1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/tools/toolutil/xmlparser.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,245 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2004-2005, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: xmlparser.h
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2004jul21
1.17 +* created by: Andy Heninger
1.18 +*
1.19 +* Tiny XML parser using ICU and intended for use in ICU tests and in build tools.
1.20 +* Not suitable for production use. Not supported.
1.21 +* Not conformant. Not efficient.
1.22 +* But very small.
1.23 +*/
1.24 +
1.25 +#ifndef __XMLPARSER_H__
1.26 +#define __XMLPARSER_H__
1.27 +
1.28 +#include "unicode/uobject.h"
1.29 +#include "unicode/unistr.h"
1.30 +#include "unicode/regex.h"
1.31 +#include "uvector.h"
1.32 +#include "hash.h"
1.33 +
1.34 +#if !UCONFIG_NO_REGULAR_EXPRESSIONS && !UCONFIG_NO_CONVERSION
1.35 +
1.36 +enum UXMLNodeType {
1.37 + /** Node type string (text contents), stored as a UnicodeString. */
1.38 + UXML_NODE_TYPE_STRING,
1.39 + /** Node type element, stored as a UXMLElement. */
1.40 + UXML_NODE_TYPE_ELEMENT,
1.41 + UXML_NODE_TYPE_COUNT
1.42 +};
1.43 +
1.44 +U_NAMESPACE_BEGIN
1.45 +
1.46 +class UXMLParser;
1.47 +
1.48 +/**
1.49 + * This class represents an element node in a parsed XML tree.
1.50 + */
1.51 +class U_TOOLUTIL_API UXMLElement : public UObject {
1.52 +public:
1.53 + /**
1.54 + * Destructor.
1.55 + */
1.56 + virtual ~UXMLElement();
1.57 +
1.58 + /**
1.59 + * Get the tag name of this element.
1.60 + */
1.61 + const UnicodeString &getTagName() const;
1.62 + /**
1.63 + * Get the text contents of the element.
1.64 + * Append the contents of all text child nodes.
1.65 + * @param recurse If TRUE, also recursively appends the contents of all
1.66 + * text child nodes of element children.
1.67 + * @return The text contents.
1.68 + */
1.69 + UnicodeString getText(UBool recurse) const;
1.70 + /**
1.71 + * Get the number of attributes.
1.72 + */
1.73 + int32_t countAttributes() const;
1.74 + /**
1.75 + * Get the i-th attribute.
1.76 + * @param i Index of the attribute.
1.77 + * @param name Output parameter, receives the attribute name.
1.78 + * @param value Output parameter, receives the attribute value.
1.79 + * @return A pointer to the attribute value (may be &value or a pointer to an
1.80 + * internal string object), or NULL if i is out of bounds.
1.81 + */
1.82 + const UnicodeString *getAttribute(int32_t i, UnicodeString &name, UnicodeString &value) const;
1.83 + /**
1.84 + * Get the value of the attribute with the given name.
1.85 + * @param name Attribute name to be looked up.
1.86 + * @return A pointer to the attribute value, or NULL if this element
1.87 + * does not have this attribute.
1.88 + */
1.89 + const UnicodeString *getAttribute(const UnicodeString &name) const;
1.90 + /**
1.91 + * Get the number of child nodes.
1.92 + */
1.93 + int32_t countChildren() const;
1.94 + /**
1.95 + * Get the i-th child node.
1.96 + * @param i Index of the child node.
1.97 + * @param type The child node type.
1.98 + * @return A pointer to the child node object, or NULL if i is out of bounds.
1.99 + */
1.100 + const UObject *getChild(int32_t i, UXMLNodeType &type) const;
1.101 + /**
1.102 + * Get the next child element node, skipping non-element child nodes.
1.103 + * @param i Enumeration index; initialize to 0 before getting the first child element.
1.104 + * @return A pointer to the next child element, or NULL if there is none.
1.105 + */
1.106 + const UXMLElement *nextChildElement(int32_t &i) const;
1.107 + /**
1.108 + * Get the immediate child element with the given name.
1.109 + * If there are multiple child elements with this name, then return
1.110 + * the first one.
1.111 + * @param name Element name to be looked up.
1.112 + * @return A pointer to the element node, or NULL if this element
1.113 + * does not have this immediate child element.
1.114 + */
1.115 + const UXMLElement *getChildElement(const UnicodeString &name) const;
1.116 +
1.117 + /**
1.118 + * ICU "poor man's RTTI", returns a UClassID for the actual class.
1.119 + */
1.120 + virtual UClassID getDynamicClassID() const;
1.121 +
1.122 + /**
1.123 + * ICU "poor man's RTTI", returns a UClassID for this class.
1.124 + */
1.125 + static UClassID U_EXPORT2 getStaticClassID();
1.126 +
1.127 +private:
1.128 + // prevent default construction etc.
1.129 + UXMLElement();
1.130 + UXMLElement(const UXMLElement &other);
1.131 + UXMLElement &operator=(const UXMLElement &other);
1.132 +
1.133 + void appendText(UnicodeString &text, UBool recurse) const;
1.134 +
1.135 + friend class UXMLParser;
1.136 +
1.137 + UXMLElement(const UXMLParser *parser, const UnicodeString *name, UErrorCode &errorCode);
1.138 +
1.139 + const UXMLParser *fParser;
1.140 + const UnicodeString *fName; // The tag name of this element (owned by the UXMLParser)
1.141 + UnicodeString fContent; // The text content of this node. All element content is
1.142 + // concatenated even when there are intervening nested elements
1.143 + // (which doesn't happen with most xml files we care about)
1.144 + // Sections of content containing only white space are dropped,
1.145 + // which gets rid the bogus white space content from
1.146 + // elements which are primarily containers for nested elements.
1.147 + UVector fAttNames; // A vector containing the names of this element's attributes
1.148 + // The names are UnicodeString objects, owned by the UXMLParser.
1.149 + UVector fAttValues; // A vector containing the attribute values for
1.150 + // this element's attributes. The order is the same
1.151 + // as that of the attribute name vector.
1.152 +
1.153 + UVector fChildren; // The child nodes of this element (a Vector)
1.154 +
1.155 + UXMLElement *fParent; // A pointer to the parent element of this element.
1.156 +};
1.157 +
1.158 +/**
1.159 + * A simple XML parser; it is neither efficient nor conformant and only useful for
1.160 + * restricted types of XML documents.
1.161 + *
1.162 + * The parse methods parse whole documents and return the parse trees via their
1.163 + * root elements.
1.164 + */
1.165 +class U_TOOLUTIL_API UXMLParser : public UObject {
1.166 +public:
1.167 + /**
1.168 + * Create an XML parser.
1.169 + */
1.170 + static UXMLParser *createParser(UErrorCode &errorCode);
1.171 + /**
1.172 + * Destructor.
1.173 + */
1.174 + virtual ~UXMLParser();
1.175 +
1.176 + /**
1.177 + * Parse an XML document, create the entire document tree, and
1.178 + * return a pointer to the root element of the parsed tree.
1.179 + * The caller must delete the element.
1.180 + */
1.181 + UXMLElement *parse(const UnicodeString &src, UErrorCode &errorCode);
1.182 + /**
1.183 + * Parse an XML file, create the entire document tree, and
1.184 + * return a pointer to the root element of the parsed tree.
1.185 + * The caller must delete the element.
1.186 + */
1.187 + UXMLElement *parseFile(const char *filename, UErrorCode &errorCode);
1.188 +
1.189 + /**
1.190 + * ICU "poor man's RTTI", returns a UClassID for the actual class.
1.191 + */
1.192 + virtual UClassID getDynamicClassID() const;
1.193 +
1.194 + /**
1.195 + * ICU "poor man's RTTI", returns a UClassID for this class.
1.196 + */
1.197 + static UClassID U_EXPORT2 getStaticClassID();
1.198 +
1.199 +private:
1.200 + // prevent default construction etc.
1.201 + UXMLParser();
1.202 + UXMLParser(const UXMLParser &other);
1.203 + UXMLParser &operator=(const UXMLParser &other);
1.204 +
1.205 + // constructor
1.206 + UXMLParser(UErrorCode &status);
1.207 +
1.208 + void parseMisc(UErrorCode &status);
1.209 + UXMLElement *createElement(RegexMatcher &mEl, UErrorCode &status);
1.210 + void error(const char *message, UErrorCode &status);
1.211 + UnicodeString scanContent(UErrorCode &status);
1.212 + void replaceCharRefs(UnicodeString &s, UErrorCode &status);
1.213 +
1.214 + const UnicodeString *intern(const UnicodeString &s, UErrorCode &errorCode);
1.215 +public:
1.216 + // public for UXMLElement only
1.217 + const UnicodeString *findName(const UnicodeString &s) const;
1.218 +private:
1.219 +
1.220 + // There is one ICU regex matcher for each of the major XML syntax items
1.221 + // that are recognized.
1.222 + RegexMatcher mXMLDecl;
1.223 + RegexMatcher mXMLComment;
1.224 + RegexMatcher mXMLSP;
1.225 + RegexMatcher mXMLDoctype;
1.226 + RegexMatcher mXMLPI;
1.227 + RegexMatcher mXMLElemStart;
1.228 + RegexMatcher mXMLElemEnd;
1.229 + RegexMatcher mXMLElemEmpty;
1.230 + RegexMatcher mXMLCharData;
1.231 + RegexMatcher mAttrValue;
1.232 + RegexMatcher mAttrNormalizer;
1.233 + RegexMatcher mNewLineNormalizer;
1.234 + RegexMatcher mAmps;
1.235 +
1.236 + Hashtable fNames; // interned element/attribute name strings
1.237 + UStack fElementStack; // Stack holds the parent elements when nested
1.238 + // elements are being parsed. All items on this
1.239 + // stack are of type UXMLElement.
1.240 + int32_t fPos; // String index of the current scan position in
1.241 + // xml source (in fSrc).
1.242 + UnicodeString fOneLF;
1.243 +};
1.244 +
1.245 +U_NAMESPACE_END
1.246 +#endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */
1.247 +
1.248 +#endif
