Introduce class TagResolver, CoreTagResolver

include/yaml-cpp/tagresolver.h

@@ -0,0 +1,151 @@
+
+#include "yaml-cpp/dll.h"
+#include "yaml-cpp/eventhandler.h"
+
+#include <string>
+
+namespace YAML {
+struct Mark;
+
+/**
+ * TagResolver is a base class for tag resolution.
+ * It acts as an event filter: it is an EventHandler itself,
+ * and it passes events on to a client EventHandler
+ * while resolving tags in the process.
+
+ * This base class passes events unmodified.
+ * A subclass is supposed to implement tag resolution
+ * according to a particular schema,
+ * as recommended by the YAML 1.2 spec:
+ * http://yaml.org/spec/1.2/spec.html#Schema
+
+ * Note that each tag in a schema also needs to provide
+ * a canonical representation of its values.
+ * Mapping values into canonical representations
+ * is not the responsibility of TagResolver though.
+ */
+
+class YAML_CPP_API TagResolver : public EventHandler
+{
+public:
+    using string = std::string;
+
+    TagResolver(EventHandler & client): m_client(client) {}
+
+    virtual void OnDocumentStart(const Mark& mark)
+    {
+        m_client.OnDocumentStart(mark);
+    }
+
+    virtual void OnDocumentEnd()
+    {
+        m_client.OnDocumentEnd();
+    }
+
+    virtual void OnNull(const Mark& mark, anchor_t anchor)
+    {
+        m_client.OnNull(mark, anchor);
+    }
+
+    virtual void OnAlias(const Mark& mark, anchor_t anchor)
+    {
+        m_client.OnAlias(mark, anchor);
+    }
+
+    virtual void OnScalar(const Mark& mark, const std::string& tag,
+                          anchor_t anchor, const std::string& value)
+    {
+        m_client.OnScalar(mark, tag, anchor, value);
+    }
+
+    virtual void OnSequenceStart(const Mark& mark, const std::string& tag,
+                                 anchor_t anchor, EmitterStyle::value style)
+    {
+        m_client.OnSequenceStart(mark, tag, anchor, style);
+    }
+
+    virtual void OnSequenceEnd()
+    {
+        m_client.OnSequenceEnd();
+    }
+
+    virtual void OnMapStart(const Mark& mark, const std::string& tag,
+                            anchor_t anchor, EmitterStyle::value style)
+    {
+        m_client.OnMapStart(mark, tag, anchor, style);
+    }
+
+    virtual void OnMapEnd()
+    {
+        m_client.OnMapEnd();
+    }
+
+    EventHandler & m_client;
+};
+
+/**
+ * Implements tag resolution according to Core Schema,
+ * as defined by YAML 1.2 spec:
+ * http://yaml.org/spec/1.2/spec.html#id2804923
+ */
+
+class YAML_CPP_API CoreTagResolver : public TagResolver
+{
+private:
+
+    static bool IsBase10Digit(char c)
+    {
+        return c >= '0' and c <= '9';
+    }
+
+    static bool IsBase8Digit(char c)
+    {
+        return c >= '0' and c <= '7';
+    }
+
+    static bool IsBase16Digit(char c)
+    {
+        return IsBase10Digit(c) || (c >= 'a' and c <= 'f') || (c >= 'A' and c <= 'F');
+    }
+
+    static bool IsBase8Int(const string & text);
+    static bool IsBase16Int(const string & text);
+    static bool isBase10Int(const string & text);
+
+public:
+    CoreTagResolver(EventHandler & client): TagResolver(client) {}
+
+    static bool TagIsNonSpecific(const string & tag)
+    {
+        return (tag == "?" || tag == "!");
+    }
+
+
+    static bool ScalarIsNull(const string & text)
+    {
+        return (text == "null" || text == "Null" || text == "NULL" || text == "~");
+    }
+
+    static bool ScalarIsBool(const string & text)
+    {
+        return text == "true" ||
+                text == "True" ||
+                text == "TRUE" ||
+                text == "false" ||
+                text == "False" ||
+                text == "FALSE";
+    }
+
+    static bool ScalarIsInt(const string & text);
+
+    static bool ScalarIsFloat(const string & text);
+
+    virtual void OnScalar(const Mark& mark, const std::string& tag,
+                          anchor_t anchor, const std::string& value);
+    virtual void OnSequenceStart(const Mark& mark, const std::string& tag,
+                                 anchor_t anchor, EmitterStyle::value style);
+    virtual void OnMapStart(const Mark& mark, const std::string& tag,
+                            anchor_t anchor, EmitterStyle::value style);
+};
+
+}

src/tagresolver.cpp

@@ -0,0 +1,187 @@
+#include "../include/yaml-cpp/tagresolver.h"
+
+#include <vector>
+#include <algorithm>
+
+namespace YAML {
+
+bool CoreTagResolver::IsBase8Int(const string & text)
+{
+    if (text.empty())
+        return false;
+    for (string::size_type i = 0; i < text.size(); ++i)
+    {
+        if (!IsBase8Digit(text[i]))
+            return false;
+    }
+    return true;
+}
+
+bool CoreTagResolver::IsBase16Int(const string & text)
+{
+    if (text.empty())
+        return false;
+    for (string::size_type i = 0; i < text.size(); ++i)
+    {
+        if (!IsBase16Digit(text[i]))
+            return false;
+    }
+    return true;
+}
+
+bool CoreTagResolver::isBase10Int(const string & text)
+{
+    using namespace std;
+
+    if (text.empty())
+        return false;
+    string::size_type pos = 0;
+    if ( text[pos] == '-' ||
+         text[pos] == '+' )
+        pos += 1;
+    if (pos == text.size())
+        return false;
+    for (; pos < text.size(); ++pos)
+    {
+        if (!IsBase10Digit(text[pos]))
+            return false;
+    }
+    return true;
+}
+
+bool CoreTagResolver::ScalarIsInt(const string & text)
+{
+    if (text.substr(0,2) == "0o")
+    {
+        return IsBase8Int(text.substr(2));
+    }
+    else if (text.substr(0,2) == "0x")
+    {
+        return IsBase16Int(text.substr(2));
+    }
+    else
+    {
+        return isBase10Int(text);
+    }
+}
+
+bool CoreTagResolver::ScalarIsFloat(const string & text)
+{
+    static std::vector<string> inf = { ".inf", ".Inf", ".INF" };
+    static std::vector<string> nan = { ".nan", ".NaN", ".NAN" };
+
+    if (text.empty())
+        return false;
+
+    if (std::find(nan.begin(), nan.end(), text) != nan.end())
+        return true;
+
+    string::size_type pos = 0;
+
+    if ( text[0] == '-' ||
+         text[0] == '+' )
+    {
+        pos += 1;
+    }
+
+    if (pos == text.size())
+        return false;
+
+    if (std::find(inf.begin(), inf.end(), text.substr(pos)) != inf.end())
+        return true;
+
+    string::size_type intStart = pos;
+    while(pos < text.size() && IsBase10Digit(text[pos]))
+    {
+        ++pos;
+    }
+    bool hasInt = pos > intStart;
+
+    if (pos == text.size())
+        return false;
+
+    if (text[pos] != '.')
+        return false;
+    ++pos;
+
+    string::size_type fracStart = pos;
+    while(pos < text.size() && IsBase10Digit(text[pos]))
+    {
+        ++pos;
+    }
+    bool hasFrac = pos > fracStart;
+
+    if (!hasInt && !hasFrac)
+        return false;
+
+    if (pos == text.size())
+        return true;
+
+    if (text[pos] != 'e' && text[pos] != 'E')
+        return false;
+    ++pos;
+
+    if ( text[pos] == '-' || text[pos] == '+' )
+        ++pos;
+
+    string::size_type expStart = pos;
+    while(pos < text.size() && IsBase10Digit(text[pos]))
+    {
+        ++pos;
+    }
+    if (pos == expStart)
+        return false;
+    if (pos != text.size())
+        return false;
+    return true;
+}
+
+void CoreTagResolver::OnScalar(const Mark& mark, const std::string& tag,
+                      anchor_t anchor, const std::string& value)
+{
+    if (tag == "!")
+    {
+        TagResolver::OnScalar(mark, "tag:yaml.org,2002:str", anchor, value);
+    }
+    else if (tag == "?")
+    {
+        string resolved_tag;
+
+        if (ScalarIsNull(value))
+            resolved_tag = "tag:yaml.org,2002:null";
+        else if (ScalarIsBool(value))
+            resolved_tag = "tag:yaml.org,2002:bool";
+        else if (ScalarIsInt(value))
+            resolved_tag = "tag:yaml.org,2002:int";
+        else if (ScalarIsFloat(value))
+            resolved_tag = "tag:yaml.org,2002:float";
+        else
+            resolved_tag = "tag:yaml.org,2002:str";
+
+        TagResolver::OnScalar(mark, resolved_tag, anchor, value);
+    }
+    else
+    {
+        TagResolver::OnScalar(mark, tag, anchor, value);
+    }
+}
+
+void CoreTagResolver::OnSequenceStart(const Mark& mark, const std::string& tag,
+                             anchor_t anchor, EmitterStyle::value style)
+{
+    if (TagIsNonSpecific(tag))
+        TagResolver::OnSequenceStart(mark, "tag:yaml.org,2002:seq", anchor, style);
+    else
+        TagResolver::OnSequenceStart(mark, tag, anchor, style);
+}
+
+void CoreTagResolver::OnMapStart(const Mark& mark, const std::string& tag,
+                        anchor_t anchor, EmitterStyle::value style)
+{
+    if (TagIsNonSpecific(tag))
+        TagResolver::OnMapStart(mark, "tag:yaml.org,2002:map", anchor, style);
+    else
+        TagResolver::OnMapStart(mark, tag, anchor, style);
+}
+
+}