trunk/conf/mjesec.yml

--- #YAML:1.0
# DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!

# encoding of this configuration file
config_encoding: 'ISO-8859-2'
# encoding in Catalyst.
catalyst_encoding: 'UTF-8'
# relative path to sites templates under Catalyst root
sites_root: 'sites'

# which indexing engine to use? (hyperestraier is default)
#use_indexer: 'hyperestraier'
use_indexer: 'hyperestraier-native'
#use_indexer: 'kinosearch'

# configuration for Hyper Estraier full text search engine
hyperestraier:
  #url: 'http://localhost:1978/node/webpac2'
  masterurl: 'http://localhost:1978'
  defaultnode: 'webpac2'
#  defaultnode: 'ps'
  defaultdepth: 1
  user: 'admin'
  passwd: 'admin'
  # don't turn this on! it will spit huge amounts of output
  #debug: 1
  #
  #
  path: 'casket/'
  # number of results on each page
  hits_on_page: 10
  # number of results to fetch for suggestion (it will fold multiple sameones)
  hits_for_suggest: 20
  #
  # options used while indexing
  #
  # which tag type to use for search engine (used while indexing)
  type: 'search'
  #

# options for pager
pager:
  # how many pages to show for navigation?
  max_pages: 20

# configuration for KinoSearch search engine library
kinosearch:
  index_path: './kinosearch/'
  database: 'unconfigured database name'
  label: 'unconfigured database label'
  encoding: 'iso-8859-2'
  # clean database before opening? (WARNING: this erases existing database)
  clean: 1
  # which field type to index?
  type: 'search'

webpac:
  # default template to use
  template: 'html_ffzg_results_short.tt'
  # path to database files
  db_path: '/data/webpac2/db'
  # path to templates used by WebPAC::Output
  template_path: '/data/webpac2/conf/output/tt'
  # default template for results
  default_template: 'html_ffzg.tt'
  # default user editable css file
  default_css: 'user.css'
  css_path: 'root/css'
  # encoding comming from webpac
  webpac_encoding: 'iso-8859-2'
  # encoding expected by Catalyst
  out_encoding: 'utf-8'
  # define different input formats (types) and perl modules to handle them
  inputs:
    isis: 'WebPAC::Input::ISIS'
    marc: 'WebPAC::Input::MARC'
    excel: 'WebPAC::Input::Excel'
  # define delimiters for validation
  delimiters:
    - ' ; '
    - ' : '
    - ' / '
    - ' = '
editor:
  # open this record when opening editor
  # (it will also be used to deduce default database and input)
  default_record_uri: 'ps/libri/1'

# directives after this are used when indexing using core WebPAC modules

databases:
  # This is empty database created only in Hyper Estraier to merge
  # all three databases
  'webpac2':
    name: 'Search all'
    links:
      - to: ffan
        credit: 10000
      - to: ffar
        credit: 10000
      - to: ffet
        credit: 10000
      - to: fffi 
        credit: 10000
      - to: fffo
        credit: 10000
      - to: ffge
        credit: 10000
      - to: ffhu 
        credit: 10000
      - to: ffiz
        credit: 10000
      - to: ffkf
        credit: 10000
      - to: ffkk 
        credit: 10000
      - to: fflo
        credit: 10000
      - to: fftu
        credit: 10000
      - to: ffpe   
        credit: 10000
      - to: ffpo
        credit: 10000
      - to: ffpu
        credit: 10000
      - to: ffps 
        credit: 10000
      - to: ffro   
        credit: 10000
      - to: ffsk
        credit: 10000
      - to: ffsf
        credit: 10000
      - to: ffso 
        credit: 10000
      - to: ffta 
        credit: 10000
      - to: ffsr 
        credit: 10000
      - to: ffmic 
        credit: 10000
      - to: vusp 
        credit: 10000
      - to: kem 
        credit: 10000
      - to: efzg 
        credit: 10000


#  hazu:
#    name: 'HAZU'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/unizg/hazu/BIB'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  sand:
#    name: 'SAND'
#    input:
#      name: casopisi
#      type: dbf
#      path: '/data/unizg/casopisi/sys/cas2000.dbf'
#      normalize:
#        path: 'conf/normalize/common.pl'
##      mapping_path: 'conf/input/dbf/cas2000.yml'
#  
#  hidra:
#    name: 'HIDRA'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/hidra/test/BIB'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  efzg:
#    name: 'EFZG'
#    input:
#      - name: crolist
#        type: marc
#        path: '/data/unizg/drustvene/efzg/unimarc.iso'
#        encoding: 'ISO-8859-2'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#  
#  ipu:
#    name: 'Institut za povijest umjetnosti'
#    input:
#      - name: bibl
#        type: isis
#        path: '/data/unizg/humanistika/ipu/20071115/bibl/BIBL'
#        encoding: 'cp852'
#        # modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  zvonar:
#    name: 'HAZU Odsjek za povijesne znanosti'
#    input:
#      - name: libri
#        type: isis
#        path: '/data/unizg/humanistika/hazu-p/LIBRI'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#      - name: peri
#        type: isis
#        path: '/data/unizg/humanistika/hazu-p/PERI'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  stross:
#    name: 'Strossmayerova galerija'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/unizg/stross/BIB'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#       
#  koncar:
#    name: 'Konèar'
#    input:
#      - name: unimar
#        type: isis
#        path: '/data/tehnika/koncar/UNIMAR'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#       
  datumi:
    name: 'Nevaljali datumi'
    input:
      - name: tablica
        type: excel
        path: '/data/isis_data/greske/datum_unosa.xls'
        encoding: 'windows-1250'
        normalize:
          - path: 'conf/normalize/common.pl'
  
#  ffps:
#    name: 'Psihologija, Filozofski fakultet u Zagrebu'
#    code: 'ps'
#    url: 'http://www.knjiznice.ffzg.hr/psihologija'
#    input:
#      - name: libri
#        type: isis
#        path: '/backup/isis_backup/sunce2/isisdata/latest/LIBRI/'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          - path: 'conf/normalize/report-sorted-txt.pl'
#    output:
##      - module: 'Excel'
##        path: '/data/isis_data/ps/reports/libri-nema-jezik.xls'
#      - module: 'Sorted'
#        path: 'out/report/mfn/ps/'
##      - module: 'Excel'
##        path: '/data/isis_data/ps/reports/autori-udk.xls'


  ffiz:
    name: 'Informacijske znanosti, Filozofski fakultet u Zagrebu'
    input:
      - name: libri
        type: isis
        path: '/backup/isis_backup/A129-2/isisdata/latest/LIBRI/LIBRI'
        encoding: 'cp852'
        modify_file: 'conf/modify/common.pl'
        normalize:
          - path: 'conf/normalize/report-sorted-txt.pl'
    output:
#      - module: 'Excel'
#        path: '/data/isis_data/iz/reports/libri-nema-jezik.xls'
      - module: 'Sorted'
        path: 'out/report/mfn/iz/'
#      - module: 'Excel'
#        path: '/data/isis_data/iz/reports/autori-udk.xls'


  casopisi:
    name: 'Èasopisi u Knji¾nicama FF-a'
    input:
      - name: ff
        type: excel
        path: '/data/isis_data/CASOPISI/svi-casopisi-FF.xls'
        ncoding: 'utf-8'
        normalize:
          - path: 'conf/normalize/ff-casopisi-xls.pl'
#          - path: 'conf/normalize/ff-casopisi-marc.pl'
    output:
      - module: 'Excel'
        path: 'out/excel/casopisi-FF-dupli-knjiznice.xls'

  fflibri:
    name: 'Filozofski fakultet u Zagrebu'
    input:
      - name: marc
        type: marc
        path: 'out/marc/fflibri.marc'
        encoding: 'ISO-8859-2'
        normalize:
          - path: 'conf/normalize/ff-libri-dupli.pl'
    output:
      - module: 'Excel'
        path: 'out/excel/knjige-duplo.xls'

1	dpavlin	612	--- #YAML:1.0
2			# DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
3
4			# encoding of this configuration file
5			config_encoding: 'ISO-8859-2'
6			# encoding in Catalyst.
7			catalyst_encoding: 'UTF-8'
8			# relative path to sites templates under Catalyst root
9			sites_root: 'sites'
10
11			# which indexing engine to use? (hyperestraier is default)
12	dpavlin	635	#use_indexer: 'hyperestraier'
13	dpavlin	1082	use_indexer: 'hyperestraier-native'
14	dpavlin	612	#use_indexer: 'kinosearch'
15
16			# configuration for Hyper Estraier full text search engine
17			hyperestraier:
18			#url: 'http://localhost:1978/node/webpac2'
19			masterurl: 'http://localhost:1978'
20			defaultnode: 'webpac2'
21			# defaultnode: 'ps'
22			defaultdepth: 1
23			user: 'admin'
24			passwd: 'admin'
25			# don't turn this on! it will spit huge amounts of output
26			#debug: 1
27			#
28	dpavlin	635	#
29			path: 'casket/'
30	dpavlin	612	# number of results on each page
31			hits_on_page: 10
32			# number of results to fetch for suggestion (it will fold multiple sameones)
33			hits_for_suggest: 20
34			#
35			# options used while indexing
36			#
37			# which tag type to use for search engine (used while indexing)
38			type: 'search'
39			#
40
41			# options for pager
42			pager:
43			# how many pages to show for navigation?
44			max_pages: 20
45
46			# configuration for KinoSearch search engine library
47			kinosearch:
48			index_path: './kinosearch/'
49			database: 'unconfigured database name'
50			label: 'unconfigured database label'
51			encoding: 'iso-8859-2'
52			# clean database before opening? (WARNING: this erases existing database)
53			clean: 1
54			# which field type to index?
55			type: 'search'
56
57			webpac:
58			# default template to use
59			template: 'html_ffzg_results_short.tt'
60			# path to database files
61			db_path: '/data/webpac2/db'
62			# path to templates used by WebPAC::Output
63			template_path: '/data/webpac2/conf/output/tt'
64			# default template for results
65			default_template: 'html_ffzg.tt'
66			# default user editable css file
67			default_css: 'user.css'
68			css_path: 'root/css'
69			# encoding comming from webpac
70			webpac_encoding: 'iso-8859-2'
71			# encoding expected by Catalyst
72			out_encoding: 'utf-8'
73			# define different input formats (types) and perl modules to handle them
74			inputs:
75			isis: 'WebPAC::Input::ISIS'
76			marc: 'WebPAC::Input::MARC'
77			excel: 'WebPAC::Input::Excel'
78	dpavlin	842	# define delimiters for validation
79			delimiters:
80			- ' ; '
81			- ' : '
82			- ' / '
83			- ' = '
84	dpavlin	612	editor:
85			# open this record when opening editor
86			# (it will also be used to deduce default database and input)
87	dpavlin	1090	default_record_uri: 'ps/libri/1'
88	dpavlin	612
89			# directives after this are used when indexing using core WebPAC modules
90
91			databases:
92			# This is empty database created only in Hyper Estraier to merge
93			# all three databases
94			'webpac2':
95			name: 'Search all'
96			links:
97	dpavlin	807	- to: ffan
98			credit: 10000
99			- to: ffar
100			credit: 10000
101			- to: ffet
102			credit: 10000
103			- to: fffi
104			credit: 10000
105			- to: fffo
106			credit: 10000
107			- to: ffge
108			credit: 10000
109			- to: ffhu
110			credit: 10000
111			- to: ffiz
112			credit: 10000
113			- to: ffkf
114			credit: 10000
115			- to: ffkk
116			credit: 10000
117			- to: fflo
118			credit: 10000
119			- to: fftu
120			credit: 10000
121			- to: ffpe
122			credit: 10000
123			- to: ffpo
124			credit: 10000
125			- to: ffpu
126			credit: 10000
127	dpavlin	806	- to: ffps
128			credit: 10000
129	dpavlin	807	- to: ffro
130	dpavlin	612	credit: 10000
131	dpavlin	807	- to: ffsk
132	dpavlin	612	credit: 10000
133	dpavlin	807	- to: ffsf
134			credit: 10000
135			- to: ffso
136			credit: 10000
137			- to: ffta
138			credit: 10000
139			- to: ffsr
140			credit: 10000
141			- to: ffmic
142			credit: 10000
143	dpavlin	848	- to: vusp
144			credit: 10000
145			- to: kem
146			credit: 10000
147	dpavlin	1090	- to: efzg
148			credit: 10000
149	dpavlin	612
150	dpavlin	807
151	dpavlin	1095	# hazu:
152			# name: 'HAZU'
153			# input:
154			# - name: bib
155			# type: isis
156			# path: '/data/unizg/hazu/BIB'
157			# encoding: 'cp852'
158			# modify_file: 'conf/modify/common.pl'
159			# normalize:
160			# path: 'conf/normalize/minimal.pl'
161			#
162			# sand:
163			# name: 'SAND'
164			# input:
165			# name: casopisi
166			# type: dbf
167			# path: '/data/unizg/casopisi/sys/cas2000.dbf'
168			# normalize:
169			# path: 'conf/normalize/common.pl'
170			## mapping_path: 'conf/input/dbf/cas2000.yml'
171			#
172			# hidra:
173			# name: 'HIDRA'
174			# input:
175			# - name: bib
176			# type: isis
177			# path: '/data/hidra/test/BIB'
178			# encoding: 'cp852'
179			# modify_file: 'conf/modify/common.pl'
180			# normalize:
181			# path: 'conf/normalize/minimal.pl'
182			#
183			# efzg:
184			# name: 'EFZG'
185			# input:
186			# - name: crolist
187			# type: marc
188			# path: '/data/unizg/drustvene/efzg/unimarc.iso'
189			# encoding: 'ISO-8859-2'
190			# normalize:
191			# path: 'conf/normalize/minimal.pl'
192			#
193			# ipu:
194			# name: 'Institut za povijest umjetnosti'
195			# input:
196			# - name: bibl
197			# type: isis
198			# path: '/data/unizg/humanistika/ipu/20071115/bibl/BIBL'
199			# encoding: 'cp852'
200			# # modify_file: 'conf/modify/common.pl'
201			# normalize:
202			# path: 'conf/normalize/minimal.pl'
203			#
204			# zvonar:
205			# name: 'HAZU Odsjek za povijesne znanosti'
206			# input:
207			# - name: libri
208			# type: isis
209			# path: '/data/unizg/humanistika/hazu-p/LIBRI'
210			# normalize:
211			# path: 'conf/normalize/minimal.pl'
212			# - name: peri
213			# type: isis
214			# path: '/data/unizg/humanistika/hazu-p/PERI'
215			# normalize:
216			# path: 'conf/normalize/minimal.pl'
217			#
218			# stross:
219			# name: 'Strossmayerova galerija'
220			# input:
221			# - name: bib
222			# type: isis
223			# path: '/data/unizg/stross/BIB'
224			# normalize:
225			# path: 'conf/normalize/minimal.pl'
226			#
227			# koncar:
228			# name: 'Konèar'
229			# input:
230			# - name: unimar
231			# type: isis
232			# path: '/data/tehnika/koncar/UNIMAR'
233			# normalize:
234			# path: 'conf/normalize/minimal.pl'
235			#
236			datumi:
237			name: 'Nevaljali datumi'
238			input:
239			- name: tablica
240			type: excel
241			path: '/data/isis_data/greske/datum_unosa.xls'
242			encoding: 'windows-1250'
243			normalize:
244			- path: 'conf/normalize/common.pl'
245
246			# ffps:
247			# name: 'Psihologija, Filozofski fakultet u Zagrebu'
248			# code: 'ps'
249			# url: 'http://www.knjiznice.ffzg.hr/psihologija'
250			# input:
251			# - name: libri
252			# type: isis
253			# path: '/backup/isis_backup/sunce2/isisdata/latest/LIBRI/'
254			# encoding: 'cp852'
255			# modify_file: 'conf/modify/common.pl'
256			# normalize:
257			# - path: 'conf/normalize/report-sorted-txt.pl'
258			# output:
259			## - module: 'Excel'
260			## path: '/data/isis_data/ps/reports/libri-nema-jezik.xls'
261			# - module: 'Sorted'
262			# path: 'out/report/mfn/ps/'
263			## - module: 'Excel'
264			## path: '/data/isis_data/ps/reports/autori-udk.xls'
265	dpavlin	807
266
267	dpavlin	1095	ffiz:
268			name: 'Informacijske znanosti, Filozofski fakultet u Zagrebu'
269	dpavlin	1082	input:
270	dpavlin	1095	- name: libri
271	dpavlin	1082	type: isis
272	dpavlin	1095	path: '/backup/isis_backup/A129-2/isisdata/latest/LIBRI/LIBRI'
273	dpavlin	1082	encoding: 'cp852'
274			modify_file: 'conf/modify/common.pl'
275			normalize:
276	dpavlin	1095	- path: 'conf/normalize/report-sorted-txt.pl'
277			output:
278			# - module: 'Excel'
279			# path: '/data/isis_data/iz/reports/libri-nema-jezik.xls'
280			- module: 'Sorted'
281			path: 'out/report/mfn/iz/'
282			# - module: 'Excel'
283			# path: '/data/isis_data/iz/reports/autori-udk.xls'
284	dpavlin	1078
285	dpavlin	1095
286			casopisi:
287			name: 'Èasopisi u Knji¾nicama FF-a'
288	dpavlin	1082	input:
289	dpavlin	1095	- name: ff
290			type: excel
291			path: '/data/isis_data/CASOPISI/svi-casopisi-FF.xls'
292			ncoding: 'utf-8'
293	dpavlin	1082	normalize:
294	dpavlin	1095	- path: 'conf/normalize/ff-casopisi-xls.pl'
295			# - path: 'conf/normalize/ff-casopisi-marc.pl'
296			output:
297			- module: 'Excel'
298			path: 'out/excel/casopisi-FF-dupli-knjiznice.xls'
299	dpavlin	1078
300	dpavlin	1095	fflibri:
301			name: 'Filozofski fakultet u Zagrebu'
302	dpavlin	1082	input:
303	dpavlin	1095	- name: marc
304	dpavlin	1090	type: marc
305	dpavlin	1095	path: 'out/marc/fflibri.marc'
306	dpavlin	1090	encoding: 'ISO-8859-2'
307	dpavlin	1082	normalize:
308	dpavlin	1095	- path: 'conf/normalize/ff-libri-dupli.pl'
309			output:
310			- module: 'Excel'
311			path: 'out/excel/knjige-duplo.xls'
312	dpavlin	1082