Merge remote-tracking branch 'origin/topic/bernhard/input-documentation'

* origin/topic/bernhard/input-documentation:
  and provide a bit of motivation to try the last example.
  and restructure it a bit
  First try at sqlite reader/writer documentation
  add check that the SQLite reader is only used in MANUAL reading mode
  rename the dbname configuration option to tablename.

CHANGES

@@ -1,4 +1,17 @@
 
+2.2-beta-80 | 2013-10-18 13:18:05 -0700
+
+  * SQLite reader/writer documentation. (Bernhard Amann)
+
+  * Check that the SQLite reader is only used in MANUAL reading mode.
+    (Bernhard Amann)
+
+  * Rename the SQLite writer "dbname" configuration option to
+    "tablename". (Bernhard Amann)
+
+  * Remove the "dbname" configuration option from the SQLite reader
+    where it wasn't used. (Bernhard Amann)
+
 2.2-beta-73 | 2013-10-14 14:28:25 -0700
 
   * Fix misc. Coverity-reported issues (leaks, potential null pointer

VERSION

@@ -1 +1 @@
-2.2-beta-73
+2.2-beta-80

doc/frameworks/input.rst

@@ -1,4 +1,6 @@
 
+.. _framework-input:
+
 ===============
 Input Framework
 ===============
@@ -260,8 +262,13 @@ to optimize the speed of the input framework. It can generate arbitrary
 amounts of semi-random data in all Bro data types supported by the input
 framework.
 
-In the future, the input framework will get support for new data sources
+Currently, Bro supports the following readers in addition to the 
-like, for example, different databases.
+aforementioned ones:
+
+.. toctree::
+   :maxdepth: 1
+
+   logging-input-sqlite
 
 Add_table options
 -----------------

doc/frameworks/logging-input-sqlite.rst

@@ -0,0 +1,239 @@
+
+============================================
+Logging To and Reading From SQLite Databases
+============================================
+
+.. rst-class:: opening
+
+   Starting with version 2.2, Bro features a SQLite logging writer
+   as well as a SQLite input reader. SQLite is a simple, file-based,
+   widely used SQL database system. Using SQLite allows Bro to write
+   and access data in a format that is easy to use in interchange with
+   other applications. Due to the transactional nature of SQLite,
+   databases can be used by several applications simultaneously. Hence,
+   they can, for example, be used to make data that changes regularly available
+   to Bro on a continuing basis.
+
+.. contents::
+
+Warning
+=======
+
+In contrast to the ASCII reader and writer, the SQLite plugins have not yet
+seen extensive use in production environments. While we are not aware
+of any issues with them, we urge to caution when using them
+in production environments. There could be lingering issues which only occur
+when the plugins are used with high amounts of data or in high-load environments.
+
+Logging Data into SQLite Databases
+==================================
+
+Logging support for SQLite is available in all Bro installations starting with
+version 2.2. There is no need to load any additional scripts or for any compile-time
+configurations.
+
+Sending data from existing logging streams to SQLite is rather straightforward. You
+have to define a filter which specifies SQLite as the writer.
+
+The following example code adds SQLite as a filter for the connection log:
+
+.. code:: bro
+
+    event bro_init()
+        {
+        local filter: Log::Filter =
+            [
+            $name="sqlite",
+            $path="/var/db/conn",
+            $config=table(["tablename"] = "conn"),
+            $writer=Log::WRITER_SQLITE
+            ];
+
+        Log::add_filter(Conn::LOG, filter);
+        }
+
+Bro will create the database file ``/var/db/conn.sqlite``, if it does not already exist.
+It will also create a table with the name ``conn`` (if it does not exist) and start
+appending connection information to the table.
+
+At the moment, SQLite databases are not rotated the same way ASCII log-files are. You
+have to take care to create them in an adequate location.
+
+If you examine the resulting SQLite database, the schema will contain the same fields
+that are present in the ASCII log files::
+
+    # sqlite3 /var/db/conn.sqlite
+
+    SQLite version 3.8.0.2 2013-09-03 17:11:13
+    Enter ".help" for instructions
+    Enter SQL statements terminated with a ";"
+    sqlite> .schema
+    CREATE TABLE conn (
+    'ts' double precision,
+    'uid' text,
+    'id.orig_h' text,
+    'id.orig_p' integer,
+    ...
+
+Note that the ASCII ``conn.log`` will still be created. To disable the ASCII writer for a
+log stream, you can remove the default filter:
+
+.. code:: bro
+
+    Log::remove_filter(Conn::LOG, "default");
+
+
+To create a custom SQLite log file, you have to create a new log stream that contains
+just the information you want to commit to the database. Please refer to the
+:ref:`framework-logging` documentation on how to create custom log streams.
+
+Reading Data from SQLite Databases
+==================================
+
+Like logging support, support for reading data from SQLite databases is built into Bro starting
+with version 2.2.
+
+Just as with the text-based input readers (please refer to the :ref:`framework-input`
+documentation for them and for basic information on how to use the input-framework), the SQLite reader
+can be used to read data - in this case the result of SQL queries - into tables or into events.
+
+Reading Data into Tables
+------------------------
+
+To read data from a SQLite database, we first have to provide Bro with the information, how
+the resulting data will be structured. For this example, we expect that we have a SQLite database,
+which contains host IP addresses and the user accounts that are allowed to log into a specific
+machine.
+
+The SQLite commands to create the schema are as follows::
+
+    create table machines_to_users (
+    host text unique not null,
+    users text not null);
+
+    insert into machines_to_users values ('192.168.17.1', 'bernhard,matthias,seth');
+    insert into machines_to_users values ('192.168.17.2', 'bernhard');
+    insert into machines_to_users values ('192.168.17.3', 'seth,matthias');
+
+After creating a file called ``hosts.sqlite`` with this content, we can read the resulting table
+into Bro:
+
+.. code:: bro
+
+    type Idx: record {
+        host: addr;
+    };
+
+    type Val: record {
+        users: set[string];
+    };
+
+    global hostslist: table[addr] of Val = table();
+
+    event bro_init() {
+        Input::add_table([$source="/var/db/hosts",
+            $name="hosts",
+            $idx=Idx,
+            $val=Val,
+            $destination=hostslist,
+            $reader=Input::READER_SQLITE,
+            $config=table(["query"] = "select * from machines_to_users;")
+        ]);
+
+    	Input::remove("hosts");
+    }
+
+    event Input::end_of_data(name: string, source: string) {
+    	if ( name != "hosts" )
+    	    return;
+
+    	# now all data is in the table
+    	print "Hosts list has been successfully imported";
+
+        # List the users of one host.
+    	print hostslist[192.168.17.1]$users;
+    }
+
+Afterwards, that table can be used to check logins into hosts against the available
+userlist.
+
+Turning Data into Events
+------------------------
+
+The second mode is to use the SQLite reader to output the input data as events. Typically there
+are two reasons to do this. First, when the structure of the input data is too complicated
+for a direct table import. In this case, the data can be read into an event which can then
+create the necessary data structures in Bro in scriptland.
+
+The second reason is, that the dataset is too big to hold it in memory. In this case, the checks
+can be performed on-demand, when Bro encounters a situation where it needs additional information.
+
+An example for this would be an internal huge database with malware hashes. Live database queries
+could be used to check the sporadically happening downloads against the database.
+
+The SQLite commands to create the schema are as follows::
+
+    create table malware_hashes (
+    hash text unique not null,
+    description text not null);
+
+    insert into malware_hashes values ('86f7e437faa5a7fce15d1ddcb9eaeaea377667b8', 'malware a');
+    insert into malware_hashes values ('e9d71f5ee7c92d6dc9e92ffdad17b8bd49418f98', 'malware b');
+    insert into malware_hashes values ('84a516841ba77a5b4648de2cd0dfcb30ea46dbb4', 'malware c');
+    insert into malware_hashes values ('3c363836cf4e16666669a25da280a1865c2d2874', 'malware d');
+    insert into malware_hashes values ('58e6b3a414a1e090dfc6029add0f3555ccba127f', 'malware e');
+    insert into malware_hashes values ('4a0a19218e082a343a1b17e5333409af9d98f0f5', 'malware f');
+    insert into malware_hashes values ('54fd1711209fb1c0781092374132c66e79e2241b', 'malware g');
+    insert into malware_hashes values ('27d5482eebd075de44389774fce28c69f45c8a75', 'malware h');
+    insert into malware_hashes values ('73f45106968ff8dc51fba105fa91306af1ff6666', 'ftp-trace');
+
+
+The following code uses the file-analysis framework to get the sha1 hashes of files that are
+transmitted over the network. For each hash, a SQL-query is run against SQLite. If the query
+returns with a result, we had a hit against our malware-database and output the matching hash.
+
+.. code:: bro
+
+    @load frameworks/files/hash-all-files
+
+    type Val: record {
+        hash: string;
+        description: string;
+    };
+
+    event line(description: Input::EventDescription, tpe: Input::Event, r: Val)
+        {
+        print fmt("malware-hit with hash %s, description %s", r$hash, r$description);
+        }
+
+    global malware_source = "/var/db/malware";
+
+    event file_hash(f: fa_file, kind: string, hash: string)
+        {
+
+        # check all sha1 hashes
+        if ( kind=="sha1" )
+            {
+	    Input::add_event(
+                [
+                $source=malware_source,
+                $name=hash,
+                $fields=Val,
+                $ev=line,
+                $want_record=T,
+                $config=table(
+                    ["query"] = fmt("select * from malware_hashes where hash='%s';", hash)
+                    ),
+                $reader=Input::READER_SQLITE
+                ]);
+            }
+        }
+
+    event Input::end_of_data(name: string, source:string)
+        {
+        if ( source == malware_source )
+            Input::remove(name);
+        }
+
+If you run this script against the trace in ``testing/btest/Traces/ftp/ipv4.trace``, you
+will get one hit.

doc/frameworks/logging.rst

@@ -387,3 +387,4 @@ Bro supports the following output formats other than ASCII:
 
    logging-dataseries
    logging-elasticsearch
+   logging-input-sqlite

src/input/readers/SQLite.cc

@@ -54,7 +54,7 @@ void SQLite::DoClose()
 		}
 	}
 
-bool SQLite::checkError( int code )
+bool SQLite::checkError(int code)
 	{
 	if ( code != SQLITE_OK && code != SQLITE_DONE )
 		{
@@ -73,23 +73,19 @@ bool SQLite::DoInit(const ReaderInfo& info, int arg_num_fields, const threading:
 		return false;
 		}
 
+	if ( Info().mode != MODE_MANUAL )
+		{
+		Error("SQLite only supports manual reading mode.");
+		return false;
+		}
+
 	started = false;
 
 	string fullpath(info.source);
 	fullpath.append(".sqlite");
 
-	string dbname;
-	map<const char*, const char*>::const_iterator it = info.config.find("dbname");
-	if ( it == info.config.end() )
-		{
-		MsgThread::Info(Fmt("dbname configuration option not found. Defaulting to source %s", info.source));
-		dbname = info.source;
-		}
-	else
-		dbname = it->second;
-
 	string query;
-	it = info.config.find("query");
+	map<const char*, const char*>::const_iterator it = info.config.find("query");
 	if ( it == info.config.end() )
 		{
 		Error(Fmt("No query specified when setting up SQLite data source. Aborting.", info.source));

src/logging/writers/SQLite.cc

@@ -124,16 +124,16 @@ bool SQLite::DoInit(const WriterInfo& info, int arg_num_fields,
 
 	string fullpath(info.path);
 	fullpath.append(".sqlite");
-	string dbname;
+	string tablename;
 
-	map<const char*, const char*>::const_iterator it = info.config.find("dbname");
+	map<const char*, const char*>::const_iterator it = info.config.find("tablename");
 	if ( it == info.config.end() )
 		{
-		MsgThread::Info(Fmt("dbname configuration option not found. Defaulting to path %s", info.path));
+		MsgThread::Info(Fmt("tablename configuration option not found. Defaulting to path %s", info.path));
-		dbname = info.path;
+		tablename = info.path;
 		}
 	else
-		dbname = it->second;
+		tablename = it->second;
 
 	if ( checkError(sqlite3_open_v2(
 					fullpath.c_str(),
@@ -145,7 +145,7 @@ bool SQLite::DoInit(const WriterInfo& info, int arg_num_fields,
 					NULL)) )
 		return false;
 
-	string create = "CREATE TABLE IF NOT EXISTS " + dbname + " (\n";
+	string create = "CREATE TABLE IF NOT EXISTS " + tablename + " (\n";
 		//"id SERIAL UNIQUE NOT NULL"; // SQLite has rowids, we do not need a counter here.
 
 	for ( unsigned int i = 0; i < num_fields; ++i )
@@ -193,7 +193,7 @@ bool SQLite::DoInit(const WriterInfo& info, int arg_num_fields,
 
 	// create the prepared statement that will be re-used forever...
 	string insert = "VALUES (";
-	string names = "INSERT INTO " + dbname + " ( ";
+	string names = "INSERT INTO " + tablename + " ( ";
 
 	for ( unsigned int i = 0; i < num_fields; i++ )
 		{

testing/btest/Baseline/scripts.base.frameworks.logging.sqlite.error/.stderr

@@ -1,3 +1,3 @@
-ssh/Log::WRITER_SQLITE: dbname configuration option not found. Defaulting to path ssh
+ssh/Log::WRITER_SQLITE: tablename configuration option not found. Defaulting to path ssh
 error: ssh/Log::WRITER_SQLITE: SQLite call failed: table ssh has no column named f
 error: ssh/Log::WRITER_SQLITE: terminating thread

testing/btest/scripts/base/frameworks/input/sqlite/basic.bro

@@ -89,7 +89,6 @@ event bro_init()
 	{
 	local config_strings: table[string] of string = {
 		 ["query"] = "select * from conn;",
-		 ["dbname"] = "conn"
 	};
 
 	outfile = open("../out");

testing/btest/scripts/base/frameworks/input/sqlite/error.bro

@@ -83,12 +83,10 @@ event bro_init()
 	{
 	local config_strings: table[string] of string = {
 		 ["query"] = "select * from ssh;",
-		 ["dbname"] = "ssh"
 	};
 	
 	local config_strings2: table[string] of string = {
 		 ["query"] = "select b, g, h from ssh;",
-		 ["dbname"] = "ssh"
 	};
 
 	outfile = open("../out");

testing/btest/scripts/base/frameworks/input/sqlite/port.bro

@@ -39,7 +39,6 @@ event bro_init()
 	{
 	local config_strings: table[string] of string = {
 		 ["query"] = "select port as p, proto from port;",
-		 ["dbname"] = "port"
 	};
 
 	outfile = open("../out");

testing/btest/scripts/base/frameworks/input/sqlite/types.bro

@@ -77,7 +77,6 @@ event bro_init()
 	{
 	local config_strings: table[string] of string = {
 		 ["query"] = "select * from ssh;",
-		 ["dbname"] = "ssh"
 	};
 
 	outfile = open("../out");