survey: add strintmap to count observed refs by prefix

Create a `struct strintmap` to count the set of observed
refs by prefix, such as "refs/heads/" or refs/remotes/origin/".
Print an "observed" section in both pretty and JSON formats.

NEEDSWORK: The new observed map overlaps a little with the
existing `prw->want_<class>` and `prs->cnt_<class>` fields,
so maybe we should refactor the older fields.

NEEDSWORK: Also, the `--all-refs` flag is a little cumbersome
as more custom classes (like `--pull` and `--prefetch`) are
added.  It might now be simpler to just have `--all-refs`
request `refs/` knowing that the counting will be handled by
the refsmap now.

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

builtin/survey.c

@@ -13,6 +13,7 @@
 #include "revision.h"
 #include "run-command.h"
 #include "strbuf.h"
+#include "strmap.h"
 #include "strvec.h"
 #include "trace2.h"
 #include "tree.h"
@@ -280,6 +281,8 @@ struct survey_stats_refs {
 	size_t len_sum_local_refnames;
 	size_t len_max_remote_refname;
 	size_t len_sum_remote_refnames;
+
+	struct strintmap refsmap;
 };
 
 /*
@@ -1046,8 +1049,11 @@ static void do_treewalk_reachable(struct ref_array *ref_array)
 static void do_calc_stats_refs(struct ref_array *ref_array)
 {
 	struct survey_stats_refs *prs = &survey_stats.refs;
+	struct strbuf buf = STRBUF_INIT;
 	int k;
 
+	strintmap_init(&prs->refsmap, 0);
+
 	for (k = 0; k < ref_array->nr; k++) {
 		struct ref_array_item *p = ref_array->items[k];
 		struct object_id peeled;
@@ -1065,29 +1071,81 @@ static void do_calc_stats_refs(struct ref_array *ref_array)
 		 */
 		switch (p->kind) {
 		case FILTER_REFS_TAGS:
+			strintmap_incr(&prs->refsmap, "refs/tags/", 1);
+
 			if (!peel_iterated_oid(&p->objectname, &peeled))
 				prs->cnt_annotated_tags++;
 			else
 				prs->cnt_lightweight_tags++;
 			break;
+
 		case FILTER_REFS_BRANCHES:
+			strintmap_incr(&prs->refsmap, "refs/heads/", 1);
+
 			prs->cnt_branches++;
 			break;
+
 		case FILTER_REFS_REMOTES:
+			/*
+			 * For the refsmap, group them by the "refs/remotes/<remote>/".
+			 * For example:
+			 *   "refs/remotes/origin/..."
+			 */
+			if (starts_with(p->refname, "refs/remotes/")) {
+				int begin = strlen("refs/remotes/");
+				size_t j;
+				strbuf_reset(&buf);
+				strbuf_addstr(&buf, p->refname);
+				for (j = begin; j < buf.len; j++) {
+					if (buf.buf[j] == '/') {
+						strbuf_setlen(&buf, j+1);
+						break;
+					}
+				}
+				strintmap_incr(&prs->refsmap, buf.buf, 1);
+			}
+
 			prs->cnt_remotes++;
 			break;
+
 		case FILTER_REFS_OTHERS:
+			/*
+			 * For the refsmap, group them by their "refs/<class>/".
+			 * For example:
+			 *   "refs/notes/..."
+			 *   "refs/stash/..."
+			 *   "refs/<custom>/..."
+			 */
+			if (starts_with(p->refname, "refs/")) {
+				int begin = strlen("refs/");
+				size_t j;
+				strbuf_reset(&buf);
+				strbuf_addstr(&buf, p->refname);
+				for (j = begin; j < buf.len; j++) {
+					if (buf.buf[j] == '/') {
+						strbuf_setlen(&buf, j+1);
+						break;
+					}
+				}
+				strintmap_incr(&prs->refsmap, buf.buf, 1);
+			}
+
 			if (starts_with(p->refname, "refs/prefetch/"))
 				prs->cnt_prefetch++;
 			else if (starts_with(p->refname, "refs/pull/"))
 				prs->cnt_pull++;
 			else
 				prs->cnt_other++;
 			break;
+
 		case FILTER_REFS_DETACHED_HEAD:
+			strintmap_incr(&prs->refsmap, p->refname, 1);
+
 			prs->cnt_detached++;
 			break;
+
 		default:
+			strintmap_incr(&prs->refsmap, p->refname, 1); /* probably "HEAD" */
 			break;
 		}
 
@@ -1416,6 +1474,26 @@ static void survey_json(struct json_writer *jw, int pretty)
 					jw_array_string(jw, survey_vec_refs_wanted.v[k]);
 			}
 			jw_end(jw);
+
+			jw_object_inline_begin_array(jw, "observed");
+			{
+				struct hashmap_iter iter;
+				struct strmap_entry *entry;
+				int value;
+
+				strintmap_for_each_entry(&prs->refsmap, &iter, entry) {
+					const char *key = entry->key;
+					intptr_t count = (intptr_t)entry->value;
+					int value = count;
+					jw_array_inline_begin_object(jw);
+					{
+						jw_object_string(jw, "class", key);
+						jw_object_intmax(jw, "count", value);
+					}
+					jw_end(jw);
+				}
+			}
+			jw_end(jw);
 		}
 		jw_end(jw);
 
@@ -1977,6 +2055,22 @@ static void pretty_print_refs(int indent)
 			fmt_txt_pair_ui64(&buf, indent3, "pull", prs->cnt_pull);
 	}
 
+	strbuf_addch(&buf, '\n');
+	fmt_txt_pair_ui64(&buf, indent1, "Observed Reference Classes", strintmap_get_size(&prs->refsmap));
+	{
+		struct hashmap_iter iter;
+		struct strmap_entry *entry;
+		int value;
+
+		strintmap_for_each_entry(&prs->refsmap, &iter, entry) {
+			const char *key = entry->key;
+			intptr_t count = (intptr_t)entry->value;
+			int value = count;
+
+			fmt_txt_pair_ui64(&buf, indent2, key, value);
+		}
+	}
+
 	strbuf_addch(&buf, '\n');
 	fmt_txt_line(&buf, indent1, "Reference Count by Storage Location");
 	fmt_txt_pair_ui64(&buf, indent2, "Loose", prs->cnt_loose);
@@ -1995,11 +2089,6 @@ static void pretty_print_refs(int indent)
 		fmt_txt_pair_ui64(&buf, indent3, "Sum", prs->len_sum_local_refnames);
 	}
 
-	/*
-	 * NEEDSWORK: Do we want to print the set of "refs/<foo>"
-	 * patterns that we requested to be in the set?
-	 */
-
 	strbuf_addch(&buf, '\n');
 	fwrite(buf.buf, 1, buf.len, stdout);
 	strbuf_release(&buf);