Refact pkg/parser/node (#2953)

* extract method processFilter()

* extract method processWhitelist()

* lint (whitespace, errors)
This commit is contained in:
mmetc 2024-04-25 17:53:10 +02:00 committed by GitHub
parent d2c4bc55fc
commit c4473839c4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 156 additions and 84 deletions

View file

@ -3,7 +3,7 @@
linters-settings:
cyclop:
# lower this after refactoring
max-complexity: 53
max-complexity: 48
gci:
sections:
@ -22,7 +22,7 @@ linters-settings:
gocyclo:
# lower this after refactoring
min-complexity: 49
min-complexity: 48
funlen:
# Checks the number of lines in a function.

View file

@ -22,69 +22,70 @@ import (
type Node struct {
FormatVersion string `yaml:"format"`
//Enable config + runtime debug of node via config o/
// Enable config + runtime debug of node via config o/
Debug bool `yaml:"debug,omitempty"`
//If enabled, the node (and its child) will report their own statistics
// If enabled, the node (and its child) will report their own statistics
Profiling bool `yaml:"profiling,omitempty"`
//Name, author, description and reference(s) for parser pattern
// Name, author, description and reference(s) for parser pattern
Name string `yaml:"name,omitempty"`
Author string `yaml:"author,omitempty"`
Description string `yaml:"description,omitempty"`
References []string `yaml:"references,omitempty"`
//if debug is present in the node, keep its specific Logger in runtime structure
// if debug is present in the node, keep its specific Logger in runtime structure
Logger *log.Entry `yaml:"-"`
//This is mostly a hack to make writing less repetitive.
//relying on stage, we know which field to parse, and we
//can also promote log to next stage on success
// This is mostly a hack to make writing less repetitive.
// relying on stage, we know which field to parse, and we
// can also promote log to next stage on success
Stage string `yaml:"stage,omitempty"`
//OnSuccess allows to tag a node to be able to move log to next stage on success
// OnSuccess allows to tag a node to be able to move log to next stage on success
OnSuccess string `yaml:"onsuccess,omitempty"`
rn string //this is only for us in debug, a random generated name for each node
//Filter is executed at runtime (with current log line as context)
//and must succeed or node is exited
rn string // this is only for us in debug, a random generated name for each node
// Filter is executed at runtime (with current log line as context)
// and must succeed or node is exited
Filter string `yaml:"filter,omitempty"`
RunTimeFilter *vm.Program `yaml:"-" json:"-"` //the actual compiled filter
//If node has leafs, execute all of them until one asks for a 'break'
RunTimeFilter *vm.Program `yaml:"-" json:"-"` // the actual compiled filter
// If node has leafs, execute all of them until one asks for a 'break'
LeavesNodes []Node `yaml:"nodes,omitempty"`
//Flag used to describe when to 'break' or return an 'error'
// Flag used to describe when to 'break' or return an 'error'
EnrichFunctions EnricherCtx
/* If the node is actually a leaf, it can have : grok, enrich, statics */
//pattern_syntax are named grok patterns that are re-utilized over several grok patterns
// pattern_syntax are named grok patterns that are re-utilized over several grok patterns
SubGroks yaml.MapSlice `yaml:"pattern_syntax,omitempty"`
//Holds a grok pattern
// Holds a grok pattern
Grok GrokPattern `yaml:"grok,omitempty"`
//Statics can be present in any type of node and is executed last
// Statics can be present in any type of node and is executed last
Statics []ExtraField `yaml:"statics,omitempty"`
//Stash allows to capture data from the log line and store it in an accessible cache
// Stash allows to capture data from the log line and store it in an accessible cache
Stash []DataCapture `yaml:"stash,omitempty"`
//Whitelists
// Whitelists
Whitelist Whitelist `yaml:"whitelist,omitempty"`
Data []*types.DataSource `yaml:"data,omitempty"`
}
func (n *Node) validate(pctx *UnixParserCtx, ectx EnricherCtx) error {
//stage is being set automagically
// stage is being set automagically
if n.Stage == "" {
return fmt.Errorf("stage needs to be an existing stage")
return errors.New("stage needs to be an existing stage")
}
/* "" behaves like continue */
if n.OnSuccess != "continue" && n.OnSuccess != "next_stage" && n.OnSuccess != "" {
return fmt.Errorf("onsuccess '%s' not continue,next_stage", n.OnSuccess)
}
if n.Filter != "" && n.RunTimeFilter == nil {
return fmt.Errorf("non-empty filter '%s' was not compiled", n.Filter)
}
if n.Grok.RunTimeRegexp != nil || n.Grok.TargetField != "" {
if n.Grok.TargetField == "" && n.Grok.ExpValue == "" {
return fmt.Errorf("grok requires 'expression' or 'apply_on'")
return errors.New("grok requires 'expression' or 'apply_on'")
}
if n.Grok.RegexpName == "" && n.Grok.RegexpValue == "" {
return fmt.Errorf("grok needs 'pattern' or 'name'")
return errors.New("grok needs 'pattern' or 'name'")
}
}
@ -93,6 +94,7 @@ func (n *Node) validate(pctx *UnixParserCtx, ectx EnricherCtx) error {
if static.ExpValue == "" {
return fmt.Errorf("static %d : when method is set, expression must be present", idx)
}
if _, ok := ectx.Registered[static.Method]; !ok {
log.Warningf("the method '%s' doesn't exist or the plugin has not been initialized", static.Method)
}
@ -100,6 +102,7 @@ func (n *Node) validate(pctx *UnixParserCtx, ectx EnricherCtx) error {
if static.Meta == "" && static.Parsed == "" && static.TargetByName == "" {
return fmt.Errorf("static %d : at least one of meta/event/target must be set", idx)
}
if static.Value == "" && static.RunTimeValue == nil {
return fmt.Errorf("static %d value or expression must be set", idx)
}
@ -110,40 +113,44 @@ func (n *Node) validate(pctx *UnixParserCtx, ectx EnricherCtx) error {
if stash.Name == "" {
return fmt.Errorf("stash %d : name must be set", idx)
}
if stash.Value == "" {
return fmt.Errorf("stash %s : value expression must be set", stash.Name)
}
if stash.Key == "" {
return fmt.Errorf("stash %s : key expression must be set", stash.Name)
}
if stash.TTL == "" {
return fmt.Errorf("stash %s : ttl must be set", stash.Name)
}
if stash.Strategy == "" {
stash.Strategy = "LRU"
}
//should be configurable
// should be configurable
if stash.MaxMapSize == 0 {
stash.MaxMapSize = 100
}
}
return nil
}
func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[string]interface{}) (bool, error) {
var NodeState bool
var NodeHasOKGrok bool
func (n *Node) processFilter(cachedExprEnv map[string]interface{}) (bool, error) {
clog := n.Logger
if n.RunTimeFilter == nil {
clog.Tracef("Node has not filter, enter")
return true, nil
}
cachedExprEnv := expressionEnv
clog.Tracef("Event entering node")
if n.RunTimeFilter != nil {
//Evaluate node's filter
// Evaluate node's filter
output, err := exprhelpers.Run(n.RunTimeFilter, cachedExprEnv, clog, n.Debug)
if err != nil {
clog.Warningf("failed to run filter : %v", err)
clog.Debugf("Event leaving node : ko")
return false, nil
}
@ -156,26 +163,26 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
default:
clog.Warningf("Expr '%s' returned non-bool, abort : %T", n.Filter, output)
clog.Debugf("Event leaving node : ko")
return false, nil
}
NodeState = true
} else {
clog.Tracef("Node has not filter, enter")
NodeState = true
}
if n.Name != "" {
NodesHits.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name}).Inc()
}
exprErr := error(nil)
return true, nil
}
func (n *Node) processWhitelist(cachedExprEnv map[string]interface{}, p *types.Event) (bool, error) {
var exprErr error
isWhitelisted := n.CheckIPsWL(p)
if !isWhitelisted {
isWhitelisted, exprErr = n.CheckExprWL(cachedExprEnv, p)
}
if exprErr != nil {
// Previous code returned nil if there was an error, so we keep this behavior
return false, nil //nolint:nilerr
}
if isWhitelisted && !p.Whitelisted {
p.Whitelisted = true
p.WhitelistReason = n.Whitelist.Reason
@ -185,18 +192,51 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
for k := range p.Overflow.Sources {
ips = append(ips, k)
}
clog.Infof("Ban for %s whitelisted, reason [%s]", strings.Join(ips, ","), n.Whitelist.Reason)
n.Logger.Infof("Ban for %s whitelisted, reason [%s]", strings.Join(ips, ","), n.Whitelist.Reason)
p.Overflow.Whitelisted = true
}
}
//Process grok if present, should be exclusive with nodes :)
return isWhitelisted, nil
}
func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[string]interface{}) (bool, error) {
var NodeHasOKGrok bool
clog := n.Logger
cachedExprEnv := expressionEnv
clog.Tracef("Event entering node")
NodeState, err := n.processFilter(cachedExprEnv)
if err != nil {
return false, err
}
if !NodeState {
return false, nil
}
if n.Name != "" {
NodesHits.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name}).Inc()
}
isWhitelisted, err := n.processWhitelist(cachedExprEnv, p)
if err != nil {
return false, err
}
// Process grok if present, should be exclusive with nodes :)
gstr := ""
if n.Grok.RunTimeRegexp != nil {
clog.Tracef("Processing grok pattern : %s : %p", n.Grok.RegexpName, n.Grok.RunTimeRegexp)
//for unparsed, parsed etc. set sensible defaults to reduce user hassle
// for unparsed, parsed etc. set sensible defaults to reduce user hassle
if n.Grok.TargetField != "" {
//it's a hack to avoid using real reflect
// it's a hack to avoid using real reflect
if n.Grok.TargetField == "Line.Raw" {
gstr = p.Line.Raw
} else if val, ok := p.Parsed[n.Grok.TargetField]; ok {
@ -211,6 +251,7 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
clog.Warningf("failed to run RunTimeValue : %v", err)
NodeState = false
}
switch out := output.(type) {
case string:
gstr = out
@ -229,12 +270,14 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
} else {
groklabel = n.Grok.RegexpName
}
grok := n.Grok.RunTimeRegexp.Parse(gstr)
if len(grok) > 0 {
/*tag explicitly that the *current* node had a successful grok pattern. it's important to know success state*/
NodeHasOKGrok = true
clog.Debugf("+ Grok '%s' returned %d entries to merge in Parsed", groklabel, len(grok))
//We managed to grok stuff, merged into parse
// We managed to grok stuff, merged into parse
for k, v := range grok {
clog.Debugf("\t.Parsed['%s'] = '%s'", k, v)
p.Parsed[k] = v
@ -246,34 +289,37 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
return false, err
}
} else {
//grok failed, node failed
// grok failed, node failed
clog.Debugf("+ Grok '%s' didn't return data on '%s'", groklabel, gstr)
NodeState = false
}
} else {
clog.Tracef("! No grok pattern : %p", n.Grok.RunTimeRegexp)
}
//Process the stash (data collection) if : a grok was present and succeeded, or if there is no grok
// Process the stash (data collection) if : a grok was present and succeeded, or if there is no grok
if NodeHasOKGrok || n.Grok.RunTimeRegexp == nil {
for idx, stash := range n.Stash {
var value string
var key string
var (
key string
value string
)
if stash.ValueExpression == nil {
clog.Warningf("Stash %d has no value expression, skipping", idx)
continue
}
if stash.KeyExpression == nil {
clog.Warningf("Stash %d has no key expression, skipping", idx)
continue
}
//collect the data
// collect the data
output, err := exprhelpers.Run(stash.ValueExpression, cachedExprEnv, clog, n.Debug)
if err != nil {
clog.Warningf("Error while running stash val expression : %v", err)
}
//can we expect anything else than a string ?
// can we expect anything else than a string ?
switch output := output.(type) {
case string:
value = output
@ -282,12 +328,12 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
continue
}
//collect the key
// collect the key
output, err = exprhelpers.Run(stash.KeyExpression, cachedExprEnv, clog, n.Debug)
if err != nil {
clog.Warningf("Error while running stash key expression : %v", err)
}
//can we expect anything else than a string ?
// can we expect anything else than a string ?
switch output := output.(type) {
case string:
key = output
@ -299,7 +345,7 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
}
}
//Iterate on leafs
// Iterate on leafs
for _, leaf := range n.LeavesNodes {
ret, err := leaf.process(p, ctx, cachedExprEnv)
if err != nil {
@ -307,7 +353,9 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
clog.Debugf("Event leaving node : ko")
return false, err
}
clog.Tracef("\tsub-node (%s) ret : %v (strategy:%s)", leaf.rn, ret, n.OnSuccess)
if ret {
NodeState = true
/* if child is successful, stop processing */
@ -328,12 +376,14 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
clog.Tracef("State after nodes : %v", NodeState)
//grok or leafs failed, don't process statics
// grok or leafs failed, don't process statics
if !NodeState {
if n.Name != "" {
NodesHitsKo.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name}).Inc()
}
clog.Debugf("Event leaving node : ko")
return NodeState, nil
}
@ -360,9 +410,10 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
if NodeState {
clog.Debugf("Event leaving node : ok")
log.Tracef("node is successful, check strategy")
if n.OnSuccess == "next_stage" {
idx := stageidx(p.Stage, ctx.Stages)
//we're at the last stage
// we're at the last stage
if idx+1 == len(ctx.Stages) {
clog.Debugf("node reached the last stage : %s", p.Stage)
} else {
@ -375,15 +426,16 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
} else {
clog.Debugf("Event leaving node : ko")
}
clog.Tracef("Node successful, continue")
return NodeState, nil
}
func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
var err error
var valid bool
valid = false
valid := false
dumpr := spew.ConfigState{MaxDepth: 1, DisablePointerAddresses: true}
n.rn = seed.Generate()
@ -393,10 +445,11 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
/* if the node has debugging enabled, create a specific logger with debug
that will be used only for processing this node ;) */
if n.Debug {
var clog = log.New()
clog := log.New()
if err = types.ConfigureLogger(clog); err != nil {
log.Fatalf("While creating bucket-specific logger : %s", err)
}
clog.SetLevel(log.DebugLevel)
n.Logger = clog.WithFields(log.Fields{
"id": n.rn,
@ -414,7 +467,7 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
n.Logger.Tracef("Compiling : %s", dumpr.Sdump(n))
//compile filter if present
// compile filter if present
if n.Filter != "" {
n.RunTimeFilter, err = expr.Compile(n.Filter, exprhelpers.GetExprOptions(map[string]interface{}{"evt": &types.Event{}})...)
if err != nil {
@ -425,12 +478,15 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
/* handle pattern_syntax and groks */
for _, pattern := range n.SubGroks {
n.Logger.Tracef("Adding subpattern '%s' : '%s'", pattern.Key, pattern.Value)
if err = pctx.Grok.Add(pattern.Key.(string), pattern.Value.(string)); err != nil {
if errors.Is(err, grokky.ErrAlreadyExist) {
n.Logger.Warningf("grok '%s' already registred", pattern.Key)
continue
}
n.Logger.Errorf("Unable to compile subpattern %s : %v", pattern.Key, err)
return err
}
}
@ -438,28 +494,36 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
/* load grok by name or compile in-place */
if n.Grok.RegexpName != "" {
n.Logger.Tracef("+ Regexp Compilation '%s'", n.Grok.RegexpName)
n.Grok.RunTimeRegexp, err = pctx.Grok.Get(n.Grok.RegexpName)
if err != nil {
return fmt.Errorf("unable to find grok '%s' : %v", n.Grok.RegexpName, err)
}
if n.Grok.RunTimeRegexp == nil {
return fmt.Errorf("empty grok '%s'", n.Grok.RegexpName)
}
n.Logger.Tracef("%s regexp: %s", n.Grok.RegexpName, n.Grok.RunTimeRegexp.String())
valid = true
} else if n.Grok.RegexpValue != "" {
if strings.HasSuffix(n.Grok.RegexpValue, "\n") {
n.Logger.Debugf("Beware, pattern ends with \\n : '%s'", n.Grok.RegexpValue)
}
n.Grok.RunTimeRegexp, err = pctx.Grok.Compile(n.Grok.RegexpValue)
if err != nil {
return fmt.Errorf("failed to compile grok '%s': %v", n.Grok.RegexpValue, err)
}
if n.Grok.RunTimeRegexp == nil {
// We shouldn't be here because compilation succeeded, so regexp shouldn't be nil
return fmt.Errorf("grok compilation failure: %s", n.Grok.RegexpValue)
}
n.Logger.Tracef("%s regexp : %s", n.Grok.RegexpValue, n.Grok.RunTimeRegexp.String())
valid = true
}
@ -473,7 +537,7 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
}
/* load grok statics */
//compile expr statics if present
// compile expr statics if present
for idx := range n.Grok.Statics {
if n.Grok.Statics[idx].ExpValue != "" {
n.Grok.Statics[idx].RunTimeValue, err = expr.Compile(n.Grok.Statics[idx].ExpValue,
@ -482,6 +546,7 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
return err
}
}
valid = true
}
@ -505,7 +570,7 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
}
logLvl := n.Logger.Logger.GetLevel()
//init the cache, does it make sense to create it here just to be sure everything is fine ?
// init the cache, does it make sense to create it here just to be sure everything is fine ?
if err = cache.CacheInit(cache.CacheCfg{
Size: n.Stash[i].MaxMapSize,
TTL: n.Stash[i].TTLVal,
@ -526,14 +591,18 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
if !n.LeavesNodes[idx].Debug && n.Debug {
n.LeavesNodes[idx].Debug = true
}
if !n.LeavesNodes[idx].Profiling && n.Profiling {
n.LeavesNodes[idx].Profiling = true
}
n.LeavesNodes[idx].Stage = n.Stage
err = n.LeavesNodes[idx].compile(pctx, ectx)
if err != nil {
return err
}
valid = true
}
@ -546,6 +615,7 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
return err
}
}
valid = true
}
@ -554,13 +624,15 @@ func (n *Node) compile(pctx *UnixParserCtx, ectx EnricherCtx) error {
if err != nil {
return err
}
valid = valid || whitelistValid
if !valid {
/* node is empty, error force return */
n.Logger.Error("Node is empty or invalid, abort")
n.Stage = ""
return fmt.Errorf("Node is empty")
return errors.New("Node is empty")
}
if err := n.validate(pctx, ectx); err != nil {