htmlclean: Add a main function for easier testing / fuzzing

2 files changed, 38 insertions, 0 deletions

diff --git a/libraries/htmlclean/main.c b/libraries/htmlclean/main.c
new file mode 100644
index 00000000..bef790cb
--- /dev/null
+++ b/libraries/htmlclean/main.c
@@ -0,0 +1,33 @@
+/* This is a CLI interface to libvilistextum to make it easier to run a fuzzer */
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "htmlclean.h"
+
+char* stdin_to_string()
+{
+	const int BUFFER_SIZE = 4096;
+	char buffer[BUFFER_SIZE];
+	size_t content_size = 1; // includes null
+	char *content = malloc(sizeof(char) * BUFFER_SIZE);
+	content[0] = '\0';
+	while(fgets(buffer, BUFFER_SIZE, stdin))
+	{
+		content_size += strlen(buffer);
+		content = realloc(content, content_size);
+		strcat(content, buffer);
+	}
+	return content;
+}
+
+int main()
+{
+	char* content = stdin_to_string();
+	char* cleaned = htmlclean_strip_html(content);
+	free(content);
+
+	printf("%s\n", cleaned);
+	free(cleaned);
+	return 0;
+}
diff --git a/libraries/htmlclean/meson.build b/libraries/htmlclean/meson.build
index 39e8341c..e1050e88 100644
--- a/libraries/htmlclean/meson.build
+++ b/libraries/htmlclean/meson.build
@@ -6,3 +6,8 @@ htmlclean_lib = static_library(
   ],
   dependencies: [ glib, gumbo ]
 )
+
+htmlclean_main = executable(
+	'htmlclean_main',
+	'main.c',
+	link_with: htmlclean_lib)