|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/***************************************************************************\ |
|
4
|
|
|
* SPIP, Systeme de publication pour l'internet * |
|
5
|
|
|
* * |
|
6
|
|
|
* Copyright (c) 2001-2018 * |
|
7
|
|
|
* Arnaud Martin, Antoine Pitrou, Philippe Riviere, Emmanuel Saint-James * |
|
8
|
|
|
* * |
|
9
|
|
|
* Ce programme est un logiciel libre distribue sous licence GNU/GPL. * |
|
10
|
|
|
* Pour plus de details voir le fichier COPYING.txt ou l'aide en ligne. * |
|
11
|
|
|
\***************************************************************************/ |
|
12
|
|
|
|
|
13
|
|
|
if (!defined('_ECRIRE_INC_VERSION')) { |
|
14
|
|
|
return; |
|
15
|
|
|
} |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Validateur XML en deux passes, fondé sur SAX pour la première |
|
19
|
|
|
* |
|
20
|
|
|
* @note |
|
21
|
|
|
* Faudrait faire deux classes car pour la première passe |
|
22
|
|
|
* on a les memes methodes et variables que l'indenteur |
|
23
|
|
|
**/ |
|
24
|
|
|
class ValidateurXML { |
|
25
|
|
|
|
|
26
|
|
|
// http://code.spip.net/@validerElement |
|
27
|
|
|
public function validerElement($phraseur, $name, $attrs) { |
|
|
|
|
|
|
28
|
|
|
if (!($p = isset($this->dtc->elements[$name]))) { |
|
29
|
|
|
if ($p = strpos($name, ':')) { |
|
30
|
|
|
$name = substr($name, $p + 1); |
|
31
|
|
|
$p = isset($this->dtc->elements[$name]); |
|
32
|
|
|
} |
|
33
|
|
|
if (!$p) { |
|
34
|
|
|
coordonnees_erreur($this, " <b>$name</b> : " |
|
35
|
|
|
. _T('zxml_inconnu_balise')); |
|
36
|
|
|
|
|
37
|
|
|
return; |
|
38
|
|
|
} |
|
39
|
|
|
} |
|
40
|
|
|
// controler les filles illegitimes, ca suffit |
|
41
|
|
|
$depth = $this->depth; |
|
42
|
|
|
$ouvrant = $this->ouvrant; |
|
43
|
|
|
#spip_log("trouve $name apres " . $ouvrant[$depth]); |
|
44
|
|
|
if (isset($ouvrant[$depth])) { |
|
45
|
|
|
if (preg_match('/^\s*(\w+)/', $ouvrant[$depth], $r)) { |
|
46
|
|
|
$pere = $r[1]; |
|
47
|
|
|
#spip_log("pere $pere"); |
|
48
|
|
|
if (isset($this->dtc->elements[$pere])) { |
|
49
|
|
|
$fils = $this->dtc->elements[$pere]; |
|
50
|
|
|
#spip_log("rejeton $name fils " . @join(',',$fils)); |
|
51
|
|
|
if (!($p = @in_array($name, $fils))) { |
|
52
|
|
View Code Duplication |
if ($p = strpos($name, ':')) { |
|
|
|
|
|
|
53
|
|
|
$p = substr($name, $p + 1); |
|
54
|
|
|
$p = @in_array($p, $fils); |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
if (!$p) { |
|
58
|
|
|
$bons_peres = @join('</b>, <b>', $this->dtc->peres[$name]); |
|
59
|
|
|
coordonnees_erreur($this, " <b>$name</b> " |
|
60
|
|
|
. _T('zxml_non_fils') |
|
61
|
|
|
. ' <b>' |
|
62
|
|
|
. $pere |
|
63
|
|
|
. '</b>' |
|
64
|
|
|
. (!$bons_peres ? '' |
|
65
|
|
|
: ('<p style="font-size: 80%"> ' . _T('zxml_mais_de') . ' <b>' . $bons_peres . '</b></p>'))); |
|
66
|
|
|
} elseif ($this->dtc->regles[$pere][0] == '/') { |
|
67
|
|
|
$frat = substr($depth, 2); |
|
68
|
|
|
if (!isset($this->fratrie[$frat])) { |
|
69
|
|
|
$this->fratrie[$frat] = ''; |
|
70
|
|
|
} |
|
71
|
|
|
$this->fratrie[$frat] .= "$name "; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
} |
|
76
|
|
|
// Init de la suite des balises a memoriser si regle difficile |
|
77
|
|
|
if ($this->dtc->regles[$name] and $this->dtc->regles[$name][0] == '/') { |
|
78
|
|
|
$this->fratrie[$depth] = ''; |
|
79
|
|
|
} |
|
80
|
|
|
if (isset($this->dtc->attributs[$name])) { |
|
81
|
|
|
foreach ($this->dtc->attributs[$name] as $n => $v) { |
|
82
|
|
|
if (($v[1] == '#REQUIRED') and (!isset($attrs[$n]))) { |
|
83
|
|
|
coordonnees_erreur($this, " <b>$n</b>" |
|
84
|
|
|
. ' : ' |
|
85
|
|
|
. _T('zxml_obligatoire_attribut') |
|
86
|
|
|
. " <b>$name</b>"); |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
// http://code.spip.net/@validerAttribut |
|
93
|
|
|
public function validerAttribut($phraseur, $name, $val, $bal) { |
|
94
|
|
|
// Si la balise est inconnue, eviter d'insister |
|
95
|
|
|
if (!isset($this->dtc->attributs[$bal])) { |
|
96
|
|
|
return; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
$a = $this->dtc->attributs[$bal]; |
|
100
|
|
|
if (!isset($a[$name])) { |
|
101
|
|
|
$bons = join(', ', array_keys($a)); |
|
102
|
|
|
if ($bons) { |
|
103
|
|
|
$bons = " title=' " . |
|
104
|
|
|
_T('zxml_connus_attributs') . |
|
105
|
|
|
' : ' . |
|
106
|
|
|
$bons . |
|
107
|
|
|
"'"; |
|
108
|
|
|
} |
|
109
|
|
|
$bons .= " style='font-weight: bold'"; |
|
110
|
|
|
coordonnees_erreur($this, " <b>$name</b> " |
|
111
|
|
|
. _T('zxml_inconnu_attribut') . ' ' . _T('zxml_de') |
|
112
|
|
|
. " <a$bons>$bal</a> (" |
|
113
|
|
|
. _T('zxml_survoler') |
|
114
|
|
|
. ")"); |
|
115
|
|
|
} else { |
|
116
|
|
|
$type = $a[$name][0]; |
|
117
|
|
|
if (!preg_match('/^\w+$/', $type)) { |
|
118
|
|
|
$this->valider_motif($phraseur, $name, $val, $bal, $type); |
|
119
|
|
|
} else { |
|
120
|
|
|
if (method_exists($this, $f = 'validerAttribut_' . $type)) { |
|
121
|
|
|
$this->$f($phraseur, $name, $val, $bal); |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
# else spip_log("$type type d'attribut inconnu"); |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
public function validerAttribut_NMTOKEN($phraseur, $name, $val, $bal) { |
|
129
|
|
|
$this->valider_motif($phraseur, $name, $val, $bal, _REGEXP_NMTOKEN); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
public function validerAttribut_NMTOKENS($phraseur, $name, $val, $bal) { |
|
133
|
|
|
$this->valider_motif($phraseur, $name, $val, $bal, _REGEXP_NMTOKENS); |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
// http://code.spip.net/@validerAttribut_ID |
|
137
|
|
|
public function validerAttribut_ID($phraseur, $name, $val, $bal) { |
|
138
|
|
|
if (isset($this->ids[$val])) { |
|
139
|
|
|
list($l, $c) = $this->ids[$val]; |
|
140
|
|
|
coordonnees_erreur($this, " <p><b>$val</b> " |
|
141
|
|
|
. _T('zxml_valeur_attribut') |
|
142
|
|
|
. " <b>$name</b> " |
|
143
|
|
|
. _T('zxml_de') |
|
144
|
|
|
. " <b>$bal</b> " |
|
145
|
|
|
. _T('zxml_vu') |
|
146
|
|
|
. " (L$l,C$c)"); |
|
147
|
|
|
} else { |
|
148
|
|
|
$this->valider_motif($phraseur, $name, $val, $bal, _REGEXP_ID); |
|
149
|
|
|
$this->ids[$val] = array(xml_get_current_line_number($phraseur), xml_get_current_column_number($phraseur)); |
|
150
|
|
|
} |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
// http://code.spip.net/@validerAttribut_IDREF |
|
154
|
|
|
public function validerAttribut_IDREF($phraseur, $name, $val, $bal) { |
|
|
|
|
|
|
155
|
|
|
$this->idrefs[] = array($val, xml_get_current_line_number($phraseur), xml_get_current_column_number($phraseur)); |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
// http://code.spip.net/@validerAttribut_IDREFS |
|
159
|
|
|
public function validerAttribut_IDREFS($phraseur, $name, $val, $bal) { |
|
|
|
|
|
|
160
|
|
|
$this->idrefss[] = array($val, xml_get_current_line_number($phraseur), xml_get_current_column_number($phraseur)); |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
// http://code.spip.net/@valider_motif |
|
164
|
|
|
public function valider_motif($phraseur, $name, $val, $bal, $motif) { |
|
|
|
|
|
|
165
|
|
|
if (!preg_match($motif, $val)) { |
|
166
|
|
|
coordonnees_erreur($this, "<b>$val</b> " |
|
167
|
|
|
. _T('zxml_valeur_attribut') |
|
168
|
|
|
. " <b>$name</b> " |
|
169
|
|
|
. _T('zxml_de') |
|
170
|
|
|
. " <b>$bal</b> " |
|
171
|
|
|
. _T('zxml_non_conforme') |
|
172
|
|
|
. "</p><p>" |
|
173
|
|
|
. "<b>" . $motif . "</b>"); |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
// http://code.spip.net/@valider_idref |
|
178
|
|
|
public function valider_idref($nom, $ligne, $col) { |
|
179
|
|
|
if (!isset($this->ids[$nom])) { |
|
180
|
|
|
$this->err[] = array(" <p><b>$nom</b> " . _T('zxml_inconnu_id'), $ligne, $col); |
|
181
|
|
|
} |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
// http://code.spip.net/@valider_passe2 |
|
185
|
|
|
public function valider_passe2() { |
|
186
|
|
|
if (!$this->err) { |
|
|
|
|
|
|
187
|
|
|
foreach ($this->idrefs as $idref) { |
|
188
|
|
|
list($nom, $ligne, $col) = $idref; |
|
189
|
|
|
$this->valider_idref($nom, $ligne, $col); |
|
190
|
|
|
} |
|
191
|
|
|
foreach ($this->idrefss as $idref) { |
|
192
|
|
|
list($noms, $ligne, $col) = $idref; |
|
193
|
|
|
foreach (preg_split('/\s+/', $noms) as $nom) { |
|
194
|
|
|
$this->valider_idref($nom, $ligne, $col); |
|
195
|
|
|
} |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
// http://code.spip.net/@debutElement |
|
201
|
|
|
public function debutElement($phraseur, $name, $attrs) { |
|
202
|
|
|
if ($this->dtc->elements) { |
|
203
|
|
|
$this->validerElement($phraseur, $name, $attrs); |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
if ($f = $this->process['debut']) { |
|
207
|
|
|
$f($this, $name, $attrs); |
|
208
|
|
|
} |
|
209
|
|
|
$depth = $this->depth; |
|
210
|
|
|
$this->debuts[$depth] = strlen($this->res); |
|
211
|
|
|
foreach ($attrs as $k => $v) { |
|
212
|
|
|
$this->validerAttribut($phraseur, $k, $v, $name); |
|
213
|
|
|
} |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
// http://code.spip.net/@finElement |
|
217
|
|
|
public function finElement($phraseur, $name) { |
|
|
|
|
|
|
218
|
|
|
$depth = $this->depth; |
|
219
|
|
|
$contenu = $this->contenu; |
|
220
|
|
|
|
|
221
|
|
|
$n = strlen($this->res); |
|
222
|
|
|
$c = strlen(trim($contenu[$depth])); |
|
223
|
|
|
$k = $this->debuts[$depth]; |
|
224
|
|
|
|
|
225
|
|
|
$regle = isset($this->dtc->regles[$name]) ? $this->dtc->regles[$name] : false; |
|
226
|
|
|
$vide = ($regle == 'EMPTY'); |
|
227
|
|
|
// controler que les balises devant etre vides le sont |
|
228
|
|
|
if ($vide) { |
|
229
|
|
|
if ($n <> ($k + $c)) { |
|
230
|
|
|
coordonnees_erreur($this, " <p><b>$name</b> " . _T('zxml_nonvide_balise')); |
|
231
|
|
|
} |
|
232
|
|
|
// pour les regles PCDATA ou iteration de disjonction, tout est fait |
|
233
|
|
|
} elseif ($regle and ($regle != '*')) { |
|
234
|
|
|
if ($regle == '+') { |
|
235
|
|
|
// iteration de disjonction non vide: 1 balise au - |
|
236
|
|
|
if ($n == $k) { |
|
237
|
|
|
coordonnees_erreur($this, "<p>\n<b>$name</b> " |
|
238
|
|
|
. _T('zxml_vide_balise')); |
|
239
|
|
|
} |
|
240
|
|
|
} else { |
|
241
|
|
|
$f = isset($this->fratrie[substr($depth, 2)]) ? $this->fratrie[substr($depth, 2)] : null; |
|
242
|
|
|
if (is_null($f) or !preg_match($regle, $f)) { |
|
243
|
|
|
coordonnees_erreur($this, |
|
244
|
|
|
" <p>\n<b>$name</b> " |
|
245
|
|
|
. _T('zxml_succession_fils_incorrecte') |
|
246
|
|
|
. ' : <b>' |
|
247
|
|
|
. $f |
|
248
|
|
|
. '</b>'); |
|
249
|
|
|
} |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
} |
|
253
|
|
|
if ($f = $this->process['fin']) { |
|
254
|
|
|
$f($this, $name, $vide); |
|
255
|
|
|
} |
|
256
|
|
|
} |
|
257
|
|
|
|
|
258
|
|
|
// http://code.spip.net/@textElement |
|
259
|
|
|
public function textElement($phraseur, $data) { |
|
|
|
|
|
|
260
|
|
|
if (trim($data)) { |
|
261
|
|
|
$d = $this->depth; |
|
262
|
|
|
$d = $this->ouvrant[$d]; |
|
263
|
|
|
preg_match('/^\s*(\S+)/', $d, $m); |
|
264
|
|
|
if (isset($this->dtc->pcdata[$m[1]]) and ($this->dtc->pcdata[$m[1]])) { |
|
265
|
|
|
coordonnees_erreur($this, " <p><b>" . $m[1] . "</b> " |
|
266
|
|
|
. _T('zxml_nonvide_balise') // message a affiner |
|
267
|
|
|
); |
|
268
|
|
|
} |
|
269
|
|
|
} |
|
270
|
|
|
if ($f = $this->process['text']) { |
|
271
|
|
|
$f($this, $data); |
|
272
|
|
|
} |
|
273
|
|
|
} |
|
274
|
|
|
|
|
275
|
|
|
public function piElement($phraseur, $target, $data) { |
|
|
|
|
|
|
276
|
|
|
if ($f = $this->process['pi']) { |
|
277
|
|
|
$f($this, $target, $data); |
|
278
|
|
|
} |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
|
|
// Denonciation des entitees XML inconnues |
|
282
|
|
|
// Pour contourner le bug de conception de SAX qui ne signale pas si elles |
|
283
|
|
|
// sont dans un attribut, les entites les plus frequentes ont ete |
|
284
|
|
|
// transcodees au prealable (sauf & < > " que SAX traite correctement). |
|
285
|
|
|
// On ne les verra donc pas passer a cette etape, contrairement a ce que |
|
286
|
|
|
// le source de la page laisse legitimement supposer. |
|
287
|
|
|
|
|
288
|
|
|
// http://code.spip.net/@defautElement |
|
289
|
|
|
public function defaultElement($phraseur, $data) { |
|
|
|
|
|
|
290
|
|
|
if (!preg_match('/^<!--/', $data) |
|
291
|
|
|
and (preg_match_all('/&([^;]*)?/', $data, $r, PREG_SET_ORDER)) |
|
292
|
|
|
) { |
|
293
|
|
|
foreach ($r as $m) { |
|
294
|
|
|
list($t, $e) = $m; |
|
|
|
|
|
|
295
|
|
|
if (!isset($this->dtc->entites[$e])) { |
|
296
|
|
|
coordonnees_erreur($this, " <b>$e</b> " |
|
297
|
|
|
. _T('zxml_inconnu_entite') |
|
298
|
|
|
. ' ' |
|
299
|
|
|
); |
|
300
|
|
|
} |
|
301
|
|
|
} |
|
302
|
|
|
} |
|
303
|
|
|
if (isset($this->process['default']) and ($f = $this->process['default'])) { |
|
304
|
|
|
$f($this, $data); |
|
305
|
|
|
} |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
// http://code.spip.net/@phraserTout |
|
309
|
|
|
public function phraserTout($phraseur, $data) { |
|
|
|
|
|
|
310
|
|
|
xml_parsestring($this, $data); |
|
311
|
|
|
|
|
312
|
|
|
if (!$this->dtc or preg_match(',^' . _MESSAGE_DOCTYPE . ',', $data)) { |
|
313
|
|
|
$this->err[] = array('DOCTYPE ?', 0, 0); |
|
314
|
|
|
} else { |
|
315
|
|
|
$this->valider_passe2($this); |
|
|
|
|
|
|
316
|
|
|
} |
|
317
|
|
|
} |
|
318
|
|
|
|
|
319
|
|
|
/** |
|
320
|
|
|
* Constructeur |
|
321
|
|
|
* |
|
322
|
|
|
* @param array $process ? |
|
323
|
|
|
**/ |
|
324
|
|
|
public function __construct($process = array()) { |
|
325
|
|
|
if (is_array($process)) { |
|
326
|
|
|
$this->process = $process; |
|
327
|
|
|
} |
|
328
|
|
|
} |
|
329
|
|
|
|
|
330
|
|
|
public $ids = array(); |
|
331
|
|
|
public $idrefs = array(); |
|
332
|
|
|
public $idrefss = array(); |
|
333
|
|
|
public $debuts = array(); |
|
334
|
|
|
public $fratrie = array(); |
|
335
|
|
|
|
|
336
|
|
|
public $dtc = null; |
|
337
|
|
|
public $sax = null; |
|
338
|
|
|
public $depth = ""; |
|
339
|
|
|
public $entete = ''; |
|
340
|
|
|
public $page = ''; |
|
341
|
|
|
public $res = ""; |
|
342
|
|
|
public $err = array(); |
|
343
|
|
|
public $contenu = array(); |
|
344
|
|
|
public $ouvrant = array(); |
|
345
|
|
|
public $reperes = array(); |
|
346
|
|
|
public $process = array( |
|
347
|
|
|
'debut' => 'xml_debutElement', |
|
348
|
|
|
'fin' => 'xml_finElement', |
|
349
|
|
|
'text' => 'xml_textElement', |
|
350
|
|
|
'pi' => 'xml_piElement', |
|
351
|
|
|
'default' => 'xml_defaultElement' |
|
352
|
|
|
); |
|
353
|
|
|
} |
|
354
|
|
|
|
|
355
|
|
|
|
|
356
|
|
|
/** |
|
357
|
|
|
* Retourne une structure ValidateurXML, dont le champ "err" est un tableau |
|
358
|
|
|
* ayant comme entrees des sous-tableaux [message, ligne, colonne] |
|
359
|
|
|
* |
|
360
|
|
|
**/ |
|
361
|
|
|
function xml_valider_dist($page, $apply = false, $process = false, $doctype = '', $charset = null) { |
|
362
|
|
|
$f = new ValidateurXML($process); |
|
363
|
|
|
$sax = charger_fonction('sax', 'xml'); |
|
364
|
|
|
|
|
365
|
|
|
return $sax($page, $apply, $f, $doctype, $charset); |
|
366
|
|
|
} |
|
367
|
|
|
|
spip /
SPIP
Loading history...
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.