Skip to content

Commit

Permalink
xml.lite uses UTF-8 from coda-oss
Browse files Browse the repository at this point in the history
  • Loading branch information
Dan Smith committed Jul 29, 2022
1 parent 32e27fc commit af27a9a
Show file tree
Hide file tree
Showing 24 changed files with 466 additions and 1,693 deletions.
4 changes: 0 additions & 4 deletions externals/coda-oss/modules/c++/xml.lite/include/xml/lite/ContentHandler.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -95,10 +95,6 @@ class ContentHandler

virtual bool vcharacters(const void/*XMLCh*/*, size_t /*length*/) // avoid XMLCh, it's specific to Xerces
{ return false; /* continue on to existing characters()*/ } /* =0 would break existing code */
virtual bool call_vcharacters() const // =0 would break existing code
{
return false; // don't call vcharacters(const void*)
}

/*!
* Receive notification of the beginning of an element.
Expand Down
63 changes: 33 additions & 30 deletions externals/coda-oss/modules/c++/xml.lite/include/xml/lite/Document.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,9 @@

#include <assert.h>

#include <utility>
#include "coda_oss/string.h"
#include "coda_oss/memory.h"

#include "xml/lite/Element.h"
#include "xml/lite/QName.h"
Expand All @@ -54,33 +56,41 @@ namespace lite
* Use the Document to access the Element nodes contained within.
* The DocumentParser will build a tree that you can use.
*/
class Document
struct Document final
{
public:
//! Constructor
Document(Element* rootNode = nullptr, bool own = true) :
mRootNode(rootNode), mOwnRoot(own)
{
}
explicit Document(std::unique_ptr<Element>&& rootNode) : // implicitly own=true
Document(rootNode.release(), true /*own*/)
{
}

/*!
* Destroy the xml tree. This deletes the nodes if they exist
* Careful, this may delete your copy if you are not careful
*/
virtual ~Document()
~Document()
{
destroy();
}

virtual Document* clone() const
std::unique_ptr<Document>& clone(std::unique_ptr<Document>& doc) const
{
Document* doc = new Document();
doc = coda_oss::make_unique<Document>();

Element* cloneRoot = new Element();
auto cloneRoot = coda_oss::make_unique<Element>();
cloneRoot->clone(*mRootNode);
doc->setRootElement(cloneRoot);
doc->setRootElement(std::move(cloneRoot));
return doc;
}
Document* clone() const
{
std::unique_ptr<Document> doc;
return clone(doc).release();
}

/*!
* Factory-type method for creating a new Element
Expand All @@ -89,21 +99,9 @@ class Document
* \param characterData The character data (if any)
* \return A new element
*/
virtual Element *createElement(const std::string & qname,
const std::string & uri,
std::string characterData = "");
#ifndef SWIG // SWIG doesn't like unique_ptr or StringEncoding
Element* createElement(const std::string& qname,
const std::string & uri,
const std::string& characterData, StringEncoding) const;
Element* createElement(const std::string& qname,
const std::string& uri,
const coda_oss::u8string& characterData) const;
std::unique_ptr<Element> createElement(const xml::lite::QName& qname, const std::string& characterData) const;
std::unique_ptr<Element> createElement(const xml::lite::QName& qname,
const std::string& characterData, StringEncoding) const;
#endif // SWIG

Element *createElement(const std::string & qname, const std::string & uri, std::string characterData = "");
std::unique_ptr<Element> createElement(const xml::lite::QName&, const std::string& characterData) const;
std::unique_ptr<Element> createElement(const xml::lite::QName&, const coda_oss::u8string& characterData) const;

/*!
* Blanket destructor. This thing deletes everything
Expand All @@ -118,13 +116,13 @@ class Document
* \param element Element to add
* \param underThis Element to add element to
*/
virtual void insert(Element * element, Element * underThis);
void insert(Element * element, Element * underThis);

/*!
* Remove an element from the tree, starting at the root
* \param toDelete The node to delete (This DOES do deletion)
*/
virtual void remove(Element * toDelete);
void remove(Element * toDelete);

/*!
* Remove an element from the tree, starting at the second param
Expand All @@ -133,13 +131,17 @@ class Document
* be an optimization depending on the task, so I allow it to remain
* public
*/
virtual void remove(Element * toDelete, Element * fromHere);
void remove(Element * toDelete, Element * fromHere);

/*!
* Sets the internal root element
* \param element The node to set.
*/
void setRootElement(Element * element, bool own = true);
void setRootElement(std::unique_ptr<Element>&& element) // implicitly own=true
{
setRootElement(element.release(), true /*own*/);
}

/*!
* Retrieves the internal root element
Expand All @@ -151,17 +153,18 @@ class Document
mOwnRoot = false;
return mRootNode;
}

std::unique_ptr<Element>& getRootElement(std::unique_ptr<Element>& rootNode) // implicitly steal=true
{
rootNode.reset(getRootElement(true /*steal*/));
return rootNode;
}
Element *getRootElement() const
{
return mRootNode;
}

protected:
//! Copy constructor
private:
Document(const Document&);

//! Assignment operator
Document& operator=(const Document&);

//! The root node element
Expand Down
126 changes: 38 additions & 88 deletions externals/coda-oss/modules/c++/xml.lite/include/xml/lite/Element.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -27,15 +27,16 @@
#include <memory>
#include <string>
#include <new> // std::nothrow_t
#include <coda_oss/string.h>

#include <io/InputStream.h>
#include <io/OutputStream.h>
#include <str/Convert.h>
#include <str/EncodedString.h>
#include "xml/lite/XMLException.h"
#include "xml/lite/Attributes.h"
#include "xml/lite/QName.h"
#include "sys/Conf.h"
#include "coda_oss/optional.h"
#include "mem/SharedPtr.h"

/*!
Expand All @@ -59,57 +60,33 @@ namespace lite
* This class stores all of the element information about an XML
* document.
*/
class Element
struct Element final
{
Element(const std::string& qname, const std::string& uri, std::nullptr_t) :
mParent(nullptr), mName(uri, qname)
{
}

public:
//! Default constructor
Element() :
mParent(nullptr)
{
}
Element() = default;

/*!
* Constructor taking the namespace prefix and the local name
* \param qname The qname of the object
* \param uri The uri of the object
* \param characterData The character data (if any)
*/
Element(const std::string& qname, const std::string& uri = "",
const std::string& characterData = "") :
Element(qname, uri, nullptr)
explicit Element(const std::string& qname, const std::string& uri = "", const std::string& characterData = "") :
mName(uri, qname)
{
setCharacterData(characterData);
}
#ifndef SWIG // SWIG doesn't like unique_ptr or StringEncoding
Element(const std::string& qname, const std::string& uri,
const std::string& characterData, StringEncoding encoding) :
Element(qname, uri, nullptr)
{
setCharacterData(characterData, encoding);
}
Element(const std::string& qname, const std::string& uri,
const coda_oss::u8string& characterData) :
Element(qname, uri, nullptr)
Element(const xml::lite::QName& qname, const coda_oss::u8string& characterData) :
mName(qname.getName(), qname.getUri().value)
{
setCharacterData(characterData);
}

// StringEncoding is assumed based on the platform: Windows-1252 or UTF-8.
static std::unique_ptr<Element> create(const std::string& qname, const std::string& uri = "", const std::string& characterData = "");
static std::unique_ptr<Element> create(const std::string& qname, const xml::lite::Uri& uri, const std::string& characterData = "");
static std::unique_ptr<Element> create(const xml::lite::QName&, const std::string& characterData = "");
static std::unique_ptr<Element> create(const xml::lite::QName&, const coda_oss::u8string&);
// Encoding of "characterData" is always UTF-8
static std::unique_ptr<Element> createU8(const xml::lite::QName&, const std::string& characterData = "");
#endif // SWIG

//! Destructor
virtual ~Element()
~Element()
{
destroyChildren();
}
Expand All @@ -118,14 +95,14 @@ class Element
void destroyChildren();

// use clone() to duplicate an Element
#if !(defined(SWIG) || defined(SWIGPYTHON) || defined(HAVE_PYTHON_H)) // SWIG needs these
//private: // encoded as part of the C++ name mangling by some compilers
#endif
#if !(defined(SWIG) || defined(SWIGPYTHON) || defined(HAVE_PYTHON_H)) // SWIG needs these
//private: // encoded as part of the C++ name mangling by some compilers
#endif
Element(const Element&);
Element& operator=(const Element&);
#if !(defined(SWIG) || defined(SWIGPYTHON) || defined(HAVE_PYTHON_H))
public:
#endif
#if !(defined(SWIG) || defined(SWIGPYTHON) || defined(HAVE_PYTHON_H))
public:
#endif

Element(Element&&) = default;
Element& operator=(Element&&) = default;
Expand Down Expand Up @@ -290,21 +267,17 @@ class Element
* \todo Add format capability
*/
void print(io::OutputStream& stream) const;

// This is another slightly goofy routine to maintain backwards compatibility.
// XML documents must be properly (UTF-8, UTF-16 or UTF-32). The legacy
// print() routine (above) can write documents with a Windows-1252 encoding
// as the string is just copied to the output.
//
// The only valid setting for StringEncoding is Utf8; but defaulting that
// could change behavior on Windows.
void prettyPrint(io::OutputStream& stream,
const std::string& formatter = " ") const;
#ifndef SWIG // SWIG doesn't like unique_ptr or StringEncoding
void print(io::OutputStream& stream, StringEncoding /*=Utf8*/) const;
void prettyPrint(io::OutputStream& stream, StringEncoding /*=Utf8*/,

// Outputs (presumablly to the console) using the **NATIVE** encoding.
// For most XML processing, **THIS IS WRONG** as output should
// always be UTF-8. However, for displaying XML on the console in Windows,
// the native (Windows-1252) encoding will work better as "special" characters
// will be displayed.
void consoleOutput_(io::OutputStream& stream) const; // be sure OutputStream is the console, not a file
void prettyConsoleOutput_(io::OutputStream& stream, // be sure OutputStream is the console, not a file
const std::string& formatter = " ") const;
#endif // SWIG

/*!
* Determines if a child element exists
Expand All @@ -329,33 +302,21 @@ class Element
* Returns the character data of this element.
* \return the charater data
*/
std::string getCharacterData() const
{
return mCharacterData;
}
#ifndef SWIG // SWIG doesn't like unique_ptr or StringEncoding
const coda_oss::optional<StringEncoding>& getEncoding() const
{
return mEncoding;
}
const coda_oss::optional<StringEncoding>& getCharacterData(std::string& result) const
{
result = getCharacterData();
return getEncoding();
}
void getCharacterData(coda_oss::u8string& result) const;
#endif // SWIG
std::string getCharacterData() const;
coda_oss::u8string& getCharacterData(coda_oss::u8string& result) const;

/*!
* Sets the character data for this element.
* \param characters The data to add to this element
*/
void setCharacterData(const std::string& characters);
#ifndef SWIG // SWIG doesn't like unique_ptr or StringEncoding
void setCharacterData_(const std::string& characters, const StringEncoding*);
void setCharacterData(const std::string& characters, StringEncoding);
void setCharacterData(const coda_oss::u8string& characters);
#endif // SWIG
void setCharacterData(const std::string&);
void setCharacterData(coda_oss::u8string s)
{
// See Item #41 in "Effective Modern C++" by Scott Meyers.
// std::basic_string<T> is "cheap to move" and "always copied"
// into mCharacterData.
mCharacterData = std::move(s);
}

/*!
* Sets the local name for this element.
Expand Down Expand Up @@ -482,8 +443,7 @@ class Element
mParent = parent;
}

protected:

private:
void changePrefix(Element* element,
const std::string& prefix,
const std::string& uri);
Expand All @@ -492,28 +452,18 @@ class Element
const std::string& prefix,
const std::string& uri);

void depthPrint(io::OutputStream& stream, int depth,
const std::string& formatter) const;
void depthPrint(io::OutputStream& stream, StringEncoding, int depth,
const std::string& formatter) const;
void depthPrint(io::OutputStream& stream, int depth, const std::string& formatter, bool isConsoleOutput = false) const;

Element* mParent;
Element* mParent = nullptr;
//! The children of this element
std::vector<Element*> mChildren;
xml::lite::QName mName;
//! The attributes for this element
xml::lite::Attributes mAttributes;
//! The character data ...
std::string mCharacterData;

private:
// ... and how that data is encoded
coda_oss::optional<StringEncoding> mEncoding;
void depthPrint(io::OutputStream& stream, bool utf8, int depth,
const std::string& formatter) const;
coda_oss::u8string mCharacterData;
};

extern Element& add(const xml::lite::QName&, const std::string& value, Element& parent);
Element& add(const xml::lite::QName&, const std::string& value, Element& parent);

#ifndef SWIG
// The (old) version of SWIG we're using doesn't like certain C++11 features.
Expand Down
Loading

0 comments on commit af27a9a

Please sign in to comment.