Gitea setup: Core Rule Set blocks attacks, not ordinary Gitea traffic (closes #198)

The Gitea setup runs the Core Rule Set in block mode. Rules that refused
ordinary Gitea requests are switched off only where they misfired:

- deploy/gitea/crs-exclusions.conf, loaded through the new setting
  SWWAF_CORE_RULE_SET_EXCLUSIONS_FILE, covers Gitea's search, file-name
  and branch-name fields, Referer, and file paths;
- smallwebwaf's defaults gain a seventh change: the registry upload
  path's from parameter skips 932260.

A file that cannot be read, does not parse or names an id from 900000
to 900999 stops servewaf at start. internal/waf's tests and
script/gitea-setup send the recorded requests in
internal/waf/testdata/gitea-requests.txt and three attacks.

Model: opus-5-5

internal/waf/waf.go

@@ -1,8 +1,9 @@
 // Package waf runs the OWASP Core Rule Set 4.25.0, through Coraza, on the
 // method, the URL with its query and the headers of a request, and on its
-// body while SWWAF_WAF_BODY_LIMIT is set, with the six changes smallwebwaf
-// makes to it, as "Attack detection" under "Settings" in SPEC.md describes
-// them. It reads no response.
+// body while SWWAF_WAF_BODY_LIMIT is set, with the seven changes smallwebwaf
+// makes to it and the exclusions in SWWAF_CORE_RULE_SET_EXCLUSIONS_FILE, as
+// "Attack detection" under "Settings" in SPEC.md describes them. It reads no
+// response.
 //
 // smallwebwaf writes only to its state directory, so Coraza is built with
 // its no_fs_access tag, as the Dockerfile and script/build build it: of a
@@ -11,10 +12,13 @@
 package waf
 
 import (
+	"errors"
 	"fmt"
 	"io"
 	"net/http"
 	"net/netip"
+	"os"
+	"regexp"
 	"slices"
 	"strconv"
 	"strings"
@@ -29,9 +33,9 @@ import (
 // paranoia level for %d, and bodyDirectives for %s while
 // SWWAF_WAF_BODY_LIMIT is set. Each rule smallwebwaf adds has an id from
 // 900000 to 900999, the ids the Core Rule Set keeps for the rules that set
-// it up, which SWWAF_WAF_DISABLED_RULES refuses, so that no setting
-// switches one off. Coraza joins a line ending in \ to the next, without
-// the spaces at the start of the next.
+// it up, which SWWAF_WAF_DISABLED_RULES and the exclusions file refuse, so
+// that no setting switches one off. Coraza joins a line ending in \ to the
+// next, without the spaces at the start of the next.
 const directives = `
 # The engine only detects. smallwebwaf compares the request's anomaly
 # score with SWWAF_WAF_ANOMALY_THRESHOLD itself, in block and detect mode
@@ -100,6 +104,13 @@ SecRuleUpdateTargetById 932260 "!ARGS:path|!ARGS:files|!ARGS:skip-to|\
 # Inspect.
 SecRuleUpdateTargetById 932340 "!REQUEST_HEADERS:Referer"
 SecRuleUpdateTargetById 944110 "!REQUEST_HEADERS:Referer"
+
+# The seventh: a container image push that reuses a layer names the image
+# it takes the layer from in the query parameter from, which is not
+# checked for command names. 932260 refused the push of a layer from
+# owner/python3-app.
+SecRule REQUEST_FILENAME "@rx ^/v2/.+/blobs/uploads/$" \
+    "id:900500,phase:1,pass,nolog,ctl:ruleRemoveTargetById=932260;ARGS:from"
 `
 
 // bodyDirectives have the Core Rule Set read the part of a request body
@@ -160,8 +171,20 @@ type Params struct {
 	// BodyLimit is the most of a request body the Core Rule Set reads
 	// (SWWAF_WAF_BODY_LIMIT), 0 while it is off and it reads none.
 	BodyLimit int64
+	// ExclusionsFile is the path of a file of exclusions loaded after the
+	// rest (SWWAF_CORE_RULE_SET_EXCLUSIONS_FILE), "" while there is none.
+	ExclusionsFile string
 }
 
+// smallwebwafRuleID matches an id from 900000 to 900999, which an
+// exclusions file may not name: those are the ids of smallwebwaf's own
+// rules.
+var smallwebwafRuleID = regexp.MustCompile(`\b900\d{3}\b`)
+
+// errSmallwebwafRuleID is the error of an exclusions file that names an
+// id of smallwebwaf's own rules.
+var errSmallwebwafRuleID = errors.New("an id of smallwebwaf's own rules")
+
 // CoreRuleSet is the Core Rule Set, ready to inspect requests. It is safe
 // for concurrent use.
 type CoreRuleSet struct {
@@ -169,9 +192,11 @@ type CoreRuleSet struct {
 	bodyLimit int64
 }
 
-// New returns the Core Rule Set with the six changes, at params'
-// paranoia level, without the rules it switches off, and reading request
-// bodies up to params' limit.
+// New returns the Core Rule Set with the seven changes, at params'
+// paranoia level, without the rules it switches off, reading request
+// bodies up to params' limit, and with params' exclusions file loaded
+// last. An exclusions file that cannot be read, does not parse or names
+// an id of smallwebwaf's own rules is an error.
 func New(params Params) (*CoreRuleSet, error) {
 	body := ""
 	if params.BodyLimit > 0 {
@@ -189,9 +214,23 @@ func New(params Params) (*CoreRuleSet, error) {
 		text += "SecRuleRemoveById " + strings.Join(ids, " ") + "\n"
 	}
 
-	waf, err := coraza.NewWAF(coraza.NewWAFConfig().
-		WithRootFS(coreruleset.FS).
-		WithDirectives(text))
+	config := coraza.NewWAFConfig().WithRootFS(coreruleset.FS).WithDirectives(text)
+
+	if params.ExclusionsFile != "" {
+		exclusions, err := os.ReadFile(params.ExclusionsFile)
+		if err != nil {
+			return nil, fmt.Errorf("read the exclusions file: %w", err)
+		}
+
+		if id := smallwebwafRuleID.Find(exclusions); id != nil {
+			return nil, fmt.Errorf("the exclusions file names %s: %w", id,
+				errSmallwebwafRuleID)
+		}
+
+		config = config.WithDirectives(string(exclusions))
+	}
+
+	waf, err := coraza.NewWAF(config)
 	if err != nil {
 		return nil, fmt.Errorf("load the Core Rule Set: %w", err)
 	}