--- documentViewer/documentViewer.py 2010/10/15 09:58:12 1.121 +++ documentViewer/documentViewer.py 2012/01/04 07:38:17 1.178 @@ -7,7 +7,7 @@ from AccessControl import getSecurityMan from Globals import package_home from Products.zogiLib.zogiLib import browserCheck -from Ft.Xml import EMPTY_NAMESPACE, Parse +from Ft.Xml import EMPTY_NAMESPACE, Parse import Ft.Xml.Domlette import os.path import sys @@ -18,7 +18,7 @@ import math import urlparse import cStringIO import re - +import string def logger(txt,method,txt2): """logging""" @@ -43,23 +43,76 @@ def getTextFromNode(nodename): rc = rc + node.data return rc -def serializeNode(node, encoding='utf-8'): +def serializeNode(node, encoding="utf-8"): """returns a string containing node as XML""" - buf = cStringIO.StringIO() - Ft.Xml.Domlette.Print(node, stream=buf, encoding=encoding) - s = buf.getvalue() - buf.close() + stream = cStringIO.StringIO() + #logging.debug("BUF: %s"%(stream)) + Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding) + s = stream.getvalue() + #logging.debug("BUF: %s"%(s)) + stream.close() return s -def getBrowserType(self): - """get browser type object""" - if self.REQUEST.SESSION.has_key('browserType'): - return self.REQUEST.SESSION['browserType'] - else: - bt = browserCheck(self) - self.REQUEST.SESSION.set('browserType', bt) - logging.debug("documentViewer (BROWSER TYPE) bt %s"%bt) - return bt +def browserCheck(self): + """check the browsers request to find out the browser type""" + bt = {} + ua = self.REQUEST.get_header("HTTP_USER_AGENT") + bt['ua'] = ua + bt['isIE'] = False + bt['isN4'] = False + bt['versFirefox']="" + bt['versIE']="" + bt['versSafariChrome']="" + bt['versOpera']="" + + if string.find(ua, 'MSIE') > -1: + bt['isIE'] = True + else: + bt['isN4'] = (string.find(ua, 'Mozilla/4.') > -1) + # Safari oder Chrome identification + try: + nav = ua[string.find(ua, '('):] + nav1=ua[string.find(ua,')'):] + nav2=nav1[string.find(nav1,'('):] + nav3=nav2[string.find(nav2,')'):] + ie = string.split(nav, "; ")[1] + ie1 =string.split(nav1, " ")[2] + ie2 =string.split(nav3, " ")[1] + ie3 =string.split(nav3, " ")[2] + if string.find(ie3, "Safari") >-1: + bt['versSafariChrome']=string.split(ie2, "/")[1] + except: pass + # IE identification + try: + nav = ua[string.find(ua, '('):] + ie = string.split(nav, "; ")[1] + if string.find(ie, "MSIE") > -1: + bt['versIE'] = string.split(ie, " ")[1] + except:pass + # Firefox identification + try: + nav = ua[string.find(ua, '('):] + nav1=ua[string.find(ua,')'):] + if string.find(ie1, "Firefox") >-1: + nav5= string.split(ie1, "/")[1] + logging.debug("FIREFOX: %s"%(nav5)) + bt['versFirefox']=nav5[0:3] + except:pass + #Opera identification + try: + if string.find(ua,"Opera") >-1: + nav = ua[string.find(ua, '('):] + nav1=nav[string.find(nav,')'):] + bt['versOpera']=string.split(nav1,"/")[2] + except:pass + + bt['isMac'] = string.find(ua, 'Macintosh') > -1 + bt['isWin'] = string.find(ua, 'Windows') > -1 + bt['isIEWin'] = bt['isIE'] and bt['isWin'] + bt['isIEMac'] = bt['isIE'] and bt['isMac'] + bt['staticHTML'] = False + + return bt def getParentDir(path): @@ -110,8 +163,6 @@ def getHttpData(url, data=None, num_trie raise IOError("ERROR fetching HTTP data from %s: %s"%(url,errmsg)) #return None - - ## ## documentViewer class ## @@ -130,10 +181,12 @@ class documentViewer(Folder): toc_text = PageTemplateFile('zpt/toc_text', globals()) toc_figures = PageTemplateFile('zpt/toc_figures', globals()) page_main_images = PageTemplateFile('zpt/page_main_images', globals()) + page_main_double = PageTemplateFile('zpt/page_main_double', globals()) page_main_text = PageTemplateFile('zpt/page_main_text', globals()) page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals()) page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals()) page_main_xml = PageTemplateFile('zpt/page_main_xml', globals()) + page_main_pureXml = PageTemplateFile('zpt/page_main_pureXml', globals()) head_main = PageTemplateFile('zpt/head_main', globals()) docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) info_xml = PageTemplateFile('zpt/info_xml', globals()) @@ -178,10 +231,18 @@ class documentViewer(Folder): """get page""" return self.template.fulltextclient.getTextPage(**args) + def getOrigPages(self, **args): + """get page""" + return self.template.fulltextclient.getOrigPages(**args) + + def getOrigPagesNorm(self, **args): + """get page""" + return self.template.fulltextclient.getOrigPagesNorm(**args) + def getQuery(self, **args): - """get query""" + """get query in search""" return self.template.fulltextclient.getQuery(**args) - + def getSearch(self, **args): """get search""" return self.template.fulltextclient.getSearch(**args) @@ -191,17 +252,9 @@ class documentViewer(Folder): return self.template.fulltextclient.getGisPlaces(**args) def getAllGisPlaces(self, **args): - """get all gis places""" - return self.template.fulltextclient.getGisPlaces(**args) - - def getNumPages(self, docinfo): - """get numpages""" - return self.template.fulltextclient.getNumPages(docinfo) - - def getNumTextPages(self, docinfo): - """get numpages text""" - return self.template.fulltextclient.getNumTextPages(docinfo) - + """get all gis places """ + return self.template.fulltextclient.getAllGisPlaces(**args) + def getTranslate(self, **args): """get translate""" return self.template.fulltextclient.getTranslate(**args) @@ -210,6 +263,14 @@ class documentViewer(Folder): """get lemma""" return self.template.fulltextclient.getLemma(**args) + def getLemmaQuery(self, **args): + """get query""" + return self.template.fulltextclient.getLemmaQuery(**args) + + def getLex(self, **args): + """get lex""" + return self.template.fulltextclient.getLex(**args) + def getToc(self, **args): """get toc""" return self.template.fulltextclient.getToc(**args) @@ -239,11 +300,13 @@ class documentViewer(Folder): self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" docinfo = self.getDocinfo(mode=mode,url=url) - pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) + #pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) + pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo) + ''' ZDES ''' pt = getattr(self.template, 'thumbs_main_rss') if viewMode=="auto": # automodus gewaehlt - if docinfo.has_key("textURL") or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert + if docinfo.has_key("textURL") or docinfo.get('textURLPath',None): #texturl gesetzt und textViewer konfiguriert viewMode="text" else: viewMode="images" @@ -251,7 +314,7 @@ class documentViewer(Folder): return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) security.declareProtected('View','index_html') - def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None, characterNormalization=""): + def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None): ''' view it @param mode: defines how to access the document behind url @@ -279,14 +342,18 @@ class documentViewer(Folder): docinfo = self.getToc(mode=tocMode, docinfo=docinfo) if viewMode=="auto": # automodus gewaehlt - if docinfo.has_key('textURL') or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert + if docinfo.has_key('textURL') or docinfo.get('textURLPath',None): #texturl gesetzt und textViewer konfiguriert viewMode="text_dict" else: viewMode="images" - pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) + pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) - pt = getattr(self.template, 'viewer_main') + if (docinfo.get('textURLPath',None)): + page = self.getTextPage(docinfo=docinfo, pageinfo=pageinfo) + pageinfo['textPage'] = page + tt = getattr(self, 'template') + pt = getattr(tt, 'viewer_main') return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) def generateMarks(self,mk): @@ -298,8 +365,14 @@ class documentViewer(Folder): for m in mk: ret+="mk=%s"%m return ret - - + + + def getBrowser(self): + """getBrowser the version of browser """ + bt = browserCheck(self) + logging.debug("BROWSER VERSION: %s"%(bt)) + return bt + def findDigilibUrl(self): """try to get the digilib URL from zogilib""" url = self.template.zogilib.getDLBaseUrl() @@ -317,41 +390,48 @@ class documentViewer(Folder): else: return style - def getLink(self,param=None,val=None): - """link to documentviewer with parameter param set to val""" - params=self.REQUEST.form.copy() + def getLink(self, param=None, val=None, params=None, baseUrl=None, paramSep='&'): + """returns URL to documentviewer with parameter param set to val or from dict params""" + # copy existing request params + urlParams=self.REQUEST.form.copy() + # change single param if param is not None: if val is None: - if params.has_key(param): - del params[param] + if urlParams.has_key(param): + del urlParams[param] else: - params[param] = str(val) + urlParams[param] = str(val) - if params.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath - params["mode"] = "imagepath" - params["url"] = getParentDir(params["url"]) + # change more params + if params is not None: + for k in params.keys(): + v = params[k] + if v is None: + # val=None removes param + if urlParams.has_key(k): + del urlParams[k] + + else: + urlParams[k] = v + + # FIXME: does this belong here? + if urlParams.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath + urlParams["mode"] = "imagepath" + urlParams["url"] = getParentDir(urlParams["url"]) - # quote values and assemble into query string - #ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) - ps = urllib.urlencode(params) - url=self.REQUEST['URL1']+"?"+ps + # quote values and assemble into query string (not escaping '/') + ps = paramSep.join(["%s=%s"%(k,urllib.quote_plus(v,'/')) for (k, v) in urlParams.items()]) + #ps = urllib.urlencode(urlParams) + if baseUrl is None: + baseUrl = self.REQUEST['URL1'] + + url = "%s?%s"%(baseUrl, ps) return url - def getLinkAmp(self,param=None,val=None): + + def getLinkAmp(self, param=None, val=None, params=None, baseUrl=None): """link to documentviewer with parameter param set to val""" - params=self.REQUEST.form.copy() - if param is not None: - if val is None: - if params.has_key(param): - del params[param] - else: - params[param] = str(val) - - # quote values and assemble into query string - logging.debug("XYXXXXX: %s"%repr(params.items())) - ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) - url=self.REQUEST['URL1']+"?"+ps - return url + return self.getLink(param, val, params, baseUrl, '&') def getInfo_xml(self,url,mode): """returns info about the document as XML""" @@ -363,6 +443,20 @@ class documentViewer(Folder): pt = getattr(self.template, 'info_xml') return pt(docinfo=docinfo) + def getOptionToggle(self, newState=None, optionName='text_options_open', initialState=True): + """returns new option state""" + if not self.REQUEST.SESSION.has_key(optionName): + # not in session -- initial + opt = {'lastState': newState, 'state': initialState} + else: + opt = self.REQUEST.SESSION.get(optionName) + if opt['lastState'] != newState: + # state in session has changed -- toggle + opt['state'] = not opt['state'] + opt['lastState'] = newState + + self.REQUEST.SESSION[optionName] = opt + return opt['state'] def isAccessible(self, docinfo): """returns if access to the resource is granted""" @@ -501,7 +595,7 @@ class documentViewer(Folder): def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): """gets bibliographical info from the index.meta file at path or given by dom""" - logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path)) + #logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path)) if docinfo is None: docinfo = {} @@ -513,7 +607,7 @@ class documentViewer(Folder): docinfo['indexMetaPath']=self.getIndexMetaPath(path); - logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path)) + #logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path)) # put in all raw bib fields as dict "bib" bib = dom.xpath("//bib/*") if bib and len(bib)>0: @@ -533,8 +627,8 @@ class documentViewer(Folder): bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC) docinfo['bib_type'] = bibtype bibmap=metaData.generateMappingForType(bibtype) - logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap)) - logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype)) + #logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap)) + #logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype)) # if there is no mapping bibmap is empty (mapping sometimes has empty fields) if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: try: @@ -546,15 +640,55 @@ class documentViewer(Folder): try: docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0]) except: pass - logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) + #logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) try: docinfo['lang']=getTextFromNode(dom.xpath("//bib/lang")[0]) except: docinfo['lang']='' - + try: + docinfo['city']=getTextFromNode(dom.xpath("//bib/city")[0]) + except: + docinfo['city']='' + try: + docinfo['number_of_pages']=getTextFromNode(dom.xpath("//bib/number_of_pages")[0]) + except: + docinfo['number_of_pages']='' + try: + docinfo['series_volume']=getTextFromNode(dom.xpath("//bib/series_volume")[0]) + except: + docinfo['series_volume']='' + try: + docinfo['number_of_volumes']=getTextFromNode(dom.xpath("//bib/number_of_volumes")[0]) + except: + docinfo['number_of_volumes']='' + try: + docinfo['translator']=getTextFromNode(dom.xpath("//bib/translator")[0]) + except: + docinfo['translator']='' + try: + docinfo['edition']=getTextFromNode(dom.xpath("//bib/edition")[0]) + except: + docinfo['edition']='' + try: + docinfo['series_author']=getTextFromNode(dom.xpath("//bib/series_author")[0]) + except: + docinfo['series_author']='' + try: + docinfo['publisher']=getTextFromNode(dom.xpath("//bib/publisher")[0]) + except: + docinfo['publisher']='' + try: + docinfo['series_title']=getTextFromNode(dom.xpath("//bib/series_title")[0]) + except: + docinfo['series_title']='' + try: + docinfo['isbn_issn']=getTextFromNode(dom.xpath("//bib/isbn_issn")[0]) + except: + docinfo['isbn_issn']='' + #logging.debug("I NEED BIBTEX %s"%docinfo) return docinfo - + def getNameFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): """gets name info from the index.meta file at path or given by dom""" if docinfo is None: @@ -564,17 +698,9 @@ class documentViewer(Folder): for x in range(cut): path=getParentDir(path) dom = self.getDomFromIndexMeta(path) - - #docinfo['indexMetaPath']=self.getIndexMetaPath(path); - - #result= dom.xpath("//result/resultPage") - #docinfo['numPages']=int(getTextFromNode(result[0])) - - #result =dom.xpath("//name") + docinfo['name']=getTextFromNode(dom.xpath("/resource/name")[0]) logging.debug("documentViewer docinfo[name] %s"%docinfo['name']) - - #logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) return docinfo def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): @@ -651,20 +777,38 @@ class documentViewer(Folder): textUrl = textUrl.replace("/mpiwg/online", '', 1) docinfo['textURL'] = textUrl - + + + #TODO: hack-DW for annalen + if (textUrl is not None) and (textUrl.startswith("/permanent/einstein/annalen")): + textUrl=textUrl.replace("/permanent/einstein/annalen/","/diverse/de/") + splitted=textUrl.split("/fulltext") + textUrl=splitted[0]+".xml" + textUrlkurz = string.split(textUrl, ".")[0] + docinfo['textURLPathkurz'] = textUrlkurz + docinfo['textURLPath'] = textUrl + logging.debug("hack") + logging.debug(textUrl) + + # new style text-url-path textUrls = dom.xpath("//texttool/text-url-path") if textUrls and (len(textUrls) > 0): textUrl = getTextFromNode(textUrls[0]) docinfo['textURLPath'] = textUrl - if not docinfo['imagePath']: + textUrlkurz = string.split(textUrl, ".")[0] + docinfo['textURLPathkurz'] = textUrlkurz + #if not docinfo['imagePath']: # text-only, no page images - docinfo = self.getNumTextPages(docinfo) + #docinfo = self.getNumTextPages(docinfo) + presentationUrls = dom.xpath("//texttool/presentation") docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag + #docinfo = self.getDownloadfromDocinfoToBibtex(url, docinfo=docinfo, dom=dom) docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom) + if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen # presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten # durch den relativen Pfad auf die presentation infos @@ -717,6 +861,7 @@ class documentViewer(Folder): #path ist the path to the images it assumes that the index.meta file is one level higher. docinfo = self.getBibinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) + #docinfo = self.getDownloadfromDocinfoToBibtex(pathorig,docinfo=docinfo,cut=cut+1) docinfo = self.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) return docinfo @@ -742,15 +887,21 @@ class documentViewer(Folder): else: logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode) raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) - + + # FIXME: fake texturlpath + if not docinfo.has_key('textURLPath'): + docinfo['textURLPath'] = None + logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo) + #logging.debug("documentViewer (getdocinfo) docinfo: %s"%) self.REQUEST.SESSION['docinfo'] = docinfo return docinfo - def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None,characterNormalization=""): + def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None): """returns pageinfo with the given parameters""" pageinfo = {} current = getInt(current) + pageinfo['current'] = current rows = int(rows or self.thumbrows) pageinfo['rows'] = rows @@ -770,16 +921,22 @@ class documentViewer(Folder): pageinfo['numgroups'] += 1 pageinfo['viewMode'] = viewMode pageinfo['tocMode'] = tocMode - #pageinfo['characterNormalization'] =characterNormalization - pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization',' ') - pageinfo['query'] = self.REQUEST.get('query',' ') - pageinfo['queryType'] = self.REQUEST.get('queryType',' ') + pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization','reg') + #pageinfo['optionToggle'] = self.REQUEST.get('optionToggle','1') + pageinfo['query'] = self.REQUEST.get('query','') + pageinfo['queryType'] = self.REQUEST.get('queryType','') pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') pageinfo['textPN'] = self.REQUEST.get('textPN','1') pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','') + + pageinfo ['highlightElementPos'] = self.REQUEST.get('highlightElementPos','') + pageinfo ['highlightElement'] = self.REQUEST.get('highlightElement','') + + pageinfo ['xpointer'] = self.REQUEST.get('xpointer','') + pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '10') - pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') + pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') toc = int (pageinfo['tocPN']) pageinfo['textPages'] =int (toc) @@ -793,7 +950,8 @@ class documentViewer(Folder): tocPages=tocSize/tocPageSize pageinfo['tocPN'] = min (tocPages,toc) pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') - pageinfo['sn'] =self.REQUEST.get('sn','') + #pageinfo['sn'] =self.REQUEST.get('sn','') + pageinfo['s'] =self.REQUEST.get('s','') return pageinfo def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None):