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