Enhance the Error type to provide more information about the error.

This introduces the following changes.:

* The `Error` code now has a `Code` field that exposes the YARA API error code to Go users, and also exposes the error codes as constants that can be used in Go code (example: `err.Code == yara.ERROR_TIMEOUT`). Sometimes is useful to get a numeric error code instead of a text string.

* When `Error` is converted to a string, it includes the name of the rule causing the error. For example, instead of getting an error like `syntax error, unexpected identifier, expecting <condition>`, you get `rule \"foo\": syntax error, unexpected identifier, expecting <condition>`.

Author: plusvic

compiler.go

@@ -10,13 +10,19 @@ package yara
 #include <yara.h>
 #include "compat.h"
 
+// rule_identifier is a union accessor function.
+static const char* rule_identifier(YR_RULE* r) {
+	return r->identifier;
+}
+
 void compilerCallback(int, char*, int, YR_RULE*, char*, void*);
 char* includeCallback(char*, char*, char*, void*);
 void freeCallback(char*, void*);
 */
 import "C"
 import (
 	"errors"
+	"fmt"
 	"os"
 	"reflect"
 	"runtime"
@@ -26,10 +32,18 @@ import (
 //export compilerCallback
 func compilerCallback(errorLevel C.int, filename *C.char, linenumber C.int, rule *C.YR_RULE, message *C.char, userData unsafe.Pointer) {
 	c := cgoHandle(*(*uintptr)(userData)).Value().(*Compiler)
+	var text string
+	if rule != nil {
+		text = fmt.Sprintf("rule \"%s\": %s",
+			C.GoString(C.rule_identifier(rule)),
+			C.GoString(message))
+	} else {
+		text = C.GoString(message)
+	}
 	msg := CompilerMessage{
 		Filename: C.GoString(filename),
 		Line:     int(linenumber),
-		Text:     C.GoString(message),
+		Text:     text,
 	}
 	if rule != nil {
 		// Rule object implicitly relies on the compiler object and is destroyed when the compiler is destroyed.

compiler_test.go

@@ -6,7 +6,9 @@
 
 package yara
 
-import "testing"
+import (
+	"testing"
+)
 
 func TestCompiler(t *testing.T) {
 	c, _ := NewCompiler()
@@ -43,6 +45,18 @@ func TestWarnings(t *testing.T) {
 	t.Logf("Recorded Errors=%#v, Warnings=%#v", c.Errors, c.Warnings)
 }
 
+func TestErrors(t *testing.T) {
+	c, _ := NewCompiler()
+	c.AddString("rule foo { bar }", "")
+	if len(c.Errors) == 0 {
+		t.Error("did not recognize error")
+	}
+	expectedError := "rule \"foo\": syntax error, unexpected identifier, expecting <condition>"
+	if c.Errors[0].Text != expectedError {
+		t.Errorf("expected error: %#v, got %#v", expectedError, c.Errors[0].Text)
+	}
+}
+
 func setupCompiler(t *testing.T) *Compiler {
 	c, err := NewCompiler()
 	if err != nil {

error.go

@@ -8,25 +8,94 @@ package yara
 
 // #include <yara.h>
 import "C"
-import "strconv"
+import (
+	"fmt"
+)
 
 // Error encapsulates the C API error codes.
-type Error int
+type Error struct {
+	// YARA error code.
+	Code int
+	// Namespace in which the error occurred, if applicable. It can be empty.
+	Namespace string
+	// Rule in which the error occurred, if applicable. It can be empty.
+	RuleIdentifier string
+	// String in which the error occurred, if applicable. It can be empty.
+	StringIdentifier string
+}
 
-func (e Error) Error() string {
-	if str, ok := errorStrings[int(e)]; ok {
+func (e Error) Error() (errorString string) {
+	if e.Namespace != "" && e.RuleIdentifier != "" {
+		errorString = fmt.Sprintf("%s caused by rule \"%s:%s\"",
+			errorCodeToString(e.Code), e.Namespace, e.RuleIdentifier)
+		if e.StringIdentifier != "" {
+			errorString += fmt.Sprintf(" string %s", e.StringIdentifier)
+		}
+	} else {
+		errorString = errorCodeToString(e.Code)
+	}
+	return errorString
+}
+
+func errorCodeToString(errorCode int) string {
+	if str, ok := errorStrings[errorCode]; ok {
 		return str
 	}
-	return "unknown YARA error " + strconv.Itoa(int(e))
+	return fmt.Sprintf("unknown error %d", errorCode)
 }
 
 func newError(code C.int) error {
-	if code != 0 {
-		return Error(code)
+	if code == C.ERROR_SUCCESS {
+		return nil
 	}
-	return nil
+	return Error{Code: int(code)}
 }
 
+const (
+	ERROR_SUCCESS                      = C.ERROR_SUCCESS
+	ERROR_INSUFFICIENT_MEMORY          = C.ERROR_INSUFFICIENT_MEMORY
+	ERROR_COULD_NOT_ATTACH_TO_PROCESS  = C.ERROR_COULD_NOT_ATTACH_TO_PROCESS
+	ERROR_COULD_NOT_OPEN_FILE          = C.ERROR_COULD_NOT_OPEN_FILE
+	ERROR_COULD_NOT_MAP_FILE           = C.ERROR_COULD_NOT_MAP_FILE
+	ERROR_INVALID_FILE                 = C.ERROR_INVALID_FILE
+	ERROR_CORRUPT_FILE                 = C.ERROR_CORRUPT_FILE
+	ERROR_UNSUPPORTED_FILE_VERSION     = C.ERROR_UNSUPPORTED_FILE_VERSION
+	ERROR_INVALID_REGULAR_EXPRESSION   = C.ERROR_INVALID_REGULAR_EXPRESSION
+	ERROR_INVALID_HEX_STRING           = C.ERROR_INVALID_HEX_STRING
+	ERROR_SYNTAX_ERROR                 = C.ERROR_SYNTAX_ERROR
+	ERROR_LOOP_NESTING_LIMIT_EXCEEDED  = C.ERROR_LOOP_NESTING_LIMIT_EXCEEDED
+	ERROR_DUPLICATED_LOOP_IDENTIFIER   = C.ERROR_DUPLICATED_LOOP_IDENTIFIER
+	ERROR_DUPLICATED_IDENTIFIER        = C.ERROR_DUPLICATED_IDENTIFIER
+	ERROR_DUPLICATED_TAG_IDENTIFIER    = C.ERROR_DUPLICATED_TAG_IDENTIFIER
+	ERROR_DUPLICATED_META_IDENTIFIER   = C.ERROR_DUPLICATED_META_IDENTIFIER
+	ERROR_DUPLICATED_STRING_IDENTIFIER = C.ERROR_DUPLICATED_STRING_IDENTIFIER
+	ERROR_UNREFERENCED_STRING          = C.ERROR_UNREFERENCED_STRING
+	ERROR_UNDEFINED_STRING             = C.ERROR_UNDEFINED_STRING
+	ERROR_UNDEFINED_IDENTIFIER         = C.ERROR_UNDEFINED_IDENTIFIER
+	ERROR_MISPLACED_ANONYMOUS_STRING   = C.ERROR_MISPLACED_ANONYMOUS_STRING
+	ERROR_INCLUDES_CIRCULAR_REFERENCE  = C.ERROR_INCLUDES_CIRCULAR_REFERENCE
+	ERROR_INCLUDE_DEPTH_EXCEEDED       = C.ERROR_INCLUDE_DEPTH_EXCEEDED
+	ERROR_WRONG_TYPE                   = C.ERROR_WRONG_TYPE
+	ERROR_EXEC_STACK_OVERFLOW          = C.ERROR_EXEC_STACK_OVERFLOW
+	ERROR_SCAN_TIMEOUT                 = C.ERROR_SCAN_TIMEOUT
+	ERROR_TOO_MANY_SCAN_THREADS        = C.ERROR_TOO_MANY_SCAN_THREADS
+	ERROR_CALLBACK_ERROR               = C.ERROR_CALLBACK_ERROR
+	ERROR_INVALID_ARGUMENT             = C.ERROR_INVALID_ARGUMENT
+	ERROR_TOO_MANY_MATCHES             = C.ERROR_TOO_MANY_MATCHES
+	ERROR_INTERNAL_FATAL_ERROR         = C.ERROR_INTERNAL_FATAL_ERROR
+	ERROR_NESTED_FOR_OF_LOOP           = C.ERROR_NESTED_FOR_OF_LOOP
+	ERROR_INVALID_FIELD_NAME           = C.ERROR_INVALID_FIELD_NAME
+	ERROR_UNKNOWN_MODULE               = C.ERROR_UNKNOWN_MODULE
+	ERROR_NOT_A_STRUCTURE              = C.ERROR_NOT_A_STRUCTURE
+	ERROR_NOT_INDEXABLE                = C.ERROR_NOT_INDEXABLE
+	ERROR_NOT_A_FUNCTION               = C.ERROR_NOT_A_FUNCTION
+	ERROR_INVALID_FORMAT               = C.ERROR_INVALID_FORMAT
+	ERROR_TOO_MANY_ARGUMENTS           = C.ERROR_TOO_MANY_ARGUMENTS
+	ERROR_WRONG_ARGUMENTS              = C.ERROR_WRONG_ARGUMENTS
+	ERROR_WRONG_RETURN_TYPE            = C.ERROR_WRONG_RETURN_TYPE
+	ERROR_DUPLICATED_STRUCTURE_MEMBER  = C.ERROR_DUPLICATED_STRUCTURE_MEMBER
+)
+
 // FIXME: This should be generated from yara/error.h
 var errorStrings = map[int]string{
 	C.ERROR_INSUFICIENT_MEMORY:             "insufficient memory",

scanner.go

@@ -41,6 +41,23 @@ type Scanner struct {
 	userData *cgoHandle
 }
 
+// Creates a new error that includes information a about the rule
+// causing the error.
+func (s *Scanner) newScanError(code C.int) error {
+	if code == C.ERROR_SUCCESS {
+		return nil
+	}
+	err := Error{Code: int(code)}
+	if rule := s.GetLastErrorRule(); rule != nil {
+		err.RuleIdentifier = rule.Identifier()
+		err.Namespace = rule.Namespace()
+	}
+	if str := s.GetLastErrorString(); str != nil {
+		err.StringIdentifier = str.Identifier()
+	}
+	return err
+}
+
 // NewScanner creates a YARA scanner.
 func NewScanner(r *Rules) (*Scanner, error) {
 	var yrScanner *C.YR_SCANNER
@@ -153,7 +170,7 @@ func (s *Scanner) ScanMem(buf []byte) (err error) {
 	}
 	s.putCallbackData()
 	C.yr_scanner_set_flags(s.cptr, s.flags.withReportFlags(s.Callback))
-	err = newError(C.yr_scanner_scan_mem(
+	err = s.newScanError(C.yr_scanner_scan_mem(
 		s.cptr,
 		ptr,
 		C.size_t(len(buf))))
@@ -176,7 +193,7 @@ func (s *Scanner) ScanFile(filename string) (err error) {
 	defer C.free(unsafe.Pointer(cfilename))
 	s.putCallbackData()
 	C.yr_scanner_set_flags(s.cptr, s.flags.withReportFlags(s.Callback))
-	err = newError(C.yr_scanner_scan_file(
+	err = s.newScanError(C.yr_scanner_scan_file(
 		s.cptr,
 		cfilename,
 	))
@@ -191,7 +208,7 @@ func (s *Scanner) ScanFile(filename string) (err error) {
 func (s *Scanner) ScanFileDescriptor(fd uintptr) (err error) {
 	s.putCallbackData()
 	C.yr_scanner_set_flags(s.cptr, s.flags.withReportFlags(s.Callback))
-	err = newError(C._yr_scanner_scan_fd(
+	err = s.newScanError(C._yr_scanner_scan_fd(
 		s.cptr,
 		C.int(fd),
 	))
@@ -206,7 +223,7 @@ func (s *Scanner) ScanFileDescriptor(fd uintptr) (err error) {
 func (s *Scanner) ScanProc(pid int) (err error) {
 	s.putCallbackData()
 	C.yr_scanner_set_flags(s.cptr, s.flags.withReportFlags(s.Callback))
-	err = newError(C.yr_scanner_scan_proc(
+	err = s.newScanError(C.yr_scanner_scan_proc(
 		s.cptr,
 		C.int(pid),
 	))
@@ -226,7 +243,7 @@ func (s *Scanner) ScanMemBlocks(mbi MemoryBlockIterator) (err error) {
 	defer ((*cgoHandle)(cmbi.context)).Delete()
 	s.putCallbackData()
 	C.yr_scanner_set_flags(s.cptr, s.flags.withReportFlags(s.Callback)|C.SCAN_FLAGS_NO_TRYCATCH)
-	err = newError(C.yr_scanner_scan_mem_blocks(
+	err = s.newScanError(C.yr_scanner_scan_mem_blocks(
 		s.cptr,
 		cmbi,
 	))