Improve handling of non-utf8 or empty text http message body

If http message body is expected to be text and turns out to
not be valid UTF-8 we fallback to using hex output.

Two bugs in that code:
1) An empty message body was considered invalid UTF-8 as it didn't
   have any valid UTF-8 characters ...
2) When restarting the output all the content already written
   to text was appended, including headers.

1 files changed, 21 insertions, 24 deletions

diff --git a/src/http_protocol.cc b/src/http_protocol.cc
index ad81050..f0e3cb3 100644
--- a/src/http_protocol.cc
+++ b/src/http_protocol.cc
@@ -326,7 +326,7 @@ private:
 class TextOutput : public OutputFilter {
 public:
   TextOutput(AttributedText* text)
-    : text_(text) {
+    : text_(text), checked_(0) {
   }
 
   void write(void const* data, size_t size, bool last) override {
@@ -334,49 +334,46 @@ public:
       hex_->write(data, size, last);
       return;
     }
-    auto d = reinterpret_cast<char const*>(data);
-    if (!buf_.empty()) {
-      buf_.append(d, size);
-      for (size_t i = 0; i < 4; ++i) {
-        if (i >= buf_.size()) break;
-        if (valid_utf8(buf_.data(), buf_.size() - i)) {
-          if (last && i > 0) break;
-          text_->append(buf_.data(), buf_.size() - i);
-          buf_.erase(0, buf_.size() - i);
-          return;
-        }
-      }
-    } else {
-      for (size_t i = 0; i < 4; ++i) {
-        if (i >= size) break;
-        if (valid_utf8(d, size - i)) {
-          if (last && i > 0) break;
-          text_->append(d, size - i);
-          if (i > 0) buf_.append(d + size - i, i);
-          return;
+
+    if (size == 0 && (!last || buf_.empty())) return;
+
+    buf_.append(reinterpret_cast<char const*>(data), size);
+    for (size_t i = 0; i < 4; ++i) {
+      if (checked_ + i >= buf_.size()) break;
+      if (valid_utf8(buf_.data() + checked_, buf_.size() - checked_ - i)) {
+        if (last) {
+          if (i > 0) break;
+          text_->append(buf_);
+          buf_.clear();
+          checked_ = 0;
+        } else {
+          checked_ = buf_.size() - i;
         }
+        return;
       }
     }
 
-    buf_.assign(text_->text());
-    buf_.append(d, size);
-    text_->clear();
     hex_.reset(new HexOutput(text_));
     hex_->write(buf_.data(), buf_.size(), last);
     buf_.clear();
   }
 
   void error() override {
+    text_->append(buf_.data(), checked_);
+    buf_.clear();
     text_->append("\nDecoding failed, invalid data\n");
   }
 
   void incomplete() override {
+    text_->append(buf_.data(), checked_);
+    buf_.clear();
     text_->append("\nNeed more data...\n");
   }
 
 private:
   AttributedText* const text_;
   std::string buf_;
+  size_t checked_;
   std::unique_ptr<HexOutput> hex_;
 };