uri: New module

Decode URI encoded string, validating both hex and that encoded
data is valid UTF-8.

1 files changed, 70 insertions, 0 deletions

diff --git a/src/uri.cc b/src/uri.cc
new file mode 100644
index 0000000..b7a3edf
--- /dev/null
+++ b/src/uri.cc
@@ -0,0 +1,70 @@
+#include "uri.hh"
+
+#include "u8.hh"
+
+#include <cstddef>
+#include <optional>
+#include <span>
+#include <string>
+#include <string_view>
+
+namespace uri {
+
+namespace {
+
+inline std::optional<uint8_t> hex(char c) {
+  if (c >= '0' && c <= '9')
+    return c - '0';
+  if (c >= 'A' && c <= 'F')
+    return 10 + (c - 'A');
+  if (c >= 'a' && c <= 'f')
+    return 10 + (c - 'a');
+  return std::nullopt;
+}
+
+}  // namespace
+
+std::optional<std::string_view> decode(std::string_view input,
+                                       std::string& dst) {
+  auto i = input.find('%');
+  if (i == std::string_view::npos)
+    return input;
+
+  dst.clear();
+  size_t last = 0;
+  bool check_utf8 = false;
+  while (true) {
+    if (input.size() - i < 3)
+      return std::nullopt;
+    auto a = hex(input[i + 1]);
+    auto b = hex(input[i + 2]);
+    if (!a.has_value() || !b.has_value())
+      return std::nullopt;
+    dst.append(input, last, i - last);
+    auto c = (a.value() << 4) | b.value();
+    if (c & 0x80)
+      check_utf8 = true;
+    dst.push_back(static_cast<char>(c));
+    last = i + 3;
+    i = input.find('%', last);
+    if (i == std::string::npos) {
+      dst.append(input, last);
+      break;
+    }
+  }
+
+  if (check_utf8) {
+    std::span<uint8_t const> data{reinterpret_cast<uint8_t const*>(dst.data()),
+                                  dst.size()};
+    auto it = data.begin();
+    while (it != data.end()) {
+      auto ret = u8::read(it, data.end());
+      if (!ret.has_value())
+        return std::nullopt;
+    }
+  }
+
+  return dst;
+}
+
+}  // namespace uri