Reference

Interface Definitions (“evt files”)

An analyzer for Zeek does more than just parsing data. Accordingly, we need to tell Zeek a couple of additional pieces about Spicy parsers we want it to provide to Zeek:

Analyzer setup

Zeek needs to know what type of analyzers we are creating, when we want Zeek to activate them, and what Spicy unit types to use as their parsing entry point.

Event definitions

We need to tell Zeek which events to provide and when to trigger them.

We define all of these through custom interface definition files that Spicy’s compiler for Zeek reads in. These files use an *.evt extension, and the following subsections discuss their content in more detail.

Generally, empty lines and comments starting with # are ignored in an *.evt.

Note

The syntax for *.evt files comes with some legacy pieces that aren’t particularly pretty. We may clean that up at some point.

Analyzer Setup

You can define protocol analyzers, packet analyzers and file analyzers in an *.evt file, per the following.

Protocol Analyzer

To define a protocol analyzer, add a new section to an *.evt file that looks like this:

protocol analyzer ANALYZER_NAME over TRANSPORT_PROTOCOL:
    PROPERTY_1,
    PROPERTY_2,
    ...
    PROPERTY_N;

Here, ANALYZER_NAME is a name to identify your analyzer inside Zeek. You can choose names arbitrarily as long as they are unique. As a convention, however, we recommend name with a spicy::* prefix (e.g., spicy::BitTorrent).

On the Zeek-side, through some normalization, these names automatically turn into tags added to Zeek’s Analyzer::Tag enum. For example, spicy::BitTorrent turns into Analyzer::ANALYZER_SPICY_BITTORRENT.

The analyzer’s name is also what goes into Zeek signatures to activate an analyzer DPD-style. If the name is spicy::BitTorrent, you’d write enable "spicy::BitTorrent" into the signature.

Note

Once you have made your analyzers available to Zeek (which we will discuss below), running zeek -NN Zeek::Spicy will show you a summary of what’s now available, including their Zeek-side names and tags.

TRANSPORT_PROTOCOL can be either tcp or udp, depending on the transport-layer protocol that your new analyzer wants to sit on top of.

Following that initial protocol analyzer ... line, a set of properties defines further specifics of your analyzer. The following properties are supported:

parse [originator|responder] with SPICY_UNIT

Specifies the top-level Spicy unit(s) the analyzer uses for parsing payload, with SPICY_UNIT being a fully-qualified Spicy-side type name (e.g. HTTP::Request). The unit type must have been declared as public in Spicy.

If originator is given, the unit is used only for parsing the connection’s originator-side payload; and if responder is given, only for responder-side payload. If neither is given, it’s used for both sides. In other words, you can use different units per side by specifying two properties parse originator with ... and parse responder with ....

port PORT or ports { PORT_1, ..., PORT_M }

Specifies one or more well-known ports for which you want Zeek to automatically activate your analyzer with corresponding connections. Each port must be specified in Spicy’s syntax for port constants (e.g., 80/tcp), or as a port range PORT_START-PORT_END where start and end port are port constants forming a closed interval. The ports’ transport protocol must match that of the analyzer.

Note

Zeek will also honor any %port meta data property that the responder-side SPICY_UNIT may define (as long as the attribute’s direction is not originator).

Note

While using port (or %port) can be convinient, for production analyzers we recommended to instead register their well-known ports from inside a Zeek script, using a snippet like this:

module MyAnalyzer;

export {
    const ports = { 12345/tcp } &redef;
}

redef likely_server_ports += { ports };

event zeek_init() &priority=5
    {
    Analyzer::register_for_ports(Analyzer::ANALYZER_MY_ANALYZER, ports);
    }

This follows the idiomatic Zeek pattern for defining well-known ports that allows users to customize them through their own site-specific scripts (e.g., redef MyAnalyzer::port += { 12346/tcp };). The package template includes such code instead of defining ports inside the EVT file.

replaces ANALYZER_NAME

Replaces a built-in analyzer that Zeek already provides with a new Spicy version by internally disabling the existing analyzer and redirecting any usage to the new Spicy analyzer instead. ANALYZER_NAME is the Zeek-side name of the analyzer. To find that name, inspect the output of zeek -NN for available analyzers:

# zeek -NN | grep '\[Analyzer\]'
...
[Analyzer] SMTP (ANALYZER_SMTP, enabled)
...

Here, SMTP is the name you would write into replaces to disable the built-in SMTP analyzer.

The replacement takes effect in most places where normally the existing Zeek analyzer would be used, or gets referenced, including in Zeek scripts (e.g., when registering well-known ports).

Note

replaces is not limited to substituting built-in analyzers; you could also replace one Spicy analyzer with another. However, there is no support for more complex scenarios like chains of analyzers replacing each other; results are undefined in that case. It’s best to stick to replacing classic, built-in analyzers, and control everything else simply through not loading any undesired Spicy analyzers in the first place.

As a full example, here’s what a new HTTP analyzer could look like:

protocol analyzer spicy::HTTP over TCP:
    parse originator with HTTP::Requests,
    parse responder with HTTP::Replies,
    port 80/tcp,
    replaces HTTP;

Packet Analyzer

Defining packet analyzers works quite similar to protocol analyzers through *.evt sections like this:

packet analyzer ANALYZER_NAME:
    PROPERTY_1,
    PROPERTY_2,
    ...
    PROPERTY_N;

Here, ANALYZER_NAME is again a name to identify your analyzer inside Zeek. On the Zeek-side, the name will be added to Zeek’s PacketAnalyzer::Tag enum.

Packet analyzers support the following properties:

parse with SPICY_UNIT

Specifies the top-level Spicy unit the analyzer uses for parsing each packet, with SPICY_UNIT being a fully-qualified Spicy-side type name. The unit type must have been declared as public in Spicy.

replaces ANALYZER_NAME

Disables an existing packet analyzer that Zeek already provides internally, allowing you to replace a built-in analyzer with a new Spicy version. ANALYZER_NAME is the Zeek-side name of the existing analyzer. To find that name, inspect the output of zeek -NN for available analyzers:

# zeek -NN | grep '\[Packet Analyzer\]'
...
[Packet Analyzer] Ethernet (ANALYZER_ETHERNET)
...

Here, Ethernet is the name you would give to replaces to disable the built-in Ethernet analyzer.

When replacing an existing packet analyzer, you still need to configure your new analyzer on the Zeek side through a spicy_init() event handler. See below for more.

Note

This feature requires Zeek >= 5.2.

As a full example, here’s what a new packet analyzer could look like:

packet analyzer spicy::RawLayer:
    parse with RawLayer::Packet;

In addition to the Spicy-side configuration, packet analyzers also need to be registered with Zeek inside a spicy_init event handler; see the Zeek documentation for more. The -NN output shows your new analyzer’s ANALYZER_* tag to use.

Note

Depending on relative loading order between your analyzer and Zeek’s scripts, in some situations you may not have access to the ANALYZER_* tag referring to your new analyzer. If you run into this, use PacketAnalyzer::try_register_packet_analyzer_by_name to register your Spicy analyzer by name (as shown by -NN). Example:

event spicy_init()
    {
    if ( ! PacketAnalyzer::try_register_packet_analyzer_by_name("Ethernet", 0x88b5, "spicy::RawLayer") )
            Reporter::error("unknown analyzer name used");
    }

File Analyzer

Defining file analyzers works quite similar to protocol analyzers, through *.evt sections like this:

file analyzer ANALYZER_NAME:
    PROPERTY_1,
    PROPERTY_2,
    ...
    PROPERTY_N;

Here, ANALYZER_NAME is again a name to identify your analyzer inside Zeek. On the Zeek-side, the name will be added to Zeek’s Files::Tag enum.

File analyzers support the following properties:

parse with SPICY_UNIT

Specifies the top-level Spicy unit the analyzer uses for parsing file content, with SPICY_UNIT being a fully-qualified Spicy-side type name. The unit type must have been declared as public in Spicy.

mime-type MIME-TYPE

Specifies a MIME type for which you want Zeek to automatically activate your analyzer when it sees a corresponding file on the network. The type is specified in standard type/subtype notion, without quotes (e.g., image/gif).

Note

Zeek will also honor any %mime-type meta data property that the SPICY_UNIT may define.

Note

Keep in mind that Zeek identifies MIME types through “content sniffing” (i.e., similar to libmagic), and usually not by protocol-level headers (e.g., not through HTTP’s Content-Type header). If in doubt, examine files.log for what it records as a file’s type.

replaces ANALYZER_NAME

Disables an existing file analyzer that Zeek already provides internally, allowing you to replace a built-in analyzer with a new Spicy version. ANALYZER_NAME is the Zeek-side name of the analyzer. To find that name, inspect the output of zeek -NN for available analyzers:

# zeek -NN | grep '\[File Analyzer\]'
...
[File Analyzer] PE (ANALYZER_PE, enabled)
...

Here, PE is the name you would write into replaces to disable the built-in PE analyzer.

Note

This feature requires Zeek >= 4.1.

As a full example, here’s what a new GIF analyzer could look like:

file analyzer spicy::GIF:
    parse with GIF::Image,
    mime-type image/gif;

Event Definitions

To define a Zeek event that you want the Spicy analyzer to trigger, you add lines of the form:

on HOOK_ID -> event EVENT_NAME(ARG_1, ..., ARG_N);

on HOOK_ID if COND -> event EVENT_NAME(ARG_1, ..., ARG_N);

Zeek automatically derives from this everything it needs to register new events with Zeek, including a mapping of the arguments’ Spicy types to corresponding Zeek types. More specifically, these are the pieces going into such an event definition:

on HOOK_ID

A Spicy-side ID that defines when you want to trigger the event. This works just like an on ... unit hook, and you can indeed use anything here that Spicy supports for those as well (except container hooks). So, e.g., on HTTP::Request::%done triggers an event whenever a HTTP::Request unit has been fully parsed, and on HTTP::Request::uri leads to an event each time the uri field has been parsed. (In the former example you may skip the %done, actually: on HTTP::Request implicitly adds it.)

EVENT_NAME

The Zeek-side name of the event you want to generate, preferably including a namespace (e.g., http::request).

ARG_1, ..., ARG_N

Arguments to pass to the event, given as arbitrary Spicy expressions. Each expression will be evaluated within the context of the unit that the on ... triggers on, similar to code running inside the body of a corresponding unit hook. That means the expression has access to self for accessing the unit instance that’s currently being parsed.

The Spicy type of the expression determines the Zeek-side type of the corresponding event argument. Most Spicy types translate over pretty naturally, the following summarizes the translation:

Type conversion from Spicy to Zeek

Spicy Type

Zeek Type

Notes

addr

addr

bool

bool

bytes

string

enum { ... }

enum { ... }

[1]

int(8|16|32|64)

int

interval

interval

list<T>

vector of T

map<V,K>

table[V] of K

optional<T>

T

[2]

port

port

real

double

set<T>

set[T]

string

string

time

time

tuple<T_1, ... ,T_N>

record { T1, ..., T_N }

[3]

uint(8|16|32|64)

count

vector<T>

vector of T

Note

[1]

A corresponding Zeek-side enum type is automatically created. See below for more.

One special-case: For convenience, the Spicy-side Protocol enum is automatically mapped to the Zeek-side transport_proto enum.

[2]

The optional value must have a value, otherwise a runtime exception will be thrown.

[3]

Must be mapped to a Zeek-side record type with matching fields.

If a tuple element is mapped to a record field with a &default or &optional attribute, a couple special cases are supported:

  • If the expression evaluates to Null, the record field is left unset.

  • If the element’s expression uses the .? operator and that fails to produce a value, the record field is likewise left unset.

In addition to full Spicy expressions, there are four reserved IDs with specific meanings when used as arguments:

$conn

Refers to the connection that’s currently being processed by Zeek. On the Zeek-side this will turn into a parameter of Zeek type connection. This ID can be used only with protocol analyzers.

$file

Refers to the file that’s currently being processed by Zeek. On the Zeek-side this will turn into a parameter of Zeek type fa_file. This ID can be used with file or protocol analyzers. For protocol analyzers it refers to the most recently opened, but not yet closed file, see zeek::file_begin and zeek::file_end.

$packet

Refers to the packet that’s currently being processed by Zeek. On the Zeek-side this will turn into a parameter of Zeek type raw_pkt_hdr, with any fields filled in that have already been parsed by Zeek’s built-in analyzers. This ID can be used only with packet analyzers. (Note that instantiation of raw_pkt_hdr values can be relatively expensive on the Zeek side, so best to limit usage of this ID to a small part of the overall traffic.)

$is_orig

A boolean indicating if the data currently being processed is coming from the originator (True) or responder (False) of the underlying connection. This turns into a corresponding boolean value on the Zeek side. This ID can be used only with protocol analyzers.

Note

Some tips:

  • If you want to force a specific type on the Zeek-side, you have a couple of options:

    1. Spicy may provide a cast operator from the actual type into the desired type (e.g., cast<uint64>(..)).

    2. Argument expressions have access to global functions defined in the Spicy source files, so you can write a conversion function taking an argument with its original type and returning it with the desired type.

  • List comprehension can be convenient to fill Zeek vectors: [some_func(i) for i in self.my_list].

if COND

If given, events are only generated if the expression COND evaluates to true. Just like event arguments, the expression is evaluated in the context of the current unit instance and has access to self.

Exporting Types

As we discuss above, the type of each event argument maps over to a corresponding Zeek type. On the Zeek side, that corresponding type needs to be known by Zeek. That is always the case for built-in atomic types (per the conversion table), but it can turn out more challenging to achieve for custom types, such as enum and record, for which you would normally need to create matching type declarations in your Zeek scripts. While that’s not necessarily hard, but it can become quite cumbersome.

Fortunately, there’s help: for most types, Zeek can instantiate corresponding types automatically as it loads the corresponding analyzer. While you will never actually see the Zeek-side type declarations, they will be available inside your Zeek scripts as if you had typed them out yourself–similar to other types that are built into Zeek itself.

To have the Zeek create a type for your analyzer automatically, you need to export the Spicy type in your EVT file. The syntax for that is:

export SPICY_ID [as ZEEK_ID];

Here, SPICY_ID is the fully-scoped type ID on the Spicy side, and ZEEK_ID is the fully-scoped type ID you want in Zeek. If you leave out the as ... part, the Zeek name will be the same as the Spicy name, including its namespace. For example, say you have a Spicy unit TFTP::Request. Adding export TFTP::Request; to your EVT file will make a record type of the same name, and with the same fields, available in Zeek. If you instead use export TFTP::Request as TheOtherTFTP::Request, it will be placed into a different namespace instead.

Exporting types generally works for most Spicy types as long as there’s an ID associated with them in your Spicy code. However, exporting is most helpful with user-defined types, such as enum and unit, because it can save you quite a bit of typing there. We discuss the more common type conversions in more detail below.

To confirm the types made available to Zeek, you can see all exports in the output of zeek -NN. With our 2nd TFTP::Request example, that looks like this:

[...]
# zeek -NN Zeek::Spicy
Zeek::Spicy - Support for Spicy parsers (*.hlto)
    [Type] TheOtherTFTP::Request
[...]

Note

Most, but not all, types can be exported automatically. For example, self-recursive types are currently not supported. Generally, if you run into trouble exporting a type, you can always fall back to declaring a corresponding Zeek version yourself in your Zeek script. Consider the export mechanism as a convenience feature that helps avoid writing a lot of boiler plate code in common cases.

Enum Types

When you export a Spicy enum type, Zeek creates a corresponding Zeek enum type. For example, assume the following Spicy declaration:

module Test;

type MyEnum = enum {
    A = 83,
    B = 84,
    C = 85
};

Using export Test::MyEnum;, Zeek will create the equivalent of the following Zeek type for use in your scripts:

module Test;

export {

  type MyEnum: enum {
      MyEnum_A = 83,
      MyEnum_B = 84,
      MyEnum_A = 85,
      MyEnum_Undef = -1
  };

}

(The odd naming is due to ID limitations on the Zeek side.)

Note

For backward compatibility, the enum type comes with an additional property: all public enum types are automatically exported, even without adding any export to your EVT file. This feature may go away at some point, and we suggest to not rely on it on new code; always use export for enum types as well.

Unit Types

When you export a Spicy unit type, Zeek creates a corresponding Zeek record type. For example, assume the following Spicy declaration:

module Test;

type MyRecord = unit {
    a: bytes &size=4;
    b: uint16;

    var c: bool;
};

Using export Test::MyRecord;, Zeek will then create the equivalent of the following Zeek type for use in your scripts:

module Test;

export {

  type MyRecord: record {
    a: string &optional;
    b: count &optional;
    c: bool;
  };

}

The individual fields map over just like event arguments do, following the the table above. For aggregate types, this works recursively: if, e.g., a field is itself of unit type and that type has been exported as well, Zeek will map it over accordingly to the corresponding record type. Note that such dependent types must be exported first in the EVT file for this to work. As a result, you cannot export self-recursive unit types.

As you can see in the example, unit fields are always declared as &optional on the Zeek-side, as they may not have been set during parsing. Unit variables are non-optional by default, unless declared as &optional in Spicy.

Controlling created Zeek types

If needed the automatic unit type exporting described above can be customized. The general syntax for this is:

export SPICY_ID [with { [SIPCY_FIELD_NAME [&log], ]... }];
export SPICY_ID [as ZEEK_ID [ with { [SIPCY_FIELD_NAME [&log], ]...]  }];
export SPICY_ID [without { [SIPCY_FIELD_NAME, ]... }];
export SPICY_ID [as ZEEK_ID [ without { [SIPCY_FIELD_NAME, ]...]  }];
export SPICY_ID &log;
export SPICY_ID as ZEEK_ID &log;

This allows

  1. including fields to export by naming them in with, e.g., export foo::A with { x, y } creates a Zeek record with the Spicy unit fields x and y added to the Zeek record type.

  2. excluding fields from export by naming them in without, e.g., export foo::A without { x } creates a Zeek record with all fields but x of the Spicy unit exported.

  3. renaming the Spicy type on export, e.g., export foo::A as bar::X exports the Spicy unit type A in Spicy module foo to a Zeek record type X in Zeek module bar.

  4. controlling &log attribute on generated Zeek record types. We can either the mark all Zeek record fields &log by marking the whole type &log with e.g., export foo::A &log or export foo::A as bar::X &log, or mark individual fields &log in the with field list.

Example: Controlling exported fields

Assume we are given a Spicy unit type foo::X:

module foo;

public type X = unit {
    x: uint8;
    y: uint8;
    z: uint8;
};

To create a Zeek type without the field x we can use the following export statement:

export foo::X without { x };

Example: Adding Zeek &log attributes to created Zeek record types

In Zeek records, data intended to be written to Zeek log streams needs to be marked &log. We can trigger creation of this attribute from Spicy either per-field or for the whole record.

The export statement

export foo::X &log;

creates a Zeek record type

module foo;

export {

  type X: record {
    x: count &optional &log;
    y: count &optional &log;
    z: count &optional &log;
  };

}

To mark individual fields &log we can use the attribute on the respective fields, e.g.,

export foo::X with { x &log, y, z };

creates a Zeek record type

module foo;

export {

  type X: record {
    x: count &optional &log;
    y: count &optional;
    z: count &optional;
  };

}

Struct Types

A Spicy struct type maps over to Zeek in the same way as unit types do, treating each field like a unit variable. See there for more information.

Tuple Types

A Spicy tuple type maps over to a Zeek record similar to how unit types do, treating each tuple element like a unit variable. See there for more information.

Exporting works only for tuple types that declare names for all their elements.

Importing Spicy Modules

Code in an *.evt file may need access to additional Spicy modules, such as when expressions for event parameters call Spicy functions defined elsewhere. To make a Spicy module available, you can insert import statements into the *.evt file that work just like in Spicy code:

import NAME

Imports Spicy module NAME.

import NAME from X.Y.Z;

Searches for the module NAME (i.e., for the filename NAME.spicy) inside a sub-directory X/Y/Z along the search path, and then imports it.

Conditional Compilation

*.evt files offer the same basic form of conditional compilation through @if/@else/@endif blocks as Spicy scripts. Zeek makes two additional identifiers available for testing to both *.evt and *.spicy code:

HAVE_ZEEK

Always set to 1 by Zeek. This can be used for feature testing from Spicy code to check if it’s being compiled for Zeek.

ZEEK_VERSION

The numerical Zeek version that’s being compiled for (see zeek -e 'print Version::number').

This is an example bracketing code by Zeek version in an EVT file:

@if ZEEK_VERSION < 30200
    <EVT code for Zeek versions older than 3.2>
@else
    <EVT code for Zeek version 3.2 or newer>
@endif

Compiling Analyzers

Once you have the *.spicy and *.evt source files for your new analyzer, you need to precompile them into an *.hlto object file containing their final executable code, which Zeek will then use. To do that, pass the relevant *.spicy and *.evt files to spicyz, then have Zeek load the output. To repeat the example from the Getting Started guide:

# spicyz -o my-http-analyzer.hlto my-http.spicy my-http.evt
# zeek -Cr request-line.pcap my-http-analyzer.hlto my-http.zeek
Zeek saw from 127.0.0.1: GET /index.html 1.0

Instead of providing the precompiled analyzer on the Zeek command line, you can also copy them into ${prefix}/lib/zeek/spicy. Zeek will automatically load any *.hlto object files it finds there. In addition, Zeek also scans its plugin directory for *.hlto files. Alternatively, you can override both of those locations by setting the environment variable ZEEK_SPICY_MODULE_PATH to a set of colon-separated directories to search instead. Zeek will then only look there. In all cases, Zeek searches any directories recursively, so it will find *.hlto also if they are nested in subfolders.

Run spicyz -h to see some additional options it provides, which are similar to https://docs.zeek.org/projects/spicy/en/latest/toolchain.html#spicy-driver.

Controlling Zeek from Spicy

Spicy grammars can import a provided library module zeek to gain access to Zeek-specific functions that call back into Zeek’s processing:

function zeek::confirm_protocol()

[Deprecated] Triggers a DPD protocol confirmation for the current connection.

This function has been deprecated and will be removed. Use spicy::accept_input instead, which will have the same effect with Zeek.

function zeek::reject_protocol(reason: string)

[Deprecated] Triggers a DPD protocol violation for the current connection.

This function has been deprecated and will be removed. Use spicy::decline_input instead, which will have the same effect with Zeek.

function zeek::weird(id: string, addl: string = "") : &cxxname="zeek::spicy::rt::weird";

Reports a “weird” to Zeek. This should be used with similar semantics as in Zeek: something quite unexpected happening at the protocol level, which however does not prevent us from continuing to process the connection.

id: the name of the weird, which (just like in Zeek) should be a static string identifying the situation reported (e.g., unexpected_command).

addl: additional information to record along with the weird

function zeek::is_orig() : bool

Returns true if we’re currently parsing the originator side of a connection.

function zeek::uid() : string

Returns the current connection’s UID.

function zeek::conn_id() : tuple<orig_h: addr, orig_p: port, resp_h: addr, resp_p: port>

Returns the current connection’s 4-tuple ID to make IP address and port information available.

function zeek::flip_roles()

Instructs Zeek to flip the directionality of the current connection.

function zeek::number_packets() : uint64

Returns the number of packets seen so far on the current side of the current connection.

function zeek::protocol_begin(analyzer: optional<string>, protocol: spicy::Protocol = spicy::Protocol::TCP)

Adds a Zeek-side child protocol analyzer to the current connection.

If the same analyzer was added previously with protocol_handle_get_or_create or protocol_begin with same argument, and not closed with protocol_handle_close or protocol_end, no new analyzer will be added.

See protocol_handle_get_or_create for the error semantics of this function.

analyzer: type of analyzer to instantiate, specified through its Zeek-side name (similar to what Zeek’s signature action enable takes)

protocol: the transport-layer protocol that the analyzer uses; only TCP is currently supported here

Note: For backwards compatibility, the analyzer argument can be left unset to add a DPD analyzer. This use is deprecated, though; use the single-argument version of protocol_begin for that instead.

function zeek::protocol_begin(protocol: spicy::Protocol = spicy::Protocol::TCP)

Adds a Zeek-side DPD child protocol analyzer performing dynamic protocol detection on subsequently provided data.

If the same DPD analyzer was added previously with protocol_handle_get_or_create or protocol_begin with same argument, and not closed with protocol_handle_close or protocol_end, no new analyzer will be added.

See protocol_handle_get_or_create for the error semantics of this function.

protocol: the transport-layer protocol on which to perform protocol detection; only TCP is currently supported here

function zeek::protocol_handle_get_or_create(analyzer: string, protocol: spicy::Protocol = spicy::Protocol::TCP) : ProtocolHandle

Gets a handle to a Zeek-side child protocol analyzer for the current connection.

If no such child exists yet it will be added; otherwise a handle to the existing child protocol analyzer will be returned.

This function will return an error if:

  • not called from a protocol analyzer, or

  • the requested child protocol analyzer is of unknown type or not support by the requested transport protocol, or

  • creation of a child analyzer of the requested type was prevented by a previous call of disable_analyzer with prevent=T

analyzer: type of analyzer to get or instantiate, specified through its Zeek-side name (similar to what Zeek’s signature action enable takes).

protocol: the transport-layer protocol that the analyser uses; only TCP is currently supported here

function zeek::protocol_data_in(is_orig: bool, data: bytes, protocol: spicy::Protocol = spicy::Protocol::TCP)

Forwards protocol data to all previously instantiated Zeek-side child protocol analyzers of a given transport-layer.

is_orig: true to feed the data to the child’s originator side, false for the responder

data: chunk of data to forward to child analyzer

protocol: the transport-layer protocol of the children to forward to; only TCP is currently supported here

function zeek::protocol_data_in(is_orig: bool, data: bytes, h: ProtocolHandle)

Forwards protocol data to a specific previously instantiated Zeek-side child analyzer.

is_orig: true to feed the data to the child’s originator side, false for the responder

data: chunk of data to forward to child analyzer

h: handle to the child analyzer to forward data into

function zeek::protocol_gap(is_orig: bool, offset: uint64, len: uint64, h: optional<ProtocolHandle> = Null)

Signals a gap in input data to all previously instantiated Zeek-side child protocol analyzers.

is_orig: true to signal gap to the child’s originator side, false for the responder

offset: start offset of gap in input stream

len: size of gap

h: optional handle to the child analyzer signal a gap to, else signal to all child analyzers

function zeek::protocol_end()

Signals end-of-data to all previously instantiated Zeek-side child protocol analyzers and removes them.

function zeek::protocol_handle_close(handle: ProtocolHandle)

Signals end-of-data to the given child analyzer and removes it.

The given handle must be live, i.e., it must not have been used in a previous protocol_handle_close call, and must not have been live when protocol_end was called. If the handle is not live a runtime error will be triggered.

handle: handle to the child analyzer to remove

function zeek::file_begin(mime_type: optional<string> = Null, fuid: optional<string> = Null) : string

Signals the beginning of a file to Zeek’s file analysis, associating it with the current connection. Optionally, a mime type can be provided. It will be passed on to Zeek’s file analysis framework. Optionally, a file ID can be provided. It will be passed on to Zeek’s file analysis framework. Returns the Zeek-side file ID of the new file.

function zeek::fuid() : string

Returns the current file’s FUID.

function zeek::terminate_session()

Terminates the currently active Zeek-side session, flushing all state. Any subsequent activity will start a new session from scratch. This can only be called from inside a protocol analyzer.

function zeek::skip_input()

Tells Zeek to skip sending any further input data to the current analyzer. This is supported for protocol and file analyzers.

function zeek::file_set_size(size: uint64, fid: optional<string> = Null)

Signals the expected size of a file to Zeek’s file analysis.

size: expected size of file fid: Zeek-side ID of the file to operate on; if not given, the file started by the most recent file_begin() will be used

function zeek::file_data_in(data: bytes, fid: optional<string> = Null)

Passes file content on to Zeek’s file analysis.

data: chunk of raw data to pass into analysis fid: Zeek-side ID of the file to operate on; if not given, the file started by the most recent file_begin() will be used

function zeek::file_data_in_at_offset(data: bytes, offset: uint64, fid: optional<string> = Null)

Passes file content at a specific offset on to Zeek’s file analysis.

data: chunk of raw data to pass into analysis offset: position in file where data starts fid: Zeek-side ID of the file to operate on; if not given, the file started by the most recent file_begin() will be used

function zeek::file_gap(offset: uint64, len: uint64, fid: optional<string> = Null)

Signals a gap in a file to Zeek’s file analysis.

offset: position in file where gap starts len: size of gap fid: Zeek-side ID of the file to operate on; if not given, the file started by the most recent file_begin() will be used

function zeek::file_end(fid: optional<string> = Null)

Signals the end of a file to Zeek’s file analysis.

fid: Zeek-side ID of the file to operate on; if not given, the file started by the most recent file_begin() will be used

function zeek::forward_packet(identifier: uint32)

Inside a packet analyzer, forwards what data remains after parsing the top-level unit on to another analyzer. The index specifies the target, per the current dispatcher table.

function zeek::network_time() : time

Gets the network time from Zeek.

function zeek::get_address(id: string) : addr

Returns the value of a global Zeek script variable of Zeek type addr. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_bool(id: string) : bool

Returns the value of a global Zeek script variable of Zeek type bool. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_count(id: string) : uint64

Returns the value of a global Zeek script variable of Zeek type count. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_double(id: string) : real

Returns the value of a global Zeek script variable of Zeek type double. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_enum(id: string) : string

Returns the value of a global Zeek script variable of Zeek type enum. The value is returned as a string containing the enum’s label name, without any scope. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_int(id: string) : int64

Returns the value of a global Zeek script variable of Zeek type int. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_interval(id: string) : interval

Returns the value of a global Zeek script variable of Zeek type interval. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_port(id: string) : port

Returns the value of a global Zeek script variable of Zeek type port. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_record(id: string) : ZeekRecord

Returns the value of a global Zeek script variable of Zeek type record. The value is returned as an opaque handle to the record, which can be used with the zeek::record_*() functions to access the record’s fields. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_set(id: string) : ZeekSet

Returns the value of a global Zeek script variable of Zeek type set. The value is returned as an opaque handle to the set, which can be used with the zeek::set_*() functions to access the set’s content. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_string(id: string) : bytes

Returns the value of a global Zeek script variable of Zeek type string. The string’s value is returned as a Spicy bytes value. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_subnet(id: string) : network

Returns the value of a global Zeek script variable of Zeek type subnet. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_table(id: string) : ZeekTable

Returns the value of a global Zeek script variable of Zeek type table. The value is returned as an opaque handle to the set, which can be used with the zeek::set_*() functions to access the set’s content. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_time(id: string) : time

Returns the value of a global Zeek script variable of Zeek type time. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_vector(id: string) : ZeekVector

Returns the value of a global Zeek script variable of Zeek type vector. The value is returned as an opaque handle to the vector, which can be used with the zeek::vector_*() functions to access the vector’s content. Throws an exception if there’s no such Zeek of that name, or if it’s not of the expected type.

id: fully-qualified name of the global Zeek variable to retrieve

function zeek::get_value(id: string) : ZeekVal

Returns an opaque handle to a global Zeek script variable. The handle can be used with the zeek::as_*() functions to access the variable’s value. Throws an exception if there’s no Zeek variable of that name.

function zeek::as_address(v: ZeekVal) : addr

Returns a Zeek addr value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_bool(v: ZeekVal) : bool

Returns a Zeek bool value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_count(v: ZeekVal) : uint64

Returns a Zeek count value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_double(v: ZeekVal) : real

Returns a Zeek double value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_enum(v: ZeekVal) : string

Returns a Zeek enum value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_int(v: ZeekVal) : int64

Returns a Zeek int value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_interval(v: ZeekVal) : interval

Returns a Zeek interval value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_port(v: ZeekVal) : port

Returns a Zeek port value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_record(v: ZeekVal) : ZeekRecord

Returns a Zeek record value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_set(v: ZeekVal) : ZeekSet

Returns a Zeek set value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_string(v: ZeekVal) : bytes

Returns a Zeek string value refereced by an opaque handle. The string’s value is returned as a Spicy bytes value. Throws an exception if the referenced value is not of the expected type.

function zeek::as_subnet(v: ZeekVal) : network

Returns a Zeek subnet value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_table(v: ZeekVal) : ZeekTable

Returns a Zeek table value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_time(v: ZeekVal) : time

Returns a Zeek time value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::as_vector(v: ZeekVal) : ZeekVector

Returns a Zeek vector value refereced by an opaque handle. Throws an exception if the referenced value is not of the expected type.

function zeek::set_contains(id: string, v: any) : bool

Returns true if a Zeek set contains a given value. Throws an exception if the given ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek set to check v: value to check for, which must be of the Spicy-side equivalent of the set’s key type

function zeek::set_contains(s: ZeekSet, v: any) : bool

Returns true if a Zeek set contains a given value. Throws an exception if the set does not have the expected type.

s: opaque handle to the Zeek set, as returned by other functions v: value to check for, which must be of the Spicy-side equivalent of the set’s key type

function zeek::table_contains(id: string, v: any) : bool

Returns true if a Zeek table contains a given value. Throws an exception if the given ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek table to check v: value to check for, which must be of the Spicy-side equivalent of the table’s key type

function zeek::table_contains(t: ZeekTable, v: any) : bool

Returns true if a Zeek table contains a given value. Throws an exception if the given ID does not exist, or does not have the expected type.

t: opaque handle to the Zeek table, as returned by other functions v: value to check for, which must be of the Spicy-side equivalent of the table’s key type

function zeek::table_lookup(id: string, v: any) : optional<ZeekVal>

Returns the value associated with a key in a Zeek table. Returns an error result if the key does not exist in the table. Throws an exception if the given table ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek table to check v: value to lookup, which must be of the Spicy-side equivalent of the table’s key type

function zeek::table_lookup(t: ZeekTable, v: any) : optional<ZeekVal>

Returns the value associated with a key in a Zeek table. Returns an error result if the key does not exist in the table. Throws an exception if the given table ID does not exist, or does not have the expected type.

t: opaque handle to the Zeek table, as returned by other functions v: value to lookup, which must be of the Spicy-side equivalent of the table’s key type

function zeek::record_has_value(id: string, field: string) : bool

Returns true if a Zeek record provides a value for a given field. This includes fields with &default values. Throws an exception if the given ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek record to check field: name of the field to check

function zeek::record_has_value(r: ZeekRecord, field: string) : bool

Returns true if a Zeek record provides a value for a given field. This includes fields with &default values.

r: opaque handle to the Zeek record, as returned by other functions field: name of the field to check

function zeek::record_has_field(id: string, field: string) : bool

Returns true if the type of a Zeek record has a field of a given name. Throws an exception if the given ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek record to check field: name of the field to check

function zeek::record_has_field(r: ZeekRecord, field: string) : bool

Returns true if the type of a Zeek record has a field of a given name.

r: opaque handle to the Zeek record, as returned by other functions field: name of the field to check

function zeek::record_field(id: string, field: string) : ZeekVal

Returns a field’s value from a Zeek record. Throws an exception if the given ID does not exist, or does not have the expected type; or if there’s no such field in the record type, or if the field does not have a value.

id: fully-qualified name of the global Zeek record to check field: name of the field to retrieve

function zeek::record_field(r: ZeekRecord, field: string) : ZeekVal

Returns a field’s value from a Zeek record. Throws an exception if the given record does not have such a field, or if the field does not have a value.

r: opaque handle to the Zeek record, as returned by other functions field: name of the field to retrieve

function zeek::vector_index(id: string, index: uint64) : ZeekVal

Returns the value of an index in a Zeek vector. Throws an exception if the given ID does not exist, or does not have the expected type; or if the index is out of bounds.

id: fully-qualified name of the global Zeek vector to check index: index of the element to retrieve

function zeek::vector_index(v: ZeekVector, index: uint64) : ZeekVal

Returns the value of an index in a Zeek vector. Throws an exception if the index is out of bounds.

v: opaque handle to the Zeek vector, as returned by other functions index: index of the element to retrieve

function zeek::vector_size(id: string) : uint64

Returns the size of a Zeek vector. Throws an exception if the given ID does not exist, or does not have the expected type.

id: fully-qualified name of the global Zeek vector to check

function zeek::vector_size(v: ZeekVector) : uint64

Returns the size of a Zeek vector.

v: opaque handle to the Zeek vector, as returned by other functions

… zeek_variables:

Accessing Zeek Variables from Spicy

You can access Zeek-side global variables from inside Spicy, which is particularly handy for configuring Spicy analyzers from Zeek script code. The zeek module facilitates this through a set of functions coverting the current value of Zeek variables into corresponding Spicy values. For example, let’s say you would like to provide a Zeek script option with a count value that your Spicy analyzer can leverage. On the Zeek side, you’d define the option like this:

module MyModule;

export {
    option my_value: count &default=42;
}

Then, in your Spicy code, you can access the value of that option like this:

import zeek;

...
local my_value = zeek::get_count("MyModule::my_value");
...

This looks up the option by its global, fully-scoped ID and returns the Zeek-side count as a Spicy uint64. If there are any errors, such as the global not existing or having the wrong type, the get_count() would abort with an exception.

There are corresponding conversion functions for most of Zeek’s built-in types, see Controlling Zeek from Spicy for the full list of get_<TYPE>() functions.

For Zeek container types (map, record, set, vector), the get_<TYPE>() functions return an opaque value that can be used to then inspect the container further. For example, you can check membership in a set like this:

# Zeek module `MyModule`
option my_set: set[count] = { 1, 2, 3 };
# Spicy
local my_set = zeek::get_set("MyModule::my_set");

if ( zeek::set_contains(my_set, 2) )
   print "Found 2 in the set";

For retrieving values from containers, there are further as_<TYPE> functions for conversion. Example accessing a record’s field:

# Zeek module `MyModule`
option my_record: record {
    a: count &default = 42;
    b: string &default = "foo";
};
# Spicy
local my_record = zeek::get_record("MyModule::my_record");
local a = zeek::as_count(zeek::record_field(my_record, "a"));
local b = zeek::as_string(zeek::record_field(my_record, "b")); # returns "bytes" (not string)

See Controlling Zeek from Spicy again for the full list of as_<TYPE>() and container accessor functions.

The API provides only read access to Zeek variables, there’s no way to change them from inside Spicy. This is both to simplify the API, and also conceptually to avoid offering a back channel into Zeek state that could end up producing a very tight coupling of Spicy and Zeek code. Use events to communicate from Spicy to Zeek instead.

Access to a Zeek global can be relatively slow as it performs the ID lookup and data conversion. Accordingly, the mechanism is primarily meant for configuration-style data, not for transferring heaps of dynamic state. However, as a way to speed up repeated access slightly, you can use zeek::get_value to cache the result of the ID lookup, then use as_<TYPE>() on the cached value to retrieve the actual value.

Note

When accessing global Zeek variables from Spicy, take into account whether their Zeek values might change over time. If you’re accessing a Zeek const, you can be sure it won’t change, so you could just store its value inside a corresponding Spicy global at initialization time, and hence avoid repeated lookups during runtime. However, if you’re accessing a Zeek option, those are designed to support runtime updates, so you should not cache their values on the Spicy side. Likewise, any Zeek global can of course change anytime. (In both cases, you can still use get_value() to cache the ID lookup.)

Dynamic Protocol Detection (DPD)

Spicy protocol analyzers support Zeek’s Dynamic Protocol Detection (DPD), i.e., analysis independent of any well-known ports. To use that with your analyzer, add two pieces:

  1. A Zeek signature to activate your analyzer based on payload patterns. Just like with any of Zeek’s standard analyzers, a signature can activate a Spicy analyzer through the enable "<name>" keyword. The name of the analyzer comes out of the EVT file: it is the ANALYZER_NAME with the double colons replaced with an underscore (e.g., spicy::HTTP turns into enable "spicy_HTTP".

  2. You should call spicy::accept_input() from a hook inside your grammar at a point when the parser can be reasonably certain that it is processing the expected protocol. Optionally, you may also call spicy::decline_input() when you’re sure the parser is not parsing the right protocol. However, Zeek will also trigger this automatically whenever your parser aborts with an error.

Configuration

Options

Zeek provides a set of script-level options to tune Spicy behavior. These all live in the Spicy:: namespace:

    ## Constant for testing if Spicy is available.
    const available = T;

    ## Show output of Spicy print statements.
    const enable_print = F &redef;

    # Record and display profiling information, if compiled into analyzer.
    const enable_profiling = F &redef;

    ## abort() instead of throwing HILTI exceptions.
    const abort_on_exceptions = F &redef;

    ## Include backtraces when reporting unhandled exceptions.
    const show_backtraces = F &redef;

    ## Maximum depth of recursive file analysis (Spicy analyzers only)
    const max_file_depth: count = 5 &redef;

Functions

Zeek also adds the following new built-in functions for Spicy, which likewise live in the Spicy:: namespace:

    ## Enable a specific Spicy protocol analyzer if not already active. If this
    ## analyzer replaces an standard analyzer, that one will automatically be
    ## disabled.
    ##
    ## tag: analyzer to toggle
    ##
    ## Returns: true if the operation succeeded
    global enable_protocol_analyzer: function(tag: Analyzer::Tag) : bool;

    ## Disable a specific Spicy protocol analyzer if not already inactive. If
    ## this analyzer replaces an standard analyzer, that one will automatically
    ## be re-enabled.
    ##
    ## tag: analyzer to toggle
    ##
    ## Returns: true if the operation succeeded
    global disable_protocol_analyzer: function(tag: Analyzer::Tag) : bool;


    ## Enable a specific Spicy file analyzer if not already active. If this
    ## analyzer replaces an standard analyzer, that one will automatically be
    ## disabled.
    ##
    ## tag: analyzer to toggle
    ##
    ## Returns: true if the operation succeeded
    global enable_file_analyzer: function(tag: Files::Tag) : bool;

    ## Disable a specific Spicy file analyzer if not already inactive. If
    ## this analyzer replaces an standard analyzer, that one will automatically
    ## be re-enabled.
    ##
    ## tag: analyzer to toggle
    ##
    ## Returns: true if the operation succeeded
    global disable_file_analyzer: function(tag: Files::Tag) : bool;

    ## Returns current resource usage as reported by the Spicy runtime system.
    global resource_usage: function() : ResourceUsage;

Debugging

If Zeek doesn’t seem to be doing the right thing with your Spicy analyzer, there are several ways to debug what’s going on. To facilitate that, compile your analyzer with spicyz -d and, if possible, use a debug version of Zeek (i.e., build Zeek with ./configure --enable-debug).

If your analyzer doesn’t seem to be active at all, first make sure Zeek actually knows about it: It should show up in the output of zeek -NN Zeek::Spicy. If it doesn’t, you might not have been using the right *.spicy or *.evt files when precompiling, or Zeek is not loading the *.hlto file. Also check your *.evt if it defines your analyzer correctly.

If Zeek knows about your analyzer and just doesn’t seem to activate it, double-check that ports or MIME types are correct in the *.evt file. If you’re using a signature instead, try a port/MIME type first, just to make sure it’s not a matter of signature mismatches.

If there’s nothing obviously wrong with your source files, you can trace what the Zeek’s Spicy support is compiling by running spicyz with -D zeek. For example, reusing the HTTP example from the Getting Started guide:

# spicyz -D zeek my-http.spicy my-http.evt -o my-http.hlt
[debug/zeek] Loading Spicy file "/Users/robin/work/spicy/main/tests/spicy/doc/my-http.spicy"
[debug/zeek] Loading EVT file "/Users/robin/work/spicy/main/doc/examples/my-http.evt"
[debug/zeek] Loading events from /Users/robin/work/spicy/main/doc/examples/my-http.evt
[debug/zeek]   Got protocol analyzer definition for spicy_MyHTTP
[debug/zeek]   Got event definition for MyHTTP::request_line
[debug/zeek] Running Spicy driver
[debug/zeek]   Got unit type 'MyHTTP::Version'
[debug/zeek]   Got unit type 'MyHTTP::RequestLine'
[debug/zeek] Adding protocol analyzer 'spicy_MyHTTP'
[debug/zeek] Adding Spicy hook 'MyHTTP::RequestLine::0x25_done' for event MyHTTP::request_line
[debug/zeek] Done with Spicy driver

You can see the main pieces in there: The files being loaded, unit types provided by them, analyzers and events being created.

If that all looks as expected, it’s time to turn to the Zeek side and see what it’s doing at runtime. You’ll need a debug version of Zeek for that, as well as a small trace with traffic that you expect your analyzer to process. Run Zeek with -B dpd (or -B file_analysis if you’re debugging a file analyzer) on your trace to record the analyzer activity into debug.log. For example, with the same HTTP example, we get:

 1# zeek -B dpd -Cr request-line.pcap my-http.hlto
 2# cat debug.log
 3[dpd] Registering analyzer SPICY_MYHTTP for port 12345/1
 4[...[
 5[dpd] Available analyzers after spicy_init():
 6[...]
 7[dpd]     spicy_MyHTTP (enabled)
 8[...]
 9[dpd] Analyzers by port:
10[dpd]     12345/tcp: SPICY_MYHTTP
11[...]
12[dpd] TCP[5] added child SPICY_MYHTTP[7]
13[dpd] 127.0.0.1:59619 > 127.0.0.1:12345 activated SPICY_MYHTTP analyzer due to port 12345
14[...]
15[dpd] SPICY_MYHTTP[7] DeliverStream(25, T) [GET /index.html HTTP/1.0\x0a]
16[dpd] SPICY_MYHTTP[7] EndOfData(T)
17[dpd] SPICY_MYHTTP[7] EndOfData(F)

The first few lines show that Zeek’s analyzer system registers the analyzer as expected. The subsequent lines show that the analyzer gets activated for processing the connection in the trace, and that it then receives the data that we know indeed constitutes its payload, before it eventually gets shutdown.

To see this from the Zeek side, set the zeek debug stream through the HILTI_DEBUG environment variable:

# HILTI_DEBUG=zeek zeek -Cr request-line.pcap my-http.hlto
[zeek] Have Spicy protocol analyzer spicy_MyHTTP
[zeek] Registering Protocol::TCP protocol analyzer spicy_MyHTTP with Zeek
[zeek]   Scheduling analyzer for port 12345/tcp
[zeek] Done with post-script initialization
[zeek] [SPICY_MYHTTP/7/orig] initial chunk: |GET /index.html HTTP/1.0\\x0a| (eod=false)
[zeek] [SPICY_MYHTTP/7/orig] -> event MyHTTP::request_line($conn, GET, /index.html, 1.0)
[zeek] [SPICY_MYHTTP/7/orig] done with parsing
[zeek] [SPICY_MYHTTP/7/orig] parsing finished, skipping further originator payload
[zeek] [SPICY_MYHTTP/7/resp] no unit specified for parsing
[zeek] [SPICY_MYHTTP/7/orig] skipping end-of-data delivery
[zeek] [SPICY_MYHTTP/7/resp] no unit specified for parsing
[zeek] [SPICY_MYHTTP/7/orig] skipping end-of-data delivery
[zeek] [SPICY_MYHTTP/7/resp] no unit specified for parsing

After the initial initialization, you see the data arriving and the event being generated for Zeek. Zeek also reports that we didn’t define a unit for the responder side—which we know in this case, but if that appears unexpectedly you probably found a problem.

So we know now that our analyzer is receiving the anticipated data to parse. At this point, we can switch to debugging the Spicy side through the usual mechanisms. In particular, setting HILTI_DEBUG=spicy tends to be helpful:

# HILTI_DEBUG=spicy zeek -Cr request-line.pcap my-http.hlto
[spicy] MyHTTP::RequestLine
[spicy]   method = GET
[spicy]   anon_2 =
[spicy]   uri = /index.html
[spicy]   anon_3 =
[spicy]   MyHTTP::Version
[spicy]     anon = HTTP/
[spicy]     number = 1.0
[spicy]   version = [$number=b"1.0"]
[spicy]   anon_4 = \n

If everything looks right with the parsing, and the right events are generated too, then the final part is to check out the events that arrive on the Zeek side. To get Zeek to see an event that Zeek raises, you need to have at least one handler implemented for it in one of your Zeek scripts. You can then load Zeek’s misc/dump-events to see them as they are being received, including their full Zeek-side values:

# zeek -Cr request-line.pcap my-http.hlto misc/dump-events
[...]
1580991211.780489 MyHTTP::request_line
              [0] c: connection      = [id=[orig_h=127.0.0.1, orig_p=59619/tcp, ...] ...]
              [1] method: string     = GET
              [2] uri: string        = /index.html
              [3] version: string    = 1.0
[...]