buf, " | %-*s", name_rev_length, pitem->name_rev->buf);
 
 	strbuf_addch(buf, '\n');
 }
@@ -1841,23 +1872,34 @@ static void fmt_large_item_vec(struct strbuf *buf,
 			       int indent,
 			       struct large_item_vec *pvec)
 {
+	int name_length = 0;
 	int name_rev_length = 10;
 	int k;
 
+	if (pvec->type != OBJ_COMMIT) {
+		/* Add "Name" column for trees and blobs. */
+		for (k = 0; k < pvec->nr_items; k++)
+			if (pvec->items[k].name && pvec->items[k].name->len > name_length)
+				name_length = pvec->items[k].name->len;
+		if (name_length)
+			if (name_length < 4) /* strlen("Name") */
+				name_length = 4;
+	}
+
 	for (k = 0; k < pvec->nr_items; k++)
 		if (pvec->items[k].name_rev->len > name_rev_length)
 			name_rev_length = pvec->items[k].name_rev->len;
 
 	strbuf_addch(buf, '\n');
 	fmt_txt_line(buf, indent, pvec->labels_pretty->dimension);
-	fmt_large_item_hr(buf, indent, name_rev_length);
-	fmt_large_item_hdr(buf, indent, name_rev_length, pvec->labels_pretty->item);
-	fmt_large_item_hr(buf, indent, name_rev_length);
+	fmt_large_item_hr(buf, indent, name_length, name_rev_length);
+	fmt_large_item_hdr(buf, indent, name_length, name_rev_length, pvec->labels_pretty->item);
+	fmt_large_item_hr(buf, indent, name_length, name_rev_length);
 
 	for (k = 0; k < pvec->nr_items; k++)
-		fmt_large_item_row(buf, indent, &pvec->items[k]);
+		fmt_large_item_row(buf, indent, name_length, name_rev_length, &pvec->items[k]);
 
-	fmt_large_item_hr(buf, indent, name_rev_length);
+	fmt_large_item_hr(buf, indent, name_length, name_rev_length);
 }
 
 static void fmt_commit_row(struct strbuf *buf,
@@ -2234,7 +2276,12 @@ static void pretty_print_trees(int indent)
 
 static void pretty_print_blobs(int indent)
 {
+	struct survey_stats_blobs *psb = &survey_stats.blobs;
+	struct survey_stats_base_object *base = &psb->base;
 	struct strbuf buf = STRBUF_INIT;
+	int indent1 = indent + 4;
+	int indent2 = indent + 8;
+	int indent3 = indent + 12;
 	int k;
 
 	const char *intro[] = {
@@ -2252,6 +2299,28 @@ static void pretty_print_blobs(int indent)
 	while (intro[k])
 		fmt_txt_line(&buf, indent, intro[k++]);
 
+	fmt_blob_row(&buf, indent1, "Total Number of Blobs", base->cnt_seen, NULL);
+
+	strbuf_addch(&buf, '\n');
+	fmt_txt_line(&buf, indent1, "Blob Count by Storage Location");
+	if (base->cnt_missing)
+		fmt_commit_row(&buf, indent2, "Missing", base->cnt_missing, NULL);
+	if (base->cnt_cached)
+		fmt_commit_row(&buf, indent2, "Cached", base->cnt_cached, NULL);
+	if (base->cnt_loose)
+		fmt_commit_row(&buf, indent2, "Loose", base->cnt_loose, NULL);
+	if (base->cnt_packed)
+		fmt_commit_row(&buf, indent2, "Packed", base->cnt_packed, NULL);
+	if (base->cnt_dbcached)
+		fmt_commit_row(&buf, indent2, "DBCached", base->cnt_dbcached, NULL);
+
+	strbuf_addch(&buf, '\n');
+	fmt_commit_row(&buf, indent1, "Total Size of Blobs", base->sum_size, NULL);
+	fmt_commit_row(&buf, indent1, "Total Disk Size of Blobs", base->sum_disk_size, NULL);
+
+	fmt_hbin(&buf, indent1, "Blob Histogram by Size in Bytes", base->size_hbin);
+	fmt_large_item_vec(&buf, indent1, psb->vec_largest_by_size_bytes);
+
 	/*
 	 * NEEDSWORK: TODO
 	 */