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
8: import urllib
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:
22: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
23: """constructor"""
24: self.id=id
25: self.title=title
26: self.timeout = timeout
27: if serverName is None:
28: self.serverUrl = serverUrl
29: else:
30: self.serverUrl = "http://%s/mpdl/interface/"%serverName
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:
41: def getSearch(self, pageinfo=None, docinfo=None, lemma=None):
42: """get search list"""
43: docpath = docinfo['textURLPath']
44: url = docinfo['url']
45: pagesize = pageinfo['queryPageSize']
46: pn = pageinfo.get('searchPN',1)
47: sn = pageinfo['sn']
48: highlightQuery = pageinfo['highlightQuery']
49: query =pageinfo['query']
50: queryType =pageinfo['queryType']
51: viewMode= pageinfo['viewMode']
52: tocMode = pageinfo['tocMode']
53: characterNormalization = pageinfo['characterNormalization']
54: optionToggle = pageinfo['optionToggle']
55: tocPN = pageinfo['tocPN']
56: selfurl = self.absolute_url()
57:
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)))
59:
60: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
61: pagedom = Parse(pagexml)
62:
63: """
64: pagedivs = pagedom.xpath("//div[@class='queryResultHits']")
65: if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")):
66: if len(pagedivs)>0:
67: docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0]))
68: s = getTextFromNode(pagedivs[0])
69: s1 = int(s)/10+1
70: try:
71: docinfo['queryResultHits'] = int(s1)
72: logging.debug("SEARCH ENTRIES: %s"%(s1))
73: except:
74: docinfo['queryResultHits'] = 0
75: """
76: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
77: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
78: if len(pagedivs)>0:
79: pagenode=pagedivs[0]
80: links=pagenode.xpath("//a")
81: for l in links:
82: hrefNode = l.getAttributeNodeNS(None, u"href")
83: if hrefNode:
84: href = hrefNode.nodeValue
85: if href.startswith('page-fragment.xql'):
86: selfurl = self.absolute_url()
87: 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))
88: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
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()
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))
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;")
107: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
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
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))
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;")
124: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
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;")
129: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
130: return serializeNode(pagenode)
131: return "no text here"
132:
133:
134: def getGisPlaces(self, docinfo=None, pageinfo=None):
135: """ Show all Gis Places of whole Page"""
136: xpath='//place'
137: docpath = docinfo.get('textURLPath',None)
138: if not docpath:
139: return None
140:
141: url = docinfo['url']
142: selfurl = self.absolute_url()
143: pn = pageinfo['current']
144: hrefList=[]
145: myList= ""
146: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
147: dom = Parse(text)
148: result = dom.xpath("//result/resultPage/place")
149: for l in result:
150: hrefNode= l.getAttributeNodeNS(None, u"id")
151: href= hrefNode.nodeValue
152: hrefList.append(href)
153: myList = ",".join(hrefList)
154: logging.debug("getGisPlaces :%s"%(myList))
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")
169:
170: for l in result:
171: hrefNode = l.getAttributeNodeNS(None, u"id")
172: href= hrefNode.nodeValue
173: hrefList.append(href)
174: myList = ",".join(hrefList)
175: logging.debug("getALLGisPlaces :%s"%(myList))
176: return myList
177:
178:
179: def getTextPage(self, docinfo=None, pageinfo=None):
180: """returns single page from fulltext"""
181: docpath = docinfo['textURLPath']
182: path = docinfo['textURLPath']
183: url = docinfo.get('url',None)
184: name = docinfo.get('name',None)
185: pn =pageinfo['current']
186: #viewMode= pageinfo['viewMode']
187: sn = pageinfo['sn']
188: highlightQuery = pageinfo['highlightQuery']
189: mode = pageinfo ['viewMode']
190: tocMode = pageinfo['tocMode']
191: characterNormalization=pageinfo['characterNormalization']
192: tocPN = pageinfo['tocPN']
193: selfurl = self.absolute_url()
194: if mode == "text_dict":
195: textmode = "textPollux"
196: else:
197: textmode = mode
198:
199: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
200: if highlightQuery is not None:
201: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
202:
203: pagexml = self.getServerData("page-fragment.xql",textParam)
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']))
212:
213: #figureEntries
214: pagedivs = dom.xpath("//div[@class='countFigureEntries']")
215: if pagedivs == dom.xpath("//div[@class='countFigureEntries']"):
216: if len(pagedivs)>0:
217: docinfo['countFigureEntries'] = getTextFromNode(pagedivs[0])
218: s = getTextFromNode(pagedivs[0])
219: if s=='0':
220: try:
221: docinfo['countFigureEntries'] = int(s)
222: logging.debug("FIGURE ENTRIES: %s"%(s))
223: except:
224: docinfo['countFigureEntries'] = 0
225: else:
226: s1 = int(s)/30+1
227: try:
228: docinfo['countFigureEntries'] = int(s1)
229: logging.debug("FIGURE ENTRIES: %s"%(s1))
230: except:
231: docinfo['countFigureEntries'] = 0
232:
233: #allPlaces
234: pagedivs = dom.xpath("//div[@class='countPlaces']")
235: if pagedivs == dom.xpath("//div[@class='countPlaces']"):
236: if len(pagedivs)>0:
237: docinfo['countPlaces']= getTextFromNode(pagedivs[0])
238: s = getTextFromNode(pagedivs[0])
239: try:
240: docinfo['countPlaces'] = int(s)
241: logging.debug("PLACES HERE: %s"%(s))
242: except:
243: docinfo['countPlaces'] = 0
244:
245: #tocEntries
246: pagedivs = dom.xpath("//div[@class='countTocEntries']")
247: if pagedivs == dom.xpath("//div[@class='countTocEntries']"):
248: if len(pagedivs)>0:
249: docinfo['countTocEntries'] = int(getTextFromNode(pagedivs[0]))
250: s = getTextFromNode(pagedivs[0])
251: if s=='0':
252: try:
253: docinfo['countTocEntries'] = int(s)
254: logging.debug("TEXT ENTRIES: %s"%(s))
255: except:
256: docinfo['countTocEntries'] = 0
257: else:
258: s1 = int(s)/30+1
259: try:
260: docinfo['countTocEntries'] = int(s1)
261: logging.debug("TEXT ENTRIES: %s"%(s1))
262: except:
263: docinfo['countTocEntries'] = 0
264:
265: #numTextPages
266: pagedivs = dom.xpath("//div[@class='countPages']")
267: if pagedivs == dom.xpath("//div[@class='countPages']"):
268: if len(pagedivs)>0:
269: docinfo['numPages'] = getTextFromNode(pagedivs[0])
270: s = getTextFromNode(pagedivs[0])
271:
272: try:
273: docinfo['numPages'] = int(s)
274: logging.debug("PAGE NUMBER: %s"%(s))
275:
276: np = docinfo['numPages']
277: pageinfo['end'] = min(pageinfo['end'], np)
278: pageinfo['numgroups'] = int(np / pageinfo['groupsize'])
279: if np % pageinfo['groupsize'] > 0:
280: pageinfo['numgroups'] += 1
281: except:
282: docinfo['numPages'] = 0
283:
284: else:
285: #no full text -- init to 0
286: docinfo['pageNumberOrig'] = 0
287: docinfo['countFigureEntries'] = 0
288: docinfo['countPlaces'] = 0
289: docinfo['countTocEntries'] = 0
290: docinfo['numPages'] = 0
291: #return docinfo
292:
293: # plain text mode
294: if mode == "text":
295: # first div contains text
296: pagedivs = dom.xpath("/div")
297: if len(pagedivs) > 0:
298: pagenode = pagedivs[0]
299: links = pagenode.xpath("//a")
300: for l in links:
301: hrefNode = l.getAttributeNodeNS(None, u"href")
302: if hrefNode:
303: href= hrefNode.nodeValue
304: if href.startswith('#note-'):
305: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
306: return serializeNode(pagenode)
307: if mode == "xml":
308: # first div contains text
309: pagedivs = dom.xpath("/div")
310: if len(pagedivs) > 0:
311: pagenode = pagedivs[0]
312: return serializeNode(pagenode)
313: if mode == "gis":
314: # first div contains text
315: pagedivs = dom.xpath("/div")
316: if len(pagedivs) > 0:
317: pagenode = pagedivs[0]
318: links =pagenode.xpath("//a")
319: for l in links:
320: hrefNode =l.getAttributeNodeNS(None, u"href")
321: if hrefNode:
322: href=hrefNode.nodeValue
323: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
324: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
325: l.setAttributeNS(None, 'target', '_blank')
326: return serializeNode(pagenode)
327:
328: if mode == "pureXml":
329: # first div contains text
330: pagedivs = dom.xpath("/div")
331: if len(pagedivs) > 0:
332: pagenode = pagedivs[0]
333: return serializeNode(pagenode)
334: # text-with-links mode
335: if mode == "text_dict":
336: # first div contains text
337: pagedivs = dom.xpath("/div")
338: if len(pagedivs) > 0:
339: pagenode = pagedivs[0]
340: # check all a-tags
341: links = pagenode.xpath("//a")
342: for l in links:
343: hrefNode = l.getAttributeNodeNS(None, u"href")
344: if hrefNode:
345: # is link with href
346: href = hrefNode.nodeValue
347: if href.startswith('lt/lex.xql'):
348: # is pollux link
349: selfurl = self.absolute_url()
350: # change href
351: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
352: # add target
353: l.setAttributeNS(None, 'target', '_blank')
354: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
355: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
356: if href.startswith('lt/lemma.xql'):
357: selfurl = self.absolute_url()
358: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
359: l.setAttributeNS(None, 'target', '_blank')
360: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
361: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
362: if href.startswith('#note-'):
363: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
364: return serializeNode(pagenode)
365: return "no text here"
366:
367: def getTranslate(self, query=None, language=None):
368: """translate into another languages"""
369: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
370: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
371: return data
372:
373: def getLemma(self, lemma=None, language=None):
374: """simular words lemma """
375: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
376: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
377: return data
378:
379: def getLemmaNew(self, query=None, language=None):
380: """simular words lemma """
381: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
382: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
383: return data
384:
385: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None):
386: #number of
387: docpath = docinfo['textURLPath']
388: pagesize = pageinfo['queryPageSize']
389: pn = pageinfo['searchPN']
390: query =pageinfo['query']
391: queryType =pageinfo['queryType']
392: tocSearch = 0
393: tocDiv = None
394:
395: 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))
396: pagedom = Parse(pagexml)
397: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
398: tocSearch = int(getTextFromNode(numdivs[0]))
399: logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch))
400: tc=int((tocSearch/10)+1)
401: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
402: return tc
403:
404: def getToc(self, mode="text", docinfo=None):
405: """loads table of contents and stores in docinfo"""
406: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
407: if mode == "none":
408: return docinfo
409: if 'tocSize_%s'%mode in docinfo:
410: # cached toc
411: return docinfo
412:
413: docpath = docinfo['textURLPath']
414: # we need to set a result set size
415: pagesize = 1000
416: pn = 1
417: if mode == "text":
418: queryType = "toc"
419: else:
420: queryType = mode
421: # number of entries in toc
422: tocSize = 0
423: tocDiv = None
424:
425: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
426:
427: # post-processing downloaded xml
428: pagedom = Parse(pagexml)
429: # get number of entries
430: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
431: if len(numdivs) > 0:
432: tocSize = int(getTextFromNode(numdivs[0]))
433: docinfo['tocSize_%s'%mode] = tocSize
434: return docinfo
435:
436: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
437: """returns single page from the table of contents"""
438: # TODO: this should use the cached TOC
439: if mode == "text":
440: queryType = "toc"
441: else:
442: queryType = mode
443: docpath = docinfo['textURLPath']
444: path = docinfo['textURLPath']
445: pagesize = pageinfo['tocPageSize']
446: pn = pageinfo['tocPN']
447: url = docinfo['url']
448: selfurl = self.absolute_url()
449: viewMode= pageinfo['viewMode']
450: characterNormalization = pageinfo ['characterNormalization']
451: optionToggle =pageinfo ['optionToggle']
452: tocMode = pageinfo['tocMode']
453: tocPN = pageinfo['tocPN']
454:
455: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn))
456: 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))
457: text = page.replace('mode=image','mode=texttool')
458: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
459:
460: return text
461:
462: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
463: """change settings"""
464: self.title=title
465: self.timeout = timeout
466: self.serverUrl = serverUrl
467: if RESPONSE is not None:
468: RESPONSE.redirect('manage_main')
469:
470: # management methods
471: def manage_addMpdlXmlTextServerForm(self):
472: """Form for adding"""
473: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
474: return pt()
475:
476: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
477: """add zogiimage"""
478: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
479: self.Destination()._setObject(id, newObj)
480: if RESPONSE is not None:
481: RESPONSE.redirect('manage_main')
482:
483:
484:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>