1 /******************************************************************************
2 *
3 * Copyright (C) 2004-2008, The Gentee Group. All rights reserved.
4 * This file is part of the Gentee open source project - http://www.gentee.com.
5 *
6 * THIS FILE IS PROVIDED UNDER THE TERMS OF THE GENTEE LICENSE ("AGREEMENT").
7 * ANY USE, REPRODUCTION OR DISTRIBUTION OF THIS FILE CONSTITUTES RECIPIENTS
8 * ACCEPTANCE OF THE AGREEMENT.
9 *
10 * Author: Alexander Krivonogov ( algen )
11 *
12 ******************************************************************************/
13
14 /*-----------------------------------------------------------------------------
15 * Id: xml L "XML"
16 *
17 * Summary: XML file processing. This library is used for XML file processing
18 and XML tree building. Neither a multibyte-character set nor a
19 document type description #b(#lgt[!DOCTYPE .....]) are handled in
20 the current version. For using this library, it is required to
21 specify the file xml.g (from lib\xml subfolder) with include
22 command. #srcg[
23 |include : $"...\gentee\lib\xml\xml.g"]
24 *
25 * List: *,xml_desc,
26 *#lng/opers#,xml_opfor,
27 *#lng/methods#,xml_addentity,xml_getroot,xml_procfile,xml_procstr,
28 *Methods of XML tree items,xmlitem_chtag,xmlitem_findtag,
29 xmlitem_getattrib,xmlitem_getchild,xmlitem_getchildtag,
30 xmlitem_getchildtext,xmlitem_getname,xmlitem_getnext,
31 xmlitem_getnexttag,xmlitem_getnexttext,xmlitem_getparent,
32 xmlitem_gettext,xmlitem_isemptytag,
33 xmlitem_ispitag,xmlitem_istag,xmlitem_istext
34 *
35 -----------------------------------------------------------------------------*/
36
37 define
38 {
39 NUMSYM = 256
40 }
41
42 type posb
43 {
44 byte state
45 byte afunc
46 byte retstate
47 }
48
49 type pos
50 {
51 int state
52 uint afunc
53 int retstate
54 uint r
55 }
56
57 operator pos = ( pos l, posb r)
58 {
59 l.state = r.state
60 l.afunc = r.afunc
61 l.retstate = r.retstate
62 return l
63 }
64
65 define {
66 TG_TEXT = 0x01 //Текст (в tgstart номер начального текста, в tgend номер конечного текста)
67 TG_TAG = 0x10 //Тэг (в tgid идентификатор имени, в tgstart номер начального атрибута, в tgend номер конечного атрибута)
68 TG_QUEST = 0x12 //<? ?>
69 TG_NOCHILD = 0x14 //< />
70
71 TX_TEXT = 0x01 //Просто текст
72 TX_SYMBOL = 0x02 //Символ, в txaddr_code код вставляемого символа
73 TX_ENTITY = 0x03 //Сущность, в txaddr_code код имени сущности в хэш таблице сущностей
74 }
75 //Элемент дерева разбора текст или тэг
76 type xmlitem {
77 uint tgtype //Тип элемента TG_*
78 uint tgid //Идентфикатор имени тэга в хэш таблице тэгов
79 uint tgstart //Номер начального атрибута/текста в таблице атрибутов/текстов
80 uint tgend //Номер конечного+1 атрибута/текста в таблице атрибутов/текстов
81 uint nparent //Номер тэга владельца
82 uint nnext //Номер следующего тэга
83 uint nchild //Номер первого потомка
84 uint xml
85 }
86 type xmltags <index=xmlitem>{
87 uint parent
88 uint cur
89 }
90
91 //Элемент массива атрибутов
92 type xattrib {
93 uint attid //Идентификатор имени атрибута
94 uint attstart //Номер начального текста(значение атрибута) в таблице текстов
95 uint attend //Номер конечного текста(значение атрибута) в таблице текстов
96 }
97
98 //Элемент массива текстов
99 type xtext {
100 uint txtype //Тип текста TX_*
101 uint txaddr_code //Адрес начала исходного текста/код символа/код имени сущности
102 uint txlen //Длина исходного теста
103 }
104
105 //Объект разбора xml текста
106 type xml
107 {
108 buf src //Исходный текст
109 arr tags of xmlitem //Массив/дерево тэгов
110 arr attribs of xattrib //Массив сущностей
111 arr texts of xtext //Массив текстов
112 hash hnames //Хэш таблица имён тэгов
113 arr names of str //Таблица строк для хэш таблицы hnames
114 hash hentities //Хэш таблица имён сущностей
115 uint err
116 }
117
118 global
119 {
120 arr tp[1,256] of pos
121
122 uint X_ia
123 uint X_curtag, X_curtext, X_curattrib
124 uint X_ncurtag, X_nparenttag, X_ncurtext, X_ncurattrib
125 uint X_maxtag, X_maxattrib, X_maxtext, X_maxstack
126 arr X_stacktags of uint
127 uint X_nstack
128 uint X_curnameoff
129 uint X_x
130 buf X_tblsrc = '\<sp.tbl>'
131 str X_sname
132 uint X_n, X_tparenttag
133 }
134
135 include {
136 "xmlfuncs.g"
137 "xmluser.g"
138 }
139
140 method xml.init()
141 {
142 uint i,j
143 arr ar[0,$NUMSYM] of posb
144 ar->buf = X_tblsrc//.read( "sp.X_tblsrc" )
145 tp.expand( (*ar/$NUMSYM)*256 )
146 fornum i=0, *ar/$NUMSYM
147 {
148 fornum j=0, $NUMSYM
149 {
150 tp[i+1,j] = ar[i,j]
151 if ar[i,j].state && ar[i,j].state !=-1
152 {
153 tp[i+1,j].state <<= 12
154 tp[i+1,j].state += tp.ptr()
155 }
156 if ar[i,j].retstate && ar[i,j].retstate !=-1
157 {
158 tp[i+1,j].retstate <<= 12
159 tp[i+1,j].retstate += tp.ptr()
160 }
161 switch ar[i,j].afunc
162 {
163 case 1 : tp[i+1,j].afunc = &f_begent
164 case 2 : tp[i+1,j].afunc = &f_endent
165 case 3 : tp[i+1,j].afunc = &f_endentnum
166 case 4 : tp[i+1,j].afunc = &f_endenthex
167 case 5 : tp[i+1,j].afunc = &f_begatrval
168 case 6 : tp[i+1,j].afunc = &f_endatrval
169 case 7 : tp[i+1,j].afunc = &f_begquest
170 case 8 : tp[i+1,j].afunc = &f_endquest
171 case 9 : tp[i+1,j].afunc = &f_endtagname
172 case 10 : tp[i+1,j].afunc = &f_begatr
173 case 11 : tp[i+1,j].afunc = &f_endatr
174 case 12 : tp[i+1,j].afunc = &f_begtag
175 case 13 : tp[i+1,j].afunc = &f_endtag
176 case 14 : tp[i+1,j].afunc = &f_endtagend
177 case 15 : tp[i+1,j].afunc = &f_begendtag
178 case 16 : tp[i+1,j].afunc = &f_begendtagend
179 case 17 : tp[i+1,j].afunc = &f_begcdata
180 case 18 : tp[i+1,j].afunc = &f_endcdata
181 case 255 : tp[i+1,j].afunc = &f_error
182 }
183 }
184 }
185 }
186
187 method uint xml.process()
188 {
189 uint off
190 arr ars[512] of uint
191 uint state, retstate
192 uint afunc
193
194 state = tp.ptr() + (1 << 12)
195 //Инициализация
196 if X_x: return 0
197 X_x = &this
198 this.err = 0
199 this.hnames.clear()
200 this.hentities.clear()
201 this.tags.clear()
202 this.texts.clear()
203 this.names.clear()
204 this.attribs.clear()
205
206 this.hnames.ignorecase()
207 this.hentities.ignorecase()
208 //this.names.reserve(100)
209 this.names.expand(1)
210 X_maxtag = max( *this.src/10, 100 )
211 this.tags.expand( X_maxtag )
212 X_maxtext = max( *this.src/20, 100 )
213 this.texts.expand(X_maxtext)
214 X_maxattrib = max( *this.src/40, 100 )
215 this.attribs.expand(X_maxattrib)
216
217 X_ncurattrib = 0
218 X_curattrib = &this.attribs[X_ncurattrib]
219
220 X_ncurtext = 0
221 X_curtext = &this.texts[X_ncurtext]
222
223 X_ncurtag = 0
224 //Корневой элемент
225 X_curtag = &this.tags[X_ncurtag]
226 X_curtag->xmlitem.tgtype = 0//$TG_TEXT
227 X_curtag->xmlitem.nchild = ++X_ncurtag
228
229 //Добавляем тэг заготовку
230 X_curtag = &this.tags[X_ncurtag]
231 X_curtag->xmlitem.nchild = 1
232 X_curtag->xmlitem.tgstart = X_ncurtext
233 X_nparenttag = 0
234
235 X_ia=this.src.ptr()
236 X_stacktags.clear()
237 X_maxstack = 100
238 X_stacktags.expand(X_maxstack)
239 X_nstack = 0
240
241 //Начать пустой текст
242 X_curtext->xtext.txaddr_code = X_ia
243 X_curtext->xtext.txtype = 0
244 uint arrs = ars.ptr()
245 fornum X_ia, this.src.ptr() + *this.src
246 {
247
248 if afunc = ((off = state + (X_ia->byte << 4)) + 4)->uint : afunc->func()
249
250 if state = off->uint
251 {
252 if retstate = (off + 8)->uint
253 {
254 arrs->uint = retstate
255 arrs += 4
256 }
257 continue
258 }
259 state = (arrs -= 4)->uint
260 }
261 if !X_curtext->xtext.txtype &&
262 X_curtext->xtext.txaddr_code != X_ia
263 { //Заканчиваем текст-текст
264 X_curtext->xtext.txlen = X_ia - X_curtext->xtext.txaddr_code
265 X_curtext->xtext.txtype = $TX_TEXT
266 X_curtag->xmlitem.tgtype = $TG_TEXT
267 X_curtag->xmlitem.tgend = X_ncurtext + 1
268 }
269 this.tags.del(X_ncurtag + 1)
270 this.texts.del(X_ncurtext + 1)
271 this.attribs.del(X_ncurattrib + 1)
272 uint i
273 fornum i = 0, *this.tags
274 {
275 this.tags[i].xml = &this
276 }
277 X_x = 0
278 if !this.err
279 {
280 this.addentity("amp","&")
281 this.addentity("quot","\"")
282 this.addentity("apos","'")
283 this.addentity("gt",">")
284 this.addentity("lt","<")
285 }
286 return !this.err
287 }
288
289 /*-----------------------------------------------------------------------------
290 * Id: xml_procfile F2
291 *
292 * Summary: Process an XML file. Reads the XML file, the name of which is
293 specified as a parameter, and process it.
294 *
295 * Params: filename - Name of the file processed.
296 *
297 * Return: #lng/retf#
298 *
299 -----------------------------------------------------------------------------*/
300
301 method uint xml.procfile( str filename )
302 {
303 if this.src.read( filename )
304 {
305 return this.process()
306 }
307 return 0
308 }
309
310 /*-----------------------------------------------------------------------------
311 * Id: xml_procstr F2
312 *
313 * Summary: Processes a string contained the XML document.
314 *
315 * Params: src - XML data string.
316 *
317 * Return: #lng/retf#
318 *
319 -----------------------------------------------------------------------------*/
320
321 method uint xml.procstr( str src )
322 {
323 this.src = src->buf
324 return this.process()
325 }
326
327 /*-----------------------------------------------------------------------------
328 ** Id: xml_desc F1
329 *
330 * Summary: A brief description of XML library. Variables of either the #b(xml)
331 and the #b(xmlitem) type (an XML tree item) are used for processing
332 XML documents. An XML tree item can be of two types: a #b(text item)
333 and a #b(tag item). There are several types of tag items:
334 #ul[
335 |tag item that contains other items #b(#lgt[tag ...].....#lgt[/tag]);
336 |tag item that contains no other items #b(#lgt[tag .../]);
337 |tag item of processing instruction #b(#lgt[?tag ...?]).
338 ]
339 #p[A tag item may contain attributes.]
340
341 #p[The sequence of operations for processing an XML document:]
342 #ul[
343 process a document (build an XML tree) with the help of the #a(xml_procfile)
344 | method or the #a(xml_procstr) method;
345 |add entity definitions, using the #a(xml_addentity) method if necessary;
346 search for the required items in the XML tree using the following methods:
347 #a(xml_getroot), #a(xmlitem_chtag), #a(xmlitem_findtag),
348 | #a(xmlitem_getnext), etc.;
349 use the #b(foreach) statement in order to process similar elements if
350 | necessary;
351 gain access to tag attributes with the help of the #a(xmlitem_getattrib)
352 | method and get a text using the #a(xmlitem_gettext) method.
353 ]
354 *
355 * Title: XML description
356 *
357 * Define:
358 *
359 -----------------------------------------------------------------------------*/
360
361 //----------------------------------------------------------------------------
Ads
Perfect Automation tool
All-In-One: Script editor, Launcher, Scheduler, Keyboard & Mouse Recorder. Try now!
CreateInstall
Freeware and commercial installers.
Cell Phone Batteries
Batteries Plus offers batteries for laptop, camcorder, cell phone,
camera.
Gentee needs your help!
How to advertise with us
laptop battery