transform2teisimple.xqm 15.3 KB
Newer Older
MRodz's avatar
MRodz committed
1
2
3
4
5
xquery version "3.1";

(:~ 
 : This modules handles the conversion of the Fontante-TEI/XML into TEISimple
 : for the edited text. The resulting TEISimple is the basis for the "Editerter
MRodz's avatar
MRodz committed
6
 : Text" view on the website and the book and represent the latest layer of text.
MRodz's avatar
MRodz committed
7
8
9
10
11
12
13
14
15
16
17
18
19
20
 : 
 : The resulting TEISimple orientates itself toward the so-called "SUB-Schema"
 : which is a subset for TEISimple that is recommended for future projects that
 : are held by the Göttingen State and University Library (SUB).
 : 
 : @author Michelle Rodzis
 : @version 0.1
 :)

module namespace fontane-simple="http://fontane-nb.dariah.eu/teisimple";

declare namespace tei="http://www.tei-c.org/ns/1.0";
declare namespace test="http://exist-db.org/xquery/xqsuite";

21
import module namespace config="http://textgrid.de/ns/SADE/config" at "../config/config.xqm";
MRodz's avatar
MRodz committed
22
23
24
25
26
27
28
29
import module namespace console="http://exist-db.org/xquery/console";

(:~ 
 : The main function initiates the transformation of a given notebook.
 : 
 : TODO: adapt to several input files?
 : 
 : :)
30
declare function fontane-simple:main($file as xs:string) as xs:string? {
MRodz's avatar
MRodz committed
31
32
33
34
35
36
37
  let $doc := 
    try {
      (doc("/db/sade-projects/textgrid/data/xml/data/" || $file))
    } catch * {
      (console:log("It was not possible to open the requested file " || $file))
    }
    
MRodz's avatar
MRodz committed
38
39
40
41
42
    let $front-covers := $doc//tei:sourceDoc/tei:surface[contains(@n, "front_cover")]/*
    let $back-covers := $doc//tei:sourceDoc/tei:surface[contains(@n, "back_cover")]/*
    let $content := $doc//tei:sourceDoc/tei:surface[not(contains(@n, "cover") 
        and matches(@n, "spine"))]
    
43
    let $tei := <TEI xmlns="http://www.tei-c.org/ns/1.0">
MRodz's avatar
MRodz committed
44
        {$doc//tei:teiHeader}
MRodz's avatar
MRodz committed
45
46
47
48
49
        <text>
            <front>{fontane-simple:transform($front-covers)}</front>
            <body>{fontane-simple:transform($content)}</body>
            <back>{fontane-simple:transform($back-covers)}</back>
        </text>
MRodz's avatar
MRodz committed
50
      </TEI>
51
    return xmldb:store("/db/apps/SADE/resources/xml/", "tei-simple.xml", $tei)
MRodz's avatar
MRodz committed
52
53
};

MRodz's avatar
MRodz committed
54
55
56
57
58
59
60
61
62
63
64
65
66
67
(:~ 
 : Recursivly iterates the passed nodes and converts them according to the 
 : requirements for the "Edierter Text". 
 : 
 : The requirements are part of the encoding documentation, c.f.
 : https://fontane-nb.dariah.eu/doku.html.
 : 
 : 
 : @author Michelle Rodzis
 : @param $nodes
 : @return $node()* a TEISimple element
 : 
 :)
declare function fontane-simple:transform($nodes as node()*) as node()* {
MRodz's avatar
MRodz committed
68
69
70
71
72
73
    for $node in $nodes
      return
        typeswitch ($node)
        case text()
          return
            $node
74
            
MRodz's avatar
MRodz committed
75
76
        case element(tei:graphic)
            return
77
                element tei:graphic {
MRodz's avatar
MRodz committed
78
79
                    $node/@*
                }
80
                
MRodz's avatar
MRodz committed
81
82
83
84
85
86
        case element(tei:del) return
            if($node/parent::tei:restore) then
                fontane-simple:transform($node/node())
            else if($node/descendant::tei:restore) then
                fontane-simple:transform($node/descendant::tei:restore)
            else
MRodz's avatar
MRodz committed
87
                ()
88
                
MRodz's avatar
MRodz committed
89
90
91
92
93
94
        case element(tei:restore) return
            if(count($node/child::*) = 1 and $node/child::tei:del
            and $node/ancestor::tei:del) then
                ()
            else
                fontane-simple:transform($node/node())
95
                
MRodz's avatar
MRodz committed
96
97
98
99
100
101
102
103
        case element(tei:retrace) return
            if($node/@rend) then
                element tei:seg {
                    $node/@rend,
                    fontane-simple:transform($node/node())
                }
            else
                fontane-simple:transform($node/node())
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
                
        case element(tei:add) return
            if($node/@type = "edited_text") then
                $node
            else if(not($node/@xml:id)) then
                fontane-simple:transform($node/node())
            else if(fontane-simple:is-transposed($node)) then
                let $corresp := //tei:metamark[matches(@target, $node/@xml:id)]
                let $bla := console:log($corresp)
                return
                    (fontane-simple:transform($corresp/node()),
                    fontane-simple:transform($node/node()))
            else
                fontane-simple:transform($node/node())
                
        case element(tei:addSpan) return
            if($node/@type = "edited_text") then
                $node
            else
                ()       
                
MRodz's avatar
MRodz committed
125
126
127
        case element(tei:fw)
            return
                ()
128
                
MRodz's avatar
MRodz committed
129
130
        case element(tei:line)
            return
131
132
133
134
135
136
                if(count($node/*) = 1 and $node/child::tei:fw) then
                    ()
                else if(not($node/@type = "item")) then
                    fontane-simple:transform($node/node())
                else if($node/@type = "item" and not($node/@xml:id)) then
                    element tei:item {
MRodz's avatar
MRodz committed
137
                        fontane-simple:transform($node/node())
138
139
140
141
142
143
144
                    }
                else if($node/@type = "item" 
                and fontane-simple:is-transposed($node)) then
                    ()
                else
                    fontane-simple:transform($node/node())
                        
MRodz's avatar
MRodz committed
145
146
147
148
149
150
151
        case element(tei:handShift)
            return
                (: tei:handShift without @new denote the duct of the current 
                writer or a change of writing medium :)
                if($node/@new and (fontane-simple:is-hand-contemporary($node/@new) or
                $node/@new = "#Friedrich_Fontane") or not($node/@new))
                    then
152
                        element tei:handShift {
MRodz's avatar
MRodz committed
153
154
155
156
157
158
159
160
161
162
                            $node/@*
                        }
                    else
                        ()
                
(:                if($node/following-sibling::*[1][count(./*) = 1 :)
(:                    and child::tei:fw or child::tei:stamp]):)
(:                    then:)
(:                        ():)
(:                    else:)
163
(:                        element tei:handShift {:)
MRodz's avatar
MRodz committed
164
165
166
167
168
(:                            $node/@*:)
(:                        }:)
        case element(tei:stamp)
            return
                ()
169
                
MRodz's avatar
MRodz committed
170
171
172
173
174
175
176
177
        case element(tei:seg)
            return
                if(count($node/*) = 1 and $node/child::tei:stamp)
                    then
                        ()
                else if($node/@type = "initials" or $node/@type = "monogram" or
                    $node/@style)
                    then
178
                        element tei:seg {
MRodz's avatar
MRodz committed
179
180
181
                            $node/@*,
                            fontane-simple:transform($node/node())
                        }
MRodz's avatar
MRodz committed
182
183
                else if($node/@type = "auction_number" 
                or $node/@type = "cancel")
MRodz's avatar
MRodz committed
184
185
186
187
                    then
                        ()
                    else
                        fontane-simple:transform($node/node())
188
                        
MRodz's avatar
MRodz committed
189
        case element(tei:surface)
MRodz's avatar
MRodz committed
190
            return 
MRodz's avatar
MRodz committed
191
192
193
194
                if(matches($node/@n, "cover"))
                    then
                        (if(not($node/@n = "outer_front_cover"))
                            then
195
                                element tei:pb {
MRodz's avatar
MRodz committed
196
197
198
                                    if($node/@n = "inner_front_cover" 
                                    or $node/@n = "outer_back_cover")
                                        then
199
                                            attribute type {"flyleaf"} 
MRodz's avatar
MRodz committed
200
                                        else
201
                                            attribute type {"cover"}
MRodz's avatar
MRodz committed
202
203
204
                                }
                            else
                                (),
205
206
207
208
                        element tei:div {
                            attribute facs {$node/@facs},
                            attribute xml:id {$node/@xml:id},
                            attribute type {$node/@n},
MRodz's avatar
MRodz committed
209
210
211
212
                            fontane-simple:transform($node/node())
                        })
                else if(matches($node/@n, "[0-9]{1,4}[rv]{1}"))
                    then
213
214
                        (element tei:pb{
                            attribute n {$node/@n},
MRodz's avatar
MRodz committed
215
216
                            if($node/@facs)
                                then
217
                                    attribute facs {$node/@facs}
MRodz's avatar
MRodz committed
218
219
220
221
222
223
                                else
                                    ()
                        }, 
                        (: if all descendant text nodes shoudln't be serialized, z.g. parent::fw or parent::stamp :)
                        if($node)
                            then
224
225
                                element tei:div{
                                    attribute type {"page"},
MRodz's avatar
MRodz committed
226
227
228
229
230
                                    fontane-simple:transform($node/node())
                                }
                            else
                                ())
                else if($node/@type = "label" and
MRodz's avatar
MRodz committed
231
232
233
234
                    (contains($node/@subtype, "Fontane")
                    or contains($node/@subtype, "Hersteller"))    
                )
                    then
235
236
                        element tei:div {
                            attribute type {"label"},
MRodz's avatar
MRodz committed
237
238
                            fontane-simple:transform($node/node())
                        }
MRodz's avatar
MRodz committed
239
240
241
242
                else if($node/@type = "fragment" 
                    and $node/descendant::text()[not(ancestor::tei:fw or ancestor::tei:stamp)])
                    then
                        fontane-simple:transform($node/node())
MRodz's avatar
MRodz committed
243
244
                    else
                        ()
245
                        
246
247
248
249
250
        case element(tei:gap)
            return
                element tei:gap {
                    $node/@*
                }
251
                
252
253
254
255
256
257
        case element(tei:metamark) return
            if($node/@function = "integrate" 
            or $node/@function = "authorial_note") then
                element tei:ab {
                    $node/@*
                }
258
259
260
261
262
            else if($node/@function = "paragraph") then
                element tei:ab {
                    attribute type {"paragraph-line-special"},
                    fontane-simple:transform($node/node())
                }
263
264
            else
                ()
265
266
267
268
269
        
        (: TODO: check if correct :)
        case element(tei:surplus) return
            ()
            
MRodz's avatar
MRodz committed
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
        case element(tei:zone) return
            if(matches($node/@style, "border-style:solid") 
            and not(matches($node/@style, "border-radius"))
            and not($node/@rend = "border-style:house")) then
                element tei:div {
                    attribute type {"frame"},
                    fontane-simple:transform($node/node())
                }
            else if(matches($node/@rend, "border-bottom-style:brace")) then
                (fontane-simple:transform($node/node()),
                element tei:ab {
                    attribute type {"bottom-brace"}
                })                    
            else if($node/@type = "cancel") then
                ()
            else if($node/@type = "marked_off") then
                element tei:seg {
                    $node/@type,
                    $node/@xml:id,
                    fontane-simple:transform($node/node())
                }
291
292
293
294
            else if(not($node/@xml:id)) then
                fontane-simple:transform($node/node())
            else if($node/@xml:id and fontane-simple:is-transposed($node)) then
                ()
MRodz's avatar
MRodz committed
295
296
            else
                fontane-simple:transform($node/node())
297
                
MRodz's avatar
MRodz committed
298
        case element(tei:figure) return
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
            if(count($node/child::*) = 1 and $node/child::tei:figDesc) then
                (: genealogy lines probably shouldn't be displayed, but I still have to 
                check that. in case they should be serialized, I leave the code :)
(:                if(matches($node/descendant::tei:ref, "Stammbaumverbindungslinie")) then:)
(:                    element tei:seg {:)
(:                        $node/@*,:)
(:                        fontane-simple:transform($node/node()):)
(:                    }:)
(:                else if(matches($node/descendant::tei:ref, "Schlusslinie")):)
                if(matches($node/descendant::tei:ref, "Schlusslinie"))
                    then
                        element tei:ab {
                            switch ($node/descendant::tei:ref)
                                case "horizontale einfache Schlusslinie" return 
                                    attribute type {"long-end-line"}                                    
                                case "Schlusslinie; horizontale Halbschleife von links oben nach rechts" return 
                                    attribute type {"long-end-line"}
                                case "horizontale einfache Schlusslinie (gewellt)" return 
                                    attribute type {"long-end-line-wavy"}
                                case "Schlusslinien; horizontale Schleife von links oben nach rechts unten" return 
                                    attribute type {"bottom-brace-short"}
                                default return 
                                    attribute type {"end-line"}  
                        }
                else if(matches($node/descendant::tei:ref, "Absatzlinie")
                (: in case of double paragraph lines the single lines are 
                encoded with "oberer" resp. "unterer Teil", but we only 
                serialize the encoding for the upper line :)
                and not(matches($node/descendant::tei:ref, "unterer Teil"))
                and not(matches($node/descendant::tei:figDesc, "unsicher"))) then
                    element tei:ab {
                        if(matches($node/descendant::tei:ref, "doppelt"))
                            then
                                attribute type {"short-paragraph-line-double"}
                            else
                                attribute type {"short-paragraph-line"}
                    }
                    else
                        ()
                else
                    $node
340
                    
MRodz's avatar
MRodz committed
341
342
        case element(tei:figDesc) return
            $node
343
            
MRodz's avatar
MRodz committed
344
345
        case element(tei:ref) return
            $node
346
347
348
349
350
351
352
            
        case element(tei:ptr) return
            let $target-id := substring-after($node/@target, "#")
            let $target := $node/ancestor::tei:TEI//*[matches(@xml:id, $target-id)]
            let $stripped-target := fontane-simple:strip-element($target)
            let $stripped := console:log($stripped-target)
            return fontane-simple:transform($stripped-target)
MRodz's avatar
MRodz committed
353
        default return
354
            fontane-simple:transform($node/node())
MRodz's avatar
MRodz committed
355
356
357
358
359
360
361
362
363
364
365
366
367
};

(: TODO: adapt to general XML :)
declare function fontane-simple:is-hand-contemporary($hand as xs:string) as xs:boolean {
    let $hand := replace($hand, "#", "")
    let $file := doc("/db/sade-projects/textgrid/data/xml/data/16b00.xml")
    let $handNote := $file//tei:handNote[@xml:id = $hand]
    return 
        if($handNote/@script = "contemporary")
            then
                true()
            else
                false()
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
};

declare function fontane-simple:is-transposed($node as node()) as xs:boolean {
    let $root := $node/ancestor::tei:TEI
    return
        if($root//tei:ptr[contains(@target, $node/@xml:id)]) then
            true()
        else
            false()
};

declare function fontane-simple:strip-element($node as node()) {
    element {$node/name()} {
        $node/@type,
        fontane-simple:transform($node/node())
    }
MRodz's avatar
MRodz committed
384
};