Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.211
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:
6: import sys
7: import logging
1.5 casties 8: import urllib
1.2 casties 9: import documentViewer
10: from documentViewer import getTextFromNode, serializeNode
11:
12: class MpdlXmlTextServer(SimpleItem):
13: """TextServer implementation for MPDL-XML eXist server"""
14: meta_type="MPDL-XML TextServer"
15:
16: manage_options=(
17: {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
18: )+SimpleItem.manage_options
19:
20: manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
21:
1.3 casties 22: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
1.2 casties 23: """constructor"""
24: self.id=id
25: self.title=title
26: self.timeout = timeout
1.3 casties 27: if serverName is None:
28: self.serverUrl = serverUrl
29: else:
30: self.serverUrl = "http://%s/mpdl/interface/"%serverName
1.2 casties 31:
32: def getHttpData(self, url, data=None):
33: """returns result from url+data HTTP request"""
34: return documentViewer.getHttpData(url,data,timeout=self.timeout)
35:
36: def getServerData(self, method, data=None):
37: """returns result from text server for method+data"""
38: url = self.serverUrl+method
39: return documentViewer.getHttpData(url,data,timeout=self.timeout)
40:
1.206 abukhman 41: def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None, characterNormalization=None, optionToggle=None):
1.2 casties 42: """get search list"""
43: docpath = docinfo['textURLPath']
44: url = docinfo['url']
45: pagesize = pageinfo['queryPageSize']
46: pn = pageinfo['searchPN']
47: sn = pageinfo['sn']
48: highlightQuery = pageinfo['highlightQuery']
1.34 abukhman 49: query =pageinfo['query']
1.2 casties 50: queryType =pageinfo['queryType']
51: viewMode= pageinfo['viewMode']
52: tocMode = pageinfo['tocMode']
1.24 abukhman 53: characterNormalization = pageinfo['characterNormalization']
1.207 abukhman 54: optionToggle = pageinfo['optionToggle']
1.2 casties 55: tocPN = pageinfo['tocPN']
56: selfurl = self.absolute_url()
57:
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: #page=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode,highlightQuery) ,outputUnicode=False)
60:
61: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
62: pagedom = Parse(pagexml)
63: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
64: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
65: if len(pagedivs)>0:
66: pagenode=pagedivs[0]
67: links=pagenode.xpath("//a")
68: for l in links:
69: hrefNode = l.getAttributeNodeNS(None, u"href")
70: if hrefNode:
71: href = hrefNode.nodeValue
72: if href.startswith('page-fragment.xql'):
73: selfurl = self.absolute_url()
1.208 abukhman 74: 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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle))
1.2 casties 75: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
76: return serializeNode(pagenode)
77: if (queryType=="fulltextMorph"):
78: pagedivs = pagedom.xpath("//div[@class='queryResult']")
79: if len(pagedivs)>0:
80: pagenode=pagedivs[0]
81: links=pagenode.xpath("//a")
82: for l in links:
83: hrefNode = l.getAttributeNodeNS(None, u"href")
84: if hrefNode:
85: href = hrefNode.nodeValue
86: if href.startswith('page-fragment.xql'):
87: selfurl = self.absolute_url()
1.208 abukhman 88: 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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle))
1.2 casties 89: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
90: if href.startswith('../lt/lemma.xql'):
91: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
92: l.setAttributeNS(None, 'target', '_blank')
93: 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 94: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 95: pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
96: return serializeNode(pagenode)
97: if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
98: pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
99: if len(pagedivs)>0:
100: pagenode=pagedivs[0]
101: links=pagenode.xpath("//a")
102: for l in links:
103: hrefNode = l.getAttributeNodeNS(None, u"href")
104: if hrefNode:
105: href = hrefNode.nodeValue
1.208 abukhman 106: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s'%(viewMode,tocMode,tocPN,pn,optionToggle))
1.2 casties 107: if href.startswith('../lt/lex.xql'):
108: hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
109: l.setAttributeNS(None, 'target', '_blank')
110: 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 111: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 112: if href.startswith('../lt/lemma.xql'):
113: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
114: l.setAttributeNS(None, 'target', '_blank')
115: 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 116: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 117: return serializeNode(pagenode)
118: return "no text here"
119:
1.58 abukhman 120: """def getNumPages(self, docinfo):
121: ""get list of pages from fulltext and put in docinfo""
1.2 casties 122: if 'numPages' in docinfo:
123: # already there
1.43 abukhman 124: return docinfo
1.2 casties 125: xquery = '//pb'
126: text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
127: docinfo['numPages'] = text.count("<pb ")
128: return docinfo
1.58 abukhman 129: """
1.43 abukhman 130: def getNumTextPages (self, docinfo):
131: """get list of pages from fulltext (texts without images) and put in docinfo"""
1.56 abukhman 132: if 'numPages' in docinfo:
1.43 abukhman 133: # allredy there
1.56 abukhman 134: return docinfo
135: xpath ='/count(//pb)'
136: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
137: dom = Parse(text)
138: result= dom.xpath("//result/resultPage")
1.202 abukhman 139:
1.56 abukhman 140: docinfo['numPages']=int(getTextFromNode(result[0]))
1.43 abukhman 141: return docinfo
1.205 abukhman 142:
1.89 abukhman 143: def getGisPlaces(self, docinfo=None, pageinfo=None):
1.58 abukhman 144: """ Show all Gis Places of whole Page"""
1.100 abukhman 145: xpath='//place'
1.89 abukhman 146: docpath = docinfo['textURLPath']
147: url = docinfo['url']
148: selfurl = self.absolute_url()
1.93 abukhman 149: pn = pageinfo['current']
1.127 abukhman 150: hrefList=[]
1.142 abukhman 151: myList= ""
1.100 abukhman 152: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
153: dom = Parse(text)
1.101 abukhman 154: result = dom.xpath("//result/resultPage/place")
1.72 abukhman 155: for l in result:
1.86 abukhman 156: hrefNode= l.getAttributeNodeNS(None, u"id")
1.108 abukhman 157: href= hrefNode.nodeValue
1.128 abukhman 158: hrefList.append(href)
1.145 abukhman 159: myList = ",".join(hrefList)
160: logging.debug("getGisPlaces :%s"%(myList))
1.143 abukhman 161: return myList
162:
163: def getAllGisPlaces (self, docinfo=None, pageinfo=None):
164: """Show all Gis Places of whole Book """
165: xpath ='//echo:place'
166: docpath =docinfo['textURLPath']
167: url = docinfo['url']
168: selfurl =self.absolute_url()
169: pn =pageinfo['current']
170: hrefList=[]
171: myList=""
172: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
173: dom =Parse(text)
174: result = dom.xpath("//result/resultPage/place")
1.205 abukhman 175:
1.143 abukhman 176: for l in result:
177: hrefNode = l.getAttributeNodeNS(None, u"id")
178: href= hrefNode.nodeValue
179: hrefList.append(href)
1.136 abukhman 180: myList = ",".join(hrefList)
1.145 abukhman 181: logging.debug("getALLGisPlaces :%s"%(myList))
182: return myList
183:
1.200 abukhman 184:
185: def getPDF (self, docinfo=None, pageinfo=None):
186: """Show and Save different Pages as PDF in Options"""
187: selfurl=self.absolute_url()
188: pn=pageinfo['current']
189:
1.201 abukhman 190: viewMode =pageinfo['viewMode']
191:
192: #text = ("page-fragment.xql","document=%s&mode=%s&pn=%s&export=%s"%(docinfo['textURLPath'], 'text', pn,'pdf'))
193: #text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],'text', pn,'pdf'))
194: #logging.debug("text :%s"%(text))
195: #dom =Parse(text)
196: #logging.debug("text :%s"%(text))
1.202 abukhman 197: #return text
198:
1.183 abukhman 199: def getOrigPages (self, docinfo=None, pageinfo=None):
1.146 abukhman 200: """Show original page """
1.183 abukhman 201: docpath = docinfo['textURLPath']
1.201 abukhman 202: logging.debug ("docinfo['textURLPath']=%s"%(docinfo['textURLPath']))
1.181 abukhman 203: #url = docinfo['url']
1.146 abukhman 204: selfurl = self.absolute_url()
1.174 abukhman 205: pn =pageinfo['current']
1.161 abukhman 206:
1.174 abukhman 207: viewMode= pageinfo['viewMode']
208: text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn))
1.146 abukhman 209: dom =Parse(text)
210: pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
1.184 abukhman 211: logging.debug("YYYYYYpagedivs :%s"%(pagedivs))
1.160 abukhman 212: if len(pagedivs)>0:
1.183 abukhman 213: originalPage= getTextFromNode(pagedivs[0])
1.181 abukhman 214: #return docinfo['originalPage']
1.179 abukhman 215: return originalPage
1.183 abukhman 216:
1.184 abukhman 217:
1.206 abukhman 218: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="", highlightQuery=None, sn=None, optionToggle=None):
1.2 casties 219: """returns single page from fulltext"""
220: docpath = docinfo['textURLPath']
221: path = docinfo['textURLPath']
222: url = docinfo['url']
1.73 abukhman 223: name = docinfo['name']
1.2 casties 224: viewMode= pageinfo['viewMode']
1.196 abukhman 225: sn = pageinfo['sn']
1.187 abukhman 226: highlightQuery = pageinfo['highlightQuery']
1.196 abukhman 227:
1.2 casties 228: tocMode = pageinfo['tocMode']
1.20 abukhman 229: characterNormalization=pageinfo['characterNormalization']
1.2 casties 230: tocPN = pageinfo['tocPN']
231: selfurl = self.absolute_url()
232: if mode == "text_dict":
233: textmode = "textPollux"
234: else:
235: textmode = mode
1.196 abukhman 236: #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
1.193 abukhman 237: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.190 abukhman 238: if highlightQuery is not None:
1.196 abukhman 239: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
1.198 abukhman 240: #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
1.38 abukhman 241: pagexml = self.getServerData("page-fragment.xql",textParam)
1.198 abukhman 242: logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
1.2 casties 243: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
244:
1.39 abukhman 245: pagedom = Parse(pagexml)
1.2 casties 246: # plain text mode
247: if mode == "text":
248: # first div contains text
249: pagedivs = pagedom.xpath("/div")
250: if len(pagedivs) > 0:
251: pagenode = pagedivs[0]
252: links = pagenode.xpath("//a")
253: for l in links:
254: hrefNode = l.getAttributeNodeNS(None, u"href")
255: if hrefNode:
256: href= hrefNode.nodeValue
257: if href.startswith('#note-'):
1.27 abukhman 258: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 259: return serializeNode(pagenode)
260: if mode == "xml":
261: # first div contains text
262: pagedivs = pagedom.xpath("/div")
263: if len(pagedivs) > 0:
264: pagenode = pagedivs[0]
265: return serializeNode(pagenode)
1.7 abukhman 266: if mode == "gis":
267: # first div contains text
268: pagedivs = pagedom.xpath("/div")
269: if len(pagedivs) > 0:
270: pagenode = pagedivs[0]
1.28 abukhman 271: links =pagenode.xpath("//a")
272: for l in links:
273: hrefNode =l.getAttributeNodeNS(None, u"href")
274: if hrefNode:
275: href=hrefNode.nodeValue
276: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
1.75 abukhman 277: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
1.62 abukhman 278: l.setAttributeNS(None, 'target', '_blank')
1.61 abukhman 279: return serializeNode(pagenode)
1.7 abukhman 280:
1.2 casties 281: if mode == "pureXml":
282: # first div contains text
283: pagedivs = pagedom.xpath("/div")
284: if len(pagedivs) > 0:
285: pagenode = pagedivs[0]
286: return serializeNode(pagenode)
287: # text-with-links mode
288: if mode == "text_dict":
289: # first div contains text
290: pagedivs = pagedom.xpath("/div")
291: if len(pagedivs) > 0:
292: pagenode = pagedivs[0]
293: # check all a-tags
294: links = pagenode.xpath("//a")
295: for l in links:
296: hrefNode = l.getAttributeNodeNS(None, u"href")
297: if hrefNode:
298: # is link with href
299: href = hrefNode.nodeValue
300: if href.startswith('lt/lex.xql'):
301: # is pollux link
302: selfurl = self.absolute_url()
303: # change href
304: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
305: # add target
306: l.setAttributeNS(None, 'target', '_blank')
307: 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 308: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 309: if href.startswith('lt/lemma.xql'):
310: selfurl = self.absolute_url()
311: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
312: l.setAttributeNS(None, 'target', '_blank')
313: 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 314: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 315: if href.startswith('#note-'):
1.19 abukhman 316: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 317: return serializeNode(pagenode)
318: return "no text here"
319:
320: def getTranslate(self, query=None, language=None):
321: """translate into another languages"""
1.5 casties 322: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 323: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
324: return data
325:
326: def getLemma(self, lemma=None, language=None):
327: """simular words lemma """
1.5 casties 328: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 329: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
330: return data
331:
332: def getLemmaNew(self, query=None, language=None):
333: """simular words lemma """
1.5 casties 334: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 335: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
336: return data
1.28 abukhman 337:
1.206 abukhman 338: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None):
1.2 casties 339: """number of"""
340: docpath = docinfo['textURLPath']
341: pagesize = pageinfo['queryPageSize']
342: pn = pageinfo['searchPN']
1.34 abukhman 343: query =pageinfo['query']
1.2 casties 344: queryType =pageinfo['queryType']
345: tocSearch = 0
346: tocDiv = None
347:
1.32 abukhman 348: 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 349: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False)
350: pagedom = Parse(pagexml)
351: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
352: tocSearch = int(getTextFromNode(numdivs[0]))
1.204 abukhman 353: logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
1.2 casties 354: tc=int((tocSearch/10)+1)
1.23 abukhman 355: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 356: return tc
357:
1.205 abukhman 358: def getQueryResultHits(self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionsClose=None):
359:
360: """number of hits in Search mode"""
361: docpath = docinfo['textURLPath']
362: pagesize = pageinfo['queryPageSize']
363: pn = pageinfo['searchPN']
364: query =pageinfo['query']
365: queryType =pageinfo['queryType']
366: tocSearch = 0
367: tocDiv = None
368:
369: 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))
370: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False)
371: pagedom = Parse(pagexml)
372: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
1.210 abukhman 373: tocSearch = int(getTextFromNode(numdivs[0]))
374: tc=int((tocSearch/10)+1)
375: return tc
1.205 abukhman 376:
377: def getQueryResultHitsText(self, docinfo=None, pageinfo=None):
378: """number of hits in Text of Contents mode"""
379:
380: docpath = docinfo['textURLPath']
381: pagesize = pageinfo['queryPageSize']
382: pn = pageinfo['searchPN']
383: query =pageinfo['query']
384: queryType =pageinfo['queryType']
385: tocSearch = 0
386: tocDiv = None
387: tocMode = pageinfo['tocMode']
388: tocPN = pageinfo['tocPN']
389: pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'toc'))
390: pagedom = Parse(pagexml)
391: logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
392: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
1.210 abukhman 393: tocSearch = int(getTextFromNode(numdivs[0]))
1.211 ! abukhman 394: tc=int((tocSearch/30)+1)
1.210 abukhman 395: return tc
1.205 abukhman 396:
397: def getQueryResultHitsFigures(self, docinfo=None, pageinfo=None):
398: """number of hits in Text of Figures mode"""
399:
400: docpath = docinfo['textURLPath']
401: pagesize = pageinfo['queryPageSize']
402: pn = pageinfo['searchPN']
403: query =pageinfo['query']
404: queryType =pageinfo['queryType']
405: tocSearch = 0
406: tocDiv = None
407: tocMode = pageinfo['tocMode']
408: tocPN = pageinfo['tocPN']
409: pagexml=self.getServerData("doc-query.xql", "document=%s&queryType=%s"%(docpath,'figures'))
410: pagedom = Parse(pagexml)
411: logging.debug("documentViewer (pagedom) pagedom: %s"%(pagedom))
412: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
1.210 abukhman 413: tocSearch = int(getTextFromNode(numdivs[0]))
1.211 ! abukhman 414: tc=int((tocSearch/30)+1)
1.210 abukhman 415: return tc
1.205 abukhman 416:
417:
1.2 casties 418: def getToc(self, mode="text", docinfo=None):
419: """loads table of contents and stores in docinfo"""
1.23 abukhman 420: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 421: if mode == "none":
422: return docinfo
423: if 'tocSize_%s'%mode in docinfo:
424: # cached toc
425: return docinfo
426:
427: docpath = docinfo['textURLPath']
428: # we need to set a result set size
429: pagesize = 1000
430: pn = 1
431: if mode == "text":
432: queryType = "toc"
433: else:
434: queryType = mode
435: # number of entries in toc
436: tocSize = 0
437: tocDiv = None
438:
439: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
440: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
441: # post-processing downloaded xml
442: pagedom = Parse(pagexml)
443: # get number of entries
444: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
445: if len(numdivs) > 0:
446: tocSize = int(getTextFromNode(numdivs[0]))
447: docinfo['tocSize_%s'%mode] = tocSize
448: return docinfo
449:
450: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
451: """returns single page from the table of contents"""
452: # TODO: this should use the cached TOC
453: if mode == "text":
454: queryType = "toc"
455: else:
456: queryType = mode
457: docpath = docinfo['textURLPath']
458: path = docinfo['textURLPath']
459: pagesize = pageinfo['tocPageSize']
460: pn = pageinfo['tocPN']
461: url = docinfo['url']
462: selfurl = self.absolute_url()
463: viewMode= pageinfo['viewMode']
1.26 abukhman 464: characterNormalization = pageinfo ['characterNormalization']
1.206 abukhman 465: optionToggle =pageinfo ['optionToggle']
1.2 casties 466: tocMode = pageinfo['tocMode']
467: tocPN = pageinfo['tocPN']
468:
1.206 abukhman 469: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s&optionToggle=%s"%(docpath,queryType, pagesize, pn,characterNormalization,optionToggle))
1.209 abukhman 470: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=%s'%(selfurl,url, viewMode, tocMode, tocPN,optionToggle))
1.2 casties 471: text = page.replace('mode=image','mode=texttool')
1.21 abukhman 472: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.19 abukhman 473: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
1.2 casties 474: return text
475:
476: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
477: """change settings"""
478: self.title=title
479: self.timeout = timeout
480: self.serverUrl = serverUrl
481: if RESPONSE is not None:
482: RESPONSE.redirect('manage_main')
483:
484: # management methods
485: def manage_addMpdlXmlTextServerForm(self):
486: """Form for adding"""
487: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
488: return pt()
489:
490: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
491: """add zogiimage"""
492: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
493: self.Destination()._setObject(id, newObj)
494: if RESPONSE is not None:
495: RESPONSE.redirect('manage_main')
496:
497:
1.4 casties 498:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>