Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.226
1.2 casties 1:
2: from OFS.SimpleItem import SimpleItem
3: from Products.PageTemplates.PageTemplateFile import PageTemplateFile
4: from Ft.Xml import EMPTY_NAMESPACE, Parse
5:
1.224 abukhman 6: import md5
1.2 casties 7: import sys
8: import logging
1.5 casties 9: import urllib
1.2 casties 10: import documentViewer
11: from documentViewer import getTextFromNode, serializeNode
12:
13: class MpdlXmlTextServer(SimpleItem):
14: """TextServer implementation for MPDL-XML eXist server"""
15: meta_type="MPDL-XML TextServer"
16:
17: manage_options=(
18: {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
19: )+SimpleItem.manage_options
20:
21: manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
22:
1.3 casties 23: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
1.2 casties 24: """constructor"""
25: self.id=id
26: self.title=title
27: self.timeout = timeout
1.3 casties 28: if serverName is None:
29: self.serverUrl = serverUrl
30: else:
31: self.serverUrl = "http://%s/mpdl/interface/"%serverName
1.2 casties 32:
33: def getHttpData(self, url, data=None):
34: """returns result from url+data HTTP request"""
35: return documentViewer.getHttpData(url,data,timeout=self.timeout)
36:
37: def getServerData(self, method, data=None):
38: """returns result from text server for method+data"""
39: url = self.serverUrl+method
40: return documentViewer.getHttpData(url,data,timeout=self.timeout)
41:
1.222 abukhman 42: def getSearch(self, pageinfo=None, docinfo=None, lemma=None):
1.2 casties 43: """get search list"""
44: docpath = docinfo['textURLPath']
45: url = docinfo['url']
46: pagesize = pageinfo['queryPageSize']
1.222 abukhman 47: pn = pageinfo.get('searchPN',1)
1.2 casties 48: sn = pageinfo['sn']
49: highlightQuery = pageinfo['highlightQuery']
1.34 abukhman 50: query =pageinfo['query']
1.2 casties 51: queryType =pageinfo['queryType']
52: viewMode= pageinfo['viewMode']
53: tocMode = pageinfo['tocMode']
1.24 abukhman 54: characterNormalization = pageinfo['characterNormalization']
1.207 abukhman 55: optionToggle = pageinfo['optionToggle']
1.2 casties 56: tocPN = pageinfo['tocPN']
57: selfurl = self.absolute_url()
1.207 abukhman 58: data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&optionToggle=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization,optionToggle ,urllib.quote(highlightQuery)))
1.2 casties 59: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
60: pagedom = Parse(pagexml)
1.222 abukhman 61:
62: """
63: pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
64: if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
65: if len(pagedivs)>0:
66: docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
67: s = getTextFromNode(pagedivs[0])
68: s1 = int(s)/10+1
69: try:
70: docinfo['queryResultHits'] = int(s1)
71: logging.debug("SEARCH ENTRIES: %s"%(s1))
72: except:
73: docinfo['queryResultHits'] = 0
74: """
1.2 casties 75: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
76: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
77: if len(pagedivs)>0:
78: pagenode=pagedivs[0]
79: links=pagenode.xpath("//a")
80: for l in links:
81: hrefNode = l.getAttributeNodeNS(None, u"href")
82: if hrefNode:
83: href = hrefNode.nodeValue
84: if href.startswith('page-fragment.xql'):
85: selfurl = self.absolute_url()
1.213 abukhman 86: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization))
1.2 casties 87: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
1.225 abukhman 88: logging.debug("PUREXML :%s"%(serializeNode(pagenode)))
1.2 casties 89: return serializeNode(pagenode)
90: if (queryType=="fulltextMorph"):
91: pagedivs = pagedom.xpath("//div[@class='queryResult']")
92: if len(pagedivs)>0:
93: pagenode=pagedivs[0]
94: links=pagenode.xpath("//a")
95: for l in links:
96: hrefNode = l.getAttributeNodeNS(None, u"href")
97: if hrefNode:
98: href = hrefNode.nodeValue
99: if href.startswith('page-fragment.xql'):
100: selfurl = self.absolute_url()
1.213 abukhman 101: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization))
1.2 casties 102: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
103: if href.startswith('../lt/lemma.xql'):
104: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
105: l.setAttributeNS(None, 'target', '_blank')
106: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 107: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 108: pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
109: return serializeNode(pagenode)
110: if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
111: pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
112: if len(pagedivs)>0:
113: pagenode=pagedivs[0]
114: links=pagenode.xpath("//a")
115: for l in links:
116: hrefNode = l.getAttributeNodeNS(None, u"href")
117: if hrefNode:
118: href = hrefNode.nodeValue
1.213 abukhman 119: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,optionToggle,characterNormalization))
1.2 casties 120: if href.startswith('../lt/lex.xql'):
121: hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
122: l.setAttributeNS(None, 'target', '_blank')
123: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 124: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 125: if href.startswith('../lt/lemma.xql'):
126: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
127: l.setAttributeNS(None, 'target', '_blank')
128: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 129: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 130: return serializeNode(pagenode)
131: return "no text here"
132:
1.222 abukhman 133:
1.89 abukhman 134: def getGisPlaces(self, docinfo=None, pageinfo=None):
1.58 abukhman 135: """ Show all Gis Places of whole Page"""
1.100 abukhman 136: xpath='//place'
1.214 casties 137: docpath = docinfo.get('textURLPath',None)
138: if not docpath:
139: return None
140:
1.89 abukhman 141: url = docinfo['url']
142: selfurl = self.absolute_url()
1.93 abukhman 143: pn = pageinfo['current']
1.127 abukhman 144: hrefList=[]
1.142 abukhman 145: myList= ""
1.100 abukhman 146: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
147: dom = Parse(text)
1.101 abukhman 148: result = dom.xpath("//result/resultPage/place")
1.72 abukhman 149: for l in result:
1.86 abukhman 150: hrefNode= l.getAttributeNodeNS(None, u"id")
1.108 abukhman 151: href= hrefNode.nodeValue
1.128 abukhman 152: hrefList.append(href)
1.145 abukhman 153: myList = ",".join(hrefList)
154: logging.debug("getGisPlaces :%s"%(myList))
1.143 abukhman 155: return myList
156:
157: def getAllGisPlaces (self, docinfo=None, pageinfo=None):
158: """Show all Gis Places of whole Book """
159: xpath ='//echo:place'
160: docpath =docinfo['textURLPath']
161: url = docinfo['url']
162: selfurl =self.absolute_url()
163: pn =pageinfo['current']
164: hrefList=[]
165: myList=""
166: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
167: dom =Parse(text)
168: result = dom.xpath("//result/resultPage/place")
1.205 abukhman 169:
1.143 abukhman 170: for l in result:
171: hrefNode = l.getAttributeNodeNS(None, u"id")
172: href= hrefNode.nodeValue
173: hrefList.append(href)
1.136 abukhman 174: myList = ",".join(hrefList)
1.145 abukhman 175: logging.debug("getALLGisPlaces :%s"%(myList))
176: return myList
1.222 abukhman 177:
1.215 abukhman 178:
1.225 abukhman 179: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None):
1.2 casties 180: """returns single page from fulltext"""
181: docpath = docinfo['textURLPath']
182: path = docinfo['textURLPath']
1.222 abukhman 183: url = docinfo.get('url',None)
184: name = docinfo.get('name',None)
185: pn =pageinfo['current']
1.223 abukhman 186: viewMode= pageinfo['viewMode']
1.196 abukhman 187: sn = pageinfo['sn']
1.187 abukhman 188: highlightQuery = pageinfo['highlightQuery']
1.225 abukhman 189: #mode = pageinfo ['viewMode']
1.2 casties 190: tocMode = pageinfo['tocMode']
1.20 abukhman 191: characterNormalization=pageinfo['characterNormalization']
1.2 casties 192: tocPN = pageinfo['tocPN']
193: selfurl = self.absolute_url()
194: if mode == "text_dict":
195: textmode = "textPollux"
196: else:
197: textmode = mode
1.222 abukhman 198:
1.193 abukhman 199: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.190 abukhman 200: if highlightQuery is not None:
1.196 abukhman 201: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
1.222 abukhman 202:
1.38 abukhman 203: pagexml = self.getServerData("page-fragment.xql",textParam)
1.222 abukhman 204: dom = Parse(pagexml)
205:
206: #original Pages
207: pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
208: if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"):
209: if len(pagedivs)>0:
210: docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0])
211: logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrig']))
1.226 ! abukhman 212:
! 213: #original Pages Norm
! 214: pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']")
! 215: if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"):
! 216: if len(pagedivs)>0:
! 217: docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0])
! 218: logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrigNorm']))
! 219:
! 220:
1.222 abukhman 221: #figureEntries
222: pagedivs = dom.xpath("//div[@class='countFigureEntries']")
223: if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
224: if len(pagedivs)>0:
225: docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
226: s = getTextFromNode(pagedivs[0])
227: if s=='0':
228: try:
229: docinfo['countFigureEntries'] = int(s)
230: logging.debug("FIGURE ENTRIES: %s"%(s))
231: except:
232: docinfo['countFigureEntries'] = 0
233: else:
234: s1 = int(s)/30+1
235: try:
236: docinfo['countFigureEntries'] = int(s1)
237: logging.debug("FIGURE ENTRIES: %s"%(s1))
238: except:
239: docinfo['countFigureEntries'] = 0
240:
241: #allPlaces
242: pagedivs = dom.xpath("//div[@class='countPlaces']")
243: if pagedivs == dom.xpath("//div[@class='countPlaces']"):
244: if len(pagedivs)>0:
245: docinfo['countPlaces']= getTextFromNode(pagedivs[0])
246: s = getTextFromNode(pagedivs[0])
247: try:
248: docinfo['countPlaces'] = int(s)
249: logging.debug("PLACES HERE: %s"%(s))
250: except:
251: docinfo['countPlaces'] = 0
252:
253: #tocEntries
254: pagedivs = dom.xpath("//div[@class='countTocEntries']")
255: if pagedivs == dom.xpath("//div[@class='countTocEntries']"):
256: if len(pagedivs)>0:
257: docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0]))
258: s = getTextFromNode(pagedivs[0])
259: if s=='0':
260: try:
261: docinfo['countTocEntries'] = int(s)
262: logging.debug("TEXT ENTRIES: %s"%(s))
263: except:
264: docinfo['countTocEntries'] = 0
265: else:
266: s1 = int(s)/30+1
267: try:
268: docinfo['countTocEntries'] = int(s1)
269: logging.debug("TEXT ENTRIES: %s"%(s1))
270: except:
271: docinfo['countTocEntries'] = 0
272:
273: #numTextPages
274: pagedivs = dom.xpath("//div[@class='countPages']")
275: if pagedivs == dom.xpath("//div[@class='countPages']"):
276: if len(pagedivs)>0:
277: docinfo['numPages'] = getTextFromNode(pagedivs[0])
278: s = getTextFromNode(pagedivs[0])
279:
280: try:
281: docinfo['numPages'] = int(s)
282: logging.debug("PAGE NUMBER: %s"%(s))
283:
284: np = docinfo['numPages']
285: pageinfo['end'] = min(pageinfo['end'], np)
286: pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
287: if np % pageinfo['groupsize'] > 0:
288: pageinfo['numgroups'] += 1
289: except:
290: docinfo['numPages'] = 0
291:
292: else:
293: #no full text -- init to 0
294: docinfo['pageNumberOrig'] = 0
295: docinfo['countFigureEntries'] = 0
296: docinfo['countPlaces'] = 0
297: docinfo['countTocEntries'] = 0
298: docinfo['numPages'] = 0
1.226 ! abukhman 299: docinfo['pageNumberOrigNorm'] = 0
1.222 abukhman 300: #return docinfo
1.2 casties 301:
302: # plain text mode
303: if mode == "text":
304: # first div contains text
1.222 abukhman 305: pagedivs = dom.xpath("/div")
1.2 casties 306: if len(pagedivs) > 0:
307: pagenode = pagedivs[0]
308: links = pagenode.xpath("//a")
309: for l in links:
310: hrefNode = l.getAttributeNodeNS(None, u"href")
311: if hrefNode:
312: href= hrefNode.nodeValue
313: if href.startswith('#note-'):
1.27 abukhman 314: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 315: return serializeNode(pagenode)
316: if mode == "xml":
317: # first div contains text
1.222 abukhman 318: pagedivs = dom.xpath("/div")
1.2 casties 319: if len(pagedivs) > 0:
320: pagenode = pagedivs[0]
321: return serializeNode(pagenode)
1.7 abukhman 322: if mode == "gis":
323: # first div contains text
1.222 abukhman 324: pagedivs = dom.xpath("/div")
1.7 abukhman 325: if len(pagedivs) > 0:
326: pagenode = pagedivs[0]
1.28 abukhman 327: links =pagenode.xpath("//a")
328: for l in links:
329: hrefNode =l.getAttributeNodeNS(None, u"href")
330: if hrefNode:
331: href=hrefNode.nodeValue
332: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
1.75 abukhman 333: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
1.62 abukhman 334: l.setAttributeNS(None, 'target', '_blank')
1.61 abukhman 335: return serializeNode(pagenode)
1.7 abukhman 336:
1.2 casties 337: if mode == "pureXml":
338: # first div contains text
1.222 abukhman 339: pagedivs = dom.xpath("/div")
1.2 casties 340: if len(pagedivs) > 0:
341: pagenode = pagedivs[0]
342: return serializeNode(pagenode)
343: # text-with-links mode
344: if mode == "text_dict":
345: # first div contains text
1.222 abukhman 346: pagedivs = dom.xpath("/div")
1.2 casties 347: if len(pagedivs) > 0:
348: pagenode = pagedivs[0]
349: # check all a-tags
350: links = pagenode.xpath("//a")
351: for l in links:
352: hrefNode = l.getAttributeNodeNS(None, u"href")
353: if hrefNode:
354: # is link with href
355: href = hrefNode.nodeValue
356: if href.startswith('lt/lex.xql'):
357: # is pollux link
358: selfurl = self.absolute_url()
359: # change href
360: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
361: # add target
362: l.setAttributeNS(None, 'target', '_blank')
363: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 364: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 365: if href.startswith('lt/lemma.xql'):
366: selfurl = self.absolute_url()
367: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
368: l.setAttributeNS(None, 'target', '_blank')
369: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 370: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 371: if href.startswith('#note-'):
1.19 abukhman 372: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 373: return serializeNode(pagenode)
374: return "no text here"
1.225 abukhman 375:
376: def getText(self, mode="text", pn=1, docinfo=None, pageinfo=None):
377: data = self.getTextPage(mode="text", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
378: logging.debug("TEXT ALL data: %s"%(data))
379: return data
380:
381: def getXML(self, mode="xml", pn=1, docinfo=None, pageinfo=None):
382: data = self.getTextPage(mode="xml", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
383: logging.debug("TEXT ALL data: %s"%(data))
384: return data
385:
386: def getDictionary(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None):
387: data = self.getTextPage(mode="text_dict", pn=1 ,docinfo=docinfo, pageinfo=pageinfo)
388: logging.debug("TEXT ALL data: %s"%(data))
389: return data
390:
1.2 casties 391: def getTranslate(self, query=None, language=None):
392: """translate into another languages"""
1.5 casties 393: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 394: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
395: return data
396:
397: def getLemma(self, lemma=None, language=None):
398: """simular words lemma """
1.5 casties 399: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 400: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
401: return data
402:
403: def getLemmaNew(self, query=None, language=None):
404: """simular words lemma """
1.5 casties 405: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 406: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
407: return data
1.28 abukhman 408:
1.206 abukhman 409: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None):
1.222 abukhman 410: #number of
1.2 casties 411: docpath = docinfo['textURLPath']
412: pagesize = pageinfo['queryPageSize']
413: pn = pageinfo['searchPN']
1.34 abukhman 414: query =pageinfo['query']
1.2 casties 415: queryType =pageinfo['queryType']
416: tocSearch = 0
417: tocDiv = None
418:
1.32 abukhman 419: 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))
1.2 casties 420: pagedom = Parse(pagexml)
421: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
422: tocSearch = int(getTextFromNode(numdivs[0]))
1.204 abukhman 423: logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
1.2 casties 424: tc=int((tocSearch/10)+1)
1.23 abukhman 425: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 426: return tc
1.222 abukhman 427:
1.2 casties 428: def getToc(self, mode="text", docinfo=None):
429: """loads table of contents and stores in docinfo"""
1.23 abukhman 430: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 431: if mode == "none":
432: return docinfo
433: if 'tocSize_%s'%mode in docinfo:
434: # cached toc
435: return docinfo
436:
437: docpath = docinfo['textURLPath']
438: # we need to set a result set size
439: pagesize = 1000
440: pn = 1
441: if mode == "text":
442: queryType = "toc"
443: else:
444: queryType = mode
445: # number of entries in toc
446: tocSize = 0
447: tocDiv = None
448:
449: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
1.222 abukhman 450:
1.2 casties 451: # post-processing downloaded xml
452: pagedom = Parse(pagexml)
453: # get number of entries
454: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
455: if len(numdivs) > 0:
456: tocSize = int(getTextFromNode(numdivs[0]))
457: docinfo['tocSize_%s'%mode] = tocSize
458: return docinfo
459:
460: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
461: """returns single page from the table of contents"""
462: # TODO: this should use the cached TOC
463: if mode == "text":
464: queryType = "toc"
465: else:
466: queryType = mode
467: docpath = docinfo['textURLPath']
468: path = docinfo['textURLPath']
469: pagesize = pageinfo['tocPageSize']
470: pn = pageinfo['tocPN']
471: url = docinfo['url']
472: selfurl = self.absolute_url()
473: viewMode= pageinfo['viewMode']
1.26 abukhman 474: characterNormalization = pageinfo ['characterNormalization']
1.206 abukhman 475: optionToggle =pageinfo ['optionToggle']
1.2 casties 476: tocMode = pageinfo['tocMode']
477: tocPN = pageinfo['tocPN']
478:
1.213 abukhman 479: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn))
480: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN))
1.2 casties 481: text = page.replace('mode=image','mode=texttool')
1.222 abukhman 482: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.225 abukhman 483: logging.debug("TEXT %s"%(text))
1.2 casties 484: return text
485:
486: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
487: """change settings"""
488: self.title=title
489: self.timeout = timeout
490: self.serverUrl = serverUrl
491: if RESPONSE is not None:
492: RESPONSE.redirect('manage_main')
493:
494: # management methods
495: def manage_addMpdlXmlTextServerForm(self):
496: """Form for adding"""
497: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
498: return pt()
499:
500: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
501: """add zogiimage"""
502: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
503: self.Destination()._setObject(id, newObj)
504: if RESPONSE is not None:
1.224 abukhman 505: RESPONSE.redirect('manage_main')
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>