1 file changed, 297 insertions, 273 deletions

diff --git a/source/ui-stats.c b/source/ui-stats.c
index 20e9a30..5436907 100644
--- a/source/ui-stats.c
+++ b/source/ui-stats.c
@@ -1,44 +1,89 @@
-/* ui-stats.c: generate stats view
- *
- * Copyright (C) 2006-2014 cgit Development Team <cgit@lists.zx2c4.com>
- *
- * Licensed under GNU General Public License v2
- *   (see LICENSE.txt for full license text)
+/*
+ * The statistics page, which counts commits per author across a window of
+ * recent weeks, months, quarters or years and breaks the tip of the branch
+ * down by the language its files are written in. The window sizes live in one
+ * table here, which cgit.c also resolves the max-stats setting against so a
+ * repository can refuse the coarser windows. Neither half of the page reads a
+ * diff or the contents of a blob, so the whole thing costs about one rev-list
+ * plus one tree read.
  */
 
 #define USE_THE_REPOSITORY_VARIABLE
 
 #include "cgit.h"
-#include "ui-stats.h"
 #include "html.h"
+#include "parsing.h"
+#include "shared.h"
 #include "ui-shared.h"
+#include "ui-stats.h"
+
+#define DEFAULT_AUTHOR_ROWS 10
+#define MAX_LANGUAGE_ROWS 6
 
+static const struct {
+	const char *ext;
+	const char *label;
+} lang_map[] = {
+	{"c", "C"}, {"h", "C"},
+	{"cpp", "C++"}, {"cc", "C++"}, {"cxx", "C++"},
+	{"hpp", "C++"}, {"hh", "C++"},
+	{"js", "JavaScript"}, {"mjs", "JavaScript"},
+	{"ts", "TypeScript"}, {"tsx", "TypeScript"},
+	{"py", "Python"}, {"lua", "Lua"},
+	{"sh", "Shell"}, {"bash", "Shell"},
+	{"go", "Go"}, {"rs", "Rust"}, {"zig", "Zig"},
+	{"java", "Java"}, {"kt", "Kotlin"}, {"cs", "C#"}, {"swift", "Swift"},
+	{"rb", "Ruby"}, {"pl", "Perl"}, {"pm", "Perl"}, {"php", "PHP"},
+	{"hs", "Haskell"}, {"el", "Lisp"}, {"ml", "OCaml"},
+	{"css", "CSS"}, {"scss", "CSS"},
+	{"html", "HTML"}, {"htm", "HTML"}, {"xml", "XML"},
+	{"md", "Markdown"}, {"rst", "Text"}, {"txt", "Text"},
+	{"json", "JSON"}, {"yml", "YAML"}, {"yaml", "YAML"}, {"toml", "TOML"},
+	{"mk", "Make"}, {"tex", "TeX"}, {"sql", "SQL"}, {"vim", "Vimscript"},
+};
+
+/*
+ * One author's share of the window. periods is keyed by the label of a
+ * period, with the commit count stored in the util field itself rather than
+ * behind another allocation.
+ */
 struct authorstat {
 	long total;
-	struct string_list list;
+	struct string_list periods;
 };
 
-#define DAY_SECS (60 * 60 * 24)
-#define WEEK_SECS (DAY_SECS * 7)
+/*
+ * Bytes at the tip of the branch per language. langs is keyed by the language
+ * name, again with the count living in the util field.
+ */
+struct lang_sizes {
+	struct string_list langs;
+	unsigned long total;
+};
 
 static void trunc_week(struct tm *tm)
 {
 	time_t t = timegm(tm);
-	t -= ((tm->tm_wday + 6) % 7) * DAY_SECS;
+
+	// tm_wday counts from Sunday, while the label comes from %V and %G,
+	// which number the ISO weeks that start on Monday.
+	t -= ((tm->tm_wday + 6) % 7) * SECONDS_PER_DAY;
 	gmtime_r(&t, tm);
 }
 
 static void dec_week(struct tm *tm)
 {
 	time_t t = timegm(tm);
-	t -= WEEK_SECS;
+
+	t -= SECONDS_PER_WEEK;
 	gmtime_r(&t, tm);
 }
 
 static void inc_week(struct tm *tm)
 {
 	time_t t = timegm(tm);
-	t += WEEK_SECS;
+
+	t += SECONDS_PER_WEEK;
 	gmtime_r(&t, tm);
 }
 
@@ -46,7 +91,11 @@ static char *pretty_week(struct tm *tm)
 {
 	static char buf[10];
 
-	strftime(buf, sizeof(buf), "W%V %G", tm);
+	// A year of five digits or more does not fit, and strftime then leaves
+	// the buffer with contents the standard says nothing about, so empty it
+	// rather than return whatever the last week left.
+	if (!strftime(buf, sizeof(buf), "W%V %G", tm))
+		buf[0] = '\0';
 	return buf;
 }
 
@@ -129,6 +178,10 @@ static char *pretty_year(struct tm *tm)
 	return cgit_fmt("%d", tm->tm_year + 1900);
 }
 
+/*
+ * The order runs from the finest window to the coarsest, because a repository
+ * caps the page by storing an index into this table as its max-stats.
+ */
 static const struct cgit_period periods[] = {
 	{'w', "week", 12, 4, trunc_week, dec_week, inc_week, pretty_week},
 	{'m', "month", 12, 4, trunc_month, dec_month, inc_month, pretty_month},
@@ -136,95 +189,58 @@ static const struct cgit_period periods[] = {
 	{'y', "year", 12, 4, trunc_year, dec_year, inc_year, pretty_year},
 };
 
-/* Given a period code or name, return a period index (1, 2, 3 or 4)
- * and update the period pointer to the correcsponding struct.
- * If no matching code is found, return 0.
- */
-int cgit_find_stats_period(const char *expr, const struct cgit_period **period)
+static void window_start(const struct cgit_period *period, struct tm *tm)
 {
-	size_t i;
-	char code = '\0';
-
-	if (!expr)
-		return 0;
-
-	if (strlen(expr) == 1)
-		code = expr[0];
-
-	for (i = 0; i < ARRAY_SIZE(periods); i++)
-		if (periods[i].code == code || !strcmp(periods[i].name, expr)) {
-			if (period)
-				*period = &periods[i];
-			return i + 1;
-		}
-	return 0;
-}
+	time_t now;
+	int i;
 
-const char *cgit_find_stats_periodname(int idx)
-{
-	if (idx > 0 && idx <= (int)ARRAY_SIZE(periods))
-		return periods[idx - 1].name;
-	else
-		return "";
+	time(&now);
+	gmtime_r(&now, tm);
+	period->trunc(tm);
+	for (i = 1; i < period->count; i++)
+		period->dec(tm);
 }
 
 static void add_commit(struct string_list *authors, struct commitinfo *info,
-	const struct cgit_period *period)
+		       const struct cgit_period *period)
 {
-	struct string_list_item *author, *item;
-	struct authorstat *authorstat;
-	struct string_list *items;
-	char *tmp;
+	struct string_list_item *author, *bucket;
+	struct authorstat *stats;
+	char *name, *label;
 	struct tm date;
-	time_t t;
-	uintptr_t *counter;
+	time_t when;
+	uintptr_t *count;
 
-	/* A commit can lack an author header, so fall back rather than
-	 * xstrdup(NULL). */
-	tmp = xstrdup(info->author ? info->author : "(unknown)");
-	author = string_list_insert(authors, tmp);
+	// A commit can lack an author header, so fall back rather than
+	// handing xstrdup a NULL.
+	name = xstrdup(info->author ? info->author : "(unknown)");
+	author = string_list_insert(authors, name);
 	if (!author->util)
 		author->util = xcalloc(1, sizeof(struct authorstat));
 	else
-		free(tmp);
-	authorstat = author->util;
-	items = &authorstat->list;
-	t = info->committer_date;
+		free(name);
+	stats = author->util;
+	when = info->committer_date;
 	// A crafted commit can carry a date outside the range gmtime_r can
-	// represent, which would leave date uninitialized and later index
-	// the month table out of bounds. Drop such a commit from the stats.
-	if (!gmtime_r(&t, &date))
+	// represent, which would leave date uninitialized and later index the
+	// month table out of bounds.
+	if (!gmtime_r(&when, &date))
 		return;
 	period->trunc(&date);
-	tmp = xstrdup(period->pretty(&date));
-	item = string_list_insert(items, tmp);
-	counter = (uintptr_t *)&item->util;
-	if (*counter)
-		free(tmp);
-	(*counter)++;
+	label = xstrdup(period->pretty(&date));
+	bucket = string_list_insert(&stats->periods, label);
+	count = (uintptr_t *)&bucket->util;
+	if (*count)
+		free(label);
+	(*count)++;
 
-	authorstat->total++;
+	stats->total++;
 }
 
-static int cmp_total_commits(const void *a1, const void *a2)
-{
-	const struct string_list_item *i1 = a1;
-	const struct string_list_item *i2 = a2;
-	const struct authorstat *auth1 = i1->util;
-	const struct authorstat *auth2 = i2->util;
-
-	// Return the sign only, since a long difference truncated to int could
-	// flip and leave the comparator inconsistent.
-	if (auth2->total > auth1->total)
-		return 1;
-	if (auth2->total < auth1->total)
-		return -1;
-	return 0;
-}
-
-/* Walk the commit DAG once for the configured window. One walk, no
- * diffs, so the cost class matches rev-list and the page cache absorbs
- * repeat views. Merge commits are skipped.
+/*
+ * Count the commits in the displayed window, returning a list of authors
+ * whose util field holds a struct authorstat. Merge commits are left out, so
+ * that pulling a branch in does not credit the merger with its commits.
  */
 static struct string_list collect_stats(const struct cgit_period *period)
 {
@@ -233,15 +249,10 @@ static struct string_list collect_stats(const struct cgit_period *period)
 	struct commit *commit;
 	const char *argv[] = {NULL, ctx.qry.head, NULL, NULL};
 	int argc = 2;
-	time_t now, since;
-	long i;
+	time_t since;
 	struct tm tm;
 
-	time(&now);
-	gmtime_r(&now, &tm);
-	period->trunc(&tm);
-	for (i = 1; i < period->count; i++)
-		period->dec(&tm);
+	window_start(period, &tm);
 	since = timegm(&tm);
 
 	if (ctx.qry.path) {
@@ -255,10 +266,12 @@ static struct string_list collect_stats(const struct cgit_period *period)
 	rev.max_parents = 1;
 	rev.verbose_header = 1;
 	rev.show_root_diff = 0;
+	// setup_revisions reads argv the way main does and ignores the first
+	// entry, so the head to walk sits at argv[1].
 	setup_revisions(argc, argv, &rev, NULL);
 	// Prune the walk to the displayed window instead of traversing the
-	// whole history and discarding older commits. The in-process check
-	// below still bounds the period edge exactly.
+	// whole history and discarding older commits. The check below still
+	// bounds the period edge exactly.
 	rev.max_age = since;
 	prepare_revision_walk(&rev);
 	memset(&authors, 0, sizeof(authors));
@@ -275,24 +288,34 @@ static struct string_list collect_stats(const struct cgit_period *period)
 	return authors;
 }
 
+static int cmp_total_commits(const void *a, const void *b)
+{
+	const struct string_list_item *first = a;
+	const struct string_list_item *second = b;
+	const struct authorstat *first_stats = first->util;
+	const struct authorstat *second_stats = second->util;
+
+	// Report the sign only, since a long difference truncated to int
+	// could flip and leave the comparator inconsistent.
+	if (second_stats->total > first_stats->total)
+		return 1;
+	if (second_stats->total < first_stats->total)
+		return -1;
+	return 0;
+}
+
 /*
- * The column labels for the displayed window, oldest first. Every table below
- * walks the same period, and each step of that walk converts a struct tm to a
- * time_t and back, so the labels are worked out once and shared. It also frees
- * the callers from pretty() handing back a buffer it reuses.
+ * The column labels for the displayed window, oldest first. pretty hands back
+ * a buffer it goes on to reuse, so the labels are copied here once and shared
+ * by every table below.
  */
 static struct string_list build_period_labels(const struct cgit_period *period)
 {
 	struct string_list labels = STRING_LIST_INIT_DUP;
 	struct tm tm;
-	time_t now;
 	int i;
 
-	time(&now);
-	gmtime_r(&now, &tm);
-	period->trunc(&tm);
-	for (i = 1; i < period->count; i++)
-		period->dec(&tm);
+	window_start(period, &tm);
 	for (i = 0; i < period->count; i++) {
 		string_list_append(&labels, period->pretty(&tm));
 		period->inc(&tm);
@@ -300,34 +323,33 @@ static struct string_list build_period_labels(const struct cgit_period *period)
 	return labels;
 }
 
-/* Sum one row across a run of authors. Taking a count rather than an end index
- * keeps an empty author list from describing a run that wraps. */
-static void print_combined_authorrow(struct string_list *authors, size_t from,
-				     size_t count, const char *name,
-				     const char *leftclass,
-				     const char *centerclass,
-				     const char *rightclass,
-				     const struct string_list *labels)
+/*
+ * The run of authors to sum is given as a start and a count rather than as
+ * two indices, so that an empty author list cannot describe a run that wraps.
+ */
+static void print_summary_row(struct string_list *authors, size_t from,
+			      size_t count, const char *label_format,
+			      const char *leftclass, const char *centerclass,
+			      const char *rightclass,
+			      const struct string_list *labels)
 {
-	struct string_list_item *author;
-	struct authorstat *authorstat;
-	struct string_list *items;
-	struct string_list_item *date;
-	size_t i, j;
+	struct authorstat *stats;
+	struct string_list_item *bucket;
+	size_t i, column;
 	long total, subtotal;
 
 	total = 0;
 	htmlf("<tr><td class='%s'>%s</td>", leftclass,
-		cgit_fmt(name, (long)count));
-	for (j = 0; j < labels->nr; j++) {
+		cgit_fmt(label_format, (long)count));
+	for (column = 0; column < labels->nr; column++) {
+		const char *label = labels->items[column].string;
+
 		subtotal = 0;
 		for (i = from; i < from + count; i++) {
-			author = &authors->items[i];
-			authorstat = author->util;
-			items = &authorstat->list;
-			date = string_list_lookup(items, labels->items[j].string);
-			if (date)
-				subtotal += (uintptr_t)date->util;
+			stats = authors->items[i].util;
+			bucket = string_list_lookup(&stats->periods, label);
+			if (bucket)
+				subtotal += (uintptr_t)bucket->util;
 		}
 		htmlf("<td class='%s'>%ld</td>", centerclass, subtotal);
 		total += subtotal;
@@ -335,91 +357,75 @@ static void print_combined_authorrow(struct string_list *authors, size_t from,
 	htmlf("<td class='%s'>%ld</td></tr>", rightclass, total);
 }
 
-static void print_authors(struct string_list *authors, int requested_top,
+static void print_authors(struct string_list *authors, int max_rows,
 			  const struct string_list *labels)
 {
-	struct string_list_item *author;
-	struct authorstat *authorstat;
-	struct string_list *items;
-	struct string_list_item *date;
-	size_t i, j, top;
+	struct string_list_item *author, *bucket;
+	struct authorstat *stats;
+	size_t i, column, rows;
 	long total;
 
 	html("<table class='stats'><tr><th>Author</th>");
-	for (j = 0; j < labels->nr; j++)
-		htmlf("<th>%s</th>", labels->items[j].string);
+	for (column = 0; column < labels->nr; column++)
+		htmlf("<th>%s</th>", labels->items[column].string);
 	html("<th>Total</th></tr>\n");
 
-	// The author count arrives through ofs, which carries -1 for "all".
-	top = (requested_top <= 0 || (size_t)requested_top > authors->nr)
-		? authors->nr : (size_t)requested_top;
+	// The row count arrives through ofs, which carries -1 for "all".
+	rows = (max_rows <= 0 || (size_t)max_rows > authors->nr)
+		? authors->nr : (size_t)max_rows;
 
-	for (i = 0; i < top; i++) {
+	for (i = 0; i < rows; i++) {
 		author = &authors->items[i];
 		html("<tr><td class='left'>");
 		html_txt(author->string);
 		html("</td>");
-		authorstat = author->util;
-		items = &authorstat->list;
+		stats = author->util;
 		total = 0;
-		for (j = 0; j < labels->nr; j++) {
-			date = string_list_lookup(items, labels->items[j].string);
-			if (!date)
+		for (column = 0; column < labels->nr; column++) {
+			const char *label = labels->items[column].string;
+
+			bucket = string_list_lookup(&stats->periods, label);
+			if (!bucket)
 				html("<td>0</td>");
 			else {
-				htmlf("<td>%lu</td>", (uintptr_t)date->util);
-				total += (uintptr_t)date->util;
+				htmlf("<td>%lu</td>", (uintptr_t)bucket->util);
+				total += (uintptr_t)bucket->util;
 			}
 		}
 		htmlf("<td class='sum'>%ld</td></tr>", total);
 	}
 
-	if (top < authors->nr)
-		print_combined_authorrow(authors, top, authors->nr - top,
+	if (rows < authors->nr)
+		print_summary_row(authors, rows, authors->nr - rows,
 			"Others (%ld)", "left", "", "sum", labels);
 
-	print_combined_authorrow(authors, 0, authors->nr, "Total",
+	print_summary_row(authors, 0, authors->nr, "Total",
 		"total", "sum", "sum", labels);
 	html("</table>");
 }
 
-/* Bytes at HEAD per language, judged by file extension. One recursive
- * tree read, sizes come from object headers without loading content. */
-static const struct {
+static const char *language_of(const char *pathname)
+{
 	const char *ext;
-	const char *label;
-} lang_map[] = {
-	{"c", "C"}, {"h", "C"},
-	{"cpp", "C++"}, {"cc", "C++"}, {"cxx", "C++"},
-	{"hpp", "C++"}, {"hh", "C++"},
-	{"js", "JavaScript"}, {"mjs", "JavaScript"},
-	{"ts", "TypeScript"}, {"tsx", "TypeScript"},
-	{"py", "Python"}, {"lua", "Lua"},
-	{"sh", "Shell"}, {"bash", "Shell"},
-	{"go", "Go"}, {"rs", "Rust"}, {"zig", "Zig"},
-	{"java", "Java"}, {"kt", "Kotlin"}, {"cs", "C#"}, {"swift", "Swift"},
-	{"rb", "Ruby"}, {"pl", "Perl"}, {"pm", "Perl"}, {"php", "PHP"},
-	{"hs", "Haskell"}, {"el", "Lisp"}, {"ml", "OCaml"},
-	{"css", "CSS"}, {"scss", "CSS"},
-	{"html", "HTML"}, {"htm", "HTML"}, {"xml", "XML"},
-	{"md", "Markdown"}, {"rst", "Text"}, {"txt", "Text"},
-	{"json", "JSON"}, {"yml", "YAML"}, {"yaml", "YAML"}, {"toml", "TOML"},
-	{"mk", "Make"}, {"tex", "TeX"}, {"sql", "SQL"}, {"vim", "Vimscript"},
-};
+	size_t i;
 
-struct lang_walk_ctx {
-	struct string_list langs;
-	unsigned long total;
-};
+	ext = strrchr(pathname, '.');
+	if (ext && ext != pathname && ext[1]) {
+		for (i = 0; i < ARRAY_SIZE(lang_map); i++)
+			if (!strcasecmp(ext + 1, lang_map[i].ext))
+				return lang_map[i].label;
+	} else if (!strcmp(pathname, "Makefile")) {
+		return "Make";
+	}
+	return "Other";
+}
 
-static int lang_walk_cb(const struct object_id *oid, struct strbuf *base,
-			const char *pathname, unsigned mode, void *cbdata)
+static int add_blob_size(const struct object_id *oid, struct strbuf *base,
+			 const char *pathname, unsigned mode, void *data)
 {
-	struct lang_walk_ctx *lw = cbdata;
-	struct string_list_item *item;
-	const char *ext, *label = NULL;
+	struct lang_sizes *sizes = data;
+	struct string_list_item *lang;
 	unsigned long size;
-	int i;
 
 	if (S_ISDIR(mode))
 		return READ_TREE_RECURSIVE;
@@ -429,35 +435,26 @@ static int lang_walk_cb(const struct object_id *oid, struct strbuf *base,
 	    || !size)
 		return 0;
 
-	ext = strrchr(pathname, '.');
-	if (ext && ext != pathname && ext[1]) {
-		for (i = 0; i < (int)ARRAY_SIZE(lang_map); i++)
-			if (!strcasecmp(ext + 1, lang_map[i].ext)) {
-				label = lang_map[i].label;
-				break;
-			}
-	} else if (!strcmp(pathname, "Makefile")) {
-		label = "Make";
-	}
-	if (!label)
-		label = "Other";
-	item = string_list_insert(&lw->langs, label);
-	item->util = (void *)((uintptr_t)item->util + size);
-	lw->total += size;
+	lang = string_list_insert(&sizes->langs, language_of(pathname));
+	lang->util = (void *)((uintptr_t)lang->util + size);
+	sizes->total += size;
 	return 0;
 }
 
-static int cmp_lang_bytes(const void *a1, const void *a2)
+static int cmp_lang_bytes(const void *a, const void *b)
 {
-	const struct string_list_item *i1 = a1;
-	const struct string_list_item *i2 = a2;
-	uintptr_t b1 = (uintptr_t)i1->util;
-	uintptr_t b2 = (uintptr_t)i2->util;
+	const struct string_list_item *first = a;
+	const struct string_list_item *second = b;
+	uintptr_t first_bytes = (uintptr_t)first->util;
+	uintptr_t second_bytes = (uintptr_t)second->util;
 
-	return b1 < b2 ? 1 : b1 > b2 ? -1 : 0;
+	return first_bytes < second_bytes ? 1 :
+	       first_bytes > second_bytes ? -1 : 0;
 }
 
-static void summarize_tree(struct lang_walk_ctx *lw)
+// Leaves the list sorted largest first, which is the order the rows print in
+// and what lets the caller stop naming languages once it reaches the tail.
+static void measure_languages(struct lang_sizes *sizes)
 {
 	struct pathspec paths = { .nr = 0 };
 	struct object_id oid;
@@ -470,9 +467,9 @@ static void summarize_tree(struct lang_walk_ctx *lw)
 		return;
 
 	read_tree(the_repository, repo_get_commit_tree(the_repository, commit),
-		  &paths, lang_walk_cb, lw);
-	qsort(lw->langs.items, lw->langs.nr, sizeof(struct string_list_item),
-	      cmp_lang_bytes);
+		  &paths, add_blob_size, sizes);
+	qsort(sizes->langs.items, sizes->langs.nr,
+	      sizeof(struct string_list_item), cmp_lang_bytes);
 }
 
 static void print_language_row(const char *label, unsigned long bytes,
@@ -489,103 +486,63 @@ static void print_language_row(const char *label, unsigned long bytes,
 	strbuf_release(&size);
 }
 
-/* Bordered like the commits-per-author table. */
-static void print_languages(struct lang_walk_ctx *lw)
+static void print_languages(const struct lang_sizes *sizes)
 {
 	unsigned long other;
 	size_t i;
 	int shown;
 
-	if (!lw->total)
+	if (!sizes->total)
 		return;
 
-	/* Show up to six named rows, everything else folds into an
-	 * Other row at the end. */
+	// The Other row is printed last whatever its size, so everything past
+	// the named languages is summed into it up front.
 	other = 0;
 	shown = 0;
-	for (i = 0; i < lw->langs.nr; i++) {
-		if (shown < 6 && strcmp(lw->langs.items[i].string, "Other")) {
+	for (i = 0; i < sizes->langs.nr; i++) {
+		if (shown < MAX_LANGUAGE_ROWS &&
+		    strcmp(sizes->langs.items[i].string, "Other")) {
 			shown++;
 			continue;
 		}
-		other += (uintptr_t)lw->langs.items[i].util;
+		other += (uintptr_t)sizes->langs.items[i].util;
 	}
 
 	html("<h2>Languages</h2>");
 	html("<table class='stats'>");
 	html("<tr><th>Language</th><th>Size</th><th>Share</th></tr>\n");
 	shown = 0;
-	for (i = 0; i < lw->langs.nr && shown < 6; i++) {
-		if (!strcmp(lw->langs.items[i].string, "Other"))
+	for (i = 0; i < sizes->langs.nr && shown < MAX_LANGUAGE_ROWS; i++) {
+		if (!strcmp(sizes->langs.items[i].string, "Other"))
 			continue;
-		print_language_row(lw->langs.items[i].string,
-				   (uintptr_t)lw->langs.items[i].util,
-				   lw->total);
+		print_language_row(sizes->langs.items[i].string,
+				   (uintptr_t)sizes->langs.items[i].util,
+				   sizes->total);
 		shown++;
 	}
 	if (other)
-		print_language_row("Other", other, lw->total);
+		print_language_row("Other", other, sizes->total);
 	html("</table>");
 }
 
-/* Create a sorted string_list with one entry per author. The util-field
- * for each author is another string_list which is used to calculate the
- * number of commits per time-interval.
- */
-void cgit_show_stats(void)
+static void print_options_form(const struct cgit_period *period, int top)
 {
-	struct string_list authors;
-	struct lang_walk_ctx lw;
-	const struct cgit_period *period;
-	int top, i;
-	const char *code = "w";
+	int choices, i;
 
-	if (ctx.qry.period)
-		code = ctx.qry.period;
-
-	i = cgit_find_stats_period(code, &period);
-	if (!i) {
-		cgit_print_error_page(404, "Not found",
-			"Unknown statistics type: %c", code[0]);
-		return;
-	}
-	if (ctx.repo->max_stats && i > ctx.repo->max_stats) {
-		cgit_print_error_page(400, "Bad request",
-			"Statistics type disabled: %s", period->name);
-		return;
-	}
-	/* Walk the tree before the history walk. Releasing commit memory
-	 * during that walk resets each commit's slab index, and a later
-	 * lookup through the commit graph would then read another
-	 * commit's slot and walk the wrong tree. */
-	memset(&lw, 0, sizeof(lw));
-	summarize_tree(&lw);
-
-	authors = collect_stats(period);
-	qsort(authors.items, authors.nr, sizeof(struct string_list_item),
-		cmp_total_commits);
-
-	top = ctx.qry.ofs;
-	if (!top)
-		top = 10;
-
-	cgit_print_layout_start();
-
-	/* The options panel floats right of the page top, the same spot
-	 * the diff controls occupy on the diff pages. */
 	html("<div class='cgit-panel'>");
 	html("<b>stat options</b>");
 	html("<form method='get'>");
 	cgit_add_hidden_formfields(1, 0, "stats");
 	html("<table><tr><td colspan='2'/></tr>");
 	if (!ctx.repo->max_stats || ctx.repo->max_stats > 1) {
-		int nperiods = ctx.repo->max_stats ?
+		choices = ctx.repo->max_stats ?
 			ctx.repo->max_stats : (int)ARRAY_SIZE(periods);
 		html("<tr><td class='label'>Period:</td>");
 		html("<td class='ctrl'><select name='period' data-autosubmit='1'>");
-		for (i = 0; i < nperiods; i++)
+		for (i = 0; i < choices; i++)
 			html_option(cgit_fmt("%c", periods[i].code),
-				    periods[i].name, cgit_fmt("%c", period->code));
+				    periods[i].name,
+				    cgit_fmt("%c", period->code));
 		html("</select></td></tr>");
 	}
 	html("<tr><td class='label'>Authors:</td>");
@@ -601,6 +558,75 @@ void cgit_show_stats(void)
 	html("</td></tr></table>");
 	html("</form>");
 	html("</div>");
+}
+
+int cgit_find_stats_period(const char *expr, const struct cgit_period **period)
+{
+	size_t i;
+	char code = '\0';
+
+	if (!expr)
+		return 0;
+
+	if (strlen(expr) == 1)
+		code = expr[0];
+
+	for (i = 0; i < ARRAY_SIZE(periods); i++)
+		if (periods[i].code == code || !strcmp(periods[i].name, expr)) {
+			if (period)
+				*period = &periods[i];
+			return i + 1;
+		}
+	return 0;
+}
+
+const char *cgit_find_stats_periodname(int idx)
+{
+	if (idx > 0 && idx <= (int)ARRAY_SIZE(periods))
+		return periods[idx - 1].name;
+	else
+		return "";
+}
+
+void cgit_show_stats(void)
+{
+	struct string_list authors, labels;
+	struct lang_sizes sizes;
+	const struct cgit_period *period;
+	int top, period_index;
+	const char *code = "w";
+
+	if (ctx.qry.period)
+		code = ctx.qry.period;
+
+	period_index = cgit_find_stats_period(code, &period);
+	if (!period_index) {
+		cgit_print_error_page(404, "Not found",
+			"Unknown statistics type: %c", code[0]);
+		return;
+	}
+	if (ctx.repo->max_stats && period_index > ctx.repo->max_stats) {
+		cgit_print_error_page(400, "Bad request",
+			"Statistics type disabled: %s", period->name);
+		return;
+	}
+	// The tree has to be measured before the history walk. That walk
+	// releases each commit as it goes, which resets the commit's slab
+	// index, so a lookup afterwards would read another commit's slot and
+	// walk the wrong tree.
+	memset(&sizes, 0, sizeof(sizes));
+	measure_languages(&sizes);
+
+	authors = collect_stats(period);
+	qsort(authors.items, authors.nr, sizeof(struct string_list_item),
+		cmp_total_commits);
+
+	top = ctx.qry.ofs;
+	if (!top)
+		top = DEFAULT_AUTHOR_ROWS;
+
+	cgit_print_layout_start();
+	print_options_form(period, top);
 
 	htmlf("<h2>Commits per author per %s", period->name);
 	if (ctx.qry.path) {
@@ -609,15 +635,13 @@ void cgit_show_stats(void)
 		html("')");
 	}
 	html("</h2>");
-	{
-		struct string_list labels = build_period_labels(period);
-		print_authors(&authors, top, &labels);
-		string_list_clear(&labels, 0);
-	}
 
-	print_languages(&lw);
-	string_list_clear(&lw.langs, 0);
+	labels = build_period_labels(period);
+	print_authors(&authors, top, &labels);
+	string_list_clear(&labels, 0);
+
+	print_languages(&sizes);
+	string_list_clear(&sizes.langs, 0);
 
 	cgit_print_layout_end();
 }
-