MINOR: uri_normalizer: Add a `percent-upper` normalizer
This normalizer uppercases the hexadecimal characters used in percent-encoding.
See GitHub Issue #714.
diff --git a/src/http_act.c b/src/http_act.c
index 1480563..06ecb9e 100644
--- a/src/http_act.c
+++ b/src/http_act.c
@@ -267,6 +267,24 @@
break;
}
+ case ACT_NORMALIZE_URI_PERCENT_UPPER:
+ case ACT_NORMALIZE_URI_PERCENT_UPPER_STRICT: {
+ const struct ist path = http_get_path(uri);
+ struct ist newpath = ist2(replace->area, replace->size);
+
+ if (!isttest(path))
+ goto leave;
+
+ err = uri_normalizer_percent_upper(path, rule->action == ACT_NORMALIZE_URI_PERCENT_UPPER_STRICT, &newpath);
+
+ if (err != URI_NORMALIZER_ERR_NONE)
+ break;
+
+ if (!http_replace_req_path(htx, newpath, 1))
+ goto fail_rewrite;
+
+ break;
+ }
}
switch (err) {
@@ -352,6 +370,21 @@
rule->action = ACT_NORMALIZE_URI_SORT_QUERY;
}
+ else if (strcmp(args[cur_arg], "percent-upper") == 0) {
+ cur_arg++;
+
+ if (strcmp(args[cur_arg], "strict") == 0) {
+ cur_arg++;
+ rule->action = ACT_NORMALIZE_URI_PERCENT_UPPER_STRICT;
+ }
+ else if (!*args[cur_arg]) {
+ rule->action = ACT_NORMALIZE_URI_PERCENT_UPPER;
+ }
+ else if (strcmp(args[cur_arg], "if") != 0 && strcmp(args[cur_arg], "unless") != 0) {
+ memprintf(err, "unknown argument '%s' for 'percent-upper' normalizer", args[cur_arg]);
+ return ACT_RET_PRS_ERR;
+ }
+ }
else {
memprintf(err, "unknown normalizer '%s'", args[cur_arg]);
return ACT_RET_PRS_ERR;
diff --git a/src/uri_normalizer.c b/src/uri_normalizer.c
index bd67542..ea9632b 100644
--- a/src/uri_normalizer.c
+++ b/src/uri_normalizer.c
@@ -15,8 +15,66 @@
#include <haproxy/api.h>
#include <haproxy/buf.h>
#include <haproxy/chunk.h>
+#include <haproxy/tools.h>
#include <haproxy/uri_normalizer.h>
+/* Uppercases letters used in percent encoding.
+ *
+ * If `strict` is set to 0 then percent characters that are not followed by a
+ * hexadecimal digit are returned as-is without modifying the following letters.
+ * If `strict` is set to 1 then `URI_NORMALIZER_ERR_INVALID_INPUT` is returned
+ * for invalid sequences.
+ */
+enum uri_normalizer_err uri_normalizer_percent_upper(const struct ist input, int strict, struct ist *dst)
+{
+ enum uri_normalizer_err err;
+
+ const size_t size = istclear(dst);
+ struct ist output = *dst;
+
+ struct ist scanner = input;
+
+ /* The output will have the same length. */
+ if (size < istlen(input)) {
+ err = URI_NORMALIZER_ERR_ALLOC;
+ goto fail;
+ }
+
+ while (istlen(scanner)) {
+ const char current = istshift(&scanner);
+
+ if (current == '%') {
+ if (istlen(scanner) >= 2) {
+ if (ishex(istptr(scanner)[0]) && ishex(istptr(scanner)[1])) {
+ output = __istappend(output, current);
+ output = __istappend(output, toupper(istshift(&scanner)));
+ output = __istappend(output, toupper(istshift(&scanner)));
+ continue;
+ }
+ }
+
+ if (strict) {
+ err = URI_NORMALIZER_ERR_INVALID_INPUT;
+ goto fail;
+ }
+ else {
+ output = __istappend(output, current);
+ }
+ }
+ else {
+ output = __istappend(output, current);
+ }
+ }
+
+ *dst = output;
+
+ return URI_NORMALIZER_ERR_NONE;
+
+ fail:
+
+ return err;
+}
+
/* Merges `/../` with preceding path segments.
*
* If `full` is set to `0` then `/../` will be printed at the start of the resulting