Function to encode strings

entities.c

@@ -9,8 +9,10 @@
 
 #include <errno.h>
 #include <stdbool.h>
+#include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <stdio.h>
 
 #define UNICODE_MAX 0x10FFFFul
 
@@ -391,3 +393,33 @@ size_t decode_html_entities_utf8(char *dest, const char *src)
 	return (size_t)(to - dest);
 }
 
+size_t encode_html_entities(char *dest, const char *src) {
+	char *to = dest;
+	for( const char *from = src ; *from ; from++ ) {
+		if ( *from>=0 ) {
+			if ( (*from>='A' && *from<='Z') || (*from>='a' && *from<='z') ) {
+				*to = *from;
+				to++;
+				continue;
+			}
+		}
+		unsigned i;
+		for( i=0 ; i<sizeof NAMED_ENTITIES / sizeof *NAMED_ENTITIES ; i++ )
+			if ( strncmp(from,NAMED_ENTITIES[i][1],strlen(NAMED_ENTITIES[i][1]))==0 ) 
+				break;
+		if ( i<sizeof NAMED_ENTITIES / sizeof *NAMED_ENTITIES ) {
+			*to = '&';
+			to++;
+			strcpy(to,NAMED_ENTITIES[i][0]);
+			to += strlen(NAMED_ENTITIES[i][0]);
+			from += strlen(NAMED_ENTITIES[i][1])-1;
+		}
+		else {
+			*to = *from;
+			to++;
+		}
+	}
+	*to = 0;
+	return (size_t)(to - dest);
+}
+

entities.h

@@ -20,5 +20,9 @@ extern size_t decode_html_entities_utf8(char *dest, const char *src);
 	The function returns the length of the decoded string.
 */
 
+
+
+size_t encode_html_entities(char *dest, const char *src);
+
 #endif
 

t-entities.c

@@ -30,6 +30,25 @@ int main(void)
 		assert(strcmp(buffer, SAMPLE) == 0);
 	}
 
+	{
+		static const char INPUT[] = ""
+			"Miguel Leitão\n"
+			"[email protected]\n"
+			"<p>Hello!!</p>";
+		static char GOAL[] = ""
+			"Miguel Leit&atilde;o\n"
+			"test&commat;example.org\n"
+			"&lt;p&gt;Hello!!&lt;/p&gt;";
+		char OUTPUT[sizeof GOAL];
+		char REVERT[sizeof GOAL];
+		assert(encode_html_entities(OUTPUT, INPUT) == sizeof GOAL - 1);
+		assert(strcmp(OUTPUT, GOAL) == 0);
+
+		decode_html_entities_utf8(REVERT, OUTPUT);
+		assert(strcmp(INPUT, REVERT) == 0);
+	}
+
+
 	fprintf(stdout, "All tests passed :-)\n");
 	return EXIT_SUCCESS;
 }