MediaType now parses extra parameters

src/http_header.cc

@@ -170,43 +170,55 @@ MediaType::parseRaw(const char* str, size_t len) {
         suffix_ = suffix;
     }
 
-    if (eof(p)) return;
-
-    if (*p == ';') ++p;
-
-    if (eof(p)) {
-        raise("Malformed Media Type");
-    }
+    // Parse parameters
+    while (!eof(p)) {
 
-    while (*p == ' ') ++p;
-
-    if (eof(p)) {
-        raise("Malformed Media Type");
-    };
-
-    Optional<Q> q = None();
+        if (*p == ';' || *p == ' ') {
+            if (eof(p + 1)) raise("Malformed Media Type");
+            ++p;
+        }
 
-    if (*p == 'q') {
-        ++p;
+        else if (*p == 'q') {
+            ++p;
 
-        if (eof(p)) {
-            raise("Invalid quality factor");
-        }
+            if (eof(p)) {
+                raise("Invalid quality factor");
+            }
 
-        if (*p == '=') {
-            char *end;
-            double val = strtod(p + 1, &end);
-            if (!eof(end) && *end != ';' && *end != ' ') {
+            if (*p == '=') {
+                char *end;
+                double val = strtod(p + 1, &end);
+                if (!eof(end) && *end != ';' && *end != ' ') {
+                    raise("Invalid quality factor");
+                }
+                q_ = Some(Q::fromFloat(val));
+                p = end;
+            }
+            else {
                 raise("Invalid quality factor");
             }
-            q = Some(Q::fromFloat(val));
         }
         else {
-            raise("Invalid quality factor");
+            const char *begin = p;
+            while (!eof(p) && *p != '=') ++p;
+
+            if (eof(p)) raise("Unfinished Media Type parameter");
+            const char *end = p;
+            ++p;
+            if (eof(p)) raise("Unfinished Media Type parameter");
+
+            std::string key(begin, end);
+
+            begin = p;
+            while (!eof(p) && *p != ' ' && *p != ';') ++p;
+
+            std::string value(begin, p);
+
+            params.insert(std::make_pair(std::move(key), std::move(value)));
+
         }
-    }
 
-    q_ = std::move(q);
+    }
 
     #undef MAX_SIZE
 
@@ -217,6 +229,21 @@ MediaType::setQuality(Q quality) {
     q_ = Some(quality);
 }
 
+Optional<std::string>
+MediaType::getParam(std::string name) const {
+    auto it = params.find(name);
+    if (it == std::end(params)) {
+        return None();
+    }
+
+    return Some(it->second);
+}
+
+void
+MediaType::setParam(std::string name, std::string value) {
+    params[name] = value;
+}
+
 std::string
 MediaType::toString() const {
 
@@ -252,6 +279,7 @@ MediaType::toString() const {
         }
     };
 
+    // @Improvement: allocating and concatenating many small strings is probably slow
     std::string res;
     res += topString(top_);
     res += "/";
@@ -265,6 +293,11 @@ MediaType::toString() const {
         res += quality.toString();
     });
 
+    for (const auto& param: params) {
+        res += "; ";
+        res += param.first + "=" + param.second;
+    }
+
     return res;
 }
 

src/http_header.h

@@ -14,6 +14,7 @@
 #include <vector>
 #include <stdexcept>
 #include <cassert>
+#include <unordered_map>
 
 #define SAFE_HEADER_CAST
 
@@ -216,6 +217,9 @@ public:
     const Optional<Q>& q() const { return q_; }
     void setQuality(Q quality);
 
+    Optional<std::string> getParam(std::string name) const;
+    void setParam(std::string name, std::string value);
+
     std::string toString() const;
 private:
     void parseRaw(const char* str, size_t len);
@@ -243,6 +247,8 @@ private:
     Index rawSubIndex;
     Index rawSuffixIndex;
 
+    std::unordered_map<std::string, std::string> params;
+
     Optional<Q> q_;
 };
 

tests/mime_test.cc

@@ -28,53 +28,92 @@ TEST(mime_test, basic_test) {
     m4.setQuality(Q::fromFloat(0.79));
 
     ASSERT_EQ(m4.toString(), "application/json+zip; q=0.79");
+
+    auto m5 = MIME(Text, Html);
+    m5.setQuality(Q::fromFloat(1.0));
+    m5.setParam("charset", "utf-8");
+    ASSERT_EQ(m5.toString(), "text/html; q=1; charset=utf-8");
+}
+
+void parse(const char* str, std::function<void (const MediaType&)> testFunc)
+{
+    auto m1 = MediaType::fromString(std::string(str));
+    testFunc(m1);
+
+    auto m2 = MediaType::fromRaw(str, strlen(str));
+    testFunc(m2);
+
+    MediaType m3(str, MediaType::DoParse);
+    testFunc(m3);
 }
 
 TEST(mime_test, valid_parsing_test) {
-    auto m1 = MediaType::fromString("application/json");
-    ASSERT_EQ(m1, MIME(Application, Json));
-    ASSERT_TRUE(m1.q().isEmpty());
-
-    auto m2 = MediaType::fromString("application/xhtml+xml");
-    ASSERT_EQ(m2, MediaType(Type::Application, Subtype::Xhtml, Suffix::Xml));
-    ASSERT_TRUE(m2.q().isEmpty());
-
-    auto m3 = MediaType::fromString("application/json; q=0.3");
-    ASSERT_EQ(m3, MIME(Application, Json));
-    ASSERT_EQ(m3.q().getOrElse(Q(0)), Q::fromFloat(0.3));
-
-    auto m4 = MediaType::fromString("application/xhtml+xml; q=0.7");
-    ASSERT_EQ(m4.top(), Type::Application);
-    ASSERT_EQ(m4.sub(), Subtype::Xhtml);
-    ASSERT_EQ(m4.suffix(), Suffix::Xml);
-    ASSERT_EQ(m4.q().getOrElse(Q(0)), Q(70));
-
-    auto m5 = MediaType::fromString("application/xhtml+xml; q=0.78");
-    ASSERT_EQ(m5.q().getOrElse(Q(0)), Q(78));
-
-    auto m6 = MediaType::fromString("application/vnd.adobe.flash-movie");
-    ASSERT_EQ(m6.top(), Type::Application);
-    ASSERT_EQ(m6.sub(), Subtype::Vendor);
-    ASSERT_EQ(m6.suffix(), Suffix::None);
-    ASSERT_EQ(m6.rawSub(), "vnd.adobe.flash-movie");
-
-    auto m7 = MediaType::fromString("application/vnd.mycompany.myapp-v2+json");
-    ASSERT_EQ(m7.top(), Type::Application);
-    ASSERT_EQ(m7.sub(), Subtype::Vendor);
-    ASSERT_EQ(m7.suffix(), Suffix::Json);
-    ASSERT_EQ(m7.rawSub(), "vnd.mycompany.myapp-v2");
-
-    auto m8 = MediaType::fromString("application/x-myapp-v1+json");
-    ASSERT_EQ(m8.top(), Type::Application);
-    ASSERT_EQ(m8.sub(), Subtype::Ext);
-    ASSERT_EQ(m8.suffix(), Suffix::Json);
-    ASSERT_EQ(m8.rawSub(), "x-myapp-v1");
-
-    auto m9 = MediaType::fromString("audio/x-my-codec");
-    ASSERT_EQ(m9.top(), Type::Audio);
-    ASSERT_EQ(m9.sub(), Subtype::Ext);
-    ASSERT_EQ(m9.suffix(), Suffix::None);
-    ASSERT_EQ(m9.rawSub(), "x-my-codec");
+    parse("application/json", [](const MediaType &m1) {
+        ASSERT_EQ(m1, MIME(Application, Json));
+        ASSERT_TRUE(m1.q().isEmpty());
+    });
+
+    parse("application/xhtml+xml", [](const MediaType &m2) {
+        ASSERT_EQ(m2, MediaType(Type::Application, Subtype::Xhtml, Suffix::Xml));
+        ASSERT_TRUE(m2.q().isEmpty());
+    });
+
+    parse("application/json; q=0.3", [](const MediaType& m3) {
+        ASSERT_EQ(m3, MIME(Application, Json));
+        ASSERT_EQ(m3.q().getOrElse(Q(0)), Q::fromFloat(0.3));
+    });
+
+    parse("application/xhtml+xml; q=0.7", [](const MediaType& m4) {
+        ASSERT_EQ(m4.top(), Type::Application);
+        ASSERT_EQ(m4.sub(), Subtype::Xhtml);
+        ASSERT_EQ(m4.suffix(), Suffix::Xml);
+        ASSERT_EQ(m4.q().getOrElse(Q(0)), Q(70));
+    });
+
+    parse("application/xhtml+xml; q=0.78", [](const MediaType& m5) {
+        ASSERT_EQ(m5.q().getOrElse(Q(0)), Q(78));
+    });
+
+    parse("application/vnd.adobe.flash-movie", [](const MediaType& m6) {
+        ASSERT_EQ(m6.top(), Type::Application);
+        ASSERT_EQ(m6.sub(), Subtype::Vendor);
+        ASSERT_EQ(m6.suffix(), Suffix::None);
+        ASSERT_EQ(m6.rawSub(), "vnd.adobe.flash-movie");
+    });
+
+    parse("application/vnd.mycompany.myapp-v2+json", [](const MediaType& m7) {
+        ASSERT_EQ(m7.top(), Type::Application);
+        ASSERT_EQ(m7.sub(), Subtype::Vendor);
+        ASSERT_EQ(m7.suffix(), Suffix::Json);
+        ASSERT_EQ(m7.rawSub(), "vnd.mycompany.myapp-v2");
+    });
+
+    parse("application/x-myapp-v1+json", [](const MediaType& m8) {
+        ASSERT_EQ(m8.top(), Type::Application);
+        ASSERT_EQ(m8.sub(), Subtype::Ext);
+        ASSERT_EQ(m8.suffix(), Suffix::Json);
+        ASSERT_EQ(m8.rawSub(), "x-myapp-v1");
+    });
+
+    parse("audio/x-my-codec", [](const MediaType& m9) {
+        ASSERT_EQ(m9.top(), Type::Audio);
+        ASSERT_EQ(m9.sub(), Subtype::Ext);
+        ASSERT_EQ(m9.suffix(), Suffix::None);
+        ASSERT_EQ(m9.rawSub(), "x-my-codec");
+    });
+
+    parse("text/html; charset=ISO-8859-4", [](const MediaType& m10) {
+        ASSERT_EQ(m10, MIME(Text, Html));
+        ASSERT_TRUE(m10.q().isEmpty());
+        auto charset = m10.getParam("charset");
+        ASSERT_EQ(charset.getOrElse(""), "ISO-8859-4");
+    });
+
+    parse("text/html; q=0.83; charset=ISO-8859-4", [](const MediaType& m11) {
+        ASSERT_EQ(m11, MIME(Text, Html));
+        ASSERT_EQ(m11.q().getOrElse(Q(0)), Q(83));
+        ASSERT_EQ(m11.getParam("charset").getOrElse(""), "ISO-8859-4");
+    });
 }
 
 
@@ -92,4 +131,9 @@ TEST(mime_test, invalid_parsing) {
     ASSERT_THROW(MediaType::fromString("text/plain;    q"), HttpError);
     ASSERT_THROW(MediaType::fromString("application/xhtml+xml;    q=a0.2"), HttpError);
     ASSERT_THROW(MediaType::fromString("application/xhtml+xml;  q=0.2b"), HttpError);
+
+    ASSERT_THROW(MediaType::fromString("text/html; q=0.21;"), HttpError);
+    ASSERT_THROW(MediaType::fromString("text/html; q=0.21; charset"), HttpError);
+    ASSERT_THROW(MediaType::fromString("text/html; q=0.21; charset="), HttpError);
+    ASSERT_THROW(MediaType::fromString("text/html; q=0.21; charset=ISO-8859-4;  "), HttpError);
 }