Core Rule Set: the package registries' searches of gitea and Forgejo pass (closes #354)

Recorded from gitea 1.27.3 and Forgejo 17.0.1 with each registry's own
client, and curl for the requests no client sends, with python3,
powershell, ls -la, .htaccess, file names and plain words. q on the
Cargo, Chef, Composer and Conan searches and NuGet's search service
joins 900603. 900605 lets /bin/sh through on Composer and Conan. New:
900609 for npm's text; 900611, in phase 2, lets one q shaped as a Conan
reference with no +, */ or --, such as zlib/*, skip 942100 if 942100
passes it with each /* comment read as a space; 900612 for NuGet's
Search(), Packages() and their counts: searchTerm, $filter, and the
names $filter, $orderby and $top for 942290.

Model: opus-5-5

SPEC.md

@@ -940,7 +940,9 @@ one entry of `apps` in `internal/coreruleset/coreruleset.go`. gitea's fixes are:
           version, then optionally `@` and a user, then optionally `/` and a
           channel. Each part is made of letters, digits, `_`, `.`, `-` and `*`.
           A part holds no `+`, since 942100 reads `+` as a space. `q` also holds
-          no `*/`, which ends an SQL comment, and no `--`, which starts one.
+          no `*/`, which ends an SQL comment, and no `--`, which starts one. And
+          942100 finds no SQL injection in `q` once each comment that `/*`
+          starts, an unclosed one included, is replaced by a space.
 
     - The path of these requests, which names a file:
         - a file's page;

internal/coreruleset/gitea.conf

@@ -274,13 +274,16 @@ SecRule REQUEST_FILENAME "@rx ^/api/packages/[^/]+/npm/-/v1/search$" \
 # and a user, then optionally / and a channel. Each part is made of letters,
 # digits, _, ., - and *. A part holds no +, since 942100 reads + as a space.
 # q must also hold no */, which ends an SQL comment, and no --, which starts
-# one. Such a q has no quote, no space and no comment between two words, so
-# SQL injection such as these stays refused:
+# one. Last, 942100 must find no SQL injection in q once each comment that /*
+# starts, an unclosed one included, is replaced by a space. So zlib/* passes,
+# and SQL injection such as these stays refused:
 #   1' OR '1'='1
 #   1/**/or/**/1
 #   1/*@*/or-1
 #   1+or+1--
 #   1-or-1--
+#   x/1.or.1
+#   1.having.1
 # This rule is in phase 2, because a phase 1 rule runs before the body is read
 # and cannot see a q there. It checks tx.app_gitea, since the skip above does
 # not reach phase 2.
@@ -290,6 +293,7 @@ SecRule REQUEST_FILENAME "@rx ^/api/packages/[^/]+/conan/v[12]/conans/search$" \
     SecRule &ARGS:q "@eq 1" "chain"
     SecRule ARGS:q "!@contains */" "chain"
     SecRule ARGS:q "!@contains --" "chain"
+    SecRule ARGS:q "!@detectSQLi" "t:none,t:replaceComments,chain"
     SecRule ARGS:q "@rx ^[\w.*-]+(?:/[\w.*-]+)?(?:@[\w.*-]+(?:/[\w.*-]+)?)?$" \
         "ctl:ruleRemoveTargetById=942100;ARGS:q"
 

internal/coreruleset/gitea_test.go

@@ -441,8 +441,8 @@ func TestGiteaRegistrySearchExclusionsAreNarrow(t *testing.T) {
 	// another operator on NuGet's searches, and $filter on NuGet's other
 	// paths. The rule for SQL injection (942100) still refuses tools/* in a
 	// field other than a Conan search's, SQL injection in a second q after a
-	// Conan pattern, SQL comments that stand in for spaces, and SQL injection
-	// holding + or --.
+	// Conan pattern, SQL comments or dots that stand in for spaces, and SQL
+	// injection holding + or --.
 	for _, target := range []string{
 		"/api/packages/alice/nuget/Search()?$where=1",
 		"/api/packages/alice/nuget/Packages()?$ne=1",
@@ -457,6 +457,9 @@ func TestGiteaRegistrySearchExclusionsAreNarrow(t *testing.T) {
 		"/api/packages/alice/conan/v2/conans/search?q=1%2F%2A%40%2A%2For-1",
 		"/api/packages/alice/conan/v2/conans/search?q=1%2Bor%2B1--",
 		"/api/packages/alice/conan/v2/conans/search?q=1-or-1--",
+		"/api/packages/alice/conan/v2/conans/search?q=x%2F1.or.1",
+		"/api/packages/alice/conan/v2/conans/search?q=x%2F1.and.1",
+		"/api/packages/alice/conan/v2/conans/search?q=1.having.1",
 	} {
 		if got := inspect(t, crs, get(target)); got.Score < threshold {
 			t.Errorf("%s let through: %+v", target, got)
@@ -476,6 +479,14 @@ func TestGiteaRegistrySearchExclusionsAreNarrow(t *testing.T) {
 		t.Errorf("POST %s %s let through: %+v", search.target, body, got)
 	}
 
+	// So is SQL injection with dots for spaces, in a q sent only in the body.
+	search.target = "/api/packages/alice/conan/v2/conans/search"
+
+	body = "q=x%2F1.or.1"
+	if got, _ := inspectBody(t, readingBodies(t), search, body); got.Score < threshold {
+		t.Errorf("POST %s %s let through: %+v", search.target, body, got)
+	}
+
 	// The rule that lets a Conan pattern through is in phase 2, where it
 	// checks tx.app_gitea itself: a hostname that does not serve gitea
 	// refuses the pattern.