From fc4547b412e28164af1bf8981234c6af959ccc0b Mon Sep 17 00:00:00 2001
From: Joel Klinghed <the_jk@spawned.biz>
Date: Tue, 13 Jun 2023 10:07:16 +0200
Subject: WIP

---
 sax/inc/sax_decoder.hh | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 57 insertions(+)
 create mode 100644 sax/inc/sax_decoder.hh

(limited to 'sax/inc/sax_decoder.hh')

diff --git a/sax/inc/sax_decoder.hh b/sax/inc/sax_decoder.hh
new file mode 100644
index 0000000..40a56c9
--- /dev/null
+++ b/sax/inc/sax_decoder.hh
@@ -0,0 +1,57 @@
+#ifndef SAX_DECODER_HH
+#define SAX_DECODER_HH
+
+#include <memory>
+#include <string>
+#include <string_view>
+
+namespace modxml {
+namespace sax {
+
+/**
+ * Decoder returned by DecoderFactory. Used by Processor to turn bytes into
+ * unicode characters.
+ */
+class Decoder {
+ public:
+  virtual ~Decoder() = default;
+
+  enum class State {
+    GOOD = 0,
+    // too little data was given to advance
+    NEED_MORE,
+    // invalid data was given to advance
+    INVALID,
+  };
+
+  /**
+   * Decode as many code points as possible from in (start at in_offset) and
+   * write them to out (start at out_offset) as UTF-8.
+   * All written code points must be valid per Unicode, so inside the
+   * range U+0 to U+10FFFF and not a surrogate pair (U+D800-U+DFFF).
+   * No partial output, only write to out if the whole UTF-8 sequence is
+   * going to fit.
+   * The is always at least 4 bytes available (out.size() - out_offset) when
+   * called.
+   * Advance in_offset for data consumed.
+   * Advance out_offset for code points written. Do NOT write past out.size().
+   * Do NOT resize out.
+   * If at least one code point is decoded and written to out, return GOOD.
+   * If it is not possible to decode a single code point, in_offset and
+   * out_offset should not be advanced and something other than GOOD returned.
+   * Do not keep any references to any of the parameters after returning, next
+   * advance() call will point to the following bytes, but all parameters
+   * may have changed as they are subject to the buffer implementations of the
+   * Processor.
+   */
+  virtual State decode(std::string_view in, std::size_t& in_offset,
+                       std::string& out, std::size_t& out_offset) = 0;
+
+ protected:
+  Decoder() = default;
+};
+
+}  // namespace sax
+}  // namespace modxml
+
+#endif  // SAX_DECODER_HH
-- 
cgit v1.2.3-70-g09d2