annotate documentViewer.py @ 225:57c2f5b6ada5

*** empty log message ***
author abukhman
date Fri, 08 Oct 2010 15:34:50 +0200
parents 58c5338749c7
children 0b614f078f7b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
46
31059e3d9338 has now also a text mode viewMode=text
dwinter
parents: 45
diff changeset
1
0
dwinter
parents:
diff changeset
2 from OFS.Folder import Folder
dwinter
parents:
diff changeset
3 from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate
52
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
4 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
0
dwinter
parents:
diff changeset
5 from AccessControl import ClassSecurityInfo
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
6 from AccessControl import getSecurityManager
0
dwinter
parents:
diff changeset
7 from Globals import package_home
dwinter
parents:
diff changeset
8
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
9 from Ft.Xml import EMPTY_NAMESPACE, Parse
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
10 import Ft.Xml.Domlette
0
dwinter
parents:
diff changeset
11 import os.path
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
12 import sys
0
dwinter
parents:
diff changeset
13 import urllib
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
14 import urllib2
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
15 import logging
61
f3d2f240692c fixed bug in calculation of group numbers
casties
parents: 59
diff changeset
16 import math
46
31059e3d9338 has now also a text mode viewMode=text
dwinter
parents: 45
diff changeset
17 import urlparse
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
18 import cStringIO
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
19 import re
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
20
52
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
21 def logger(txt,method,txt2):
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
22 """logging"""
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
23 logging.info(txt+ txt2)
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
24
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
25
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
26 def getInt(number, default=0):
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
27 """returns always an int (0 in case of problems)"""
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
28 try:
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
29 return int(number)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
30 except:
62
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
31 return int(default)
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
32
0
dwinter
parents:
diff changeset
33 def getTextFromNode(nodename):
46
31059e3d9338 has now also a text mode viewMode=text
dwinter
parents: 45
diff changeset
34 """get the cdata content of a node"""
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
35 if nodename is None:
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
36 return ""
0
dwinter
parents:
diff changeset
37 nodelist=nodename.childNodes
dwinter
parents:
diff changeset
38 rc = ""
dwinter
parents:
diff changeset
39 for node in nodelist:
dwinter
parents:
diff changeset
40 if node.nodeType == node.TEXT_NODE:
dwinter
parents:
diff changeset
41 rc = rc + node.data
dwinter
parents:
diff changeset
42 return rc
dwinter
parents:
diff changeset
43
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
44 def serializeNode(node, encoding='utf-8'):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
45 """returns a string containing node as XML"""
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
46 buf = cStringIO.StringIO()
136
a2114cd4673c fixed broken imports (silly me)
casties
parents: 135
diff changeset
47 Ft.Xml.Domlette.Print(node, stream=buf, encoding=encoding)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
48 s = buf.getvalue()
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
49 buf.close()
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
50 return s
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
51
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
52
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
53 def getParentDir(path):
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
54 """returns pathname shortened by one"""
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
55 return '/'.join(path.split('/')[0:-1])
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
56
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
57
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
58 def getHttpData(url, data=None, num_tries=3, timeout=10):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
59 """returns result from url+data HTTP request"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
60 # we do GET (by appending data to url)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
61 if isinstance(data, str) or isinstance(data, unicode):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
62 # if data is string then append
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
63 url = "%s?%s"%(url,data)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
64 elif isinstance(data, dict) or isinstance(data, list) or isinstance(data, tuple):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
65 # urlencode
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
66 url = "%s?%s"%(url,urllib.urlencode(data))
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
67
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
68 response = None
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
69 errmsg = None
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
70 for cnt in range(num_tries):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
71 try:
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
72 logging.debug("getHttpData(#%s %ss) url=%s"%(cnt+1,timeout,url))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
73 if sys.version_info < (2, 6):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
74 # set timeout on socket -- ugly :-(
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
75 import socket
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
76 socket.setdefaulttimeout(float(timeout))
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
77 response = urllib2.urlopen(url)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
78 else:
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
79 response = urllib2.urlopen(url,timeout=float(timeout))
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
80 # check result?
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
81 break
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
82 except urllib2.HTTPError, e:
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
83 logging.error("getHttpData: HTTP error(%s): %s"%(e.code,e))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
84 errmsg = str(e)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
85 # stop trying
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
86 break
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
87 except urllib2.URLError, e:
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
88 logging.error("getHttpData: URLLIB error(%s): %s"%(e.reason,e))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
89 errmsg = str(e)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
90 # stop trying
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
91 #break
0
dwinter
parents:
diff changeset
92
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
93 if response is not None:
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
94 data = response.read()
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
95 response.close()
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
96 return data
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
97
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
98 raise IOError("ERROR fetching HTTP data from %s: %s"%(url,errmsg))
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
99 #return None
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
100
0
dwinter
parents:
diff changeset
101
dwinter
parents:
diff changeset
102
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
103 ##
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
104 ## documentViewer class
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
105 ##
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
106 class documentViewer(Folder):
0
dwinter
parents:
diff changeset
107 """document viewer"""
dwinter
parents:
diff changeset
108 meta_type="Document viewer"
dwinter
parents:
diff changeset
109
dwinter
parents:
diff changeset
110 security=ClassSecurityInfo()
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
111 manage_options=Folder.manage_options+(
0
dwinter
parents:
diff changeset
112 {'label':'main config','action':'changeDocumentViewerForm'},
dwinter
parents:
diff changeset
113 )
dwinter
parents:
diff changeset
114
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
115 # templates and forms
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
116 viewer_main = PageTemplateFile('zpt/viewer_main', globals())
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
117 toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals())
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
118 toc_text = PageTemplateFile('zpt/toc_text', globals())
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
119 toc_figures = PageTemplateFile('zpt/toc_figures', globals())
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
120 page_main_images = PageTemplateFile('zpt/page_main_images', globals())
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
121 page_main_text = PageTemplateFile('zpt/page_main_text', globals())
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
122 page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals())
140
6345583d809a Gis mode
abukhman
parents: 137
diff changeset
123 page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals())
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
124 page_main_xml = PageTemplateFile('zpt/page_main_xml', globals())
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
125 head_main = PageTemplateFile('zpt/head_main', globals())
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
126 docuviewer_css = PageTemplateFile('css/docuviewer.css', globals())
57
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
127 info_xml = PageTemplateFile('zpt/info_xml', globals())
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
128
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
129
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
130 thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals())
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
131 security.declareProtected('View management screens','changeDocumentViewerForm')
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
132 changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals())
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
133
0
dwinter
parents:
diff changeset
134
95
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
135 def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"):
0
dwinter
parents:
diff changeset
136 """init document viewer"""
dwinter
parents:
diff changeset
137 self.id=id
dwinter
parents:
diff changeset
138 self.title=title
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
139 self.thumbcols = thumbcols
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
140 self.thumbrows = thumbrows
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
141 # authgroups is list of authorized groups (delimited by ,)
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
142 self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
143 # create template folder so we can always use template.something
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
144
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
145 templateFolder = Folder('template')
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
146 #self['template'] = templateFolder # Zope-2.12 style
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
147 self._setObject('template',templateFolder) # old style
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
148 try:
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
149 import MpdlXmlTextServer
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 130
diff changeset
150 textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
151 #templateFolder['fulltextclient'] = xmlRpcClient
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
152 templateFolder._setObject('fulltextclient',textServer)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
153 except Exception, e:
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
154 logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e))
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
155 try:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
156 from Products.zogiLib.zogiLib import zogiLib
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
157 zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
158 #templateFolder['zogilib'] = zogilib
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
159 templateFolder._setObject('zogilib',zogilib)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
160 except Exception, e:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
161 logging.error("Unable to create zogiLib for zogilib: "+str(e))
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
162
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
163
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
164 # proxy text server methods to fulltextclient
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
165 def getTextPage(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
166 """get page"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
167 return self.template.fulltextclient.getTextPage(**args)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
168
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
169 def getQuery(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
170 """get query"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
171 return self.template.fulltextclient.getQuery(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
172
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
173 def getSearch(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
174 """get search"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
175 return self.template.fulltextclient.getSearch(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
176
133
40730cfbcb8e fixed problem with getnumpages and **args
casties
parents: 132
diff changeset
177 def getNumPages(self, docinfo):
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
178 """get numpages"""
133
40730cfbcb8e fixed problem with getnumpages and **args
casties
parents: 132
diff changeset
179 return self.template.fulltextclient.getNumPages(docinfo)
191
cffaec9bc5ad *** empty log message ***
abukhman
parents: 174
diff changeset
180
cffaec9bc5ad *** empty log message ***
abukhman
parents: 174
diff changeset
181 def getNumTextPages(self, docinfo):
cffaec9bc5ad *** empty log message ***
abukhman
parents: 174
diff changeset
182 """get numpages text"""
cffaec9bc5ad *** empty log message ***
abukhman
parents: 174
diff changeset
183 return self.template.fulltextclient.getNumTextPages(docinfo)
cffaec9bc5ad *** empty log message ***
abukhman
parents: 174
diff changeset
184
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
185 def getTranslate(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
186 """get translate"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
187 return self.template.fulltextclient.getTranslate(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
188
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
189 def getLemma(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
190 """get lemma"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
191 return self.template.fulltextclient.getLemma(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
192
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
193 def getToc(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
194 """get toc"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
195 return self.template.fulltextclient.getToc(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
196
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
197 def getTocPage(self, **args):
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
198 """get tocpage"""
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
199 return self.template.fulltextclient.getTocPage(**args)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
200
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
201
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
202 security.declareProtected('View','thumbs_rss')
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
203 def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1):
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
204 '''
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
205 view it
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
206 @param mode: defines how to access the document behind url
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
207 @param url: url which contains display information
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
208 @param viewMode: if images display images, if text display text, default is images (text,images or auto)
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
209
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
210 '''
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
211 logging.debug("HHHHHHHHHHHHHH:load the rss")
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
212 logger("documentViewer (index)", logging.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn))
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
213
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
214 if not hasattr(self, 'template'):
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
215 # create template folder if it doesn't exist
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
216 self.manage_addFolder('template')
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
217
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
218 if not self.digilibBaseUrl:
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
219 self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
220
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
221 docinfo = self.getDocinfo(mode=mode,url=url)
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
222 pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo)
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
223 pt = getattr(self.template, 'thumbs_main_rss')
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
224
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
225 if viewMode=="auto": # automodus gewaehlt
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
226 if docinfo.has_key("textURL") or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
227 viewMode="text"
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
228 else:
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
229 viewMode="images"
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
230
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
231 return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode)
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
232
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
233 security.declareProtected('View','index_html')
170
caf66e1fccd3 characterNormalization
abukhman
parents: 169
diff changeset
234 def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None, characterNormalization=""):
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
235 '''
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
236 view it
57
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
237 @param mode: defines how to access the document behind url
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
238 @param url: url which contains display information
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
239 @param viewMode: if images display images, if text display text, default is auto (text,images or auto)
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
240 @param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none)
141
39c7d6164da8 characterNormalization (reg,norm, none)
abukhman
parents: 140
diff changeset
241 @param characterNormalization type of text display (reg, norm, none)
100
137459e3b378 Last update
abukhman
parents: 99
diff changeset
242 @param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
243 '''
0
dwinter
parents:
diff changeset
244
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
245 logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn))
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
246
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
247 if not hasattr(self, 'template'):
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
248 # this won't work
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
249 logging.error("template folder missing!")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
250 return "ERROR: template folder missing!"
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
251
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
252 if not getattr(self, 'digilibBaseUrl', None):
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 130
diff changeset
253 self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary"
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
254
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
255 docinfo = self.getDocinfo(mode=mode,url=url)
97
2b8fd19432fb Last update
abukhman
parents: 96
diff changeset
256
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
257 if tocMode != "thumbs":
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
258 # get table of contents
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
259 docinfo = self.getToc(mode=tocMode, docinfo=docinfo)
97
2b8fd19432fb Last update
abukhman
parents: 96
diff changeset
260
51
c5d3aabbf61b textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents: 50
diff changeset
261 if viewMode=="auto": # automodus gewaehlt
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 130
diff changeset
262 if docinfo.has_key('textURL') or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert
127
49f381f896d0 try to make text_dict default for texts
casties
parents: 126
diff changeset
263 viewMode="text_dict"
51
c5d3aabbf61b textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents: 50
diff changeset
264 else:
c5d3aabbf61b textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents: 50
diff changeset
265 viewMode="images"
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
266
170
caf66e1fccd3 characterNormalization
abukhman
parents: 169
diff changeset
267 pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode)
127
49f381f896d0 try to make text_dict default for texts
casties
parents: 126
diff changeset
268
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
269 pt = getattr(self.template, 'viewer_main')
75
9673218e155b minorCVS: ----------------------------------------------------------------------
dwinter
parents: 74
diff changeset
270 return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk))
0
dwinter
parents:
diff changeset
271
74
dwinter
parents: 73
diff changeset
272 def generateMarks(self,mk):
dwinter
parents: 73
diff changeset
273 ret=""
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
274 if mk is None:
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
275 return ""
134
6a33aa624ba4 fixed more oopsies
casties
parents: 133
diff changeset
276 if not isinstance(mk, list):
132
39167e96e582 fixed creation of text server on __init__
casties
parents: 130
diff changeset
277 mk=[mk]
74
dwinter
parents: 73
diff changeset
278 for m in mk:
75
9673218e155b minorCVS: ----------------------------------------------------------------------
dwinter
parents: 74
diff changeset
279 ret+="mk=%s"%m
74
dwinter
parents: 73
diff changeset
280 return ret
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
281
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
282
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
283 def findDigilibUrl(self):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
284 """try to get the digilib URL from zogilib"""
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
285 url = self.template.zogilib.getDLBaseUrl()
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
286 return url
126
c6dcf12e3807 added method to get url of viewer instance
casties
parents: 124
diff changeset
287
c6dcf12e3807 added method to get url of viewer instance
casties
parents: 124
diff changeset
288 def getDocumentViewerURL(self):
c6dcf12e3807 added method to get url of viewer instance
casties
parents: 124
diff changeset
289 """returns the URL of this instance"""
c6dcf12e3807 added method to get url of viewer instance
casties
parents: 124
diff changeset
290 return self.absolute_url()
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
291
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
292 def getStyle(self, idx, selected, style=""):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
293 """returns a string with the given style and append 'sel' if path == selected."""
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
294 #logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style))
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
295 if idx == selected:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
296 return style + 'sel'
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
297 else:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
298 return style
74
dwinter
parents: 73
diff changeset
299
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
300 def getLink(self,param=None,val=None):
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
301 """link to documentviewer with parameter param set to val"""
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
302 params=self.REQUEST.form.copy()
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
303 if param is not None:
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
304 if val is None:
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
305 if params.has_key(param):
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
306 del params[param]
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
307 else:
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
308 params[param] = str(val)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
309
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
310 if params.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
311 params["mode"] = "imagepath"
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
312 params["url"] = getParentDir(params["url"])
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
313
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
314 # quote values and assemble into query string
135
123248b3bb34 fixed more oopsies
casties
parents: 134
diff changeset
315 #ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()])
123248b3bb34 fixed more oopsies
casties
parents: 134
diff changeset
316 ps = urllib.urlencode(params)
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
317 url=self.REQUEST['URL1']+"?"+ps
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
318 return url
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
319
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
320 def getLinkAmp(self,param=None,val=None):
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
321 """link to documentviewer with parameter param set to val"""
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
322 params=self.REQUEST.form.copy()
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
323 if param is not None:
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
324 if val is None:
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
325 if params.has_key(param):
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
326 del params[param]
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
327 else:
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
328 params[param] = str(val)
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
329
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
330 # quote values and assemble into query string
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
331 logging.debug("XYXXXXX: %s"%repr(params.items()))
68
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
332 ps = "&amp;".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()])
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
333 url=self.REQUEST['URL1']+"?"+ps
b8457fc33446 piclens rss/support
dwinter
parents: 65
diff changeset
334 return url
81
fae97f071724 fixed problem with info.xml when url without index.meta
casties
parents: 79
diff changeset
335
57
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
336 def getInfo_xml(self,url,mode):
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
337 """returns info about the document as XML"""
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
338
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
339 if not self.digilibBaseUrl:
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
340 self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
341
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
342 docinfo = self.getDocinfo(mode=mode,url=url)
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
343 pt = getattr(self.template, 'info_xml')
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
344 return pt(docinfo=docinfo)
7cdb0fc34a92 added getInfo_xml method
casties
parents: 55
diff changeset
345
0
dwinter
parents:
diff changeset
346
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
347 def isAccessible(self, docinfo):
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
348 """returns if access to the resource is granted"""
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
349 access = docinfo.get('accessType', None)
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
350 logging.debug("documentViewer (accessOK) access type %s"%access)
45
0391fe75aef3 fixed handling of documents with missing access tag
casties
parents: 43
diff changeset
351 if access is not None and access == 'free':
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
352 logging.debug("documentViewer (accessOK) access is free")
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
353 return True
45
0391fe75aef3 fixed handling of documents with missing access tag
casties
parents: 43
diff changeset
354 elif access is None or access in self.authgroups:
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
355 # only local access -- only logged in users
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
356 user = getSecurityManager().getUser()
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
357 logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr()))
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
358 if user is not None:
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
359 #print "user: ", user
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
360 return (user.getUserName() != "Anonymous User")
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
361 else:
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
362 return False
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
363
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
364 logging.error("documentViewer (accessOK) unknown access type %s"%access)
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
365 return False
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
366
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
367
73
dwinter
parents: 71
diff changeset
368 def getDirinfoFromDigilib(self,path,docinfo=None,cut=0):
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
369 """gibt param von dlInfo aus"""
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
370 if docinfo is None:
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
371 docinfo = {}
73
dwinter
parents: 71
diff changeset
372
dwinter
parents: 71
diff changeset
373 for x in range(cut):
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
374
73
dwinter
parents: 71
diff changeset
375 path=getParentDir(path)
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
376
40
749ee5389892 try to ignore superfluous /mpiwg/online in urls
casties
parents: 39
diff changeset
377 infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
378
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
379 logging.debug("documentViewer (getparamfromdigilib) dirInfo from %s"%(infoUrl))
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
380
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
381 txt = getHttpData(infoUrl)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
382 if txt is None:
40
749ee5389892 try to ignore superfluous /mpiwg/online in urls
casties
parents: 39
diff changeset
383 raise IOError("Unable to get dir-info from %s"%(infoUrl))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
384
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
385 dom = Parse(txt)
37
ead830ce45d6 better error messages
casties
parents: 35
diff changeset
386 sizes=dom.xpath("//dir/size")
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
387 logging.debug("documentViewer (getparamfromdigilib) dirInfo:size"%sizes)
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
388
37
ead830ce45d6 better error messages
casties
parents: 35
diff changeset
389 if sizes:
ead830ce45d6 better error messages
casties
parents: 35
diff changeset
390 docinfo['numPages'] = int(getTextFromNode(sizes[0]))
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
391 else:
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
392 docinfo['numPages'] = 0
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
393
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
394 # TODO: produce and keep list of image names and numbers
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
395
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
396 return docinfo
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
397
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
398 def getIndexMetaPath(self,url):
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
399 """gib nur den Pfad zurueck"""
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
400 regexp = re.compile(r".*(experimental|permanent)/(.*)")
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
401 regpath = regexp.match(url)
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
402 if (regpath==None):
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
403 return ""
224
58c5338749c7 *** empty log message ***
abukhman
parents: 223
diff changeset
404 logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(2))
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
405 return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2))
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
406
225
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
407 def getIndexMetaName(self,url):
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
408 """gib nur den Pfad zurueck"""
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
409 #regexp = re.compile(r".*(experimental|permanent)/(.*)")
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
410 regpath = regexp.match(url)
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
411 if (regpath==None):
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
412 return ""
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
413 logging.debug("(getIndexMetaName): XAXA: %s"%regpath.group(2))
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
414 return regpath
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
415
57c2f5b6ada5 *** empty log message ***
abukhman
parents: 224
diff changeset
416
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
417 def getIndexMetaUrl(self,url):
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
418 """returns utr of index.meta document at url"""
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
419
39
1dd90aabd366 added retry when reading index meta from texter applet
casties
parents: 38
diff changeset
420 metaUrl = None
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
421 if url.startswith("http://"):
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
422 # real URL
39
1dd90aabd366 added retry when reading index meta from texter applet
casties
parents: 38
diff changeset
423 metaUrl = url
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
424 else:
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
425 # online path
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
426 server=self.digilibBaseUrl+"/servlet/Texter?fn="
40
749ee5389892 try to ignore superfluous /mpiwg/online in urls
casties
parents: 39
diff changeset
427 metaUrl=server+url.replace("/mpiwg/online","")
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
428 if not metaUrl.endswith("index.meta"):
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
429 metaUrl += "/index.meta"
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
430
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
431 return metaUrl
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
432
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
433 def getDomFromIndexMeta(self, url):
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
434 """get dom from index meta"""
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
435 dom = None
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
436 metaUrl = self.getIndexMetaUrl(url)
39
1dd90aabd366 added retry when reading index meta from texter applet
casties
parents: 38
diff changeset
437
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
438 logging.debug("(getDomFromIndexMeta): METAURL: %s"%metaUrl)
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
439 txt=getHttpData(metaUrl)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
440 if txt is None:
39
1dd90aabd366 added retry when reading index meta from texter applet
casties
parents: 38
diff changeset
441 raise IOError("Unable to read index meta from %s"%(url))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
442
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
443 dom = Parse(txt)
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
444 return dom
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
445
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
446 def getPresentationInfoXML(self, url):
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
447 """returns dom of info.xml document at url"""
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
448 dom = None
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
449 metaUrl = None
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
450 if url.startswith("http://"):
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
451 # real URL
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
452 metaUrl = url
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
453 else:
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
454 # online path
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
455 server=self.digilibBaseUrl+"/servlet/Texter?fn="
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
456 metaUrl=server+url.replace("/mpiwg/online","")
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
457
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
458 txt=getHttpData(metaUrl)
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
459 if txt is None:
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
460 raise IOError("Unable to read infoXMLfrom %s"%(url))
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
461
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
462 dom = Parse(txt)
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
463 return dom
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
464
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
465
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
466 def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
467 """gets authorization info from the index.meta file at path or given by dom"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
468 logging.debug("documentViewer (getauthinfofromindexmeta) path: %s"%(path))
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
469
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
470 access = None
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
471
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
472 if docinfo is None:
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
473 docinfo = {}
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
474
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
475 if dom is None:
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
476 for x in range(cut):
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
477 path=getParentDir(path)
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
478 dom = self.getDomFromIndexMeta(path)
46
31059e3d9338 has now also a text mode viewMode=text
dwinter
parents: 45
diff changeset
479
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
480 acctype = dom.xpath("//access-conditions/access/@type")
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
481 if acctype and (len(acctype)>0):
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
482 access=acctype[0].value
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
483 if access in ['group', 'institution']:
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
484 access = getTextFromNode(dom.xpath("//access-conditions/access/name")[0]).lower()
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
485
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
486 docinfo['accessType'] = access
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
487 return docinfo
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
488
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
489
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
490 def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
35
2d9261aea8f3 version 0.2.4
casties
parents: 32
diff changeset
491 """gets bibliographical info from the index.meta file at path or given by dom"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
492 logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path))
20
9884703dae70 new modi
dwinter
parents: 0
diff changeset
493
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
494 if docinfo is None:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
495 docinfo = {}
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
496
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
497 if dom is None:
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
498 for x in range(cut):
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
499 path=getParentDir(path)
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
500 dom = self.getDomFromIndexMeta(path)
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
501
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
502 docinfo['indexMetaPath']=self.getIndexMetaPath(path);
79
df6952ac93e9 bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents: 78
diff changeset
503
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
504 logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path))
59
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
505 # put in all raw bib fields as dict "bib"
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
506 bib = dom.xpath("//bib/*")
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
507 if bib and len(bib)>0:
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
508 bibinfo = {}
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
509 for e in bib:
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
510 bibinfo[e.localName] = getTextFromNode(e)
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
511 docinfo['bib'] = bibinfo
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
512
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
513 # extract some fields (author, title, year) according to their mapping
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
514 metaData=self.metadata.main.meta.bib
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
515 bibtype=dom.xpath("//bib/@type")
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
516 if bibtype and (len(bibtype)>0):
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
517 bibtype=bibtype[0].value
20
9884703dae70 new modi
dwinter
parents: 0
diff changeset
518 else:
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
519 bibtype="generic"
59
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
520
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
521 bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC)
59
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
522 docinfo['bib_type'] = bibtype
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
523 bibmap=metaData.generateMappingForType(bibtype)
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
524 logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap))
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
525 logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype))
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
526 # if there is no mapping bibmap is empty (mapping sometimes has empty fields)
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
527 if len(bibmap) > 0 and len(bibmap['author'][0]) > 0:
63
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
528 try:
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
529 docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0])
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
530 except: pass
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
531 try:
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
532 docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0])
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
533 except: pass
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
534 try:
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
535 docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0])
4a17b755bfc7 added more try/excepts to bib-meta reading code
casties
parents: 62
diff changeset
536 except: pass
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
537 logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype)
52
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
538 try:
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
539 docinfo['lang']=getTextFromNode(dom.xpath("//bib/lang")[0])
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
540 except:
92047eaa6272 zLOG exchanged by logging
dwinter
parents: 51
diff changeset
541 docinfo['lang']=''
59
996b61d71351 added all fields from bib tag to docinfo
casties
parents: 57
diff changeset
542
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
543 return docinfo
83
ec12a2440daa My last update Bukhman Andrey
abukhman
parents: 82
diff changeset
544
218
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
545
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
546 def getNameFromIndexMeta(self,path,docinfo=None,dom=None,cut=0):
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
547 """gets name info from the index.meta file at path or given by dom"""
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
548 if docinfo is None:
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
549 docinfo = {}
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
550
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
551 if dom is None:
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
552 for x in range(cut):
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
553 path=getParentDir(path)
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
554 dom = self.getDomFromIndexMeta(path)
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
555
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
556 docinfo['indexMetaPath']=self.getIndexMetaPath(path);
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
557
219
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
558 #result= dom.xpath("//result/resultPage")
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
559 #docinfo['numPages']=int(getTextFromNode(result[0]))
218
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
560
219
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
561 if len(name) > 0:
218
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
562 try:
219
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
563 result =dom.xpath("//name")
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
564 docinfo['name']=getTextFromNode(result[0])
17b205ca60a5 *** empty log message ***
abukhman
parents: 218
diff changeset
565 logging.debug("documentViewer docinfo[name] %s"%docinfo[name])
218
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
566 except: pass
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
567 #logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype)
66f334ec3256 *** empty log message ***
abukhman
parents: 205
diff changeset
568 return docinfo
83
ec12a2440daa My last update Bukhman Andrey
abukhman
parents: 82
diff changeset
569
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
570 def getDocinfoFromTextTool(self, url, dom=None, docinfo=None):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
571 """parse texttool tag in index meta"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
572 logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url))
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
573 if docinfo is None:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
574 docinfo = {}
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
575 if docinfo.get('lang', None) is None:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
576 docinfo['lang'] = '' # default keine Sprache gesetzt
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
577 if dom is None:
174
5780092e4989 *** empty log message ***
dwinter
parents: 173
diff changeset
578 dom = self.getDomFromIndexMeta(url)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
579
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
580 archivePath = None
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
581 archiveName = None
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
582
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
583 archiveNames = dom.xpath("//resource/name")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
584 if archiveNames and (len(archiveNames) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
585 archiveName = getTextFromNode(archiveNames[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
586 else:
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
587 logging.warning("documentViewer (getdocinfofromtexttool) resource/name missing in: %s" % (url))
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
588
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
589 archivePaths = dom.xpath("//resource/archive-path")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
590 if archivePaths and (len(archivePaths) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
591 archivePath = getTextFromNode(archivePaths[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
592 # clean up archive path
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
593 if archivePath[0] != '/':
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
594 archivePath = '/' + archivePath
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
595 if archiveName and (not archivePath.endswith(archiveName)):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
596 archivePath += "/" + archiveName
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
597 else:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
598 # try to get archive-path from url
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
599 logging.warning("documentViewer (getdocinfofromtexttool) resource/archive-path missing in: %s" % (url))
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
600 if (not url.startswith('http')):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
601 archivePath = url.replace('index.meta', '')
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
602
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
603 if archivePath is None:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
604 # we balk without archive-path
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
605 raise IOError("Missing archive-path (for text-tool) in %s" % (url))
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
606
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
607 imageDirs = dom.xpath("//texttool/image")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
608 if imageDirs and (len(imageDirs) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
609 imageDir = getTextFromNode(imageDirs[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
610
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
611 else:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
612 # we balk with no image tag / not necessary anymore because textmode is now standard
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
613 #raise IOError("No text-tool info in %s"%(url))
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
614 imageDir = ""
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
615 #xquery="//pb"
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
616 docinfo['imagePath'] = "" # keine Bilder
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
617 docinfo['imageURL'] = ""
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
618
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
619 if imageDir and archivePath:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
620 #print "image: ", imageDir, " archivepath: ", archivePath
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
621 imageDir = os.path.join(archivePath, imageDir)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
622 imageDir = imageDir.replace("/mpiwg/online", '')
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
623 docinfo = self.getDirinfoFromDigilib(imageDir, docinfo=docinfo)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
624 docinfo['imagePath'] = imageDir
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
625
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
626 docinfo['imageURL'] = self.digilibBaseUrl + "/servlet/Scaler?fn=" + imageDir
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
627
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
628 viewerUrls = dom.xpath("//texttool/digiliburlprefix")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
629 if viewerUrls and (len(viewerUrls) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
630 viewerUrl = getTextFromNode(viewerUrls[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
631 docinfo['viewerURL'] = viewerUrl
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
632
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
633 # old style text URL
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
634 textUrls = dom.xpath("//texttool/text")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
635 if textUrls and (len(textUrls) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
636 textUrl = getTextFromNode(textUrls[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
637 if urlparse.urlparse(textUrl)[0] == "": #keine url
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
638 textUrl = os.path.join(archivePath, textUrl)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
639 # fix URLs starting with /mpiwg/online
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
640 if textUrl.startswith("/mpiwg/online"):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
641 textUrl = textUrl.replace("/mpiwg/online", '', 1)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
642
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
643 docinfo['textURL'] = textUrl
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
644
130
5c779d7b5f71 more modular version with separate object MpdlXmlTextServer
casties
parents: 128
diff changeset
645 # new style text-url-path
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
646 textUrls = dom.xpath("//texttool/text-url-path")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
647 if textUrls and (len(textUrls) > 0):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
648 textUrl = getTextFromNode(textUrls[0])
102
27513c893ba9 trying to fix text-only documents
casties
parents: 101
diff changeset
649 docinfo['textURLPath'] = textUrl
27513c893ba9 trying to fix text-only documents
casties
parents: 101
diff changeset
650 if not docinfo['imagePath']:
27513c893ba9 trying to fix text-only documents
casties
parents: 101
diff changeset
651 # text-only, no page images
205
1fc7d73ce7ce *** empty log message ***
abukhman
parents: 204
diff changeset
652 docinfo = self.getNumTextPages(docinfo)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
653
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
654 presentationUrls = dom.xpath("//texttool/presentation")
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
655 docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag
221
f10ed44d5f05 *** empty log message ***
abukhman
parents: 220
diff changeset
656 #docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom)
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
657
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
658 if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
659 # presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
660 # durch den relativen Pfad auf die presentation infos
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
661 presentationPath = getTextFromNode(presentationUrls[0])
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
662 if url.endswith("index.meta"):
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
663 presentationUrl = url.replace('index.meta', presentationPath)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
664 else:
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
665 presentationUrl = url + "/" + presentationPath
102
27513c893ba9 trying to fix text-only documents
casties
parents: 101
diff changeset
666
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
667 docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom)
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
668
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
669 docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
670
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
671 return docinfo
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
672
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
673
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
674 def getBibinfoFromTextToolPresentation(self,url,docinfo=None,dom=None):
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
675 """gets the bibliographical information from the preseantion entry in texttools
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
676 """
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
677 dom=self.getPresentationInfoXML(url)
62
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
678 try:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
679 docinfo['author']=getTextFromNode(dom.xpath("//author")[0])
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
680 except:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
681 pass
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
682 try:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
683 docinfo['title']=getTextFromNode(dom.xpath("//title")[0])
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
684 except:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
685 pass
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
686 try:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
687 docinfo['year']=getTextFromNode(dom.xpath("//date")[0])
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
688 except:
8a16ea8db858 fixed bug in getInt
casties
parents: 61
diff changeset
689 pass
50
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
690 return docinfo
6c0f20cecc60 added evaluation of the presentation/info.xml in texttools
dwinter
parents: 49
diff changeset
691
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
692 def getDocinfoFromImagePath(self,path,docinfo=None,cut=0):
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
693 """path ist the path to the images it assumes that the index.meta file is one level higher."""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
694 logging.debug("documentViewer (getdocinfofromimagepath) path: %s"%(path))
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
695 if docinfo is None:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
696 docinfo = {}
29
e1bed068b351 small fixes
casties
parents: 28
diff changeset
697 path=path.replace("/mpiwg/online","")
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
698 docinfo['imagePath'] = path
73
dwinter
parents: 71
diff changeset
699 docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo,cut=cut)
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
700
79
df6952ac93e9 bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents: 78
diff changeset
701 pathorig=path
78
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
702 for x in range(cut):
70ab234a18dc bugs in filepath mode fixes
dwinter
parents: 75
diff changeset
703 path=getParentDir(path)
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
704 logging.debug("documentViewer (getdocinfofromimagepath) PATH:"+path)
31
c6451e8d5d23 more small fixes - now version 0.2.2
casties
parents: 29
diff changeset
705 imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
706 docinfo['imageURL'] = imageUrl
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
707
79
df6952ac93e9 bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents: 78
diff changeset
708 #path ist the path to the images it assumes that the index.meta file is one level higher.
df6952ac93e9 bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents: 78
diff changeset
709 docinfo = self.getBibinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1)
df6952ac93e9 bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents: 78
diff changeset
710 docinfo = self.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
711 return docinfo
20
9884703dae70 new modi
dwinter
parents: 0
diff changeset
712
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
713
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
714 def getDocinfo(self, mode, url):
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
715 """returns docinfo depending on mode"""
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
716 logging.debug("documentViewer (getdocinfo) mode: %s, url: %s"%(mode,url))
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
717 # look for cached docinfo in session
51
c5d3aabbf61b textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents: 50
diff changeset
718 if self.REQUEST.SESSION.has_key('docinfo'):
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
719 docinfo = self.REQUEST.SESSION['docinfo']
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
720 # check if its still current
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
721 if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url:
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
722 logging.debug("documentViewer (getdocinfo) docinfo in session: %s"%docinfo)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
723 return docinfo
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
724 # new docinfo
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
725 docinfo = {'mode': mode, 'url': url}
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
726 if mode=="texttool": #index.meta with texttool information
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
727 docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo)
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
728 elif mode=="imagepath":
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
729 docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo)
70
0049d64aa464 filepath introduced
dwinter
parents: 68
diff changeset
730 elif mode=="filepath":
75
9673218e155b minorCVS: ----------------------------------------------------------------------
dwinter
parents: 74
diff changeset
731 docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
732 else:
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
733 logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode)
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
734 raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode))
37
ead830ce45d6 better error messages
casties
parents: 35
diff changeset
735
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
736 logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo)
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
737 self.REQUEST.SESSION['docinfo'] = docinfo
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
738 return docinfo
128
4db49915d825 Last lemma
abukhman
parents: 127
diff changeset
739
170
caf66e1fccd3 characterNormalization
abukhman
parents: 169
diff changeset
740 def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None,characterNormalization=""):
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
741 """returns pageinfo with the given parameters"""
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
742 pageinfo = {}
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
743 current = getInt(current)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
744 pageinfo['current'] = current
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
745 rows = int(rows or self.thumbrows)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
746 pageinfo['rows'] = rows
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
747 cols = int(cols or self.thumbcols)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
748 pageinfo['cols'] = cols
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
749 grpsize = cols * rows
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
750 pageinfo['groupsize'] = grpsize
61
f3d2f240692c fixed bug in calculation of group numbers
casties
parents: 59
diff changeset
751 start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1)))
f3d2f240692c fixed bug in calculation of group numbers
casties
parents: 59
diff changeset
752 # int(current / grpsize) * grpsize +1))
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
753 pageinfo['start'] = start
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
754 pageinfo['end'] = start + grpsize
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
755 if (docinfo is not None) and ('numPages' in docinfo):
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
756 np = int(docinfo['numPages'])
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
757 pageinfo['end'] = min(pageinfo['end'], np)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
758 pageinfo['numgroups'] = int(np / grpsize)
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
759 if np % grpsize > 0:
128
4db49915d825 Last lemma
abukhman
parents: 127
diff changeset
760 pageinfo['numgroups'] += 1
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
761 pageinfo['viewMode'] = viewMode
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
762 pageinfo['tocMode'] = tocMode
173
9fa61da63e5e characterNormalization
abukhman
parents: 170
diff changeset
763 #pageinfo['characterNormalization'] =characterNormalization
9fa61da63e5e characterNormalization
abukhman
parents: 170
diff changeset
764 pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization',' ')
95
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
765 pageinfo['query'] = self.REQUEST.get('query',' ')
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
766 pageinfo['queryType'] = self.REQUEST.get('queryType',' ')
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
767 pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext')
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
768 pageinfo['textPN'] = self.REQUEST.get('textPN','1')
106
9f11913bc132 *** empty log message ***
abukhman
parents: 105
diff changeset
769 pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','')
95
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
770 pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30')
105
6bffb612cd6a *** empty log message ***
abukhman
parents: 104
diff changeset
771 pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '10')
90
6a4a72033d58 new version with new full-text infrastructure and some more changed templates
casties
parents: 84
diff changeset
772 pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1')
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
773 toc = int (pageinfo['tocPN'])
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
774 pageinfo['textPages'] =int (toc)
158
4f4fe4e56ffe characterNormalization
abukhman
parents: 157
diff changeset
775
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
776 if 'tocSize_%s'%tocMode in docinfo:
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
777 tocSize = int(docinfo['tocSize_%s'%tocMode])
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
778 tocPageSize = int(pageinfo['tocPageSize'])
128
4db49915d825 Last lemma
abukhman
parents: 127
diff changeset
779 # cached toc
99
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
780 if tocSize%tocPageSize>0:
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
781 tocPages=tocSize/tocPageSize+1
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
782 else:
4738a696d265 Search function update
abukhman
parents: 97
diff changeset
783 tocPages=tocSize/tocPageSize
128
4db49915d825 Last lemma
abukhman
parents: 127
diff changeset
784 pageinfo['tocPN'] = min (tocPages,toc)
95
db6d594aa4d9 Last update with search function (getSearch)
abukhman
parents: 90
diff changeset
785 pageinfo['searchPN'] =self.REQUEST.get('searchPN','1')
112
d48c1d507556 *** empty log message ***
abukhman
parents: 109
diff changeset
786 pageinfo['sn'] =self.REQUEST.get('sn','')
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
787 return pageinfo
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
788
128
4db49915d825 Last lemma
abukhman
parents: 127
diff changeset
789 def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None):
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
790 """init document viewer"""
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
791 self.title=title
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
792 self.digilibBaseUrl = digilibBaseUrl
25
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
793 self.thumbrows = thumbrows
e93fb8cadd3a new, less preliminary version 0.2
casties
parents: 22
diff changeset
794 self.thumbcols = thumbcols
32
b25c89d693cf version 0.2.3 - first version with access control!
casties
parents: 31
diff changeset
795 self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
796 if RESPONSE is not None:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
797 RESPONSE.redirect('manage_main')
0
dwinter
parents:
diff changeset
798
dwinter
parents:
diff changeset
799 def manage_AddDocumentViewerForm(self):
dwinter
parents:
diff changeset
800 """add the viewer form"""
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
801 pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self)
0
dwinter
parents:
diff changeset
802 return pt()
dwinter
parents:
diff changeset
803
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
804 def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None):
0
dwinter
parents:
diff changeset
805 """add the viewer"""
84
a6e4f9b6729a first version with new full-text infrastructure and slightly changed templates
casties
parents: 83
diff changeset
806 newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName)
0
dwinter
parents:
diff changeset
807 self._setObject(id,newObj)
dwinter
parents:
diff changeset
808
dwinter
parents:
diff changeset
809 if RESPONSE is not None:
dwinter
parents:
diff changeset
810 RESPONSE.redirect('manage_main')
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
811
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
812 ## DocumentViewerTemplate class
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
813 class DocumentViewerTemplate(ZopePageTemplate):
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
814 """Template for document viewer"""
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
815 meta_type="DocumentViewer Template"
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
816
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
817
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
818 def manage_addDocumentViewerTemplateForm(self):
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
819 """Form for adding"""
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
820 pt=PageTemplateFile('zpt/addDocumentViewerTemplate', globals()).__of__(self)
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
821 return pt()
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
822
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
823 def manage_addDocumentViewerTemplate(self, id='viewer_main', title=None, text=None,
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
824 REQUEST=None, submit=None):
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
825 "Add a Page Template with optional file content."
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
826
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
827 self._setObject(id, DocumentViewerTemplate(id))
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
828 ob = getattr(self, id)
53
f4e0af8c281d NEW - # 44: ECHO - vollst?ndige bibliographische Angabe
dwinter
parents: 52
diff changeset
829 txt=file(os.path.join(package_home(globals()),'zpt/viewer_main.zpt'),'r').read()
167
7e2b97941a66 characterNormalization
abukhman
parents: 166
diff changeset
830 logging.info("txt %s:"%txt)
53
f4e0af8c281d NEW - # 44: ECHO - vollst?ndige bibliographische Angabe
dwinter
parents: 52
diff changeset
831 ob.pt_edit(txt,"text/html")
22
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
832 if title:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
833 ob.pt_setTitle(title)
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
834 try:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
835 u = self.DestinationURL()
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
836 except AttributeError:
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
837 u = REQUEST['URL1']
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
838
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
839 u = "%s/%s" % (u, urllib.quote(id))
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
840 REQUEST.RESPONSE.redirect(u+'/manage_main')
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
841 return ''
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
842
b139f9937e97 preliminary version 0.2
casties
parents: 20
diff changeset
843
41
0c8ee8fcfd76 some more logging
casties
parents: 40
diff changeset
844