From d0a1d32488f45de9362a71da225e401c3b33b390 Mon Sep 17 00:00:00 2001
From: Jason Volk <jason@zemos.net>
Date: Wed, 17 Nov 2021 17:07:24 -0800
Subject: [PATCH] ircd::db: Add SST scan interface w/ console suite.

---
 include/ircd/db/sst.h |  9 +++++
 ircd/db.h             |  1 +
 ircd/db_database.cc   | 93 ++++++++++++++++++++++++++++++++++++++++++-
 modules/console.cc    | 70 ++++++++++++++++++++++++++++++++
 4 files changed, 172 insertions(+), 1 deletion(-)

diff --git a/include/ircd/db/sst.h b/include/ircd/db/sst.h
index cef528357..9d0e97ba5 100644
--- a/include/ircd/db/sst.h
+++ b/include/ircd/db/sst.h
@@ -15,6 +15,7 @@ struct ircd::db::database::sst
 {
 	struct info;
 	struct dump;
+	struct scan;
 
 	[[deprecated]]
 	static void tool(const vector_view<const string_view> &args);
@@ -97,3 +98,11 @@ struct ircd::db::database::sst::dump
 	dump(dump &&) = delete;
 	dump(const dump &) = delete;
 };
+
+struct ircd::db::database::sst::scan
+{
+	using closure = std::function<bool (const string_view &, const string_view &)>;
+
+	scan(db::database &, const string_view &path, const closure &);
+	scan(db::database &, const string_view &path); // VerifyChecksum
+};
diff --git a/ircd/db.h b/ircd/db.h
index 974e227d3..b0790e7d2 100644
--- a/ircd/db.h
+++ b/ircd/db.h
@@ -50,6 +50,7 @@
 #include <rocksdb/filter_policy.h>
 #include <rocksdb/table.h>
 #include <rocksdb/sst_file_manager.h>
+#include <rocksdb/sst_file_reader.h>
 #include <rocksdb/sst_dump_tool.h>
 #include <rocksdb/compaction_filter.h>
 #include <rocksdb/wal_filter.h>
diff --git a/ircd/db_database.cc b/ircd/db_database.cc
index bf93f41bd..17de80ea8 100644
--- a/ircd/db_database.cc
+++ b/ircd/db_database.cc
@@ -3615,7 +3615,98 @@ ircd::db::database::sst::tool(const vector_view<const string_view> &args)
 }
 
 //
-// sst::dump::dump
+// sst::scan
+//
+
+ircd::db::database::sst::scan::scan(database &d,
+                                    const string_view &fpath)
+{
+	const auto &opts
+	{
+		d.d->GetOptions()
+	};
+
+	rocksdb::SstFileReader reader
+	{
+		opts
+	};
+
+	const string_view path_parts[]
+	{
+		fs::base::db, db::name(d), fpath
+	};
+
+	const auto path
+	{
+		fs::path_string(path_parts)
+	};
+
+	throw_on_error
+	{
+		reader.Open(path)
+	};
+
+	db::gopts gopts;
+	throw_on_error
+	{
+		reader.VerifyChecksum(make_opts(gopts))
+	};
+}
+
+ircd::db::database::sst::scan::scan(database &d,
+                                    const string_view &fpath,
+                                    const closure &call)
+{
+	const auto &opts
+	{
+		d.d->GetOptions()
+	};
+
+	rocksdb::SstFileReader reader
+	{
+		opts
+	};
+
+	const string_view path_parts[]
+	{
+		fs::base::db, db::name(d), fpath
+	};
+
+	const auto path
+	{
+		fs::path_string(path_parts)
+	};
+
+	throw_on_error
+	{
+		reader.Open(path)
+	};
+
+	db::gopts gopts;
+	std::unique_ptr<rocksdb::Iterator> it
+	{
+		reader.NewIterator(make_opts(gopts))
+	};
+
+	for(it->SeekToFirst(); db::valid(*it); it->Next())
+	{
+		const auto &key
+		{
+			slice(it->key())
+		};
+
+		const auto &val
+		{
+			slice(it->value())
+		};
+
+		if(!call(key, val))
+			break;
+	}
+}
+
+//
+// sst::dump
 //
 
 ircd::db::database::sst::dump::dump(db::column column,
diff --git a/modules/console.cc b/modules/console.cc
index 412fa7282..2b13929ad 100644
--- a/modules/console.cc
+++ b/modules/console.cc
@@ -4711,6 +4711,76 @@ console_cmd__db__sst__dump(opt &out, const string_view &line)
 	return true;
 }
 
+bool
+console_cmd__db__sst__scan(opt &out, const string_view &line)
+{
+	const params param{line, " ",
+	{
+		"dbname", "path",
+	}};
+
+	const auto dbname
+	{
+		param.at("dbname")
+	};
+
+	const auto path
+	{
+		param.at("path")
+	};
+
+	auto &database
+	{
+		db::database::get(dbname)
+	};
+
+	db::database::sst::scan
+	{
+		database, path
+	};
+
+	out << "Completed without error." << std::endl;
+	return true;
+}
+
+bool
+console_cmd__db__sst__scan__count(opt &out, const string_view &line)
+{
+	const params param{line, " ",
+	{
+		"dbname", "path",
+	}};
+
+	const auto dbname
+	{
+		param.at("dbname")
+	};
+
+	const auto path
+	{
+		param.at("path")
+	};
+
+	auto &database
+	{
+		db::database::get(dbname)
+	};
+
+	size_t i(0);
+	db::database::sst::scan
+	{
+		database, path, [&out, &i]
+		(const auto &key, const auto &val)
+		{
+			++i;
+			return true;
+		}
+	};
+
+	out << "Found " << i << " entries." << std::endl;
+	return true;
+}
+
 bool
 console_cmd__db__wal(opt &out, const string_view &line)
 try