new(libsinsp): add len() filter transformer

Signed-off-by: Luca Guerra <luca@guerra.sh>

userspace/libsinsp/filter/parser.cpp

@@ -90,12 +90,11 @@ static const std::vector<std::string> s_binary_list_ops = {
 
 static constexpr const char* s_field_transformer_val = "val(";
 
-static const std::vector<std::string> s_field_transformers = {
-        "tolower(",
-        "toupper(",
-        "b64(",
-        "basename(",
-};
+static const std::vector<std::string> s_field_transformers = {"tolower(",
+                                                              "toupper(",
+                                                              "b64(",
+                                                              "basename(",
+                                                              "len("};
 
 static inline void update_pos(const char c, ast::pos_info& pos) {
 	pos.col++;

userspace/libsinsp/filter/parser.h

@@ -69,7 +69,7 @@ class RE2;
 //                             | 'startswith ' | 'bstartswith ' | 'endswith '
 //     ListOperator        ::= 'intersects' | 'in' | 'pmatch'
 //     FieldTransformerVal    ::= 'val('
-//     FieldTransformerType   ::= 'tolower(' | 'toupper(' | 'b64(' | 'basename('
+//     FieldTransformerType   ::= 'tolower(' | 'toupper(' | 'b64(' | 'basename(' | 'len('
 //
 // Tokens (Regular Expressions):
 //     Identifier          ::= [a-zA-Z]+[a-zA-Z0-9_]*

userspace/libsinsp/sinsp_filter_transformer.cpp

@@ -63,15 +63,16 @@ bool sinsp_filter_transformer::string_transformer(std::vector<extract_value_t>&
 	return true;
 }
 
-bool sinsp_filter_transformer::transform_type(ppm_param_type& t) const {
+bool sinsp_filter_transformer::transform_type(ppm_param_type& t, uint32_t& flags) const {
+	bool is_list = flags & EPF_IS_LIST;
 	switch(m_type) {
 	case FTR_TOUPPER: {
 		switch(t) {
 		case PT_CHARBUF:
 		case PT_FSPATH:
 		case PT_FSRELPATH:
 			// for TOUPPER, the transformed type is the same as the input type
-			return true;
+			return !is_list;
 		default:
 			return false;
 		}
@@ -82,7 +83,7 @@ bool sinsp_filter_transformer::transform_type(ppm_param_type& t) const {
 		case PT_FSPATH:
 		case PT_FSRELPATH:
 			// for TOLOWER, the transformed type is the same as the input type
-			return true;
+			return !is_list;
 		default:
 			return false;
 		}
@@ -92,7 +93,7 @@ bool sinsp_filter_transformer::transform_type(ppm_param_type& t) const {
 		case PT_CHARBUF:
 		case PT_BYTEBUF:
 			// for BASE64, the transformed type is the same as the input type
-			return true;
+			return !is_list;
 		default:
 			return false;
 		}
@@ -107,6 +108,23 @@ bool sinsp_filter_transformer::transform_type(ppm_param_type& t) const {
 		case PT_FSPATH:
 		case PT_FSRELPATH:
 			// for BASENAME, the transformed type is the same as the input type
+			return !is_list;
+		default:
+			return false;
+		}
+	}
+	case FTR_LEN: {
+		if(is_list) {
+			t = PT_UINT64;
+			flags = 0;
+			return true;
+		}
+		switch(t) {
+		case PT_CHARBUF:
+		case PT_BYTEBUF:
+		case PT_FSPATH:
+		case PT_FSRELPATH:
+			t = PT_UINT64;
 			return true;
 		default:
 			return false;
@@ -119,8 +137,11 @@ bool sinsp_filter_transformer::transform_type(ppm_param_type& t) const {
 }
 
 bool sinsp_filter_transformer::transform_values(std::vector<extract_value_t>& vec,
-                                                ppm_param_type& t) {
-	if(!transform_type(t)) {
+                                                ppm_param_type& t,
+                                                uint32_t& flags) {
+	bool is_list = flags & EPF_IS_LIST;
+	ppm_param_type original_type = t;
+	if(!transform_type(t, flags)) {
 		throw_type_incompatibility_err(t, filter_transformer_type_str(m_type));
 	}
 
@@ -180,6 +201,50 @@ bool sinsp_filter_transformer::transform_values(std::vector<extract_value_t>& ve
 			return true;
 		});
 	}
+	case FTR_LEN: {
+		assert((void("len() type must be PT_UINT64"), t == PT_UINT64));
+		if(is_list) {
+			uint64_t len = static_cast<uint64_t>(vec.size());
+			auto stored_val = store_scalar(len);
+			vec.clear();
+			vec.push_back(stored_val);
+			return true;
+		}
+
+		// not a list: could be string or buffer
+		bool is_string = false;
+		switch(original_type) {
+		case PT_CHARBUF:
+		case PT_FSPATH:
+		case PT_FSRELPATH:
+			is_string = true;
+			break;
+		case PT_BYTEBUF:
+			is_string = false;
+			break;
+		default:
+			return false;
+		}
+
+		for(std::size_t i = 0; i < vec.size(); i++) {
+			uint64_t len;
+			if(vec[i].ptr == nullptr) {
+				vec[i] = store_scalar(0);
+				continue;
+			}
+
+			if(is_string) {
+				len = static_cast<uint64_t>(
+				        strnlen(reinterpret_cast<const char*>(vec[i].ptr), vec[i].len));
+				vec[i] = store_scalar(len);
+				continue;
+			}
+
+			len = static_cast<uint64_t>(vec[i].len);
+			vec[i] = store_scalar(len);
+		}
+		return true;
+	}
 	default:
 		throw_unsupported_err(m_type);
 		return false;

userspace/libsinsp/sinsp_filter_transformer.h

@@ -28,6 +28,7 @@ enum filter_transformer_type : uint8_t {
 	FTR_BASE64 = 2,
 	FTR_STORAGE = 3,  // This transformer is only used internally
 	FTR_BASENAME = 4,
+	FTR_LEN = 5
 };
 
 static inline std::string filter_transformer_type_str(filter_transformer_type m) {
@@ -42,6 +43,8 @@ static inline std::string filter_transformer_type_str(filter_transformer_type m)
 		return "storage";
 	case FTR_BASENAME:
 		return "basename";
+	case FTR_LEN:
+		return "len";
 	default:
 		throw sinsp_exception("unknown field transfomer id " + std::to_string(m));
 	}
@@ -63,6 +66,9 @@ static inline filter_transformer_type filter_transformer_from_str(const std::str
 	if(str == "basename") {
 		return filter_transformer_type::FTR_BASENAME;
 	}
+	if(str == "len") {
+		return filter_transformer_type::FTR_LEN;
+	}
 	throw sinsp_exception("unknown field transfomer '" + str + "'");
 }
 
@@ -72,9 +78,9 @@ class sinsp_filter_transformer {
 
 	sinsp_filter_transformer(filter_transformer_type t): m_type(t) {};
 
-	bool transform_type(ppm_param_type& t) const;
+	bool transform_type(ppm_param_type& t, uint32_t& flags) const;
 
-	bool transform_values(std::vector<extract_value_t>& vals, ppm_param_type& t);
+	bool transform_values(std::vector<extract_value_t>& vals, ppm_param_type& t, uint32_t& flags);
 
 private:
 	using str_transformer_func_t = std::function<bool(std::string_view in, storage_t& out)>;
@@ -83,6 +89,13 @@ class sinsp_filter_transformer {
 	                        ppm_param_type t,
 	                        str_transformer_func_t mod);
 
+	template<class T>
+	extract_value_t store_scalar(T value) {
+		uint8_t* bytes = reinterpret_cast<uint8_t*>(&value);
+		storage_t& stored_val = m_storage_values.emplace_back(bytes, bytes + sizeof(T));
+		return {static_cast<uint8_t*>(stored_val.data()), static_cast<uint32_t>(stored_val.size())};
+	}
+
 	filter_transformer_type m_type;
 	std::vector<storage_t> m_storage_values;
 };

userspace/libsinsp/sinsp_filtercheck.cpp

@@ -448,7 +448,7 @@ char* sinsp_filter_check::tostring(sinsp_evt* evt) {
 	}
 
 	auto ftype = get_transformed_field_info()->m_type;
-	if(m_field->m_flags & EPF_IS_LIST) {
+	if(get_transformed_field_info()->m_flags & EPF_IS_LIST) {
 		std::string res = "(";
 		for(auto& val : m_extracted_values) {
 			if(res.size() > 1) {
@@ -478,7 +478,7 @@ Json::Value sinsp_filter_check::tojson(sinsp_evt* evt) {
 		}
 
 		auto ftype = get_transformed_field_info()->m_type;
-		if(m_field->m_flags & EPF_IS_LIST) {
+		if(get_transformed_field_info()->m_flags & EPF_IS_LIST) {
 			for(auto& val : m_extracted_values) {
 				jsonval.append(rawval_to_json(val.ptr, ftype, m_field->m_print_format, val.len));
 			}
@@ -1045,11 +1045,12 @@ void sinsp_filter_check::add_transformer(filter_transformer_type trtype) {
 	// apply type transformation, both as a feasibility check and
 	// as an information to be returned later on
 	sinsp_filter_transformer tr(trtype);
-	if(!tr.transform_type(m_transformed_field->m_type)) {
+	if(!tr.transform_type(m_transformed_field->m_type, m_transformed_field->m_flags)) {
 		throw sinsp_exception("can't add field transformer: type '" +
 		                      std::string(param_type_to_string(m_transformed_field->m_type)) +
 		                      "' is not supported by '" + filter_transformer_type_str(trtype) +
-		                      "' transformer applied on field '" +
+		                      "' transformer applied on " +
+		                      (m_transformed_field->is_list() ? "list " : "") + "field '" +
 		                      std::string(get_field_info()->m_name) + "'");
 	}
 
@@ -1062,9 +1063,11 @@ void sinsp_filter_check::add_transformer(filter_transformer_type trtype) {
 }
 
 bool sinsp_filter_check::apply_transformers(std::vector<extract_value_t>& values) {
-	auto type = get_field_info()->m_type;
+	const filtercheck_field_info* field_info = get_field_info();
+	auto field_type = field_info->m_type;
+	auto field_flags = field_info->m_flags;
 	for(auto& tr : m_transformers) {
-		if(!tr.transform_values(values, type)) {
+		if(!tr.transform_values(values, field_type, field_flags)) {
 			return false;
 		}
 	}

userspace/libsinsp/sinsp_filtercheck.h

@@ -137,7 +137,7 @@ class sinsp_filter_check {
 	//
 	// Extract the field from the event. If sanitize_strings is true, any
 	// string values are sanitized to remove nonprintable characters.
-	// By default, this fills the vector with only one value, retireved by calling the single-result
+	// By default, this fills the vector with only one value, retrieved by calling the single-result
 	// extract method.
 	// If a NULL value is returned by extract, the vector is emptied.
 	// Subclasses are meant to either override this, or the single-valued extract method.

userspace/libsinsp/sinsp_filtercheck_fd.cpp

@@ -293,7 +293,7 @@ static const filtercheck_field_info sinsp_filter_check_fd_fields[] = {
          "FD full name raw. Just like fd.name, but only used if fd is a file path. File path is "
          "kept raw with limited sanitization and without deriving the absolute path."},
         {PT_CHARBUF,
-         EPF_IS_LIST | EPF_ARG_ALLOWED | EPF_NO_RHS | EPF_NO_TRANSFORMER,
+         EPF_IS_LIST | EPF_ARG_ALLOWED | EPF_NO_RHS,
          PF_DEC,
          "fd.types",
          "FD Type",