sfeed

commit 7664faf66555c5c5b1221322bff7e49722ea36c7
parent 134a1ac3372fe1eae6bc5c6acd12666c17e82696
Author: Hiltjo Posthuma <hiltjo@codemadness.org>
Date:   Sun,  1 Nov 2020 16:53:40 +0100

sfeed_xmlenc: be more paranoid in printing encoding names

sfeed_xmlenc is used automatically in sfeed_update for detecting the encoding.

In particular do not allow slashes anymore either.  For example "//IGNORE" and
"//TRANSLIT" which are normally allowed.

Some iconv implementation might allow other funky names or even pathnames too,
so disallow that.

See also the notes about the "frommap" for the "-f" option.
https://pubs.opengroup.org/onlinepubs/9699919799/utilities/iconv.html

+ some minor parsing handling improvements.

Diffstat:
M
sfeed_xmlenc.1
 | 
5
+++--
M
sfeed_xmlenc.c
 | 
16
+++++++++++++---

2 files changed, 16 insertions(+), 5 deletions(-)
diff --git a/sfeed_xmlenc.1 b/sfeed_xmlenc.1
@@ -1,4 +1,4 @@
-.Dd March 15, 2020
+.Dd November 1, 2020
 .Dt SFEED_XMLENC 1
 .Os
 .Sh NAME
@@ -9,7 +9,8 @@
 .Sh DESCRIPTION
 .Nm
 reads XML data from stdin and writes the found text\-encoding to stdout.
-It translates the characters to lowercase and strips control characters.
+It converts characters in the encoding-name to lowercase and strips characters
+which are not allowed in an encoding-name.
 .Sh EXIT STATUS
 .Ex -std
 .Sh EXAMPLES
diff --git a/sfeed_xmlenc.c b/sfeed_xmlenc.c
@@ -14,7 +14,7 @@ static void
 xmltagstart(XMLParser *p, const char *t, size_t tl)
 {
 	/* optimization: try to find a processing instruction only at the
-	   start of the data. */
+	   start of the data at the first few starting tags. */
 	if (tags++ > 3)
 		exit(0);
 }
@@ -26,11 +26,19 @@ xmlattr(XMLParser *p, const char *t, size_t tl, const char *n, size_t nl,
 	if (strcasecmp(t, "?xml") || strcasecmp(n, "encoding"))
 		return;
 
-	/* output lowercase, no control characters */
 	for (; *v; v++) {
-		if (!iscntrl((unsigned char)*v))
+		if (isalpha((unsigned char)*v) ||
+		    isdigit((unsigned char)*v) ||
+		    *v == '.' || *v == ':' || *v == '-' || *v == '_')
 			putchar(tolower((unsigned char)*v));
 	}
+}
+
+static void
+xmlattrend(XMLParser *p, const char *t, size_t tl, const char *n, size_t nl)
+{
+	if (strcasecmp(t, "?xml") || strcasecmp(n, "encoding"))
+		return;
 	putchar('\n');
 	exit(0);
 }
@@ -42,6 +50,8 @@ main(void)
 		err(1, "pledge");
 
 	parser.xmlattr = xmlattr;
+	parser.xmlattrentity = xmlattr; /* no entity conversion */
+	parser.xmlattrend = xmlattrend;
 	parser.xmltagstart = xmltagstart;
 
 	/* NOTE: getnext is defined in xml.h for inline optimization */