version 1.1, 2005/12/18 12:35:02
|
version 1.157, 2010/11/16 11:31:57
|
Line 1
|
Line 1
|
|
|
genericDigilib="http://nausikaa2.rz-berlin.mpg.de/digitallibrary/" |
|
|
|
from OFS.Folder import Folder |
from OFS.Folder import Folder |
from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate |
from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate |
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
from AccessControl import ClassSecurityInfo |
from AccessControl import ClassSecurityInfo |
|
from AccessControl import getSecurityManager |
from Globals import package_home |
from Globals import package_home |
|
from Products.zogiLib.zogiLib import browserCheck |
|
|
from Ft.Xml.Domlette import NonvalidatingReader |
from Ft.Xml import EMPTY_NAMESPACE, Parse |
from Ft.Xml.Domlette import PrettyPrint, Print |
import Ft.Xml.Domlette |
from Ft.Xml import EMPTY_NAMESPACE |
|
|
|
import Ft.Xml.XPath |
|
|
|
import os.path |
import os.path |
import cgi |
import sys |
import urllib |
import urllib |
|
import urllib2 |
|
import logging |
|
import math |
|
import urlparse |
|
import cStringIO |
|
import re |
|
import string |
|
|
|
def logger(txt,method,txt2): |
|
"""logging""" |
|
logging.info(txt+ txt2) |
|
|
|
|
|
def getInt(number, default=0): |
|
"""returns always an int (0 in case of problems)""" |
|
try: |
|
return int(number) |
|
except: |
|
return int(default) |
|
|
def getTextFromNode(nodename): |
def getTextFromNode(nodename): |
|
"""get the cdata content of a node""" |
|
if nodename is None: |
|
return "" |
nodelist=nodename.childNodes |
nodelist=nodename.childNodes |
rc = "" |
rc = "" |
for node in nodelist: |
for node in nodelist: |
Line 25 def getTextFromNode(nodename):
|
Line 43 def getTextFromNode(nodename):
|
rc = rc + node.data |
rc = rc + node.data |
return rc |
return rc |
|
|
|
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() |
|
return s |
|
|
|
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 |
|
if string.find(ua, 'MSIE') > -1: |
|
bt['isIE'] = True |
|
else: |
|
bt['isN4'] = (string.find(ua, 'Mozilla/4.') > -1) |
|
|
|
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 |
|
|
|
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): |
|
"""returns pathname shortened by one""" |
|
return '/'.join(path.split('/')[0:-1]) |
|
|
|
|
|
def getHttpData(url, data=None, num_tries=3, timeout=10): |
|
"""returns result from url+data HTTP request""" |
|
# we do GET (by appending data to url) |
|
if isinstance(data, str) or isinstance(data, unicode): |
|
# if data is string then append |
|
url = "%s?%s"%(url,data) |
|
elif isinstance(data, dict) or isinstance(data, list) or isinstance(data, tuple): |
|
# urlencode |
|
url = "%s?%s"%(url,urllib.urlencode(data)) |
|
|
|
response = None |
|
errmsg = None |
|
for cnt in range(num_tries): |
|
try: |
|
logging.debug("getHttpData(#%s %ss) url=%s"%(cnt+1,timeout,url)) |
|
if sys.version_info < (2, 6): |
|
# set timeout on socket -- ugly :-( |
import socket |
import socket |
|
socket.setdefaulttimeout(float(timeout)) |
|
response = urllib2.urlopen(url) |
|
else: |
|
response = urllib2.urlopen(url,timeout=float(timeout)) |
|
# check result? |
|
break |
|
except urllib2.HTTPError, e: |
|
logging.error("getHttpData: HTTP error(%s): %s"%(e.code,e)) |
|
errmsg = str(e) |
|
# stop trying |
|
break |
|
except urllib2.URLError, e: |
|
logging.error("getHttpData: URLLIB error(%s): %s"%(e.reason,e)) |
|
errmsg = str(e) |
|
# stop trying |
|
#break |
|
|
|
if response is not None: |
|
data = response.read() |
|
response.close() |
|
return data |
|
|
|
raise IOError("ERROR fetching HTTP data from %s: %s"%(url,errmsg)) |
|
#return None |
|
|
def urlopen(url): |
|
"""urlopen mit timeout""" |
|
socket.setdefaulttimeout(2) |
|
ret=urllib.urlopen(url) |
|
socket.setdefaulttimeout(5) |
|
return ret |
|
|
|
def getParamFromDigilib(path,param): |
|
"""gibt param von dlInfo aus""" |
|
imageUrl=genericDigilib+"/dlInfo-xml.jsp?fn="+path |
|
|
|
try: |
## |
dom = NonvalidatingReader.parseUri(imageUrl) |
## documentViewer class |
except: |
## |
return None |
class documentViewer(Folder): |
|
"""document viewer""" |
|
meta_type="Document viewer" |
|
|
|
security=ClassSecurityInfo() |
|
manage_options=Folder.manage_options+( |
|
{'label':'main config','action':'changeDocumentViewerForm'}, |
|
) |
|
|
params=dom.xpath("//document-parameters/parameter[@name='%s']/@value"%param) |
# templates and forms |
|
viewer_main = PageTemplateFile('zpt/viewer_main', globals()) |
|
toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals()) |
|
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_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()) |
|
|
if params: |
|
return params[0].value |
|
|
|
def parseUrlTextTool(url): |
thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals()) |
"""parse index meta""" |
security.declareProtected('View management screens','changeDocumentViewerForm') |
|
changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) |
|
|
|
|
|
def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"): |
|
"""init document viewer""" |
|
self.id=id |
|
self.title=title |
|
self.thumbcols = thumbcols |
|
self.thumbrows = thumbrows |
|
# authgroups is list of authorized groups (delimited by ,) |
|
self.authgroups = [s.strip().lower() for s in authgroups.split(',')] |
|
# create template folder so we can always use template.something |
|
|
|
templateFolder = Folder('template') |
|
#self['template'] = templateFolder # Zope-2.12 style |
|
self._setObject('template',templateFolder) # old style |
try: |
try: |
dom = NonvalidatingReader.parseUri(url) |
import MpdlXmlTextServer |
except: |
textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName) |
zLOG.LOG("documentViewer (parseUrlTexttool)", zLOG.INFO,"%s (%s)"%sys.exc_info()[0:2]) |
#templateFolder['fulltextclient'] = xmlRpcClient |
return (None,None,None) |
templateFolder._setObject('fulltextclient',textServer) |
|
except Exception, e: |
|
logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e)) |
|
try: |
|
from Products.zogiLib.zogiLib import zogiLib |
|
zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book") |
|
#templateFolder['zogilib'] = zogilib |
|
templateFolder._setObject('zogilib',zogilib) |
|
except Exception, e: |
|
logging.error("Unable to create zogiLib for zogilib: "+str(e)) |
|
|
|
|
|
# proxy text server methods to fulltextclient |
|
def getTextPage(self, **args): |
|
"""get page""" |
|
return self.template.fulltextclient.getTextPage(**args) |
|
|
|
def getQuery(self, **args): |
|
"""get query""" |
|
return self.template.fulltextclient.getQuery(**args) |
|
|
|
def getQueryResultHits(self, **args): |
|
"""get query""" |
|
return self.template.fulltextclient.getQueryResultHits(**args) |
|
|
|
def getQueryResultHitsText(self, **args): |
|
"""get query""" |
|
return self.template.fulltextclient.getQueryResultHitsText(**args) |
|
|
|
def getQueryResultHitsFigures(self, **args): |
|
"""get query""" |
|
return self.template.fulltextclient.getQueryResultHitsFigures(**args) |
|
|
|
def getPDF(self, **args): |
|
"""get query""" |
|
return self.template.fulltextclient.getPDF(**args) |
|
|
|
def getSearch(self, **args): |
|
"""get search""" |
|
return self.template.fulltextclient.getSearch(**args) |
|
|
|
def getGisPlaces(self, **args): |
|
"""get gis places""" |
|
return self.template.fulltextclient.getGisPlaces(**args) |
|
|
|
def getAllGisPlaces(self, **args): |
|
"""get all gis places """ |
|
return self.template.fulltextclient.getAllGisPlaces(**args) |
|
|
|
def getOrigPages(self, **args): |
|
"""get original page number """ |
|
return self.template.fulltextclient.getOrigPages(**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) |
|
|
|
def getTranslate(self, **args): |
|
"""get translate""" |
|
return self.template.fulltextclient.getTranslate(**args) |
|
|
|
def getLemma(self, **args): |
|
"""get lemma""" |
|
return self.template.fulltextclient.getLemma(**args) |
|
|
|
def getToc(self, **args): |
|
"""get toc""" |
|
return self.template.fulltextclient.getToc(**args) |
|
|
|
def getTocPage(self, **args): |
|
"""get tocpage""" |
|
return self.template.fulltextclient.getTocPage(**args) |
|
|
archivePaths=dom.xpath("//resource/archive-path") |
|
|
|
if archivePaths and (len(archivePaths)>0): |
security.declareProtected('View','thumbs_rss') |
archivePath=getTextFromNode(archivePaths[0]) |
def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1): |
|
''' |
|
view it |
|
@param mode: defines how to access the document behind url |
|
@param url: url which contains display information |
|
@param viewMode: if images display images, if text display text, default is images (text,images or auto) |
|
|
|
''' |
|
logging.debug("HHHHHHHHHHHHHH:load the rss") |
|
logger("documentViewer (index)", logging.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
|
|
|
if not hasattr(self, 'template'): |
|
# create template folder if it doesn't exist |
|
self.manage_addFolder('template') |
|
|
|
if not self.digilibBaseUrl: |
|
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) |
|
''' 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 |
|
viewMode="text" |
else: |
else: |
archivePath=None |
viewMode="images" |
|
|
|
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) |
|
|
images=dom.xpath("//texttool/image") |
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="", optionToggle=None): |
|
''' |
|
view it |
|
@param mode: defines how to access the document behind url |
|
@param url: url which contains display information |
|
@param viewMode: if images display images, if text display text, default is auto (text,images or auto) |
|
@param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none) |
|
@param characterNormalization type of text display (reg, norm, none) |
|
@param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma) |
|
''' |
|
|
if images and (len(images)>0): |
logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
image=getTextFromNode(images[0]) |
|
else: |
if not hasattr(self, 'template'): |
image=None |
# this won't work |
|
logging.error("template folder missing!") |
|
return "ERROR: template folder missing!" |
|
|
if image and archivePath: |
if not getattr(self, 'digilibBaseUrl', None): |
image=os.path.join(archivePath,image) |
self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary" |
image=image.replace("/mpiwg/online",'') |
|
pt=getParamFromDigilib(image,'pt') |
|
|
|
|
docinfo = self.getDocinfo(mode=mode,url=url) |
|
|
|
if tocMode != "thumbs": |
|
# get table of contents |
|
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 |
|
viewMode="text_dict" |
else: |
else: |
image=None |
viewMode="images" |
|
|
viewerUrls=dom.xpath("//texttool/digiliburlprefix") |
pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) |
|
|
if viewerUrls and (len(viewerUrls)>0): |
pt = getattr(self.template, 'viewer_main') |
viewerUrl=getTextFromNode(viewerUrls[0]) |
return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) |
|
|
|
def generateMarks(self,mk): |
|
ret="" |
|
if mk is None: |
|
return "" |
|
if not isinstance(mk, list): |
|
mk=[mk] |
|
for m in mk: |
|
ret+="mk=%s"%m |
|
return ret |
|
|
|
|
|
def getBrowser(self): |
|
"""getBrowser the version of browser """ |
|
names="" |
|
names = browserCheck(self) |
|
#logging.debug("XXXXXXXXXXXXXXXX: %s"%names) |
|
return names |
|
|
|
def findDigilibUrl(self): |
|
"""try to get the digilib URL from zogilib""" |
|
url = self.template.zogilib.getDLBaseUrl() |
|
return url |
|
|
|
def getDocumentViewerURL(self): |
|
"""returns the URL of this instance""" |
|
return self.absolute_url() |
|
|
|
def getStyle(self, idx, selected, style=""): |
|
"""returns a string with the given style and append 'sel' if path == selected.""" |
|
#logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) |
|
if idx == selected: |
|
return style + 'sel' |
else: |
else: |
viewerUrl=None |
return style |
|
|
|
def getLink(self,param=None,val=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) |
|
|
textUrls=dom.xpath("//texttool/text") |
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"]) |
|
|
|
# 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 |
|
return url |
|
|
if textUrls and (len(textUrls)>0): |
def getLinkAmp(self,param=None,val=None): |
textUrl=getTextFromNode(textUrls[0]) |
"""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: |
else: |
textUrl=None |
params[param] = str(val) |
return viewerUrl,(image,pt),textUrl |
|
|
|
|
# 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 |
|
|
class documentViewer(ZopePageTemplate): |
def getInfo_xml(self,url,mode): |
"""document viewer""" |
"""returns info about the document as XML""" |
|
|
meta_type="Document viewer" |
if not self.digilibBaseUrl: |
|
self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" |
|
|
security=ClassSecurityInfo() |
docinfo = self.getDocinfo(mode=mode,url=url) |
manage_options=ZopePageTemplate.manage_options+( |
pt = getattr(self.template, 'info_xml') |
{'label':'main config','action':'changeDocumentViewerForm'}, |
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""" |
|
access = docinfo.get('accessType', None) |
|
logging.debug("documentViewer (accessOK) access type %s"%access) |
|
if access is not None and access == 'free': |
|
logging.debug("documentViewer (accessOK) access is free") |
|
return True |
|
elif access is None or access in self.authgroups: |
|
# only local access -- only logged in users |
|
user = getSecurityManager().getUser() |
|
logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr())) |
|
if user is not None: |
|
#print "user: ", user |
|
return (user.getUserName() != "Anonymous User") |
|
else: |
|
return False |
|
|
_default_content_fn = os.path.join(package_home(globals()),'zpt','documentViewer_template.zpt') |
logging.error("documentViewer (accessOK) unknown access type %s"%access) |
|
return False |
|
|
def __init__(self,id,imageViewerUrl,title=""): |
|
"""init document viewer""" |
|
self.id=id |
|
self.title=title |
|
self.imageViewerUrl=imageViewerUrl |
|
|
|
security.declareProtected('View management screens','changeDocumentViewerForm') |
def getDirinfoFromDigilib(self,path,docinfo=None,cut=0): |
def changeDocumentViewerForm(self): |
"""gibt param von dlInfo aus""" |
"""change it""" |
if docinfo is None: |
pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','changeDocumentViewer.zpt')).__of__(self) |
docinfo = {} |
return pt() |
|
|
|
|
for x in range(cut): |
|
|
def changeDocumentViewer(self,imageViewerUrl,title="",RESPONSE=None): |
path=getParentDir(path) |
"""init document viewer""" |
|
self.title=title |
|
self.imageViewerUrl=imageViewerUrl |
|
|
|
if RESPONSE is not None: |
infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path |
RESPONSE.redirect('manage_main') |
|
|
|
|
logging.debug("documentViewer (getparamfromdigilib) dirInfo from %s"%(infoUrl)) |
|
|
def imageLink(self,nr): |
txt = getHttpData(infoUrl) |
"""link hinter den images""" |
if txt is None: |
paramsTmp=cgi.parse_qs(self.REQUEST['QUERY_STRING']) |
raise IOError("Unable to get dir-info from %s"%(infoUrl)) |
params={} |
|
for x in paramsTmp.iteritems(): |
|
params[x[0]]=x[1][0] |
|
|
|
params['pn']=nr |
|
newUrl=self.REQUEST['URL']+"?"+urllib.urlencode(params) |
|
return newUrl |
|
|
|
|
dom = Parse(txt) |
|
sizes=dom.xpath("//dir/size") |
|
logging.debug("documentViewer (getparamfromdigilib) dirInfo:size"%sizes) |
|
|
def thumbruler(self,cols,rows,start,maximum): |
if sizes: |
"""ruler for thumbs""" |
docinfo['numPages'] = int(getTextFromNode(sizes[0])) |
ret="" |
else: |
paramsTmp=cgi.parse_qs(self.REQUEST['QUERY_STRING']) |
docinfo['numPages'] = 0 |
params={} |
|
for x in paramsTmp.iteritems(): |
|
|
|
if not x[0]=="start": |
|
params[x[0]]=x[1][0] |
|
|
|
newUrlSelect=self.REQUEST['URL']+"?"+urllib.urlencode(params) |
|
if start>0: |
|
newStart=max(start-cols*rows,0) |
|
params['start']=newStart |
|
newUrl=self.REQUEST['URL']+"?"+urllib.urlencode(params) |
|
ret+="""<a href="%s">prev</a>"""%newUrl |
|
|
|
|
|
ret+="""<select onChange="location.href='%s&start='+this.options[this.selectedIndex].value" """%newUrlSelect |
|
nr,rest=divmod(maximum,cols*rows) |
|
if rest > 0: |
|
nr+=1 |
|
for i in range(nr): |
|
nr=i*cols*rows |
|
|
|
if (start >= nr) and (start < nr+cols*rows): |
|
ret+="""<option value="%s" selected>%s</option>"""%(nr,nr) |
|
else: |
|
ret+="""<option value="%s">%s</option>"""%(nr,nr) |
|
ret+="</select>" |
|
|
|
if start<maximum: |
|
newStart=min(start+cols*rows,maximum) |
|
params['start']=newStart |
|
newUrl=self.REQUEST['URL']+"?"+urllib.urlencode(params) |
|
ret+="""<a href="%s">next</a>"""%newUrl |
|
|
|
return ret |
# TODO: produce and keep list of image names and numbers |
|
|
def textToolThumb(self,url,start=0): |
return docinfo |
"""understands the texttool format |
|
@param url: url to index.meta with texttool tag |
|
""" |
|
(viewerUrl,imagepath,textpath)=parseUrlTextTool(url) |
|
|
|
imageUrl=genericDigilib+"/servlet/Scaler?fn=%s"%imagepath[0] |
def getIndexMetaPath(self,url): |
|
"""gib nur den Pfad zurueck""" |
|
regexp = re.compile(r".*(experimental|permanent)/(.*)") |
|
regpath = regexp.match(url) |
|
if (regpath==None): |
|
return "" |
|
logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(2)) |
|
return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2)) |
|
|
pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','thumbs.zpt')).__of__(self) |
|
return pt(imageUrl=imageUrl,pt=imagepath[1],start=start) |
|
|
|
def text(self,mode,url,pn): |
|
"""give text""" |
|
if mode=="texttool": #index.meta with texttool information |
|
(viewerUrl,imagepath,textpath)=parseUrlTextTool(url) |
|
|
|
print textpath |
def getIndexMetaUrl(self,url): |
try: |
"""returns utr of index.meta document at url""" |
dom = NonvalidatingReader.parseUri(textpath) |
|
except: |
|
return None |
|
|
|
list=[] |
metaUrl = None |
nodes=dom.xpath("//pb") |
if url.startswith("http://"): |
|
# real URL |
|
metaUrl = url |
|
else: |
|
# online path |
|
server=self.digilibBaseUrl+"/servlet/Texter?fn=" |
|
metaUrl=server+url.replace("/mpiwg/online","") |
|
if not metaUrl.endswith("index.meta"): |
|
metaUrl += "/index.meta" |
|
|
|
return metaUrl |
|
|
|
def getDomFromIndexMeta(self, url): |
|
"""get dom from index meta""" |
|
dom = None |
|
metaUrl = self.getIndexMetaUrl(url) |
|
|
|
logging.debug("(getDomFromIndexMeta): METAURL: %s"%metaUrl) |
|
txt=getHttpData(metaUrl) |
|
if txt is None: |
|
raise IOError("Unable to read index meta from %s"%(url)) |
|
|
|
dom = Parse(txt) |
|
return dom |
|
|
|
def getPresentationInfoXML(self, url): |
|
"""returns dom of info.xml document at url""" |
|
dom = None |
|
metaUrl = None |
|
if url.startswith("http://"): |
|
# real URL |
|
metaUrl = url |
|
else: |
|
# online path |
|
server=self.digilibBaseUrl+"/servlet/Texter?fn=" |
|
metaUrl=server+url.replace("/mpiwg/online","") |
|
|
|
txt=getHttpData(metaUrl) |
|
if txt is None: |
|
raise IOError("Unable to read infoXMLfrom %s"%(url)) |
|
|
|
dom = Parse(txt) |
|
return dom |
|
|
|
|
|
def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
|
"""gets authorization info from the index.meta file at path or given by dom""" |
|
logging.debug("documentViewer (getauthinfofromindexmeta) path: %s"%(path)) |
|
|
|
access = None |
|
|
|
if docinfo is None: |
|
docinfo = {} |
|
|
|
if dom is None: |
|
for x in range(cut): |
|
path=getParentDir(path) |
|
dom = self.getDomFromIndexMeta(path) |
|
|
|
acctype = dom.xpath("//access-conditions/access/@type") |
|
if acctype and (len(acctype)>0): |
|
access=acctype[0].value |
|
if access in ['group', 'institution']: |
|
access = getTextFromNode(dom.xpath("//access-conditions/access/name")[0]).lower() |
|
|
|
docinfo['accessType'] = access |
|
return docinfo |
|
|
|
|
|
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)) |
|
|
|
if docinfo is None: |
|
docinfo = {} |
|
|
|
if dom is None: |
|
for x in range(cut): |
|
path=getParentDir(path) |
|
dom = self.getDomFromIndexMeta(path) |
|
|
|
docinfo['indexMetaPath']=self.getIndexMetaPath(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: |
|
bibinfo = {} |
|
for e in bib: |
|
bibinfo[e.localName] = getTextFromNode(e) |
|
docinfo['bib'] = bibinfo |
|
|
|
# extract some fields (author, title, year) according to their mapping |
|
metaData=self.metadata.main.meta.bib |
|
bibtype=dom.xpath("//bib/@type") |
|
if bibtype and (len(bibtype)>0): |
|
bibtype=bibtype[0].value |
|
else: |
|
bibtype="generic" |
|
|
node=nodes[int(pn)-1] |
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)) |
|
# if there is no mapping bibmap is empty (mapping sometimes has empty fields) |
|
if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: |
|
try: |
|
docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0]) |
|
except: pass |
|
try: |
|
docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0]) |
|
except: pass |
|
try: |
|
docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0]) |
|
except: pass |
|
logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) |
|
try: |
|
docinfo['lang']=getTextFromNode(dom.xpath("//bib/lang")[0]) |
|
except: |
|
docinfo['lang']='' |
|
|
p=node |
return docinfo |
|
|
while p.tagName!="p": |
|
p=p.parentNode |
|
|
|
|
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: |
|
docinfo = {} |
|
|
|
if dom is None: |
|
for x in range(cut): |
|
path=getParentDir(path) |
|
dom = self.getDomFromIndexMeta(path) |
|
|
|
docinfo['name']=getTextFromNode(dom.xpath("/resource/name")[0]) |
|
logging.debug("documentViewer docinfo[name] %s"%docinfo['name']) |
|
return docinfo |
|
|
|
def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): |
|
"""parse texttool tag in index meta""" |
|
logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url)) |
|
if docinfo is None: |
|
docinfo = {} |
|
if docinfo.get('lang', None) is None: |
|
docinfo['lang'] = '' # default keine Sprache gesetzt |
|
if dom is None: |
|
dom = self.getDomFromIndexMeta(url) |
|
|
endNode=nodes[int(pn)] |
archivePath = None |
|
archiveName = None |
|
|
|
archiveNames = dom.xpath("//resource/name") |
|
if archiveNames and (len(archiveNames) > 0): |
|
archiveName = getTextFromNode(archiveNames[0]) |
|
else: |
|
logging.warning("documentViewer (getdocinfofromtexttool) resource/name missing in: %s" % (url)) |
|
|
e=endNode |
archivePaths = dom.xpath("//resource/archive-path") |
|
if archivePaths and (len(archivePaths) > 0): |
|
archivePath = getTextFromNode(archivePaths[0]) |
|
# clean up archive path |
|
if archivePath[0] != '/': |
|
archivePath = '/' + archivePath |
|
if archiveName and (not archivePath.endswith(archiveName)): |
|
archivePath += "/" + archiveName |
|
else: |
|
# try to get archive-path from url |
|
logging.warning("documentViewer (getdocinfofromtexttool) resource/archive-path missing in: %s" % (url)) |
|
if (not url.startswith('http')): |
|
archivePath = url.replace('index.meta', '') |
|
|
|
if archivePath is None: |
|
# we balk without archive-path |
|
raise IOError("Missing archive-path (for text-tool) in %s" % (url)) |
|
|
|
imageDirs = dom.xpath("//texttool/image") |
|
if imageDirs and (len(imageDirs) > 0): |
|
imageDir = getTextFromNode(imageDirs[0]) |
|
|
while e.tagName!="p": |
else: |
e=e.parentNode |
# we balk with no image tag / not necessary anymore because textmode is now standard |
|
#raise IOError("No text-tool info in %s"%(url)) |
|
imageDir = "" |
|
#xquery="//pb" |
|
docinfo['imagePath'] = "" # keine Bilder |
|
docinfo['imageURL'] = "" |
|
|
|
if imageDir and archivePath: |
|
#print "image: ", imageDir, " archivepath: ", archivePath |
|
imageDir = os.path.join(archivePath, imageDir) |
|
imageDir = imageDir.replace("/mpiwg/online", '') |
|
docinfo = self.getDirinfoFromDigilib(imageDir, docinfo=docinfo) |
|
docinfo['imagePath'] = imageDir |
|
|
|
docinfo['imageURL'] = self.digilibBaseUrl + "/servlet/Scaler?fn=" + imageDir |
|
|
next=node.parentNode |
viewerUrls = dom.xpath("//texttool/digiliburlprefix") |
|
if viewerUrls and (len(viewerUrls) > 0): |
|
viewerUrl = getTextFromNode(viewerUrls[0]) |
|
docinfo['viewerURL'] = viewerUrl |
|
|
#sammle s |
# old style text URL |
while next and (next!=endNode.parentNode): |
textUrls = dom.xpath("//texttool/text") |
list.append(next) |
if textUrls and (len(textUrls) > 0): |
next=next.nextSibling |
textUrl = getTextFromNode(textUrls[0]) |
list.append(endNode.parentNode) |
if urlparse.urlparse(textUrl)[0] == "": #keine url |
|
textUrl = os.path.join(archivePath, textUrl) |
|
# fix URLs starting with /mpiwg/online |
|
if textUrl.startswith("/mpiwg/online"): |
|
textUrl = textUrl.replace("/mpiwg/online", '', 1) |
|
|
if p==e:# beide im selben paragraphen |
docinfo['textURL'] = 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']: |
|
# text-only, no page images |
|
docinfo = self.getNumTextPages(docinfo) |
|
|
|
presentationUrls = dom.xpath("//texttool/presentation") |
|
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
|
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 |
|
presentationPath = getTextFromNode(presentationUrls[0]) |
|
if url.endswith("index.meta"): |
|
presentationUrl = url.replace('index.meta', presentationPath) |
else: |
else: |
next=p |
presentationUrl = url + "/" + presentationPath |
while next!=e: |
|
print next,e |
|
list.append(next) |
|
next=next.nextSibling |
|
|
|
for x in list: |
docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom) |
PrettyPrint(x) |
|
|
|
return list |
docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info |
|
|
def image(self,mode,url,pn): |
return docinfo |
"""give image out""" |
|
if mode=="texttool": #index.meta with texttool information |
|
(viewerUrl,imagepath,textpath)=parseUrlTextTool(url) |
|
url=viewerUrl+"pn=%s&fn=%s"%(pn,imagepath[0]) |
|
ret="""<iframe height="100%%" width="100%%" src="%s"/>"""%url |
|
return url |
|
|
|
def thumbs(self,mode,url,start): |
|
"""give thumbs out""" |
|
if mode=="texttool": #index.meta with texttool information |
|
return self.textToolThumb(url,int(start)) |
|
|
|
security.declareProtected('View','index_html') |
|
|
|
|
def getBibinfoFromTextToolPresentation(self,url,docinfo=None,dom=None): |
|
"""gets the bibliographical information from the preseantion entry in texttools |
|
""" |
|
dom=self.getPresentationInfoXML(url) |
|
try: |
|
docinfo['author']=getTextFromNode(dom.xpath("//author")[0]) |
|
except: |
|
pass |
|
try: |
|
docinfo['title']=getTextFromNode(dom.xpath("//title")[0]) |
|
except: |
|
pass |
|
try: |
|
docinfo['year']=getTextFromNode(dom.xpath("//date")[0]) |
|
except: |
|
pass |
|
return docinfo |
|
|
def index_html(self,mode,url,start=0,pn=0): |
def getDocinfoFromImagePath(self,path,docinfo=None,cut=0): |
''' |
"""path ist the path to the images it assumes that the index.meta file is one level higher.""" |
view it |
logging.debug("documentViewer (getdocinfofromimagepath) path: %s"%(path)) |
@param mode: defines which type of document is behind url |
if docinfo is None: |
@param url: url which contains display information |
docinfo = {} |
''' |
path=path.replace("/mpiwg/online","") |
|
docinfo['imagePath'] = path |
|
docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo,cut=cut) |
|
|
|
pathorig=path |
|
for x in range(cut): |
|
path=getParentDir(path) |
|
logging.debug("documentViewer (getdocinfofromimagepath) PATH:"+path) |
|
imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path |
|
docinfo['imageURL'] = imageUrl |
|
|
|
#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.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) |
|
return docinfo |
|
|
|
|
|
def getDocinfo(self, mode, url): |
|
"""returns docinfo depending on mode""" |
|
logging.debug("documentViewer (getdocinfo) mode: %s, url: %s"%(mode,url)) |
|
# look for cached docinfo in session |
|
if self.REQUEST.SESSION.has_key('docinfo'): |
|
docinfo = self.REQUEST.SESSION['docinfo'] |
|
# check if its still current |
|
if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url: |
|
logging.debug("documentViewer (getdocinfo) docinfo in session: %s"%docinfo) |
|
return docinfo |
|
# new docinfo |
|
docinfo = {'mode': mode, 'url': url} |
|
if mode=="texttool": #index.meta with texttool information |
|
docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo) |
|
elif mode=="imagepath": |
|
docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo) |
|
elif mode=="filepath": |
|
docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1) |
|
else: |
|
logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode) |
|
raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) |
|
|
|
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=""): |
|
"""returns pageinfo with the given parameters""" |
|
pageinfo = {} |
|
current = getInt(current) |
|
|
|
pageinfo['current'] = current |
|
rows = int(rows or self.thumbrows) |
|
pageinfo['rows'] = rows |
|
cols = int(cols or self.thumbcols) |
|
pageinfo['cols'] = cols |
|
grpsize = cols * rows |
|
pageinfo['groupsize'] = grpsize |
|
start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1))) |
|
# int(current / grpsize) * grpsize +1)) |
|
pageinfo['start'] = start |
|
pageinfo['end'] = start + grpsize |
|
if (docinfo is not None) and ('numPages' in docinfo): |
|
np = int(docinfo['numPages']) |
|
pageinfo['end'] = min(pageinfo['end'], np) |
|
pageinfo['numgroups'] = int(np / grpsize) |
|
if np % grpsize > 0: |
|
pageinfo['numgroups'] += 1 |
|
pageinfo['viewMode'] = viewMode |
|
pageinfo['tocMode'] = tocMode |
|
pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization','') |
|
pageinfo['optionToggle'] = self.REQUEST.get('optionToggle','') |
|
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['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') |
|
pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '10') |
|
pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
|
toc = int (pageinfo['tocPN']) |
|
pageinfo['textPages'] =int (toc) |
|
|
pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','documentViewer_template.zpt')).__of__(self) |
|
return pt(mode=mode,url=url,start=start,pn=pn) |
|
|
|
|
|
|
if 'tocSize_%s'%tocMode in docinfo: |
|
tocSize = int(docinfo['tocSize_%s'%tocMode]) |
|
tocPageSize = int(pageinfo['tocPageSize']) |
|
# cached toc |
|
if tocSize%tocPageSize>0: |
|
tocPages=tocSize/tocPageSize+1 |
|
else: |
|
tocPages=tocSize/tocPageSize |
|
pageinfo['tocPN'] = min (tocPages,toc) |
|
pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') |
|
pageinfo['sn'] =self.REQUEST.get('sn','') |
|
return pageinfo |
|
|
# security.declareProtected('View management screens','renameImageForm') |
def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): |
|
"""init document viewer""" |
|
self.title=title |
|
self.digilibBaseUrl = digilibBaseUrl |
|
self.thumbrows = thumbrows |
|
self.thumbcols = thumbcols |
|
self.authgroups = [s.strip().lower() for s in authgroups.split(',')] |
|
if RESPONSE is not None: |
|
RESPONSE.redirect('manage_main') |
|
|
def manage_AddDocumentViewerForm(self): |
def manage_AddDocumentViewerForm(self): |
"""add the viewer form""" |
"""add the viewer form""" |
pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','addDocumentViewer.zpt')).__of__(self) |
pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self) |
return pt() |
return pt() |
|
|
def manage_AddDocumentViewer(self,id,imageViewerUrl="",title="",RESPONSE=None): |
def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None): |
"""add the viewer""" |
"""add the viewer""" |
newObj=documentViewer(id,imageViewerUrl,title) |
newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName) |
self._setObject(id,newObj) |
self._setObject(id,newObj) |
|
|
if RESPONSE is not None: |
if RESPONSE is not None: |
RESPONSE.redirect('manage_main') |
RESPONSE.redirect('manage_main') |
|
|
|
|
|
## DocumentViewerTemplate class |
|
class DocumentViewerTemplate(ZopePageTemplate): |
|
"""Template for document viewer""" |
|
meta_type="DocumentViewer Template" |
|
|
|
|
|
def manage_addDocumentViewerTemplateForm(self): |
|
"""Form for adding""" |
|
pt=PageTemplateFile('zpt/addDocumentViewerTemplate', globals()).__of__(self) |
|
return pt() |
|
|
|
def manage_addDocumentViewerTemplate(self, id='viewer_main', title=None, text=None, |
|
REQUEST=None, submit=None): |
|
"Add a Page Template with optional file content." |
|
|
|
self._setObject(id, DocumentViewerTemplate(id)) |
|
ob = getattr(self, id) |
|
txt=file(os.path.join(package_home(globals()),'zpt/viewer_main.zpt'),'r').read() |
|
logging.info("txt %s:"%txt) |
|
ob.pt_edit(txt,"text/html") |
|
if title: |
|
ob.pt_setTitle(title) |
|
try: |
|
u = self.DestinationURL() |
|
except AttributeError: |
|
u = REQUEST['URL1'] |
|
|
|
u = "%s/%s" % (u, urllib.quote(id)) |
|
REQUEST.RESPONSE.redirect(u+'/manage_main') |
|
return '' |
|
|
|
|
|
|