Merge remote-tracking branch 'origin/topic/christian/gh-1982'

* origin/topic/christian/gh-1982:
  The is_num(), is_alpha(), and is_alnum() BiFs now return F on empty string
  Expand testcases around is_num(), is_alpha(), is_alnum(), is_ascii() BiFs
  Clarify is_ascii() BiF docstring re behavior on empty strings
  Fix minor indentation bugs in strings.bif

CHANGES

@@ -1,3 +1,9 @@
+5.0.0-dev.156 | 2022-03-02 08:23:50 +0000
+
+  * The is_num(), is_alpha(), and is_alnum() BiFs now return F on empty string.
+    The testcases for these functions, and for is_ascii() were expanded. The documentation of is_ascii()
+    concerning behavior of an empty string was clarified (Christian Kreibich, Corelight)
+
 5.0.0-dev.151 | 2022-03-02 08:09:28 +0000
 
   * SSL: rudimentary decryption for TLS 1.2 (Florian Wilkens, Johanna Amann)

NEWS

@@ -15,6 +15,8 @@ New Functionality
 Changed Functionality
 ---------------------
 
+- The is_num(), is_alpha(), and is_alnum() BiFs now return F for the empty string.
+
 Deprecated Functionality
 ------------------------
 

VERSION

@@ -1 +1 @@
-5.0.0-dev.151
+5.0.0-dev.156

src/strings.bif

@@ -541,7 +541,7 @@ function to_lower%(str: string%): string
 			*ls++ = s[i];
 		}
 
-    *ls++ = '\0';
+	*ls++ = '\0';
 
 	return zeek::make_intrusive<zeek::StringVal>(new zeek::String(1, lower_s, n));
 	%}
@@ -570,7 +570,7 @@ function to_upper%(str: string%): string
 			*us++ = s[i];
 		}
 
-    *us++ = '\0';
+	*us++ = '\0';
 
 	return zeek::make_intrusive<zeek::StringVal>(new zeek::String(1, upper_s, n));
 	%}
@@ -614,6 +614,7 @@ function to_string_literal%(str: string%): string
 	%}
 
 ## Determines whether a given string contains only ASCII characters.
+## The empty string is ASCII.
 ##
 ## str: The string to examine.
 ##
@@ -1253,13 +1254,17 @@ function ends_with%(str: string, sub: string%) : bool
 	return zeek::val_mgr->Bool(s.rfind(sub_s) == (s.size() - sub_s.size()));
 	%}
 
-## Returns whether an entire string consists only of digits.
+## Returns whether a string consists entirely of digits.
+## The empty string is not numeric.
 ##
 function is_num%(str: string%) : bool
 	%{
 	// Python's version of this method (which this is based on) just checks to see if every
 	// character in the string is a numeric value. If something more than this is desired, we
 	// could use something like std::from_chars or std::strto{ul,f} to check it.
+	if ( str->Len() == 0 )
+		return zeek::val_mgr->False();
+
 	const char* s = str->CheckString();
 	for ( int i = 0; i < str->Len(); i++ )
 		if ( ! std::isdigit(s[i]) )
@@ -1268,10 +1273,14 @@ function is_num%(str: string%) : bool
 	return zeek::val_mgr->True();
 	%}
 
-## Returns whether an entire string is alphabetic characters.
+## Returns whether a string consists entirely of alphabetic characters.
+## The empty string is not alphabetic.
 ##
 function is_alpha%(str: string%) : bool
 	%{
+	if ( str->Len() == 0 )
+		return zeek::val_mgr->False();
+
 	const char* s = str->CheckString();
 	for ( int i = 0; i < str->Len(); i++ )
 		if ( ! std::isalpha(s[i]) )
@@ -1280,10 +1289,14 @@ function is_alpha%(str: string%) : bool
 	return zeek::val_mgr->True();
 	%}
 
-## Returns whether an entire string is alphanumeric characters
+## Returns whether a string consists entirely of alphanumeric characters.
+## The empty string is not alphanumeric.
 ##
 function is_alnum%(str: string%) : bool
 	%{
+	if ( str->Len() == 0 )
+		return zeek::val_mgr->False();
+
 	const char* s = str->CheckString();
 	for ( int i = 0; i < str->Len(); i++ )
 		if ( ! std::isalnum(s[i]) )

testing/btest/Baseline/bifs.is_ascii/out

@@ -1,3 +1,4 @@
 ### BTest baseline data generated by btest-diff. Do not edit. Use "btest -U/-u" to update. Requires BTest >= 0.63.
 F
 T
+T

testing/btest/Baseline/bifs.string_utils/out

@@ -17,14 +17,17 @@ Content checking
 ----------------
 is_num abc   : 0
 is_num 123   : 1
+is_num ''    : 0
 is_alpha ab  : 1
 is_alpha 1a  : 0
 is_alpha a1  : 0
+is_alpha ''  : 0
 is_alnum ab  : 1
 is_alnum 1a  : 1
 is_alnum a1  : 1
 is_alnum 12  : 1
 is_alnum ##12: 0
+is_alnum ''  : 0
 
 String counting (input str 'aabbaa')
 ------------------------------------

testing/btest/bifs/is_ascii.zeek

@@ -9,4 +9,5 @@ event zeek_init()
 
 	print is_ascii(a);
 	print is_ascii(b);
+	print is_ascii("");
 	}

testing/btest/bifs/string_utils.zeek

@@ -23,14 +23,17 @@ event zeek_init()
 	print "----------------";
 	print fmt("is_num abc   : %d", is_num("abc"));
 	print fmt("is_num 123   : %d", is_num("123"));
+	print fmt("is_num ''    : %d", is_num(""));
 	print fmt("is_alpha ab  : %d", is_alpha("ab"));
 	print fmt("is_alpha 1a  : %d", is_alpha("1a"));
 	print fmt("is_alpha a1  : %d", is_alpha("a1"));
+	print fmt("is_alpha ''  : %d", is_alpha(""));
 	print fmt("is_alnum ab  : %d", is_alnum("ab"));
 	print fmt("is_alnum 1a  : %d", is_alnum("1a"));
 	print fmt("is_alnum a1  : %d", is_alnum("a1"));
 	print fmt("is_alnum 12  : %d", is_alnum("12"));
 	print fmt("is_alnum ##12: %d", is_alnum("##12"));
+	print fmt("is_alnum ''  : %d", is_alnum(""));
 	print "";
 
 	print "String counting (input str 'aabbaa')";