1:
2: from OFS.Folder import Folder
3: from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate
4: from Products.PageTemplates.PageTemplateFile import PageTemplateFile
5: from AccessControl import ClassSecurityInfo
6: from AccessControl import getSecurityManager
7: from Globals import package_home
8:
9: from Ft.Xml import EMPTY_NAMESPACE, Parse
10: import Ft.Xml.Domlette
11: import os.path
12: import sys
13: import urllib
14: import urllib2
15: import logging
16: import math
17: import urlparse
18: import cStringIO
19: import re
20:
21: def logger(txt,method,txt2):
22: """logging"""
23: logging.info(txt+ txt2)
24:
25:
26: def getInt(number, default=0):
27: """returns always an int (0 in case of problems)"""
28: try:
29: return int(number)
30: except:
31: return int(default)
32:
33: def getTextFromNode(nodename):
34: """get the cdata content of a node"""
35: if nodename is None:
36: return ""
37: nodelist=nodename.childNodes
38: rc = ""
39: for node in nodelist:
40: if node.nodeType == node.TEXT_NODE:
41: rc = rc + node.data
42: return rc
43:
44: def serializeNode(node, encoding='utf-8'):
45: """returns a string containing node as XML"""
46: buf = cStringIO.StringIO()
47: Ft.Xml.Domlette.Print(node, stream=buf, encoding=encoding)
48: s = buf.getvalue()
49: buf.close()
50: return s
51:
52:
53: def getParentDir(path):
54: """returns pathname shortened by one"""
55: return '/'.join(path.split('/')[0:-1])
56:
57:
58: def getHttpData(url, data=None, num_tries=3, timeout=10):
59: """returns result from url+data HTTP request"""
60: # we do GET (by appending data to url)
61: if isinstance(data, str) or isinstance(data, unicode):
62: # if data is string then append
63: url = "%s?%s"%(url,data)
64: elif isinstance(data, dict) or isinstance(data, list) or isinstance(data, tuple):
65: # urlencode
66: url = "%s?%s"%(url,urllib.urlencode(data))
67:
68: response = None
69: errmsg = None
70: for cnt in range(num_tries):
71: try:
72: logging.debug("getHttpData(#%s %ss) url=%s"%(cnt+1,timeout,url))
73: if sys.version_info < (2, 6):
74: # set timeout on socket -- ugly :-(
75: import socket
76: socket.setdefaulttimeout(float(timeout))
77: response = urllib2.urlopen(url)
78: else:
79: response = urllib2.urlopen(url,timeout=float(timeout))
80: # check result?
81: break
82: except urllib2.HTTPError, e:
83: logging.error("getHttpData: HTTP error(%s): %s"%(e.code,e))
84: errmsg = str(e)
85: # stop trying
86: break
87: except urllib2.URLError, e:
88: logging.error("getHttpData: URLLIB error(%s): %s"%(e.reason,e))
89: errmsg = str(e)
90: # stop trying
91: #break
92:
93: if response is not None:
94: data = response.read()
95: response.close()
96: return data
97:
98: raise IOError("ERROR fetching HTTP data from %s: %s"%(url,errmsg))
99: #return None
100:
101:
102:
103: ##
104: ## documentViewer class
105: ##
106: class documentViewer(Folder):
107: """document viewer"""
108: meta_type="Document viewer"
109:
110: security=ClassSecurityInfo()
111: manage_options=Folder.manage_options+(
112: {'label':'main config','action':'changeDocumentViewerForm'},
113: )
114:
115: # templates and forms
116: viewer_main = PageTemplateFile('zpt/viewer_main', globals())
117: toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals())
118: toc_text = PageTemplateFile('zpt/toc_text', globals())
119: toc_figures = PageTemplateFile('zpt/toc_figures', globals())
120: page_main_images = PageTemplateFile('zpt/page_main_images', globals())
121: page_main_text = PageTemplateFile('zpt/page_main_text', globals())
122: page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals())
123: page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals())
124: page_main_xml = PageTemplateFile('zpt/page_main_xml', globals())
125: head_main = PageTemplateFile('zpt/head_main', globals())
126: docuviewer_css = PageTemplateFile('css/docuviewer.css', globals())
127: info_xml = PageTemplateFile('zpt/info_xml', globals())
128:
129:
130: thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals())
131: security.declareProtected('View management screens','changeDocumentViewerForm')
132: changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals())
133:
134:
135: def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"):
136: """init document viewer"""
137: self.id=id
138: self.title=title
139: self.thumbcols = thumbcols
140: self.thumbrows = thumbrows
141: # authgroups is list of authorized groups (delimited by ,)
142: self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
143: # create template folder so we can always use template.something
144:
145: templateFolder = Folder('template')
146: #self['template'] = templateFolder # Zope-2.12 style
147: self._setObject('template',templateFolder) # old style
148: try:
149: import MpdlXmlTextServer
150: textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName)
151: #templateFolder['fulltextclient'] = xmlRpcClient
152: templateFolder._setObject('fulltextclient',textServer)
153: except Exception, e:
154: logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e))
155: try:
156: from Products.zogiLib.zogiLib import zogiLib
157: zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book")
158: #templateFolder['zogilib'] = zogilib
159: templateFolder._setObject('zogilib',zogilib)
160: except Exception, e:
161: logging.error("Unable to create zogiLib for zogilib: "+str(e))
162:
163:
164: # proxy text server methods to fulltextclient
165: def getTextPage(self, **args):
166: """get page"""
167: return self.template.fulltextclient.getTextPage(**args)
168:
169: def getQuery(self, **args):
170: """get query"""
171: return self.template.fulltextclient.getQuery(**args)
172:
173: def getSearch(self, **args):
174: """get search"""
175: return self.template.fulltextclient.getSearch(**args)
176:
177: def getNumPages(self, docinfo):
178: """get numpages"""
179: return self.template.fulltextclient.getNumPages(docinfo)
180:
181: def getNumTextPages(self, docinfo):
182: """get numpages text"""
183: return self.template.fulltextclient.getNumTextPages(docinfo)
184:
185: def getTranslate(self, **args):
186: """get translate"""
187: return self.template.fulltextclient.getTranslate(**args)
188:
189: def getLemma(self, **args):
190: """get lemma"""
191: return self.template.fulltextclient.getLemma(**args)
192:
193: def getToc(self, **args):
194: """get toc"""
195: return self.template.fulltextclient.getToc(**args)
196:
197: def getTocPage(self, **args):
198: """get tocpage"""
199: return self.template.fulltextclient.getTocPage(**args)
200:
201:
202: security.declareProtected('View','thumbs_rss')
203: def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1):
204: '''
205: view it
206: @param mode: defines how to access the document behind url
207: @param url: url which contains display information
208: @param viewMode: if images display images, if text display text, default is images (text,images or auto)
209:
210: '''
211: logging.debug("HHHHHHHHHHHHHH:load the rss")
212: logger("documentViewer (index)", logging.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn))
213:
214: if not hasattr(self, 'template'):
215: # create template folder if it doesn't exist
216: self.manage_addFolder('template')
217:
218: if not self.digilibBaseUrl:
219: self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
220:
221: docinfo = self.getDocinfo(mode=mode,url=url)
222: pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo)
223: pt = getattr(self.template, 'thumbs_main_rss')
224:
225: if viewMode=="auto": # automodus gewaehlt
226: if docinfo.has_key("textURL") or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert
227: viewMode="text"
228: else:
229: viewMode="images"
230:
231: return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode)
232:
233: security.declareProtected('View','index_html')
234: def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None, characterNormalization=""):
235: '''
236: view it
237: @param mode: defines how to access the document behind url
238: @param url: url which contains display information
239: @param viewMode: if images display images, if text display text, default is auto (text,images or auto)
240: @param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none)
241: @param characterNormalization type of text display (reg, norm, none)
242: @param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma)
243: '''
244:
245: logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn))
246:
247: if not hasattr(self, 'template'):
248: # this won't work
249: logging.error("template folder missing!")
250: return "ERROR: template folder missing!"
251:
252: if not getattr(self, 'digilibBaseUrl', None):
253: self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary"
254:
255: docinfo = self.getDocinfo(mode=mode,url=url)
256:
257: if tocMode != "thumbs":
258: # get table of contents
259: docinfo = self.getToc(mode=tocMode, docinfo=docinfo)
260:
261: if viewMode=="auto": # automodus gewaehlt
262: if docinfo.has_key('textURL') or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert
263: viewMode="text_dict"
264: else:
265: viewMode="images"
266:
267: pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode)
268:
269: pt = getattr(self.template, 'viewer_main')
270: return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk))
271:
272: def generateMarks(self,mk):
273: ret=""
274: if mk is None:
275: return ""
276: if not isinstance(mk, list):
277: mk=[mk]
278: for m in mk:
279: ret+="mk=%s"%m
280: return ret
281:
282:
283: def findDigilibUrl(self):
284: """try to get the digilib URL from zogilib"""
285: url = self.template.zogilib.getDLBaseUrl()
286: return url
287:
288: def getDocumentViewerURL(self):
289: """returns the URL of this instance"""
290: return self.absolute_url()
291:
292: def getStyle(self, idx, selected, style=""):
293: """returns a string with the given style and append 'sel' if path == selected."""
294: #logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style))
295: if idx == selected:
296: return style + 'sel'
297: else:
298: return style
299:
300: def getLink(self,param=None,val=None):
301: """link to documentviewer with parameter param set to val"""
302: params=self.REQUEST.form.copy()
303: if param is not None:
304: if val is None:
305: if params.has_key(param):
306: del params[param]
307: else:
308: params[param] = str(val)
309:
310: if params.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath
311: params["mode"] = "imagepath"
312: params["url"] = getParentDir(params["url"])
313:
314: # quote values and assemble into query string
315: #ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()])
316: ps = urllib.urlencode(params)
317: url=self.REQUEST['URL1']+"?"+ps
318: return url
319:
320: def getLinkAmp(self,param=None,val=None):
321: """link to documentviewer with parameter param set to val"""
322: params=self.REQUEST.form.copy()
323: if param is not None:
324: if val is None:
325: if params.has_key(param):
326: del params[param]
327: else:
328: params[param] = str(val)
329:
330: # quote values and assemble into query string
331: logging.debug("XYXXXXX: %s"%repr(params.items()))
332: ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()])
333: url=self.REQUEST['URL1']+"?"+ps
334: return url
335:
336: def getInfo_xml(self,url,mode):
337: """returns info about the document as XML"""
338:
339: if not self.digilibBaseUrl:
340: self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
341:
342: docinfo = self.getDocinfo(mode=mode,url=url)
343: pt = getattr(self.template, 'info_xml')
344: return pt(docinfo=docinfo)
345:
346:
347: def isAccessible(self, docinfo):
348: """returns if access to the resource is granted"""
349: access = docinfo.get('accessType', None)
350: logging.debug("documentViewer (accessOK) access type %s"%access)
351: if access is not None and access == 'free':
352: logging.debug("documentViewer (accessOK) access is free")
353: return True
354: elif access is None or access in self.authgroups:
355: # only local access -- only logged in users
356: user = getSecurityManager().getUser()
357: logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr()))
358: if user is not None:
359: #print "user: ", user
360: return (user.getUserName() != "Anonymous User")
361: else:
362: return False
363:
364: logging.error("documentViewer (accessOK) unknown access type %s"%access)
365: return False
366:
367:
368: def getDirinfoFromDigilib(self,path,docinfo=None,cut=0):
369: """gibt param von dlInfo aus"""
370: if docinfo is None:
371: docinfo = {}
372:
373: for x in range(cut):
374:
375: path=getParentDir(path)
376:
377: infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path
378:
379: logging.debug("documentViewer (getparamfromdigilib) dirInfo from %s"%(infoUrl))
380:
381: txt = getHttpData(infoUrl)
382: if txt is None:
383: raise IOError("Unable to get dir-info from %s"%(infoUrl))
384:
385: dom = Parse(txt)
386: sizes=dom.xpath("//dir/size")
387: logging.debug("documentViewer (getparamfromdigilib) dirInfo:size"%sizes)
388:
389: if sizes:
390: docinfo['numPages'] = int(getTextFromNode(sizes[0]))
391: else:
392: docinfo['numPages'] = 0
393:
394: # TODO: produce and keep list of image names and numbers
395:
396: return docinfo
397:
398: def getIndexMetaPath(self,url):
399: """gib nur den Pfad zurueck"""
400: regexp = re.compile(r".*(experimental|permanent)/(.*)")
401: regpath = regexp.match(url)
402: if (regpath==None):
403: return ""
404: logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(3))
405: return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2))
406:
407: def getIndexMetaUrl(self,url):
408: """returns utr of index.meta document at url"""
409:
410: metaUrl = None
411: if url.startswith("http://"):
412: # real URL
413: metaUrl = url
414: else:
415: # online path
416: server=self.digilibBaseUrl+"/servlet/Texter?fn="
417: metaUrl=server+url.replace("/mpiwg/online","")
418: if not metaUrl.endswith("index.meta"):
419: metaUrl += "/index.meta"
420:
421: return metaUrl
422:
423: def getDomFromIndexMeta(self, url):
424: """get dom from index meta"""
425: dom = None
426: metaUrl = self.getIndexMetaUrl(url)
427:
428: logging.debug("(getDomFromIndexMeta): METAURL: %s"%metaUrl)
429: txt=getHttpData(metaUrl)
430: if txt is None:
431: raise IOError("Unable to read index meta from %s"%(url))
432:
433: dom = Parse(txt)
434: return dom
435:
436: def getPresentationInfoXML(self, url):
437: """returns dom of info.xml document at url"""
438: dom = None
439: metaUrl = None
440: if url.startswith("http://"):
441: # real URL
442: metaUrl = url
443: else:
444: # online path
445: server=self.digilibBaseUrl+"/servlet/Texter?fn="
446: metaUrl=server+url.replace("/mpiwg/online","")
447:
448: txt=getHttpData(metaUrl)
449: if txt is None:
450: raise IOError("Unable to read infoXMLfrom %s"%(url))
451:
452: dom = Parse(txt)
453: return dom
454:
455:
456: def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
457: """gets authorization info from the index.meta file at path or given by dom"""
458: logging.debug("documentViewer (getauthinfofromindexmeta) path: %s"%(path))
459:
460: access = None
461:
462: if docinfo is None:
463: docinfo = {}
464:
465: if dom is None:
466: for x in range(cut):
467: path=getParentDir(path)
468: dom = self.getDomFromIndexMeta(path)
469:
470: acctype = dom.xpath("//access-conditions/access/@type")
471: if acctype and (len(acctype)>0):
472: access=acctype[0].value
473: if access in ['group', 'institution']:
474: access = getTextFromNode(dom.xpath("//access-conditions/access/name")[0]).lower()
475:
476: docinfo['accessType'] = access
477: return docinfo
478:
479:
480: def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
481: """gets bibliographical info from the index.meta file at path or given by dom"""
482: logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path))
483:
484: if docinfo is None:
485: docinfo = {}
486:
487: if dom is None:
488: for x in range(cut):
489: path=getParentDir(path)
490: dom = self.getDomFromIndexMeta(path)
491:
492: docinfo['indexMetaPath']=self.getIndexMetaPath(path);
493:
494: logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path))
495: # put in all raw bib fields as dict "bib"
496: bib = dom.xpath("//bib/*")
497: if bib and len(bib)>0:
498: bibinfo = {}
499: for e in bib:
500: bibinfo[e.localName] = getTextFromNode(e)
501: docinfo['bib'] = bibinfo
502:
503: # extract some fields (author, title, year) according to their mapping
504: metaData=self.metadata.main.meta.bib
505: bibtype=dom.xpath("//bib/@type")
506: if bibtype and (len(bibtype)>0):
507: bibtype=bibtype[0].value
508: else:
509: bibtype="generic"
510:
511: bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC)
512: docinfo['bib_type'] = bibtype
513: bibmap=metaData.generateMappingForType(bibtype)
514: logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap))
515: logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype))
516: # if there is no mapping bibmap is empty (mapping sometimes has empty fields)
517: if len(bibmap) > 0 and len(bibmap['author'][0]) > 0:
518: try:
519: docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0])
520: except: pass
521: try:
522: docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0])
523: except: pass
524: try:
525: docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0])
526: except: pass
527: logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype)
528: try:
529: docinfo['lang']=getTextFromNode(dom.xpath("//bib/lang")[0])
530: except:
531: docinfo['lang']=''
532:
533: return docinfo
534:
535:
536: def getNameFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
537: """gets name info from the index.meta file at path or given by dom"""
538: if docinfo is None:
539: docinfo = {}
540:
541: if dom is None:
542: for x in range(cut):
543: path=getParentDir(path)
544: dom = self.getDomFromIndexMeta(path)
545:
546: docinfo['indexMetaPath']=self.getIndexMetaPath(path);
547:
548: #result= dom.xpath("//result/resultPage")
549: #docinfo['numPages']=int(getTextFromNode(result[0]))
550:
551: if len(name) > 0:
552: try:
553: result =dom.xpath("//name")
554: docinfo['name']=getTextFromNode(result[0])
555: logging.debug("documentViewer docinfo[name] %s"%docinfo[name])
556: except: pass
557: #logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype)
558: return docinfo
559:
560: def getDocinfoFromTextTool(self, url, dom=None, docinfo=None):
561: """parse texttool tag in index meta"""
562: logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url))
563: if docinfo is None:
564: docinfo = {}
565: if docinfo.get('lang', None) is None:
566: docinfo['lang'] = '' # default keine Sprache gesetzt
567: if dom is None:
568: dom = self.getDomFromIndexMeta(url)
569:
570: archivePath = None
571: archiveName = None
572:
573: archiveNames = dom.xpath("//resource/name")
574: if archiveNames and (len(archiveNames) > 0):
575: archiveName = getTextFromNode(archiveNames[0])
576: else:
577: logging.warning("documentViewer (getdocinfofromtexttool) resource/name missing in: %s" % (url))
578:
579: archivePaths = dom.xpath("//resource/archive-path")
580: if archivePaths and (len(archivePaths) > 0):
581: archivePath = getTextFromNode(archivePaths[0])
582: # clean up archive path
583: if archivePath[0] != '/':
584: archivePath = '/' + archivePath
585: if archiveName and (not archivePath.endswith(archiveName)):
586: archivePath += "/" + archiveName
587: else:
588: # try to get archive-path from url
589: logging.warning("documentViewer (getdocinfofromtexttool) resource/archive-path missing in: %s" % (url))
590: if (not url.startswith('http')):
591: archivePath = url.replace('index.meta', '')
592:
593: if archivePath is None:
594: # we balk without archive-path
595: raise IOError("Missing archive-path (for text-tool) in %s" % (url))
596:
597: imageDirs = dom.xpath("//texttool/image")
598: if imageDirs and (len(imageDirs) > 0):
599: imageDir = getTextFromNode(imageDirs[0])
600:
601: else:
602: # we balk with no image tag / not necessary anymore because textmode is now standard
603: #raise IOError("No text-tool info in %s"%(url))
604: imageDir = ""
605: #xquery="//pb"
606: docinfo['imagePath'] = "" # keine Bilder
607: docinfo['imageURL'] = ""
608:
609: if imageDir and archivePath:
610: #print "image: ", imageDir, " archivepath: ", archivePath
611: imageDir = os.path.join(archivePath, imageDir)
612: imageDir = imageDir.replace("/mpiwg/online", '')
613: docinfo = self.getDirinfoFromDigilib(imageDir, docinfo=docinfo)
614: docinfo['imagePath'] = imageDir
615:
616: docinfo['imageURL'] = self.digilibBaseUrl + "/servlet/Scaler?fn=" + imageDir
617:
618: viewerUrls = dom.xpath("//texttool/digiliburlprefix")
619: if viewerUrls and (len(viewerUrls) > 0):
620: viewerUrl = getTextFromNode(viewerUrls[0])
621: docinfo['viewerURL'] = viewerUrl
622:
623: # old style text URL
624: textUrls = dom.xpath("//texttool/text")
625: if textUrls and (len(textUrls) > 0):
626: textUrl = getTextFromNode(textUrls[0])
627: if urlparse.urlparse(textUrl)[0] == "": #keine url
628: textUrl = os.path.join(archivePath, textUrl)
629: # fix URLs starting with /mpiwg/online
630: if textUrl.startswith("/mpiwg/online"):
631: textUrl = textUrl.replace("/mpiwg/online", '', 1)
632:
633: docinfo['textURL'] = textUrl
634:
635: # new style text-url-path
636: textUrls = dom.xpath("//texttool/text-url-path")
637: if textUrls and (len(textUrls) > 0):
638: textUrl = getTextFromNode(textUrls[0])
639: docinfo['textURLPath'] = textUrl
640: if not docinfo['imagePath']:
641: # text-only, no page images
642: docinfo = self.getNumTextPages(docinfo)
643:
644: presentationUrls = dom.xpath("//texttool/presentation")
645: docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag
646: #docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom)
647:
648: if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen
649: # presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten
650: # durch den relativen Pfad auf die presentation infos
651: presentationPath = getTextFromNode(presentationUrls[0])
652: if url.endswith("index.meta"):
653: presentationUrl = url.replace('index.meta', presentationPath)
654: else:
655: presentationUrl = url + "/" + presentationPath
656:
657: docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom)
658:
659: docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info
660:
661: return docinfo
662:
663:
664: def getBibinfoFromTextToolPresentation(self,url,docinfo=None,dom=None):
665: """gets the bibliographical information from the preseantion entry in texttools
666: """
667: dom=self.getPresentationInfoXML(url)
668: try:
669: docinfo['author']=getTextFromNode(dom.xpath("//author")[0])
670: except:
671: pass
672: try:
673: docinfo['title']=getTextFromNode(dom.xpath("//title")[0])
674: except:
675: pass
676: try:
677: docinfo['year']=getTextFromNode(dom.xpath("//date")[0])
678: except:
679: pass
680: return docinfo
681:
682: def getDocinfoFromImagePath(self,path,docinfo=None,cut=0):
683: """path ist the path to the images it assumes that the index.meta file is one level higher."""
684: logging.debug("documentViewer (getdocinfofromimagepath) path: %s"%(path))
685: if docinfo is None:
686: docinfo = {}
687: path=path.replace("/mpiwg/online","")
688: docinfo['imagePath'] = path
689: docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo,cut=cut)
690:
691: pathorig=path
692: for x in range(cut):
693: path=getParentDir(path)
694: logging.debug("documentViewer (getdocinfofromimagepath) PATH:"+path)
695: imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path
696: docinfo['imageURL'] = imageUrl
697:
698: #path ist the path to the images it assumes that the index.meta file is one level higher.
699: docinfo = self.getBibinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1)
700: docinfo = self.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1)
701: return docinfo
702:
703:
704: def getDocinfo(self, mode, url):
705: """returns docinfo depending on mode"""
706: logging.debug("documentViewer (getdocinfo) mode: %s, url: %s"%(mode,url))
707: # look for cached docinfo in session
708: if self.REQUEST.SESSION.has_key('docinfo'):
709: docinfo = self.REQUEST.SESSION['docinfo']
710: # check if its still current
711: if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url:
712: logging.debug("documentViewer (getdocinfo) docinfo in session: %s"%docinfo)
713: return docinfo
714: # new docinfo
715: docinfo = {'mode': mode, 'url': url}
716: if mode=="texttool": #index.meta with texttool information
717: docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo)
718: elif mode=="imagepath":
719: docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo)
720: elif mode=="filepath":
721: docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1)
722: else:
723: logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode)
724: raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode))
725:
726: logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo)
727: self.REQUEST.SESSION['docinfo'] = docinfo
728: return docinfo
729:
730: def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None,characterNormalization=""):
731: """returns pageinfo with the given parameters"""
732: pageinfo = {}
733: current = getInt(current)
734: pageinfo['current'] = current
735: rows = int(rows or self.thumbrows)
736: pageinfo['rows'] = rows
737: cols = int(cols or self.thumbcols)
738: pageinfo['cols'] = cols
739: grpsize = cols * rows
740: pageinfo['groupsize'] = grpsize
741: start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1)))
742: # int(current / grpsize) * grpsize +1))
743: pageinfo['start'] = start
744: pageinfo['end'] = start + grpsize
745: if (docinfo is not None) and ('numPages' in docinfo):
746: np = int(docinfo['numPages'])
747: pageinfo['end'] = min(pageinfo['end'], np)
748: pageinfo['numgroups'] = int(np / grpsize)
749: if np % grpsize > 0:
750: pageinfo['numgroups'] += 1
751: pageinfo['viewMode'] = viewMode
752: pageinfo['tocMode'] = tocMode
753: #pageinfo['characterNormalization'] =characterNormalization
754: pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization',' ')
755: pageinfo['query'] = self.REQUEST.get('query',' ')
756: pageinfo['queryType'] = self.REQUEST.get('queryType',' ')
757: pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext')
758: pageinfo['textPN'] = self.REQUEST.get('textPN','1')
759: pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','')
760: pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30')
761: pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '10')
762: pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1')
763: toc = int (pageinfo['tocPN'])
764: pageinfo['textPages'] =int (toc)
765:
766: if 'tocSize_%s'%tocMode in docinfo:
767: tocSize = int(docinfo['tocSize_%s'%tocMode])
768: tocPageSize = int(pageinfo['tocPageSize'])
769: # cached toc
770: if tocSize%tocPageSize>0:
771: tocPages=tocSize/tocPageSize+1
772: else:
773: tocPages=tocSize/tocPageSize
774: pageinfo['tocPN'] = min (tocPages,toc)
775: pageinfo['searchPN'] =self.REQUEST.get('searchPN','1')
776: pageinfo['sn'] =self.REQUEST.get('sn','')
777: return pageinfo
778:
779: def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None):
780: """init document viewer"""
781: self.title=title
782: self.digilibBaseUrl = digilibBaseUrl
783: self.thumbrows = thumbrows
784: self.thumbcols = thumbcols
785: self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
786: if RESPONSE is not None:
787: RESPONSE.redirect('manage_main')
788:
789: def manage_AddDocumentViewerForm(self):
790: """add the viewer form"""
791: pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self)
792: return pt()
793:
794: def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None):
795: """add the viewer"""
796: newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName)
797: self._setObject(id,newObj)
798:
799: if RESPONSE is not None:
800: RESPONSE.redirect('manage_main')
801:
802: ## DocumentViewerTemplate class
803: class DocumentViewerTemplate(ZopePageTemplate):
804: """Template for document viewer"""
805: meta_type="DocumentViewer Template"
806:
807:
808: def manage_addDocumentViewerTemplateForm(self):
809: """Form for adding"""
810: pt=PageTemplateFile('zpt/addDocumentViewerTemplate', globals()).__of__(self)
811: return pt()
812:
813: def manage_addDocumentViewerTemplate(self, id='viewer_main', title=None, text=None,
814: REQUEST=None, submit=None):
815: "Add a Page Template with optional file content."
816:
817: self._setObject(id, DocumentViewerTemplate(id))
818: ob = getattr(self, id)
819: txt=file(os.path.join(package_home(globals()),'zpt/viewer_main.zpt'),'r').read()
820: logging.info("txt %s:"%txt)
821: ob.pt_edit(txt,"text/html")
822: if title:
823: ob.pt_setTitle(title)
824: try:
825: u = self.DestinationURL()
826: except AttributeError:
827: u = REQUEST['URL1']
828:
829: u = "%s/%s" % (u, urllib.quote(id))
830: REQUEST.RESPONSE.redirect(u+'/manage_main')
831: return ''
832:
833:
834:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>