From 8c8ccf6a2e7dbd746b73c0e564eb3fafc432e58c Mon Sep 17 00:00:00 2001
From: "Sascha L. Teichmann" <sascha.teichmann@intevation.de>
Date: Sun, 22 Jan 2023 20:55:26 +0100
Subject: [PATCH] Extract more than one string from expr: category fields.

---
 cmd/csaf_aggregator/mirror.go |  9 ++++---
 cmd/csaf_provider/actions.go  |  5 ++--
 docs/csaf_provider.md         |  6 +++++
 util/json.go                  | 50 +++++++++++++++++++++++++++++++++++
 4 files changed, 64 insertions(+), 6 deletions(-)

diff --git a/cmd/csaf_aggregator/mirror.go b/cmd/csaf_aggregator/mirror.go
index 7fe71d7..3ea4837 100644
--- a/cmd/csaf_aggregator/mirror.go
+++ b/cmd/csaf_aggregator/mirror.go
@@ -451,18 +451,19 @@ func (w *worker) extractCategories(label string, advisory any) error {
 		w.categories[label] = cats
 	}
 
-	var result string
-	matcher := util.StringMatcher(&result)
-
 	const exprPrefix = "expr:"
 
 	for _, cat := range categories {
 		if strings.HasPrefix(cat, exprPrefix) {
 			expr := cat[len(exprPrefix):]
+			var results []string
+			matcher := util.StringTreeMatcher(&results)
 			if err := w.expr.Extract(expr, matcher, true, advisory); err != nil {
 				return err
 			}
-			cats[result] = true
+			for _, result := range results {
+				cats[result] = true
+			}
 		} else { // Normal
 			cats[cat] = true
 		}
diff --git a/cmd/csaf_provider/actions.go b/cmd/csaf_provider/actions.go
index ea099e3..b1ad0a9 100644
--- a/cmd/csaf_provider/actions.go
+++ b/cmd/csaf_provider/actions.go
@@ -195,9 +195,10 @@ func (c *controller) upload(r *http.Request) (any, error) {
 	var dynamicCategories []string
 	if catExprs := c.cfg.DynamicCategories(); len(catExprs) > 0 {
 		var err error
-		if dynamicCategories, err = pe.Strings(catExprs, true, content); err != nil {
+		if dynamicCategories, err = pe.StringsFromTree(
+			catExprs, true, content); err != nil {
 			// XXX: Should we die here?
-			log.Printf("eval of dynamic catecory expressions failed: %v\n", err)
+			log.Printf("eval of dynamic catergory expressions failed: %v\n", err)
 		}
 	}
 
diff --git a/docs/csaf_provider.md b/docs/csaf_provider.md
index 81a45fa..464f0db 100644
--- a/docs/csaf_provider.md
+++ b/docs/csaf_provider.md
@@ -106,6 +106,12 @@ The following example file documents all available configuration options:
 # If a list item starts with `expr:`
 #   the rest of the string is used as a JsonPath expression
 #   to extract a string from the incoming advisories.
+#   If the result of the expression is a string this string
+#   is used. If the result is an array each element of
+#   this array is tested if it is a string or an array.
+#   If this test fails the expression fails. If the
+#   test succeeds the rules are applied recursively to
+#   collect all strings in the result.
 # Strings not starting with `expr:` are taken verbatim.
 # By default no category documents are created.
 # This example provides an overview over the syntax,
diff --git a/util/json.go b/util/json.go
index ed0fd70..48d0e09 100644
--- a/util/json.go
+++ b/util/json.go
@@ -101,6 +101,37 @@ func StringMatcher(dst *string) func(any) error {
 	}
 }
 
+// StringTreeMatcher returns a matcher which add strings from
+// stringss and recursively from arrays from strings.
+func StringTreeMatcher(strings *[]string) func(any) error {
+	// Only add unique strings.
+	unique := func(s string) {
+		for _, t := range *strings {
+			if s == t {
+				return
+			}
+		}
+		*strings = append(*strings, s)
+	}
+	var recurse func(any) error
+	recurse = func(x any) error {
+		switch y := x.(type) {
+		case string:
+			unique(y)
+		case []any:
+			for _, z := range y {
+				if err := recurse(z); err != nil {
+					return err
+				}
+			}
+		default:
+			return fmt.Errorf("unsupported type: %T", x)
+		}
+		return nil
+	}
+	return recurse
+}
+
 // TimeMatcher stores a time with a given format.
 func TimeMatcher(dst *time.Time, format string) func(any) error {
 	return func(x any) error {
@@ -147,6 +178,25 @@ func (pe *PathEval) Match(matcher []PathEvalMatcher, doc any) error {
 	return nil
 }
 
+// StringsFromTree returns strings from the given exprs.
+//  1. If a expression results a string this string is used.
+//  2. if a expression results in an array the elements
+//     of this array are recursively treated with 1. and 2.
+func (pe *PathEval) StringsFromTree(
+	exprs []string,
+	optional bool,
+	doc any,
+) ([]string, error) {
+	results := make([]string, 0, len(exprs))
+	matcher := StringTreeMatcher(&results)
+	for _, expr := range exprs {
+		if err := pe.Extract(expr, matcher, optional, doc); err != nil {
+			return nil, err
+		}
+	}
+	return results, nil
+}
+
 // Strings searches the given document for the given set of expressions
 // and returns the corresponding strings. The optional flag indicates
 // if the expression evaluation have to succseed or not.