1 | |
---|
2 | from OFS.SimpleItem import SimpleItem |
---|
3 | from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
---|
4 | |
---|
5 | from Ft.Xml import EMPTY_NAMESPACE, Parse |
---|
6 | from Ft.Xml.Domlette import NonvalidatingReader |
---|
7 | import Ft.Xml.Domlette |
---|
8 | import cStringIO |
---|
9 | |
---|
10 | import xml.etree.ElementTree as ET |
---|
11 | |
---|
12 | import re |
---|
13 | import logging |
---|
14 | import urllib |
---|
15 | |
---|
16 | from SrvTxtUtils import getInt, getText, getHttpData |
---|
17 | |
---|
18 | def serialize(node): |
---|
19 | """returns a string containing an XML snippet of node""" |
---|
20 | s = ET.tostring(node, 'UTF-8') |
---|
21 | # snip off XML declaration |
---|
22 | if s.startswith('<?xml'): |
---|
23 | i = s.find('?>') |
---|
24 | return s[i+3:] |
---|
25 | |
---|
26 | return s |
---|
27 | |
---|
28 | |
---|
29 | def getTextFromNode(node): |
---|
30 | """get the cdata content of a node""" |
---|
31 | if node is None: |
---|
32 | return "" |
---|
33 | # ET: |
---|
34 | # text = node.text or "" |
---|
35 | # for e in node: |
---|
36 | # text += gettext(e) |
---|
37 | # if e.tail: |
---|
38 | # text += e.tail |
---|
39 | |
---|
40 | # 4Suite: |
---|
41 | nodelist=node.childNodes |
---|
42 | text = "" |
---|
43 | for n in nodelist: |
---|
44 | if n.nodeType == node.TEXT_NODE: |
---|
45 | text = text + n.data |
---|
46 | |
---|
47 | return text |
---|
48 | |
---|
49 | def serializeNode(node, encoding="utf-8"): |
---|
50 | """returns a string containing node as XML""" |
---|
51 | #s = ET.tostring(node) |
---|
52 | |
---|
53 | # 4Suite: |
---|
54 | stream = cStringIO.StringIO() |
---|
55 | Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding) |
---|
56 | s = stream.getvalue() |
---|
57 | stream.close() |
---|
58 | |
---|
59 | return s |
---|
60 | |
---|
61 | |
---|
62 | class MpdlXmlTextServer(SimpleItem): |
---|
63 | """TextServer implementation for MPDL-XML eXist server""" |
---|
64 | meta_type="MPDL-XML TextServer" |
---|
65 | |
---|
66 | manage_options=( |
---|
67 | {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'}, |
---|
68 | )+SimpleItem.manage_options |
---|
69 | |
---|
70 | manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals()) |
---|
71 | |
---|
72 | def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40): |
---|
73 | """constructor""" |
---|
74 | self.id=id |
---|
75 | self.title=title |
---|
76 | self.timeout = timeout |
---|
77 | if serverName is None: |
---|
78 | self.serverUrl = serverUrl |
---|
79 | else: |
---|
80 | self.serverUrl = "http://%s/mpdl/interface/"%serverName |
---|
81 | |
---|
82 | def getHttpData(self, url, data=None): |
---|
83 | """returns result from url+data HTTP request""" |
---|
84 | return getHttpData(url,data,timeout=self.timeout) |
---|
85 | |
---|
86 | def getServerData(self, method, data=None): |
---|
87 | """returns result from text server for method+data""" |
---|
88 | url = self.serverUrl+method |
---|
89 | return getHttpData(url,data,timeout=self.timeout) |
---|
90 | |
---|
91 | # WTF: what does this really do? can it be integrated in getPage? |
---|
92 | def getSearch(self, pageinfo=None, docinfo=None): |
---|
93 | """get search list""" |
---|
94 | logging.debug("getSearch()") |
---|
95 | docpath = docinfo['textURLPath'] |
---|
96 | url = docinfo['url'] |
---|
97 | pagesize = pageinfo['queryPageSize'] |
---|
98 | pn = pageinfo.get('searchPN',1) |
---|
99 | sn = pageinfo['sn'] |
---|
100 | highlightQuery = pageinfo['highlightQuery'] |
---|
101 | query =pageinfo['query'] |
---|
102 | queryType =pageinfo['queryType'] |
---|
103 | viewMode= pageinfo['viewMode'] |
---|
104 | tocMode = pageinfo['tocMode'] |
---|
105 | characterNormalization = pageinfo['characterNormalization'] |
---|
106 | #optionToggle = pageinfo['optionToggle'] |
---|
107 | tocPN = pageinfo['tocPN'] |
---|
108 | selfurl = self.absolute_url() |
---|
109 | data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization, urllib.quote(highlightQuery))) |
---|
110 | pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) |
---|
111 | pagedom = Parse(pagexml) |
---|
112 | |
---|
113 | """ |
---|
114 | pagedivs = pagedom.xpath("//div[@class='queryResultHits']") |
---|
115 | if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")): |
---|
116 | if len(pagedivs)>0: |
---|
117 | docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0])) |
---|
118 | s = getTextFromNode(pagedivs[0]) |
---|
119 | s1 = int(s)/10+1 |
---|
120 | try: |
---|
121 | docinfo['queryResultHits'] = int(s1) |
---|
122 | logging.debug("SEARCH ENTRIES: %s"%(s1)) |
---|
123 | except: |
---|
124 | docinfo['queryResultHits'] = 0 |
---|
125 | """ |
---|
126 | if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
---|
127 | pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
---|
128 | if len(pagedivs)>0: |
---|
129 | pagenode=pagedivs[0] |
---|
130 | links=pagenode.xpath("//a") |
---|
131 | for l in links: |
---|
132 | hrefNode = l.getAttributeNodeNS(None, u"href") |
---|
133 | if hrefNode: |
---|
134 | href = hrefNode.nodeValue |
---|
135 | if href.startswith('page-fragment.xql'): |
---|
136 | selfurl = self.absolute_url() |
---|
137 | pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN, characterNormalization)) |
---|
138 | hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
---|
139 | #logging.debug("PUREXML :%s"%(serializeNode(pagenode))) |
---|
140 | return serializeNode(pagenode) |
---|
141 | if (queryType=="fulltextMorph"): |
---|
142 | pagedivs = pagedom.xpath("//div[@class='queryResult']") |
---|
143 | if len(pagedivs)>0: |
---|
144 | pagenode=pagedivs[0] |
---|
145 | links=pagenode.xpath("//a") |
---|
146 | for l in links: |
---|
147 | hrefNode = l.getAttributeNodeNS(None, u"href") |
---|
148 | if hrefNode: |
---|
149 | href = hrefNode.nodeValue |
---|
150 | if href.startswith('page-fragment.xql'): |
---|
151 | selfurl = self.absolute_url() |
---|
152 | pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,characterNormalization)) |
---|
153 | hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
---|
154 | if href.startswith('../lt/lemma.xql'): |
---|
155 | hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_query'%(selfurl)) |
---|
156 | l.setAttributeNS(None, 'target', '_blank') |
---|
157 | l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
---|
158 | l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
---|
159 | pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']") |
---|
160 | return serializeNode(pagenode) |
---|
161 | if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): |
---|
162 | pagedivs= pagedom.xpath("//div[@class='queryResultPage']") |
---|
163 | if len(pagedivs)>0: |
---|
164 | pagenode=pagedivs[0] |
---|
165 | links=pagenode.xpath("//a") |
---|
166 | for l in links: |
---|
167 | hrefNode = l.getAttributeNodeNS(None, u"href") |
---|
168 | if hrefNode: |
---|
169 | href = hrefNode.nodeValue |
---|
170 | hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,characterNormalization)) |
---|
171 | if href.startswith('../lt/lex.xql'): |
---|
172 | hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_lex'%selfurl) |
---|
173 | l.setAttributeNS(None, 'target', '_blank') |
---|
174 | l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
---|
175 | l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
---|
176 | if href.startswith('../lt/lemma.xql'): |
---|
177 | hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%(selfurl)) |
---|
178 | l.setAttributeNS(None, 'target', '_blank') |
---|
179 | l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
---|
180 | l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
---|
181 | return serializeNode(pagenode) |
---|
182 | return "no text here" |
---|
183 | |
---|
184 | def getGisPlaces(self, docinfo=None, pageinfo=None): |
---|
185 | """ Show all Gis Places of whole Page""" |
---|
186 | xpath='//place' |
---|
187 | docpath = docinfo.get('textURLPath',None) |
---|
188 | if not docpath: |
---|
189 | return None |
---|
190 | |
---|
191 | url = docinfo['url'] |
---|
192 | selfurl = self.absolute_url() |
---|
193 | pn = pageinfo['current'] |
---|
194 | hrefList=[] |
---|
195 | myList= "" |
---|
196 | text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn)) |
---|
197 | dom = ET.fromstring(text) |
---|
198 | result = dom.findall(".//result/resultPage/place") |
---|
199 | for l in result: |
---|
200 | href = l.get("id") |
---|
201 | hrefList.append(href) |
---|
202 | # WTF: what does this do? |
---|
203 | myList = ",".join(hrefList) |
---|
204 | #logging.debug("getGisPlaces :%s"%(myList)) |
---|
205 | return myList |
---|
206 | |
---|
207 | def getAllGisPlaces (self, docinfo=None, pageinfo=None): |
---|
208 | """Show all Gis Places of whole Book """ |
---|
209 | xpath ='//echo:place' |
---|
210 | docpath =docinfo['textURLPath'] |
---|
211 | url = docinfo['url'] |
---|
212 | selfurl =self.absolute_url() |
---|
213 | pn =pageinfo['current'] |
---|
214 | hrefList=[] |
---|
215 | myList="" |
---|
216 | text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath)) |
---|
217 | dom = ET.fromstring(text) |
---|
218 | result = dom.findall(".//result/resultPage/place") |
---|
219 | |
---|
220 | for l in result: |
---|
221 | href = l.get("id") |
---|
222 | hrefList.append(href) |
---|
223 | # WTF: what does this do? |
---|
224 | myList = ",".join(hrefList) |
---|
225 | #logging.debug("getALLGisPlaces :%s"%(myList)) |
---|
226 | return myList |
---|
227 | |
---|
228 | def processPageInfo(self, dom, docinfo, pageinfo): |
---|
229 | """processes page info divs from dom and stores in docinfo and pageinfo""" |
---|
230 | # assume first second level div is pageMeta |
---|
231 | alldivs = dom.find("div") |
---|
232 | for div in alldivs: |
---|
233 | dc = div.get('class') |
---|
234 | |
---|
235 | # pageNumberOrig |
---|
236 | if dc == 'pageNumberOrig': |
---|
237 | pageinfo['pageNumberOrig'] = div.text |
---|
238 | |
---|
239 | # pageNumberOrigNorm |
---|
240 | elif dc == 'pageNumberOrigNorm': |
---|
241 | pageinfo['pageNumberOrigNorm'] = div.text |
---|
242 | |
---|
243 | # pageNumberOrigNorm |
---|
244 | elif dc == 'countFigureEntries': |
---|
245 | docinfo['numFigureEntries'] = getInt(div.text) |
---|
246 | |
---|
247 | # pageNumberOrigNorm |
---|
248 | elif dc == 'countTocEntries': |
---|
249 | # WTF: s1 = int(s)/30+1 |
---|
250 | docinfo['numTocEntries'] = getInt(div.text) |
---|
251 | |
---|
252 | # pageHeaderTitle |
---|
253 | elif dc == 'pageHeaderTitle': |
---|
254 | docinfo['pageHeaderTitle'] = div.text |
---|
255 | |
---|
256 | # numTextPages |
---|
257 | elif dc == 'countPages': |
---|
258 | np = getInt(div.text) |
---|
259 | if np > 0: |
---|
260 | docinfo['numTextPages'] = np |
---|
261 | if docinfo.get('numPages', 0) == 0: |
---|
262 | # seems to be text-only - update page count |
---|
263 | docinfo['numPages'] = np |
---|
264 | pageinfo['end'] = min(pageinfo['end'], np) |
---|
265 | pageinfo['numgroups'] = int(np / pageinfo['groupsize']) |
---|
266 | if np % pageinfo['groupsize'] > 0: |
---|
267 | pageinfo['numgroups'] += 1 |
---|
268 | |
---|
269 | return |
---|
270 | |
---|
271 | |
---|
272 | def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None): |
---|
273 | """returns single page from fulltext""" |
---|
274 | logging.debug("getTextPage mode=%s, pn=%s"%(mode,pn)) |
---|
275 | # check for cached text -- but this shouldn't be called twice |
---|
276 | if pageinfo.has_key('textPage'): |
---|
277 | logging.debug("getTextPage: using cached text") |
---|
278 | return pageinfo['textPage'] |
---|
279 | |
---|
280 | docpath = docinfo['textURLPath'] |
---|
281 | # just checking |
---|
282 | if pageinfo['current'] != pn: |
---|
283 | logging.warning("getTextPage: current!=pn!") |
---|
284 | |
---|
285 | # stuff for constructing full urls |
---|
286 | url = docinfo['url'] |
---|
287 | urlmode = docinfo['mode'] |
---|
288 | sn = pageinfo.get('sn', None) |
---|
289 | highlightQuery = pageinfo.get('highlightQuery', None) |
---|
290 | tocMode = pageinfo.get('tocMode', None) |
---|
291 | tocPN = pageinfo.get('tocPN',None) |
---|
292 | characterNormalization = pageinfo.get('characterNormalization', None) |
---|
293 | selfurl = docinfo['viewerUrl'] |
---|
294 | |
---|
295 | if mode == "text_dict": |
---|
296 | # text_dict is called textPollux in the backend |
---|
297 | textmode = "textPollux" |
---|
298 | else: |
---|
299 | textmode = mode |
---|
300 | |
---|
301 | textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
---|
302 | if highlightQuery: |
---|
303 | textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
---|
304 | |
---|
305 | # fetch the page |
---|
306 | pagexml = self.getServerData("page-fragment.xql",textParam) |
---|
307 | dom = ET.fromstring(pagexml) |
---|
308 | # extract additional info |
---|
309 | self.processPageInfo(dom, docinfo, pageinfo) |
---|
310 | # page content is in <div class="pageContent"> |
---|
311 | pagediv = None |
---|
312 | # ElementTree 1.2 in Python 2.6 can't do div[@class='pageContent'] |
---|
313 | # so we look at the second level divs |
---|
314 | alldivs = dom.findall("div") |
---|
315 | for div in alldivs: |
---|
316 | dc = div.get('class') |
---|
317 | # page content div |
---|
318 | if dc == 'pageContent': |
---|
319 | pagediv = div |
---|
320 | break |
---|
321 | |
---|
322 | # plain text mode |
---|
323 | if mode == "text": |
---|
324 | if pagediv: |
---|
325 | links = pagediv.findall(".//a") |
---|
326 | for l in links: |
---|
327 | href = l.get('href') |
---|
328 | if href and href.startswith('#note-'): |
---|
329 | href = href.replace('#note-',"?mode=%s&url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn)) |
---|
330 | l.set('href', href) |
---|
331 | |
---|
332 | return serialize(pagediv) |
---|
333 | |
---|
334 | # text-with-links mode |
---|
335 | elif mode == "text_dict": |
---|
336 | if pagediv: |
---|
337 | # check all a-tags |
---|
338 | links = pagediv.findall(".//a") |
---|
339 | for l in links: |
---|
340 | href = l.get('href') |
---|
341 | |
---|
342 | if href: |
---|
343 | # is link with href |
---|
344 | if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'): |
---|
345 | # is pollux link |
---|
346 | selfurl = self.absolute_url() |
---|
347 | # change href |
---|
348 | l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl)) |
---|
349 | # add target |
---|
350 | l.set('target', '_blank') |
---|
351 | |
---|
352 | if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'): |
---|
353 | selfurl = self.absolute_url() |
---|
354 | l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl)) |
---|
355 | l.set('target', '_blank') |
---|
356 | l.set('onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;") |
---|
357 | l.set('ondblclick', 'popupWin.focus();') |
---|
358 | |
---|
359 | if href.startswith('#note-'): |
---|
360 | l.set('href', href.replace('#note-',"?mode=%s&url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn))) |
---|
361 | |
---|
362 | return serialize(pagediv) |
---|
363 | |
---|
364 | # xml mode |
---|
365 | elif mode == "xml": |
---|
366 | if pagediv: |
---|
367 | return serialize(pagediv) |
---|
368 | |
---|
369 | # pureXml mode |
---|
370 | elif mode == "pureXml": |
---|
371 | if pagediv: |
---|
372 | return serialize(pagediv) |
---|
373 | |
---|
374 | # gis mode |
---|
375 | elif mode == "gis": |
---|
376 | name = docinfo['name'] |
---|
377 | if pagediv: |
---|
378 | # check all a-tags |
---|
379 | links = pagediv.findall(".//a") |
---|
380 | for l in links: |
---|
381 | href = l.get('href') |
---|
382 | if href: |
---|
383 | if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): |
---|
384 | l.set('href', href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)) |
---|
385 | l.set('target', '_blank') |
---|
386 | |
---|
387 | return serialize(pagediv) |
---|
388 | |
---|
389 | return "no text here" |
---|
390 | |
---|
391 | # WTF: is this needed? |
---|
392 | def getOrigPages(self, docinfo=None, pageinfo=None): |
---|
393 | logging.debug("CALLED: getOrigPages!") |
---|
394 | if not pageinfo.has_key('pageNumberOrig'): |
---|
395 | logging.warning("getOrigPages: not in pageinfo!") |
---|
396 | return None |
---|
397 | |
---|
398 | return pageinfo['pageNumberOrig'] |
---|
399 | |
---|
400 | # WTF: is this needed? |
---|
401 | def getOrigPagesNorm(self, docinfo=None, pageinfo=None): |
---|
402 | logging.debug("CALLED: getOrigPagesNorm!") |
---|
403 | if not pageinfo.has_key('pageNumberOrigNorm'): |
---|
404 | logging.warning("getOrigPagesNorm: not in pageinfo!") |
---|
405 | return None |
---|
406 | |
---|
407 | return pageinfo['pageNumberOrigNorm'] |
---|
408 | |
---|
409 | # TODO: should be getWordInfo |
---|
410 | def getTranslate(self, word=None, language=None): |
---|
411 | """translate into another languages""" |
---|
412 | data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html") |
---|
413 | return data |
---|
414 | |
---|
415 | # WTF: what does this do? |
---|
416 | def getLemma(self, lemma=None, language=None): |
---|
417 | """simular words lemma """ |
---|
418 | data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html") |
---|
419 | return data |
---|
420 | |
---|
421 | # WTF: what does this do? |
---|
422 | def getLemmaQuery(self, query=None, language=None): |
---|
423 | """simular words lemma """ |
---|
424 | data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html") |
---|
425 | return data |
---|
426 | |
---|
427 | # WTF: what does this do? |
---|
428 | def getLex(self, query=None, language=None): |
---|
429 | #simular words lemma |
---|
430 | data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) |
---|
431 | return data |
---|
432 | |
---|
433 | # WTF: what does this do? |
---|
434 | def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1): |
---|
435 | #number of |
---|
436 | docpath = docinfo['textURLPath'] |
---|
437 | pagesize = pageinfo['queryPageSize'] |
---|
438 | pn = pageinfo['searchPN'] |
---|
439 | query =pageinfo['query'] |
---|
440 | queryType =pageinfo['queryType'] |
---|
441 | tocSearch = 0 |
---|
442 | tocDiv = None |
---|
443 | |
---|
444 | pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn)) |
---|
445 | pagedom = Parse(pagexml) |
---|
446 | numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
---|
447 | tocSearch = int(getTextFromNode(numdivs[0])) |
---|
448 | tc=int((tocSearch/10)+1) |
---|
449 | return tc |
---|
450 | |
---|
451 | def getToc(self, mode="text", docinfo=None): |
---|
452 | """loads table of contents and stores XML in docinfo""" |
---|
453 | logging.debug("getToc mode=%s"%mode) |
---|
454 | if mode == "none": |
---|
455 | return docinfo |
---|
456 | |
---|
457 | if 'tocSize_%s'%mode in docinfo: |
---|
458 | # cached toc |
---|
459 | return docinfo |
---|
460 | |
---|
461 | docpath = docinfo['textURLPath'] |
---|
462 | # we need to set a result set size |
---|
463 | pagesize = 1000 |
---|
464 | pn = 1 |
---|
465 | if mode == "text": |
---|
466 | queryType = "toc" |
---|
467 | else: |
---|
468 | queryType = mode |
---|
469 | # number of entries in toc |
---|
470 | tocSize = 0 |
---|
471 | tocDiv = None |
---|
472 | # fetch full toc |
---|
473 | pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) |
---|
474 | dom = ET.fromstring(pagexml) |
---|
475 | # page content is in <div class="queryResultPage"> |
---|
476 | pagediv = None |
---|
477 | # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage'] |
---|
478 | alldivs = dom.findall("div") |
---|
479 | for div in alldivs: |
---|
480 | dc = div.get('class') |
---|
481 | # page content div |
---|
482 | if dc == 'queryResultPage': |
---|
483 | pagediv = div |
---|
484 | |
---|
485 | elif dc == 'queryResultHits': |
---|
486 | docinfo['tocSize_%s'%mode] = getInt(div.text) |
---|
487 | |
---|
488 | if pagediv: |
---|
489 | # store XML in docinfo |
---|
490 | docinfo['tocXML_%s'%mode] = ET.tostring(pagediv, 'UTF-8') |
---|
491 | |
---|
492 | return docinfo |
---|
493 | |
---|
494 | def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
---|
495 | """returns single page from the table of contents""" |
---|
496 | logging.debug("getTocPage mode=%s, pn=%s"%(mode,pn)) |
---|
497 | if mode == "text": |
---|
498 | queryType = "toc" |
---|
499 | else: |
---|
500 | queryType = mode |
---|
501 | |
---|
502 | # check for cached TOC |
---|
503 | if not docinfo.has_key('tocXML_%s'%mode): |
---|
504 | self.getToc(mode=mode, docinfo=docinfo) |
---|
505 | |
---|
506 | tocxml = docinfo.get('tocXML_%s'%mode, None) |
---|
507 | if not tocxml: |
---|
508 | logging.error("getTocPage: unable to find tocXML") |
---|
509 | return "No ToC" |
---|
510 | |
---|
511 | pagesize = int(pageinfo['tocPageSize']) |
---|
512 | url = docinfo['url'] |
---|
513 | urlmode = docinfo['mode'] |
---|
514 | selfurl = docinfo['viewerUrl'] |
---|
515 | viewMode= pageinfo['viewMode'] |
---|
516 | tocMode = pageinfo['tocMode'] |
---|
517 | tocPN = int(pageinfo['tocPN']) |
---|
518 | pn = tocPN |
---|
519 | |
---|
520 | fulltoc = ET.fromstring(tocxml) |
---|
521 | |
---|
522 | if fulltoc: |
---|
523 | # paginate |
---|
524 | start = (pn - 1) * pagesize * 2 |
---|
525 | len = pagesize * 2 |
---|
526 | del fulltoc[:start] |
---|
527 | del fulltoc[len:] |
---|
528 | tocdivs = fulltoc |
---|
529 | |
---|
530 | # check all a-tags |
---|
531 | links = tocdivs.findall(".//a") |
---|
532 | for l in links: |
---|
533 | href = l.get('href') |
---|
534 | if href: |
---|
535 | # take pn from href |
---|
536 | m = re.match(r'page-fragment\.xql.*pn=(\d+)', href) |
---|
537 | if m is not None: |
---|
538 | # and create new url |
---|
539 | l.set('href', '%s?mode=%s&url=%s&viewMode=%s&pn=%s&tocMode=%s&tocPN=%s'%(selfurl, urlmode, url, viewMode, m.group(1), tocMode, tocPN)) |
---|
540 | else: |
---|
541 | logging.warning("getTocPage: Problem with link=%s"%href) |
---|
542 | |
---|
543 | return serialize(tocdivs) |
---|
544 | |
---|
545 | |
---|
546 | def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
---|
547 | """change settings""" |
---|
548 | self.title=title |
---|
549 | self.timeout = timeout |
---|
550 | self.serverUrl = serverUrl |
---|
551 | if RESPONSE is not None: |
---|
552 | RESPONSE.redirect('manage_main') |
---|
553 | |
---|
554 | # management methods |
---|
555 | def manage_addMpdlXmlTextServerForm(self): |
---|
556 | """Form for adding""" |
---|
557 | pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self) |
---|
558 | return pt() |
---|
559 | |
---|
560 | def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
---|
561 | #def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None): |
---|
562 | """add zogiimage""" |
---|
563 | newObj = MpdlXmlTextServer(id,title,serverUrl,timeout) |
---|
564 | self.Destination()._setObject(id, newObj) |
---|
565 | if RESPONSE is not None: |
---|
566 | RESPONSE.redirect('manage_main') |
---|
567 | |
---|
568 | |
---|