Add a url escaper next to the scanner

2 files changed, 35 insertions, 1 deletions

diff --git a/scan_urlencoded_query.c b/scan_urlencoded_query.c
index 50aced3..cf0c5b1 100644
--- a/scan_urlencoded_query.c
+++ b/scan_urlencoded_query.c
@@ -20,7 +20,7 @@
 */
 
 /* This matrix holds for each ascii character the information,
-   whether it is a non-terminating character for on of the three
+   whether it is a non-terminating character for one of the three
    scan states we are in, that is 'path', 'param' and 'value' from
   /path?param=value&param=value, it is encoded in bit 0, 1 and 2
   respectively
@@ -159,3 +159,29 @@ ssize_t scan_fixed_int(char *data, size_t len, int *tmp) {
     *tmp = -*tmp;
   return len;
 }
+
+size_t fmt_urlencoded(char *dest, const uint8_t *src, size_t len) {
+  static const char safe[]     = "$/.=~-_";
+  static const char hexdigit[] = "0123456789abcdef";
+  size_t            written    = 0, i;
+
+  for (i = 0; i < len; ++i) {
+    uint8_t c = src[i];
+
+    /* strchr would find the terminator, so a zero byte never counts as safe */
+    if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || (c && strchr(safe, c))) {
+      if (dest)
+        dest[written] = (char)c;
+      ++written;
+    } else {
+      if (dest) {
+        dest[written]     = '%';
+        dest[written + 1] = hexdigit[c >> 4];
+        dest[written + 2] = hexdigit[c & 15];
+      }
+      written += 3;
+    }
+  }
+
+  return written;
+}
diff --git a/scan_urlencoded_query.h b/scan_urlencoded_query.h
index 74246e7..6f12c03 100644
--- a/scan_urlencoded_query.h
+++ b/scan_urlencoded_query.h
@@ -6,6 +6,8 @@
 #ifndef SCAN_URLENCODED_QUERY_H__
 #define SCAN_URLENCODED_QUERY_H__
 
+#include <stddef.h>
+#include <stdint.h>
 #include <sys/types.h>
 
 typedef struct {
@@ -52,4 +54,10 @@ void    scan_urlencoded_skipvalue(char **string);
  */
 ssize_t scan_fixed_int(char *data, size_t len, int *number);
 
+/* The other direction: percent encode len bytes from src into dest and return
+   how many characters were written, or how many would be needed when dest is
+   NULL. dest needs room for three times len.
+ */
+size_t  fmt_urlencoded(char *dest, const uint8_t *src, size_t len);
+
 #endif