Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.67
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:
41: def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None):
42: """get search list"""
43: docpath = docinfo['textURLPath']
44: url = docinfo['url']
1.23 abukhman 45: logging.debug("documentViewer (gettoc) docpath: %s"%(docpath))
46: logging.debug("documentViewer (gettoc) url: %s"%(url))
1.2 casties 47: pagesize = pageinfo['queryPageSize']
48: pn = pageinfo['searchPN']
49: sn = pageinfo['sn']
50: highlightQuery = pageinfo['highlightQuery']
1.34 abukhman 51: query =pageinfo['query']
1.2 casties 52: queryType =pageinfo['queryType']
53: viewMode= pageinfo['viewMode']
54: tocMode = pageinfo['tocMode']
1.24 abukhman 55: characterNormalization = pageinfo['characterNormalization']
1.2 casties 56: tocPN = pageinfo['tocPN']
57: selfurl = self.absolute_url()
58:
1.41 abukhman 59: data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,urllib.quote(highlightQuery)))
1.2 casties 60: #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)
61:
62: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
63: pagedom = Parse(pagexml)
64: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
65: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
66: if len(pagedivs)>0:
67: pagenode=pagedivs[0]
68: links=pagenode.xpath("//a")
69: for l in links:
70: hrefNode = l.getAttributeNodeNS(None, u"href")
71: if hrefNode:
72: href = hrefNode.nodeValue
73: if href.startswith('page-fragment.xql'):
74: selfurl = self.absolute_url()
1.30 abukhman 75: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
1.2 casties 76: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
77: return serializeNode(pagenode)
78: if (queryType=="fulltextMorph"):
79: pagedivs = pagedom.xpath("//div[@class='queryResult']")
80: if len(pagedivs)>0:
81: pagenode=pagedivs[0]
82: links=pagenode.xpath("//a")
83: for l in links:
84: hrefNode = l.getAttributeNodeNS(None, u"href")
85: if hrefNode:
86: href = hrefNode.nodeValue
87: if href.startswith('page-fragment.xql'):
88: selfurl = self.absolute_url()
1.30 abukhman 89: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
1.2 casties 90: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
91: if href.startswith('../lt/lemma.xql'):
92: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
93: l.setAttributeNS(None, 'target', '_blank')
94: 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 95: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 96: pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
97: return serializeNode(pagenode)
98: if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
99: pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
100: if len(pagedivs)>0:
101: pagenode=pagedivs[0]
102: links=pagenode.xpath("//a")
103: for l in links:
104: hrefNode = l.getAttributeNodeNS(None, u"href")
105: if hrefNode:
106: href = hrefNode.nodeValue
1.18 abukhman 107: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
1.2 casties 108: if href.startswith('../lt/lex.xql'):
109: hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
110: l.setAttributeNS(None, 'target', '_blank')
111: 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 112: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 113: if href.startswith('../lt/lemma.xql'):
114: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
115: l.setAttributeNS(None, 'target', '_blank')
116: 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 117: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 118: return serializeNode(pagenode)
119: return "no text here"
120:
1.58 abukhman 121: """def getNumPages(self, docinfo):
122: ""get list of pages from fulltext and put in docinfo""
1.2 casties 123: if 'numPages' in docinfo:
124: # already there
1.43 abukhman 125: return docinfo
1.2 casties 126: xquery = '//pb'
127: text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
128: docinfo['numPages'] = text.count("<pb ")
129: return docinfo
1.58 abukhman 130: """
1.43 abukhman 131: def getNumTextPages (self, docinfo):
132: """get list of pages from fulltext (texts without images) and put in docinfo"""
1.56 abukhman 133: if 'numPages' in docinfo:
1.43 abukhman 134: # allredy there
1.56 abukhman 135: return docinfo
136: xpath ='/count(//pb)'
137: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
138: dom = Parse(text)
139: result= dom.xpath("//result/resultPage")
140: docinfo['numPages']=int(getTextFromNode(result[0]))
1.43 abukhman 141: return docinfo
1.58 abukhman 142:
143: def getGisPlases(self, docinfo):
144: """ Show all Gis Places of whole Page"""
145: xpath='//place'
146: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
147: pagedom = Parse(text)
148: result =pagedom.xpath("//result/resultPage/place")
149: docinfo['place']=getTextFromNode(result[0])
150: return docinfo
151:
1.25 abukhman 152: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
1.2 casties 153: """returns single page from fulltext"""
154: docpath = docinfo['textURLPath']
155: path = docinfo['textURLPath']
156: url = docinfo['url']
157: viewMode= pageinfo['viewMode']
158: tocMode = pageinfo['tocMode']
1.20 abukhman 159: characterNormalization=pageinfo['characterNormalization']
1.2 casties 160: tocPN = pageinfo['tocPN']
161: selfurl = self.absolute_url()
162: if mode == "text_dict":
163: textmode = "textPollux"
164: else:
165: textmode = mode
1.19 abukhman 166: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.26 abukhman 167: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.2 casties 168: if highlightQuery is not None:
1.40 abukhman 169: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
1.2 casties 170:
1.38 abukhman 171: pagexml = self.getServerData("page-fragment.xql",textParam)
1.2 casties 172: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
173:
1.39 abukhman 174: pagedom = Parse(pagexml)
1.2 casties 175: # plain text mode
176: if mode == "text":
177: # first div contains text
178: pagedivs = pagedom.xpath("/div")
179: if len(pagedivs) > 0:
180: pagenode = pagedivs[0]
181: links = pagenode.xpath("//a")
182: for l in links:
183: hrefNode = l.getAttributeNodeNS(None, u"href")
184: if hrefNode:
185: href= hrefNode.nodeValue
186: if href.startswith('#note-'):
1.27 abukhman 187: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 188: return serializeNode(pagenode)
189: if mode == "xml":
190: # first div contains text
191: pagedivs = pagedom.xpath("/div")
192: if len(pagedivs) > 0:
193: pagenode = pagedivs[0]
194: return serializeNode(pagenode)
1.7 abukhman 195: if mode == "gis":
196: # first div contains text
197: pagedivs = pagedom.xpath("/div")
198: if len(pagedivs) > 0:
199: pagenode = pagedivs[0]
1.28 abukhman 200: links =pagenode.xpath("//a")
201: for l in links:
202: hrefNode =l.getAttributeNodeNS(None, u"href")
203: if hrefNode:
204: href=hrefNode.nodeValue
205: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
1.67 ! abukhman 206: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl%s'%url)
1.62 abukhman 207:
208: #hrefNode.nodeValue=href.replace('chgis','')
209: l.setAttributeNS(None, 'target', '_blank')
1.61 abukhman 210: return serializeNode(pagenode)
1.7 abukhman 211:
1.2 casties 212: if mode == "pureXml":
213: # first div contains text
214: pagedivs = pagedom.xpath("/div")
215: if len(pagedivs) > 0:
216: pagenode = pagedivs[0]
217: return serializeNode(pagenode)
218: # text-with-links mode
219: if mode == "text_dict":
220: # first div contains text
221: pagedivs = pagedom.xpath("/div")
222: if len(pagedivs) > 0:
223: pagenode = pagedivs[0]
224: # check all a-tags
225: links = pagenode.xpath("//a")
226: for l in links:
227: hrefNode = l.getAttributeNodeNS(None, u"href")
228: if hrefNode:
229: # is link with href
230: href = hrefNode.nodeValue
231: if href.startswith('lt/lex.xql'):
232: # is pollux link
233: selfurl = self.absolute_url()
234: # change href
235: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
236: # add target
237: l.setAttributeNS(None, 'target', '_blank')
238: 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 239: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 240: if href.startswith('lt/lemma.xql'):
241: selfurl = self.absolute_url()
242: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
243: l.setAttributeNS(None, 'target', '_blank')
244: 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 245: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 246: if href.startswith('#note-'):
1.19 abukhman 247: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 248: return serializeNode(pagenode)
249: return "no text here"
250:
251: def getTranslate(self, query=None, language=None):
252: """translate into another languages"""
1.5 casties 253: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 254: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
255: return data
256:
257: def getLemma(self, lemma=None, language=None):
258: """simular words lemma """
1.5 casties 259: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 260: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
261: return data
262:
263: def getLemmaNew(self, query=None, language=None):
264: """simular words lemma """
1.5 casties 265: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 266: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
267: return data
1.28 abukhman 268:
1.2 casties 269: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
270: """number of"""
271: docpath = docinfo['textURLPath']
272: pagesize = pageinfo['queryPageSize']
273: pn = pageinfo['searchPN']
1.34 abukhman 274: query =pageinfo['query']
1.2 casties 275: queryType =pageinfo['queryType']
276: tocSearch = 0
277: tocDiv = None
278:
1.32 abukhman 279: 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 280: #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)
281: pagedom = Parse(pagexml)
282: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
283: tocSearch = int(getTextFromNode(numdivs[0]))
284: tc=int((tocSearch/10)+1)
1.23 abukhman 285: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 286: return tc
287:
288: def getToc(self, mode="text", docinfo=None):
289: """loads table of contents and stores in docinfo"""
1.23 abukhman 290: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 291: if mode == "none":
292: return docinfo
293: if 'tocSize_%s'%mode in docinfo:
294: # cached toc
295: return docinfo
296:
297: docpath = docinfo['textURLPath']
298: # we need to set a result set size
299: pagesize = 1000
300: pn = 1
301: if mode == "text":
302: queryType = "toc"
303: else:
304: queryType = mode
305: # number of entries in toc
306: tocSize = 0
307: tocDiv = None
308:
309: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
310: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
311: # post-processing downloaded xml
312: pagedom = Parse(pagexml)
313: # get number of entries
314: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
315: if len(numdivs) > 0:
316: tocSize = int(getTextFromNode(numdivs[0]))
317: docinfo['tocSize_%s'%mode] = tocSize
318: return docinfo
319:
320: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
321: """returns single page from the table of contents"""
322: # TODO: this should use the cached TOC
323: if mode == "text":
324: queryType = "toc"
325: else:
326: queryType = mode
327: docpath = docinfo['textURLPath']
328: path = docinfo['textURLPath']
329: pagesize = pageinfo['tocPageSize']
330: pn = pageinfo['tocPN']
331: url = docinfo['url']
332: selfurl = self.absolute_url()
333: viewMode= pageinfo['viewMode']
1.26 abukhman 334: characterNormalization = pageinfo ['characterNormalization']
1.2 casties 335: tocMode = pageinfo['tocMode']
336: tocPN = pageinfo['tocPN']
337:
1.23 abukhman 338: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
339: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
1.2 casties 340: text = page.replace('mode=image','mode=texttool')
1.21 abukhman 341: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.19 abukhman 342: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
1.2 casties 343: return text
344:
345: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
346: """change settings"""
347: self.title=title
348: self.timeout = timeout
349: self.serverUrl = serverUrl
350: if RESPONSE is not None:
351: RESPONSE.redirect('manage_main')
352:
353: # management methods
354: def manage_addMpdlXmlTextServerForm(self):
355: """Form for adding"""
356: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
357: return pt()
358:
359: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
360: """add zogiimage"""
361: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
362: self.Destination()._setObject(id, newObj)
363: if RESPONSE is not None:
364: RESPONSE.redirect('manage_main')
365:
366:
1.4 casties 367:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>