| 1 |
|
|---|
| 2 | /*
|
|---|
| 3 | *@@sourcefile xml.h:
|
|---|
| 4 | * header file for xml.c (XML parsing).
|
|---|
| 5 | *
|
|---|
| 6 | * See remarks there.
|
|---|
| 7 | *
|
|---|
| 8 | *@@added V0.9.6 (2000-10-29) [umoeller]
|
|---|
| 9 | *@@include #include <os2.h>
|
|---|
| 10 | *@@include #include "expat\expat.h" // must come before xml.h
|
|---|
| 11 | *@@include #include "helpers\linklist.h"
|
|---|
| 12 | *@@include #include "helpers\tree.h"
|
|---|
| 13 | *@@include #include "helpers\xstring.h"
|
|---|
| 14 | *@@include #include "helpers\xml.h"
|
|---|
| 15 | */
|
|---|
| 16 |
|
|---|
| 17 | #if __cplusplus
|
|---|
| 18 | extern "C" {
|
|---|
| 19 | #endif
|
|---|
| 20 |
|
|---|
| 21 | #ifndef XML_HEADER_INCLUDED
|
|---|
| 22 | #define XML_HEADER_INCLUDED
|
|---|
| 23 |
|
|---|
| 24 | // define some basic things to make this work even with standard C
|
|---|
| 25 | #if (!defined OS2_INCLUDED) && (!defined _OS2_H) && (!defined __SIMPLES_DEFINED) // changed V0.9.0 (99-10-22) [umoeller]
|
|---|
| 26 | typedef unsigned long BOOL;
|
|---|
| 27 | typedef unsigned long ULONG;
|
|---|
| 28 | typedef unsigned char *PSZ;
|
|---|
| 29 | #define TRUE (BOOL)1
|
|---|
| 30 | #define FALSE (BOOL)0
|
|---|
| 31 |
|
|---|
| 32 | #ifdef __IBMCPP__ // added V0.9.0 (99-10-22) [umoeller]
|
|---|
| 33 | #define APIENTRY _System
|
|---|
| 34 | #endif
|
|---|
| 35 |
|
|---|
| 36 | #define __SIMPLES_DEFINED
|
|---|
| 37 | #endif
|
|---|
| 38 |
|
|---|
| 39 | /* ******************************************************************
|
|---|
| 40 | *
|
|---|
| 41 | * Error handling
|
|---|
| 42 | *
|
|---|
| 43 | ********************************************************************/
|
|---|
| 44 |
|
|---|
| 45 | // ERROR_XML_FIRST is defined in expat\expat.h;
|
|---|
| 46 | // the range up to ERROR_XML_FIRST + 24 is used
|
|---|
| 47 | // by expat
|
|---|
| 48 |
|
|---|
| 49 | // START MATCHING ERROR MESSAGES (xmlDescribeError)
|
|---|
| 50 | // validity errors:
|
|---|
| 51 | #define ERROR_DOM_UNDECLARED_ELEMENT (ERROR_XML_FIRST + 24)
|
|---|
| 52 | // invalidity: element is undeclared
|
|---|
| 53 | #define ERROR_DOM_ROOT_ELEMENT_MISNAMED (ERROR_XML_FIRST + 25)
|
|---|
| 54 | #define ERROR_DOM_INVALID_ROOT_ELEMENT (ERROR_XML_FIRST + 26)
|
|---|
| 55 | #define ERROR_DOM_INVALID_SUBELEMENT (ERROR_XML_FIRST + 27)
|
|---|
| 56 | // subelement may not appear in its parent element
|
|---|
| 57 | #define ERROR_DOM_DUPLICATE_ELEMENT_DECL (ERROR_XML_FIRST + 28)
|
|---|
| 58 | // more than one declaration for an element type
|
|---|
| 59 | #define ERROR_DOM_DUPLICATE_ATTRIBUTE_DECL (ERROR_XML_FIRST + 29)
|
|---|
| 60 | // more than one declaration for an attribute type
|
|---|
| 61 | #define ERROR_DOM_UNDECLARED_ATTRIBUTE (ERROR_XML_FIRST + 30)
|
|---|
| 62 | #define ERROR_ELEMENT_CANNOT_HAVE_CONTENT (ERROR_XML_FIRST + 31)
|
|---|
| 63 | // element was declared "empty" and contains text anyway,
|
|---|
| 64 | // or was declared "children" and contains something other
|
|---|
| 65 | // than whitespace
|
|---|
| 66 | #define ERROR_DOM_INVALID_ATTRIB_VALUE (ERROR_XML_FIRST + 32)
|
|---|
| 67 | #define ERROR_DOM_REQUIRED_ATTRIBUTE_MISSING (ERROR_XML_FIRST + 33)
|
|---|
| 68 | #define ERROR_DOM_SUBELEMENT_IN_EMPTY_ELEMENT (ERROR_XML_FIRST + 34)
|
|---|
| 69 | // END MATCHING ERROR MESSAGES (xmlDescribeError)
|
|---|
| 70 |
|
|---|
| 71 | // error categories:
|
|---|
| 72 | #define ERROR_DOM_PARSING (ERROR_XML_FIRST + 35)
|
|---|
| 73 | #define ERROR_DOM_VALIDITY (ERROR_XML_FIRST + 36)
|
|---|
| 74 |
|
|---|
| 75 | // additional DOM errors
|
|---|
| 76 | #define ERROR_DOM_NODETYPE_NOT_SUPPORTED (ERROR_XML_FIRST + 37)
|
|---|
| 77 | // invalid node type in xmlCreateDomNode
|
|---|
| 78 | #define ERROR_DOM_NO_DOCUMENT (ERROR_XML_FIRST + 38)
|
|---|
| 79 | // cannot find document node
|
|---|
| 80 | #define ERROR_DOM_NO_ELEMENT (ERROR_XML_FIRST + 39)
|
|---|
| 81 | #define ERROR_DOM_DUPLICATE_DOCTYPE (ERROR_XML_FIRST + 40)
|
|---|
| 82 | #define ERROR_DOM_DOCTYPE_ROOT_NAMES_MISMATCH (ERROR_XML_FIRST + 41)
|
|---|
| 83 | // DOCTYPE is given and root element name does not match doctype name
|
|---|
| 84 | #define ERROR_DOM_INTEGRITY (ERROR_XML_FIRST + 42)
|
|---|
| 85 | #define ERROR_DOM_DUPLICATE_ATTRIBUTE (ERROR_XML_FIRST + 43)
|
|---|
| 86 |
|
|---|
| 87 | #define ERROR_DOM_VALIDATE_INVALID_ELEMENT (ERROR_XML_FIRST + 44)
|
|---|
| 88 | #define ERROR_DOM_ELEMENT_DECL_OUTSIDE_DOCTYPE (ERROR_XML_FIRST + 45)
|
|---|
| 89 | #define ERROR_DOM_ATTLIST_DECL_OUTSIDE_DOCTYPE (ERROR_XML_FIRST + 46)
|
|---|
| 90 |
|
|---|
| 91 | #define ERROR_DOM_INCOMPLETE_ENCODING_MAP (ERROR_XML_FIRST + 47)
|
|---|
| 92 | // callback to UnknownEncodingHandler has provided
|
|---|
| 93 | // an incomplete encoding map V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 94 |
|
|---|
| 95 | #define ERROR_DOM_INVALID_EXTERNAL_HANDLER (ERROR_XML_FIRST + 48)
|
|---|
| 96 |
|
|---|
| 97 | #define ERROR_XML_LAST (ERROR_XML_FIRST + 48)
|
|---|
| 98 |
|
|---|
| 99 | const char* xmlDescribeError(int code);
|
|---|
| 100 |
|
|---|
| 101 | /* ******************************************************************
|
|---|
| 102 | *
|
|---|
| 103 | * Most basic node management
|
|---|
| 104 | *
|
|---|
| 105 | ********************************************************************/
|
|---|
| 106 |
|
|---|
| 107 | // content model node types:
|
|---|
| 108 | typedef enum _NODEBASETYPE
|
|---|
| 109 | {
|
|---|
| 110 | TYPE_UNKNOWN,
|
|---|
| 111 |
|
|---|
| 112 | DOMNODE_ELEMENT, // node is a DOM ELEMENT
|
|---|
| 113 | DOMNODE_ATTRIBUTE, // node is a DOM ATTRIBUTE
|
|---|
| 114 | DOMNODE_TEXT, // node is a DOM TEXT node
|
|---|
| 115 | // DOMNODE_CDATA_SECTION 4
|
|---|
| 116 | // DOMNODE_ENTITY_REFERENCE 5
|
|---|
| 117 | // DOMNODE_ENTITY 6
|
|---|
| 118 | DOMNODE_PROCESSING_INSTRUCTION, // node is a DOM PI
|
|---|
| 119 | DOMNODE_COMMENT, // node is a DOM COMMENT
|
|---|
| 120 | DOMNODE_DOCUMENT, // node is a DOM document
|
|---|
| 121 | DOMNODE_DOCUMENT_TYPE, // node is a DOM DOCUMENTTYPE
|
|---|
| 122 | // #define DOMNODE_DOCUMENT_FRAGMENT 11
|
|---|
| 123 | // #define DOMNODE_NOTATION 12
|
|---|
| 124 |
|
|---|
| 125 | // the following are all CMELEMENTPARTICLE nodes
|
|---|
| 126 | ELEMENTPARTICLE_EMPTY,
|
|---|
| 127 | ELEMENTPARTICLE_ANY,
|
|---|
| 128 | ELEMENTPARTICLE_MIXED,
|
|---|
| 129 | ELEMENTPARTICLE_CHOICE,
|
|---|
| 130 | ELEMENTPARTICLE_SEQ,
|
|---|
| 131 | ELEMENTPARTICLE_NAME,
|
|---|
| 132 |
|
|---|
| 133 | ATTRIBUTE_DECLARATION_BASE, // node is a CMATTRIBUTEDECLBASE
|
|---|
| 134 | ATTRIBUTE_DECLARATION, // node is a CMATTRIBUTEDECL
|
|---|
| 135 | ATTRIBUTE_DECLARATION_ENUM // node is a plain NODEBASE, part of an attr value enum
|
|---|
| 136 | } NODEBASETYPE;
|
|---|
| 137 |
|
|---|
| 138 | /*
|
|---|
| 139 | *@@ NODEBASE:
|
|---|
| 140 | * root class of all nodes used in the
|
|---|
| 141 | * DOM tree.
|
|---|
| 142 | *
|
|---|
| 143 | * The first item is a TREE to make this insertable
|
|---|
| 144 | * into string maps via the functions in tree.c.
|
|---|
| 145 | *
|
|---|
| 146 | * A NODEBASE is also the first member of a DOMNODE
|
|---|
| 147 | * so this is effectively a simulation of inheritance
|
|---|
| 148 | * in plain C... DOMNODE inherits from NODEBASE,
|
|---|
| 149 | * and some other types inherit from DOMNODE.
|
|---|
| 150 | *
|
|---|
| 151 | * This contains the node name, which is also used
|
|---|
| 152 | * by the DOMNODE's (e.g. for element names).
|
|---|
| 153 | * However, only DOMNODE defines the node value
|
|---|
| 154 | * string.
|
|---|
| 155 | *
|
|---|
| 156 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 157 | */
|
|---|
| 158 |
|
|---|
| 159 | typedef struct _NODEBASE
|
|---|
| 160 | {
|
|---|
| 161 | TREE Tree; // tree.c
|
|---|
| 162 | // ulKey simply points to the strNodeName
|
|---|
| 163 | // member always V0.9.13 (2001-06-27) [umoeller]
|
|---|
| 164 |
|
|---|
| 165 | NODEBASETYPE ulNodeType; // class type; this is precious,
|
|---|
| 166 | // all xml* functions make assumptions
|
|---|
| 167 | // from this value
|
|---|
| 168 |
|
|---|
| 169 | XSTRING strNodeName;
|
|---|
| 170 | // node name;
|
|---|
| 171 | // -- for the various DOMNODE_* items, see _DOMNODE;
|
|---|
| 172 | // -- for CMELEMENTPARTICLE nodes, this is the particle's name
|
|---|
| 173 | // -- for CMELEMENTDECLNODE nodes, element name being declared
|
|---|
| 174 | // -- for CMATTRIBUTEDECLBASE nodes, name of element to which this
|
|---|
| 175 | // attrib decl belongs
|
|---|
| 176 | // -- for CMATTRIBUTEDECL nodes, name of attribute;
|
|---|
| 177 | // -- for ATTRIBUTE_DECLARATION_ENUM, attribute value in the
|
|---|
| 178 | // possible values list.
|
|---|
| 179 |
|
|---|
| 180 | } NODEBASE, *PNODEBASE;
|
|---|
| 181 |
|
|---|
| 182 | /* ******************************************************************
|
|---|
| 183 | *
|
|---|
| 184 | * DOM level 1
|
|---|
| 185 | *
|
|---|
| 186 | ********************************************************************/
|
|---|
| 187 |
|
|---|
| 188 | /*
|
|---|
| 189 | *@@ DOMNODE:
|
|---|
| 190 | * this represents one @DOM node in an @XML document.
|
|---|
| 191 | *
|
|---|
| 192 | * The document itself is represented by a node with the
|
|---|
| 193 | * DOMNODE_DOCUMENT type, which is the root of a tree as
|
|---|
| 194 | * shown with xmlParse.
|
|---|
| 195 | *
|
|---|
| 196 | * The contents of the members vary according
|
|---|
| 197 | * to ulNodeType (0 specifies that the field does not
|
|---|
| 198 | * apply to that type).
|
|---|
| 199 | *
|
|---|
| 200 | * The first member of a DOMNODE is a NODEBASE to allow
|
|---|
| 201 | * inserting these things in a tree. NODEBASE.ulNodeType
|
|---|
| 202 | * _always_ specifies the various types that are using
|
|---|
| 203 | * that structure to allow for type-safety (if we watch out).
|
|---|
| 204 | * This is for faking inheritance.
|
|---|
| 205 | *
|
|---|
| 206 | * Note that we also implement specialized sub-structures of
|
|---|
| 207 | * DOMNODE, whose first member is the DOMNODE (and therefore
|
|---|
| 208 | * a NODEBASE as well):
|
|---|
| 209 | *
|
|---|
| 210 | * -- DOCUMENT nodes are given a _DOMDOCUMENTNODE structure.
|
|---|
| 211 | *
|
|---|
| 212 | * -- DOCTYPE nodes are given a _DOMDOCTYPENODE structure.
|
|---|
| 213 | *
|
|---|
| 214 | * Overview of member fields usage:
|
|---|
| 215 | +
|
|---|
| 216 | + ulNodeType | strNodeName | strNodeValue | llChildren | AttributesMap
|
|---|
| 217 | + (NODEBASE) | (NODEBASE) | (DOMNODE) | (DOMNODE) | (DOMNODE)
|
|---|
| 218 | + =======================================================================
|
|---|
| 219 | + | | | |
|
|---|
| 220 | + DOCUMENT | name from | 0 | 1 root | 0
|
|---|
| 221 | + | DOCTYPE or | | ELEMENT |
|
|---|
| 222 | + | NULL | | |
|
|---|
| 223 | + | | | |
|
|---|
| 224 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 225 | + | | | |
|
|---|
| 226 | + ELEMENT | tag name | 0 | ELEMENT | ATTRIBUTE
|
|---|
| 227 | + | | | nodes | nodes
|
|---|
| 228 | + | | | |
|
|---|
| 229 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 230 | + | | | |
|
|---|
| 231 | + ATTRIBUTE | attribute | attribute | 0 | 0
|
|---|
| 232 | + | name | value | |
|
|---|
| 233 | + | | | |
|
|---|
| 234 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 235 | + | | | |
|
|---|
| 236 | + TEXT | 0 | text | 0 | 0
|
|---|
| 237 | + | | contents | |
|
|---|
| 238 | + | | | |
|
|---|
| 239 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 240 | + | | | |
|
|---|
| 241 | + COMMENT | 0 | comment | 0 | 0
|
|---|
| 242 | + | | contents | |
|
|---|
| 243 | + | | | |
|
|---|
| 244 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 245 | + | | | |
|
|---|
| 246 | + PI | PI target | PI data | 0 | 0
|
|---|
| 247 | + | | | |
|
|---|
| 248 | + | | | |
|
|---|
| 249 | + --------------+-------------+--------------+------------+--------------
|
|---|
| 250 | + | | | |
|
|---|
| 251 | + DOCTYPE | doctype | | 0 | 0
|
|---|
| 252 | + | name | | |
|
|---|
| 253 | + | | | |
|
|---|
| 254 | +
|
|---|
| 255 | * The xwphelpers implementation does not implement CDATA sections,
|
|---|
| 256 | * for which we have no need because @expat properly converts these
|
|---|
| 257 | * into plain @content.
|
|---|
| 258 | *
|
|---|
| 259 | * In addition, W3C DOM specifies that the "node name" members contain
|
|---|
| 260 | * "#document", "#text", and "#comment" strings for DOCUMENT,
|
|---|
| 261 | * TEXT, and COMMENT nodes, respectively. I see no point in this other
|
|---|
| 262 | * than consuming memory, so these fields are empty with this implementation.
|
|---|
| 263 | */
|
|---|
| 264 |
|
|---|
| 265 | typedef struct _DOMNODE
|
|---|
| 266 | {
|
|---|
| 267 | NODEBASE NodeBase;
|
|---|
| 268 |
|
|---|
| 269 | PXSTRING pstrNodeValue; // ptr is NULL if none
|
|---|
| 270 |
|
|---|
| 271 | struct _DOMNODE *pParentNode;
|
|---|
| 272 | // the parent node;
|
|---|
| 273 | // NULL for DOCUMENT, DOCUMENT_FRAGMENT.
|
|---|
| 274 | // The DOM spec says that attribs have no parent,
|
|---|
| 275 | // but even though the attribute is not added to
|
|---|
| 276 | // the "children" list of an element (but to the
|
|---|
| 277 | // attributes map instead), we specify the element
|
|---|
| 278 | // as the attribute's parent here.
|
|---|
| 279 |
|
|---|
| 280 | struct _DOMNODE *pDocumentNode;
|
|---|
| 281 | // the document node, unless this is a DOCUMENT in itself.
|
|---|
| 282 |
|
|---|
| 283 | LINKLIST llChildren; // of DOMNODE* pointers, no auto-free
|
|---|
| 284 |
|
|---|
| 285 | TREE *AttributesMap; // of DOMNODE* pointers
|
|---|
| 286 |
|
|---|
| 287 | } DOMNODE, *PDOMNODE;
|
|---|
| 288 |
|
|---|
| 289 | /*
|
|---|
| 290 | *@@ DOMDOCTYPENODE:
|
|---|
| 291 | * specific _DOMNODE replacement structure which
|
|---|
| 292 | * is used for DOCTYPE nodes.
|
|---|
| 293 | *
|
|---|
| 294 | * The DOMDOCTYPENODE is special (other than having
|
|---|
| 295 | * extra fields) in that it is stored both in
|
|---|
| 296 | * the document node's children list and in its
|
|---|
| 297 | * pDocType field.
|
|---|
| 298 | *
|
|---|
| 299 | * DOMNODE.pstrNodeName is set to the name in the
|
|---|
| 300 | * DOCTYPE statement by xmlCreateDocumentTypeNode,
|
|---|
| 301 | * or is NULL if there's no DOCTYPE.
|
|---|
| 302 | *
|
|---|
| 303 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 304 | */
|
|---|
| 305 |
|
|---|
| 306 | typedef struct _DOMDOCTYPENODE
|
|---|
| 307 | {
|
|---|
| 308 | DOMNODE DomNode;
|
|---|
| 309 |
|
|---|
| 310 | XSTRING strPublicID;
|
|---|
| 311 | XSTRING strSystemID;
|
|---|
| 312 |
|
|---|
| 313 | BOOL fHasInternalSubset;
|
|---|
| 314 |
|
|---|
| 315 | TREE *ElementDeclsTree;
|
|---|
| 316 | // tree with pointers to _CMELEMENTDECLNODE nodes
|
|---|
| 317 |
|
|---|
| 318 | TREE *AttribDeclBasesTree;
|
|---|
| 319 | // tree with pointers to _CMATTRIBUTEDECLBASE nodes
|
|---|
| 320 |
|
|---|
| 321 | } DOMDOCTYPENODE, *PDOMDOCTYPENODE;
|
|---|
| 322 |
|
|---|
| 323 | /*
|
|---|
| 324 | *@@ DOMDOCUMENTNODE:
|
|---|
| 325 | * specific _DOMNODE replacement structure which
|
|---|
| 326 | * is used for DOCUMENT nodes.
|
|---|
| 327 | *
|
|---|
| 328 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 329 | */
|
|---|
| 330 |
|
|---|
| 331 | typedef struct _DOMDOCUMENTNODE
|
|---|
| 332 | {
|
|---|
| 333 | DOMNODE DomNode;
|
|---|
| 334 |
|
|---|
| 335 | PDOMDOCTYPENODE pDocType;
|
|---|
| 336 | // != NULL if DOCTYPE was found
|
|---|
| 337 |
|
|---|
| 338 | } DOMDOCUMENTNODE, *PDOMDOCUMENTNODE;
|
|---|
| 339 |
|
|---|
| 340 | APIRET xmlCreateDomNode(PDOMNODE pParentNode,
|
|---|
| 341 | NODEBASETYPE ulNodeType,
|
|---|
| 342 | const char *pcszNodeName,
|
|---|
| 343 | ULONG ulNodeNameLength,
|
|---|
| 344 | PDOMNODE *ppNew);
|
|---|
| 345 |
|
|---|
| 346 | VOID xmlDeleteNode(PNODEBASE pNode);
|
|---|
| 347 |
|
|---|
| 348 | /* ******************************************************************
|
|---|
| 349 | *
|
|---|
| 350 | * Specific DOM node constructors
|
|---|
| 351 | *
|
|---|
| 352 | ********************************************************************/
|
|---|
| 353 |
|
|---|
| 354 | APIRET xmlCreateElementNode(PDOMNODE pParent,
|
|---|
| 355 | const char *pcszElement,
|
|---|
| 356 | PDOMNODE *ppNew);
|
|---|
| 357 |
|
|---|
| 358 | APIRET xmlCreateAttributeNode(PDOMNODE pElement,
|
|---|
| 359 | const char *pcszName,
|
|---|
| 360 | const char *pcszValue,
|
|---|
| 361 | PDOMNODE *ppNew);
|
|---|
| 362 |
|
|---|
| 363 | APIRET xmlCreateTextNode(PDOMNODE pParent,
|
|---|
| 364 | const char *pcszText,
|
|---|
| 365 | ULONG ulLength,
|
|---|
| 366 | PDOMNODE *ppNew);
|
|---|
| 367 |
|
|---|
| 368 | APIRET xmlCreateCommentNode(PDOMNODE pParent,
|
|---|
| 369 | const char *pcszText,
|
|---|
| 370 | PDOMNODE *ppNew);
|
|---|
| 371 |
|
|---|
| 372 | APIRET xmlCreatePINode(PDOMNODE pParent,
|
|---|
| 373 | const char *pcszTarget,
|
|---|
| 374 | const char *pcszData,
|
|---|
| 375 | PDOMNODE *ppNew);
|
|---|
| 376 |
|
|---|
| 377 | APIRET xmlCreateDocumentTypeNode(PDOMDOCUMENTNODE pDocumentNode,
|
|---|
| 378 | const char *pcszDoctypeName,
|
|---|
| 379 | const char *pcszSysid,
|
|---|
| 380 | const char *pcszPubid,
|
|---|
| 381 | int fHasInternalSubset,
|
|---|
| 382 | PDOMDOCTYPENODE *ppNew);
|
|---|
| 383 |
|
|---|
| 384 | /* ******************************************************************
|
|---|
| 385 | *
|
|---|
| 386 | * DOM level 3 content models
|
|---|
| 387 | *
|
|---|
| 388 | ********************************************************************/
|
|---|
| 389 |
|
|---|
| 390 | // data types (XML schemes):
|
|---|
| 391 | #define STRING_DATATYPE 1
|
|---|
| 392 | #define BOOLEAN_DATATYPE 2
|
|---|
| 393 | #define FLOAT_DATATYPE 3
|
|---|
| 394 | #define DOUBLE_DATATYPE 4
|
|---|
| 395 | #define LONG_DATATYPE 5
|
|---|
| 396 | #define INT_DATATYPE 6
|
|---|
| 397 | #define SHORT_DATATYPE 7
|
|---|
| 398 | #define BYTE_DATATYPE 8
|
|---|
| 399 |
|
|---|
| 400 | /*
|
|---|
| 401 | *@@ CMELEMENTPARTICLE:
|
|---|
| 402 | * element declaration particle in a
|
|---|
| 403 | * _CMELEMENTDECLNODE.
|
|---|
| 404 | *
|
|---|
| 405 | * One of these structures is a full
|
|---|
| 406 | * (non-pointer) member in _CMELEMENTDECLNODE.
|
|---|
| 407 | * This struct in turn has a linked list with
|
|---|
| 408 | * possible subnodes. See _CMELEMENTDECLNODE.
|
|---|
| 409 | *
|
|---|
| 410 | *@@added V0.9.9 (2001-02-16) [umoeller]
|
|---|
| 411 | */
|
|---|
| 412 |
|
|---|
| 413 | typedef struct _CMELEMENTPARTICLE
|
|---|
| 414 | {
|
|---|
| 415 | NODEBASE NodeBase; // has TREE* as first item in turn
|
|---|
| 416 | // NODEBASE.ulNodeType may be one of these:
|
|---|
| 417 | // -- ELEMENTPARTICLE_EMPTY:
|
|---|
| 418 | // ulRepeater will be XML_CQUANT_NONE, rest is NULL
|
|---|
| 419 | // -- ELEMENTPARTICLE_ANY:
|
|---|
| 420 | // ulRepeater will be XML_CQUANT_NONE, rest is NULL
|
|---|
| 421 | // -- ELEMENTPARTICLE_MIXED:
|
|---|
| 422 | // mixed content (with PCDATA); if the list contains
|
|---|
| 423 | // something, the element may have PCDATA and sub-elements
|
|---|
| 424 | // mixed
|
|---|
| 425 | // -- ELEMENTPARTICLE_CHOICE:
|
|---|
| 426 | // list is a choicelist
|
|---|
| 427 | // -- ELEMENTPARTICLE_SEQ:
|
|---|
| 428 | // list is a seqlist
|
|---|
| 429 | // -- ELEMENTPARTICLE_NAME:
|
|---|
| 430 | // used for terminal particles in a parent particle's
|
|---|
| 431 | // list, which finally specifies the name of a sub-particle.
|
|---|
| 432 | // This can never appear in a root particle.
|
|---|
| 433 |
|
|---|
| 434 | ULONG ulRepeater;
|
|---|
| 435 | // one of:
|
|---|
| 436 | // -- XML_CQUANT_NONE --> all fields below are NULL
|
|---|
| 437 | // -- XML_CQUANT_OPT, question mark
|
|---|
| 438 | // -- XML_CQUANT_REP, asterisk
|
|---|
| 439 | // -- XML_CQUANT_PLUS plus sign
|
|---|
| 440 |
|
|---|
| 441 | struct _CMELEMENTPARTICLE *pParentParticle; // or NULL if this is in the
|
|---|
| 442 | // CMELEMENTDECLNODE
|
|---|
| 443 |
|
|---|
| 444 | PLINKLIST pllSubNodes;
|
|---|
| 445 | // linked list of sub-CMELEMENTPARTICLE structs
|
|---|
| 446 | // (for mixed, choice, seq types);
|
|---|
| 447 | // if NULL, there's no sub-CMELEMENTPARTICLE
|
|---|
| 448 |
|
|---|
| 449 | } CMELEMENTPARTICLE, *PCMELEMENTPARTICLE;
|
|---|
| 450 |
|
|---|
| 451 | /*
|
|---|
| 452 | *@@ CMELEMENTDECLNODE:
|
|---|
| 453 | * representation of an @element_declaration within a
|
|---|
| 454 | * _DOMDOCTYPENODE (a document @DTD).
|
|---|
| 455 | *
|
|---|
| 456 | * This is complicated because element declarations
|
|---|
| 457 | * are complicated with nested lists and content
|
|---|
| 458 | * particles. For this, we introduce the representation
|
|---|
| 459 | * of a _CMELEMENTPARTICLE, which is contained in the
|
|---|
| 460 | * "Particle" member.
|
|---|
| 461 | *
|
|---|
| 462 | * For minimal memory consumption, the _CMELEMENTDECLNODE
|
|---|
| 463 | * is an _CMELEMENTPARTICLE with extra fields, while the
|
|---|
| 464 | * list in _CMELEMENTPARTICLE points to plain
|
|---|
| 465 | * _CMELEMENTPARTICLE structs only.
|
|---|
| 466 | *
|
|---|
| 467 | * For the "root" element declaration in the DTD,
|
|---|
| 468 | * Particle.NODEBASE.ulNodeType will always be one of the following:
|
|---|
| 469 | *
|
|---|
| 470 | * -- ELEMENTPARTICLE_EMPTY: element must be empty.
|
|---|
| 471 | *
|
|---|
| 472 | * -- ELEMENTPARTICLE_ANY: element can have any content.
|
|---|
| 473 | *
|
|---|
| 474 | * -- ELEMENTPARTICLE_CHOICE: _CMELEMENTPARTICLE has a choicelist with
|
|---|
| 475 | * more _CMELEMENTPARTICLE structs.
|
|---|
| 476 | *
|
|---|
| 477 | * -- ELEMENTPARTICLE_SEQ: _CMELEMENTPARTICLE has a seqlist with
|
|---|
| 478 | * more _CMELEMENTPARTICLE structs.
|
|---|
| 479 | *
|
|---|
| 480 | * -- ELEMENTPARTICLE_MIXED: element can have mixed content including #PCDATA.
|
|---|
| 481 | * If there is no content particle list, then the element may
|
|---|
| 482 | * ONLY have PCDATA. If there's a content particle list, then the
|
|---|
| 483 | * element may have both sub-elements and PCDATA. Oh my.
|
|---|
| 484 | *
|
|---|
| 485 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 486 | */
|
|---|
| 487 |
|
|---|
| 488 | typedef struct _CMELEMENTDECLNODE
|
|---|
| 489 | {
|
|---|
| 490 | CMELEMENTPARTICLE Particle;
|
|---|
| 491 | // root particle for this element decl; this may contain
|
|---|
| 492 | // sub-particles...
|
|---|
| 493 | // this has a NODEBASE as first member, which has TREE* as
|
|---|
| 494 | // first item in turn
|
|---|
| 495 |
|
|---|
| 496 | TREE *ParticleNamesTree;
|
|---|
| 497 | // tree sorted by element names with all sub-particles,
|
|---|
| 498 | // no matter how deeply nested; this is just for quickly
|
|---|
| 499 | // checking if an element name is allowed as a sub-element
|
|---|
| 500 | // at all. Tree items are _CMELEMENTPARTICLE nodes.
|
|---|
| 501 |
|
|---|
| 502 | } CMELEMENTDECLNODE, *PCMELEMENTDECLNODE;
|
|---|
| 503 |
|
|---|
| 504 | typedef enum _ATTRIBCONSTRAINT
|
|---|
| 505 | {
|
|---|
| 506 | CMAT_IMPLIED,
|
|---|
| 507 | CMAT_REQUIRED,
|
|---|
| 508 | CMAT_DEFAULT_VALUE,
|
|---|
| 509 | CMAT_FIXED_VALUE
|
|---|
| 510 | } ATTRIBCONSTRAINT;
|
|---|
| 511 |
|
|---|
| 512 | typedef enum _ATTRIBTYPE
|
|---|
| 513 | {
|
|---|
| 514 | CMAT_CDATA,
|
|---|
| 515 | CMAT_ID,
|
|---|
| 516 | CMAT_IDREF,
|
|---|
| 517 | CMAT_IDREFS,
|
|---|
| 518 | CMAT_ENTITY,
|
|---|
| 519 | CMAT_ENTITIES,
|
|---|
| 520 | CMAT_NMTOKEN,
|
|---|
| 521 | CMAT_NMTOKENS,
|
|---|
| 522 | CMAT_ENUM
|
|---|
| 523 | } ATTRIBTYPE;
|
|---|
| 524 |
|
|---|
| 525 | /*
|
|---|
| 526 | *@@ CMATTRIBUTEDECL:
|
|---|
| 527 | * single attribute declaration within the attribute
|
|---|
| 528 | * declarations tree in _CMATTRIBUTEDECLBASE.
|
|---|
| 529 | *
|
|---|
| 530 | *@@added V0.9.9 (2001-02-16) [umoeller]
|
|---|
| 531 | */
|
|---|
| 532 |
|
|---|
| 533 | typedef struct _CMATTRIBUTEDECL
|
|---|
| 534 | {
|
|---|
| 535 | NODEBASE NodeBase; // has TREE* as first item in turn
|
|---|
| 536 | // NodeBase.strName is attribute name
|
|---|
| 537 |
|
|---|
| 538 | ATTRIBTYPE ulAttrType;
|
|---|
| 539 | // one of:
|
|---|
| 540 | // -- CMAT_CDATA
|
|---|
| 541 | // -- CMAT_ID
|
|---|
| 542 | // -- CMAT_IDREF
|
|---|
| 543 | // -- CMAT_IDREFS
|
|---|
| 544 | // -- CMAT_ENTITY
|
|---|
| 545 | // -- CMAT_ENTITIES
|
|---|
| 546 | // -- CMAT_NMTOKEN
|
|---|
| 547 | // -- CMAT_NMTOKENS
|
|---|
| 548 | // -- CMAT_ENUM: pllEnum lists the allowed values.
|
|---|
| 549 | TREE *ValuesTree;
|
|---|
| 550 | // enumeration of allowed values, if CMAT_ENUM;
|
|---|
| 551 | // tree entries are plain NODEBASEs with
|
|---|
| 552 | // ATTRIBUTE_DECLARATION_ENUM type
|
|---|
| 553 |
|
|---|
| 554 | ATTRIBCONSTRAINT ulConstraint;
|
|---|
| 555 | // one of:
|
|---|
| 556 | // -- CMAT_IMPLIED: attrib can have any value.
|
|---|
| 557 | // -- CMAT_REQUIRED: attrib must be specified.
|
|---|
| 558 | // -- CMAT_DEFAULT_VALUE: attrib is optional and has default
|
|---|
| 559 | // value as in pstrDefaultValue.
|
|---|
| 560 | // -- CMAT_FIXED_VALUE: attrib is optional, but must have
|
|---|
| 561 | // fixed value as in pstrDefaultValue.
|
|---|
| 562 | PXSTRING pstrDefaultValue;
|
|---|
| 563 | // default value of this attribute; NULL with implied or required
|
|---|
| 564 |
|
|---|
| 565 | } CMATTRIBUTEDECL, *PCMATTRIBUTEDECL;
|
|---|
| 566 |
|
|---|
| 567 | /*
|
|---|
| 568 | *@@ CMATTRIBUTEDECLBASE:
|
|---|
| 569 | * representation of an @attribute_declaration.
|
|---|
| 570 | *
|
|---|
| 571 | * I'd love to have stored the attribute declarations with
|
|---|
| 572 | * the element specifications, but the XML spec says that
|
|---|
| 573 | * attribute declarations are allowed even if no element
|
|---|
| 574 | * declaration exists for the element to which the attribute
|
|---|
| 575 | * belongs. Now, whatever this is good for... anyway, this
|
|---|
| 576 | * forces us to do a second tree in the _DOMDOCTYPENODE node
|
|---|
| 577 | * according to attribute's element names.
|
|---|
| 578 | *
|
|---|
| 579 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 580 | */
|
|---|
| 581 |
|
|---|
| 582 | typedef struct _CMATTRIBUTEDECLBASE
|
|---|
| 583 | {
|
|---|
| 584 | NODEBASE NodeBase; // has TREE* as first item in turn
|
|---|
| 585 | // NodeBase.strName is element name
|
|---|
| 586 |
|
|---|
| 587 | TREE *AttribDeclsTree;
|
|---|
| 588 | // root of tree with CMATTRIBUTEDECL;
|
|---|
| 589 |
|
|---|
| 590 | } CMATTRIBUTEDECLBASE, *PCMATTRIBUTEDECLBASE;
|
|---|
| 591 |
|
|---|
| 592 | /*
|
|---|
| 593 | *@@ CMENTITYDECLNODE:
|
|---|
| 594 | *
|
|---|
| 595 | * See @entity_declaration.
|
|---|
| 596 | *
|
|---|
| 597 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 598 | */
|
|---|
| 599 |
|
|---|
| 600 | typedef struct _CMENTITYDECLNODE
|
|---|
| 601 | {
|
|---|
| 602 | NODEBASE NodeBase;
|
|---|
| 603 | } CMENTITYDECLNODE, *PCMENTITYDECLNODE;
|
|---|
| 604 |
|
|---|
| 605 | /*
|
|---|
| 606 | *@@ CMNOTATIONDECLNODE:
|
|---|
| 607 | *
|
|---|
| 608 | * See @notation_declaration.
|
|---|
| 609 | *
|
|---|
| 610 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 611 | */
|
|---|
| 612 |
|
|---|
| 613 | typedef struct _CMNOTATIONDECLNODE
|
|---|
| 614 | {
|
|---|
| 615 | NODEBASE NodeBase;
|
|---|
| 616 | } CMNOTATIONDECLNODE, *PCMNOTATIONDECLNODE;
|
|---|
| 617 |
|
|---|
| 618 | APIRET xmlCreateElementDecl(const char *pcszName,
|
|---|
| 619 | PXMLCONTENT pModel,
|
|---|
| 620 | PCMELEMENTDECLNODE *ppNew);
|
|---|
| 621 |
|
|---|
| 622 | /* ******************************************************************
|
|---|
| 623 | *
|
|---|
| 624 | * DOM parser APIs
|
|---|
| 625 | *
|
|---|
| 626 | ********************************************************************/
|
|---|
| 627 |
|
|---|
| 628 | typedef struct _XMLDOM *PXMLDOM;
|
|---|
| 629 |
|
|---|
| 630 | typedef int APIENTRY FNGETCPDATA(PXMLDOM pDom, ULONG ulCP, int *piMap);
|
|---|
| 631 | typedef FNGETCPDATA *PFNGETCPDATA;
|
|---|
| 632 |
|
|---|
| 633 | typedef APIRET APIENTRY FNEXTERNALHANDLER(PXMLDOM pDom,
|
|---|
| 634 | XML_Parser pSubParser,
|
|---|
| 635 | const char *pcszSystemID,
|
|---|
| 636 | const char *pcszPublicID);
|
|---|
| 637 | typedef FNEXTERNALHANDLER *PFNEXTERNALHANDLER;
|
|---|
| 638 |
|
|---|
| 639 | /*
|
|---|
| 640 | *@@ STATICSYSTEMID:
|
|---|
| 641 | * specification of a supported static system
|
|---|
| 642 | * ID, which is most helpful with predefined
|
|---|
| 643 | * DTD's.
|
|---|
| 644 | *
|
|---|
| 645 | * An array of this structure can be passed to
|
|---|
| 646 | * xmlCreateDOM to avoid having to supply an
|
|---|
| 647 | * external entity reference callback.
|
|---|
| 648 | *
|
|---|
| 649 | * This has one system ID with a corresponding
|
|---|
| 650 | * contents (normally a DTD) so the
|
|---|
| 651 | * implementation's external entity handler can
|
|---|
| 652 | * parse the doctype automatically.
|
|---|
| 653 | *
|
|---|
| 654 | *@@added V0.9.20 (2002-07-06) [umoeller]
|
|---|
| 655 | */
|
|---|
| 656 |
|
|---|
| 657 | typedef struct _STATICSYSTEMID
|
|---|
| 658 | {
|
|---|
| 659 | PCSZ pcszSystemId;
|
|---|
| 660 | // possible system ID specified in DTD, e.g. "warpin.dtd"
|
|---|
| 661 | // (without quotes)
|
|---|
| 662 |
|
|---|
| 663 | PCSZ pcszContent;
|
|---|
| 664 | // corresponding external DTD subset without square brackets,
|
|---|
| 665 | // e.g. "<!ELEMENT job EMPTY >\n"
|
|---|
| 666 | // (without quotes)
|
|---|
| 667 |
|
|---|
| 668 | } STATICSYSTEMID, *PSTATICSYSTEMID;
|
|---|
| 669 |
|
|---|
| 670 | /*
|
|---|
| 671 | *@@ XMLDOM:
|
|---|
| 672 | * DOM instance returned by xmlCreateDOM.
|
|---|
| 673 | *
|
|---|
| 674 | *@@added V0.9.9 (2001-02-14) [umoeller]
|
|---|
| 675 | */
|
|---|
| 676 |
|
|---|
| 677 | typedef struct _XMLDOM
|
|---|
| 678 | {
|
|---|
| 679 | /*
|
|---|
| 680 | * Public fields (should be read only)
|
|---|
| 681 | */
|
|---|
| 682 |
|
|---|
| 683 | PDOMDOCUMENTNODE pDocumentNode;
|
|---|
| 684 | // document node (contains all the elements)
|
|---|
| 685 |
|
|---|
| 686 | PDOMDOCTYPENODE pDocTypeNode;
|
|---|
| 687 | // != NULL only if the document has a DOCTYPE
|
|---|
| 688 |
|
|---|
| 689 | // error handling
|
|---|
| 690 | APIRET arcDOM; // validation errors etc.;
|
|---|
| 691 | // if != 0, this has a detailed
|
|---|
| 692 | // expat or DOM error code
|
|---|
| 693 | BOOL fInvalid; // TRUE if validation failed
|
|---|
| 694 | // (parser error otherwise)
|
|---|
| 695 |
|
|---|
| 696 | const char *pcszErrorDescription; // error description
|
|---|
| 697 | PXSTRING pxstrSystemID; // system ID of external entity
|
|---|
| 698 | // where error occurred, or NULL
|
|---|
| 699 | // if in main document
|
|---|
| 700 | ULONG ulErrorLine; // line where error occurred
|
|---|
| 701 | ULONG ulErrorColumn; // column where error occurred
|
|---|
| 702 | PXSTRING pxstrFailingNode; // element or attribute name
|
|---|
| 703 | // or NULL
|
|---|
| 704 |
|
|---|
| 705 | /*
|
|---|
| 706 | * Private fields (for xml* functions)
|
|---|
| 707 | */
|
|---|
| 708 |
|
|---|
| 709 | // params copied from xmlCreateDOM
|
|---|
| 710 | ULONG flParserFlags;
|
|---|
| 711 | PFNGETCPDATA pfnGetCPData;
|
|---|
| 712 | PFNEXTERNALHANDLER pfnExternalHandler;
|
|---|
| 713 | PVOID pvCallbackUser;
|
|---|
| 714 | const STATICSYSTEMID *paSystemIds;
|
|---|
| 715 | ULONG cSystemIds;
|
|---|
| 716 |
|
|---|
| 717 | XML_Parser pParser;
|
|---|
| 718 | // expat parser instance
|
|---|
| 719 |
|
|---|
| 720 | LINKLIST llElementStack;
|
|---|
| 721 | // stack for maintaining the current items;
|
|---|
| 722 | // these point to DOMSTACKITEMs (auto-free)
|
|---|
| 723 |
|
|---|
| 724 | PDOMNODE pLastWasTextNode;
|
|---|
| 725 |
|
|---|
| 726 | PCMATTRIBUTEDECLBASE pAttListDeclCache;
|
|---|
| 727 | // cache for attribute declarations according
|
|---|
| 728 | // to attdecl element name
|
|---|
| 729 | } XMLDOM;
|
|---|
| 730 |
|
|---|
| 731 | #define DF_PARSECOMMENTS 0x0001
|
|---|
| 732 | #define DF_PARSEDTD 0x0002
|
|---|
| 733 | #define DF_FAIL_IF_NO_DTD 0x0004
|
|---|
| 734 | #define DF_DROP_WHITESPACE 0x0008
|
|---|
| 735 |
|
|---|
| 736 | APIRET xmlCreateDOM(ULONG flParserFlags,
|
|---|
| 737 | const STATICSYSTEMID *paSystemIds,
|
|---|
| 738 | ULONG cSystemIds,
|
|---|
| 739 | PFNGETCPDATA pfnGetCPData,
|
|---|
| 740 | PFNEXTERNALHANDLER pfnExternalHandler,
|
|---|
| 741 | PVOID pvCallbackUser,
|
|---|
| 742 | PXMLDOM *ppDom);
|
|---|
| 743 |
|
|---|
| 744 | APIRET xmlParse(PXMLDOM pDom,
|
|---|
| 745 | const char *pcszBuf,
|
|---|
| 746 | ULONG cb,
|
|---|
| 747 | BOOL fIsLast);
|
|---|
| 748 |
|
|---|
| 749 | VOID xmlDump(PXMLDOM pDom);
|
|---|
| 750 |
|
|---|
| 751 | APIRET xmlFreeDOM(PXMLDOM pDom);
|
|---|
| 752 |
|
|---|
| 753 | /* ******************************************************************
|
|---|
| 754 | *
|
|---|
| 755 | * DOM lookup
|
|---|
| 756 | *
|
|---|
| 757 | ********************************************************************/
|
|---|
| 758 |
|
|---|
| 759 | PCMELEMENTDECLNODE xmlFindElementDecl(PXMLDOM pDom,
|
|---|
| 760 | const XSTRING *pcszElementName);
|
|---|
| 761 |
|
|---|
| 762 | PCMATTRIBUTEDECLBASE xmlFindAttribDeclBase(PXMLDOM pDom,
|
|---|
| 763 | const XSTRING *pstrElementName);
|
|---|
| 764 |
|
|---|
| 765 | PCMATTRIBUTEDECL xmlFindAttribDecl(PXMLDOM pDom,
|
|---|
| 766 | const XSTRING *pstrElementName,
|
|---|
| 767 | const XSTRING *pstrAttribName,
|
|---|
| 768 | PCMATTRIBUTEDECLBASE *ppAttribDeclBase);
|
|---|
| 769 |
|
|---|
| 770 | PDOMNODE xmlGetRootElement(PXMLDOM pDom);
|
|---|
| 771 |
|
|---|
| 772 | PDOMNODE xmlGetFirstChild(PDOMNODE pDomNode);
|
|---|
| 773 |
|
|---|
| 774 | PDOMNODE xmlGetLastChild(PDOMNODE pDomNode);
|
|---|
| 775 |
|
|---|
| 776 | PDOMNODE xmlGetFirstText(PDOMNODE pElement);
|
|---|
| 777 |
|
|---|
| 778 | PLINKLIST xmlGetElementsByTagName(PDOMNODE pParent,
|
|---|
| 779 | const char *pcszName);
|
|---|
| 780 |
|
|---|
| 781 | const XSTRING* xmlGetAttribute(PDOMNODE pElement,
|
|---|
| 782 | const char *pcszAttribName);
|
|---|
| 783 |
|
|---|
| 784 | /* ******************************************************************
|
|---|
| 785 | *
|
|---|
| 786 | * DOM build
|
|---|
| 787 | *
|
|---|
| 788 | ********************************************************************/
|
|---|
| 789 |
|
|---|
| 790 | APIRET xmlCreateDocument(const char *pcszRootElementName,
|
|---|
| 791 | PDOMDOCUMENTNODE *ppDocument,
|
|---|
| 792 | PDOMNODE *ppRootElement);
|
|---|
| 793 |
|
|---|
| 794 | APIRET xmlWriteDocument(PDOMDOCUMENTNODE pDocument,
|
|---|
| 795 | const char *pcszEncoding,
|
|---|
| 796 | const char *pcszDoctype,
|
|---|
| 797 | PXSTRING pxstr);
|
|---|
| 798 | #endif
|
|---|
| 799 |
|
|---|
| 800 | #if __cplusplus
|
|---|
| 801 | }
|
|---|
| 802 | #endif
|
|---|
| 803 |
|
|---|