index : cgit

/*
 * 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 "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 periods;
};

/*
 * 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);

	// 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 -= SECONDS_PER_WEEK;
	gmtime_r(&t, tm);
}

static void inc_week(struct tm *tm)
{
	time_t t = timegm(tm);

	t += SECONDS_PER_WEEK;
	gmtime_r(&t, tm);
}

static char *pretty_week(struct tm *tm)
{
	static char buf[10];

	// 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;
}

static void trunc_month(struct tm *tm)
{
	tm->tm_mday = 1;
}

static void dec_month(struct tm *tm)
{
	tm->tm_mon--;
	if (tm->tm_mon < 0) {
		tm->tm_year--;
		tm->tm_mon = 11;
	}
}

static void inc_month(struct tm *tm)
{
	tm->tm_mon++;
	if (tm->tm_mon > 11) {
		tm->tm_year++;
		tm->tm_mon = 0;
	}
}

static char *pretty_month(struct tm *tm)
{
	static const char *months[] = {
		"Jan", "Feb", "Mar", "Apr", "May", "Jun",
		"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
	};
	return cgit_fmt("%s %d", months[tm->tm_mon], tm->tm_year + 1900);
}

static void trunc_quarter(struct tm *tm)
{
	trunc_month(tm);
	while (tm->tm_mon % 3 != 0)
		dec_month(tm);
}

static void dec_quarter(struct tm *tm)
{
	dec_month(tm);
	dec_month(tm);
	dec_month(tm);
}

static void inc_quarter(struct tm *tm)
{
	inc_month(tm);
	inc_month(tm);
	inc_month(tm);
}

static char *pretty_quarter(struct tm *tm)
{
	return cgit_fmt("Q%d %d", tm->tm_mon / 3 + 1, tm->tm_year + 1900);
}

static void trunc_year(struct tm *tm)
{
	trunc_month(tm);
	tm->tm_mon = 0;
}

static void dec_year(struct tm *tm)
{
	tm->tm_year--;
}

static void inc_year(struct tm *tm)
{
	tm->tm_year++;
}

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},
	{'q', "quarter", 12, 4, trunc_quarter, dec_quarter, inc_quarter, pretty_quarter},
	{'y', "year", 12, 4, trunc_year, dec_year, inc_year, pretty_year},
};

static void window_start(const struct cgit_period *period, 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);
}

static void add_commit(struct string_list *authors, struct commitinfo *info,
		       const struct cgit_period *period)
{
	struct string_list_item *author, *bucket;
	struct authorstat *stats;
	char *name, *label;
	struct tm date;
	time_t when;
	uintptr_t *count;

	// 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(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.
	if (!gmtime_r(&when, &date))
		return;
	period->trunc(&date);
	label = xstrdup(period->pretty(&date));
	bucket = string_list_insert(&stats->periods, label);
	count = (uintptr_t *)&bucket->util;
	if (*count)
		free(label);
	(*count)++;

	stats->total++;
}

/*
 * 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)
{
	struct string_list authors;
	struct rev_info rev;
	struct commit *commit;
	const char *argv[] = {NULL, ctx.qry.head, NULL, NULL};
	int argc = 2;
	time_t since;
	struct tm tm;

	window_start(period, &tm);
	since = timegm(&tm);

	if (ctx.qry.path) {
		argv[2] = "--";
		argv[3] = ctx.qry.path;
		argc += 2;
	}
	repo_init_revisions(the_repository, &rev, NULL);
	rev.abbrev = DEFAULT_ABBREV;
	rev.commit_format = CMIT_FMT_DEFAULT;
	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 check below still
	// bounds the period edge exactly.
	rev.max_age = since;
	prepare_revision_walk(&rev);
	memset(&authors, 0, sizeof(authors));
	while ((commit = get_revision(&rev)) != NULL) {
		struct commitinfo *info = cgit_parse_commit(commit);

		if ((time_t)info->committer_date >= since)
			add_commit(&authors, info, period);

		cgit_free_commitinfo(info);
		release_commit_memory(the_repository->parsed_objects, commit);
		commit->parents = NULL;
	}
	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. 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;
	int i;

	window_start(period, &tm);
	for (i = 0; i < period->count; i++) {
		string_list_append(&labels, period->pretty(&tm));
		period->inc(&tm);
	}
	return 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 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(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++) {
			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;
	}
	htmlf("<td class='%s'>%ld</td></tr>\n", rightclass, total);
}

static void print_authors(struct string_list *authors, int max_rows,
			  const struct string_list *labels)
{
	struct string_list_item *author, *bucket;
	struct authorstat *stats;
	size_t i, column, rows;
	long total;

	html("<table class='stats'>\n<tr><th>Author</th>");
	for (column = 0; column < labels->nr; column++)
		htmlf("<th>%s</th>", labels->items[column].string);
	html("<th>Total</th></tr>\n");

	// 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 < rows; i++) {
		author = &authors->items[i];
		html("<tr><td class='left'>");
		html_txt(author->string);
		html("</td>");
		stats = author->util;
		total = 0;
		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)bucket->util);
				total += (uintptr_t)bucket->util;
			}
		}
		htmlf("<td class='sum'>%ld</td></tr>\n", total);
	}

	if (rows < authors->nr)
		print_summary_row(authors, rows, authors->nr - rows,
			"Others (%ld)", "left", "", "sum", labels);

	print_summary_row(authors, 0, authors->nr, "Total",
		"total", "sum", "sum", labels);
	html("</table>\n");
}

static const char *language_of(const char *pathname)
{
	const char *ext;
	size_t i;

	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 add_blob_size(const struct object_id *oid, struct strbuf *base,
			 const char *pathname, unsigned mode, void *data)
{
	struct lang_sizes *sizes = data;
	struct string_list_item *lang;
	unsigned long size;

	if (S_ISDIR(mode))
		return READ_TREE_RECURSIVE;
	if (!S_ISREG(mode))
		return 0;
	if (odb_read_object_info(the_repository->objects, oid, &size) != OBJ_BLOB
	    || !size)
		return 0;

	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 *a, const void *b)
{
	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 first_bytes < second_bytes ? 1 :
	       first_bytes > second_bytes ? -1 : 0;
}

// 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;
	struct commit *commit;

	if (repo_get_oid(the_repository, ctx.qry.head, &oid))
		return;
	commit = lookup_commit_reference(the_repository, &oid);
	if (!commit || repo_parse_commit(the_repository, commit))
		return;

	read_tree(the_repository, repo_get_commit_tree(the_repository, commit),
		  &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,
			       unsigned long total)
{
	struct strbuf size = STRBUF_INIT;
	int tenths;

	strbuf_humanise_bytes(&size, bytes);
	html("<tr><td class='left'>");
	html_txt(label);
	html("</td><td>");
	html_txt(size.buf);
	// Printed from integers so a Lua filter switching LC_NUMERIC to a
	// comma-decimal locale cannot change the output.
	tenths = (int)(1000.0 * bytes / total + 0.5);
	htmlf("</td><td>%d.%d%%</td></tr>\n", tenths / 10, tenths % 10);
	strbuf_release(&size);
}

static void print_languages(const struct lang_sizes *sizes)
{
	unsigned long other;
	size_t i;
	int shown;

	if (!sizes->total)
		return;

	// 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 < sizes->langs.nr; i++) {
		if (shown < MAX_LANGUAGE_ROWS &&
		    strcmp(sizes->langs.items[i].string, "Other")) {
			shown++;
			continue;
		}
		other += (uintptr_t)sizes->langs.items[i].util;
	}

	html("<h2>Languages</h2>\n");
	html("<table class='stats'>\n");
	html("<tr><th>Language</th><th>Size</th><th>Share</th></tr>\n");
	shown = 0;
	for (i = 0; i < sizes->langs.nr && shown < MAX_LANGUAGE_ROWS; i++) {
		if (!strcmp(sizes->langs.items[i].string, "Other"))
			continue;
		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, sizes->total);
	html("</table>\n");
}

static void print_options_form(const struct cgit_period *period, int top)
{
	int choices, i;

	html("<div class='cgit-panel'>\n");
	html("<strong>stat options</strong>");
	html("<form method='get'>");
	cgit_add_hidden_formfields(1, 0, "stats");
	html("<table><tr><td colspan='2'></td></tr>\n");
	if (!ctx.repo->max_stats || ctx.repo->max_stats > 1) {
		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 < choices; i++)
			html_option(cgit_fmt("%c", periods[i].code),
				    periods[i].name,
				    cgit_fmt("%c", period->code));
		html("</select></td></tr>\n");
	}
	html("<tr><td class='label'>Authors:</td>");
	html("<td class='ctrl'><select name='ofs' data-autosubmit='1'>");
	html_intoption(10, "10", top);
	html_intoption(25, "25", top);
	html_intoption(50, "50", top);
	html_intoption(100, "100", top);
	html_intoption(-1, "all", top);
	html("</select></td></tr>\n");
	html("<tr><td></td><td class='ctrl'>");
	html("<noscript><input type='submit' value='Reload'></noscript>");
	html("</td></tr></table>");
	html("</form>\n");
	html("</div>\n");
}

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) {
		html(" (path '");
		html_txt(ctx.qry.path);
		html("')");
	}
	html("</h2>\n");

	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();
}