trunk/conf/mjesec-isi.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'
    dbf: 'WebPAC::Input::DBF'
    isi: 'WebPAC::Input::ISI'
  # 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/peri/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: isi
        credit: 10000
  
  citirani:
    name: 'CITIRANI' 
    input:
      - name: radovi
        type: excel
        path: '/data/citiranje/proba.xls'
        encoding: 'windows-1250'
        # modify_file: 'conf/modify/common.pl'
        normalize:
          path: 'conf/normalize/isi2xls-citirani.pl'
    output:
      - module: 'Excel'
        path: '/data/citiranje/citirani.xls'
  
  isi:
    name: 'ISI' 
    input:
      - name: isi
        type: isi
        path: '/home/mglavica/citiranje/citirajuci_clanci/sverkob.txt'
        encoding: 'ISO-8859-1'
        # modify_file: 'conf/modify/common.pl'
        normalize:
          path: 'conf/normalize/isi2xls.pl'
    output:
      - module: 'Excel'
        path: '/data/citiranje/sverko.xls'
1	--- #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	#use_indexer: 'hyperestraier'
13	#use_indexer: 'hyperestraier-native'
14	#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	#
29	path: 'casket/'
30	# 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	dbf: 'WebPAC::Input::DBF'
79	isi: 'WebPAC::Input::ISI'
80	# define delimiters for validation
81	delimiters:
82	- ' ; '
83	- ' : '
84	- ' / '
85	- ' = '
86	editor:
87	# open this record when opening editor
88	# (it will also be used to deduce default database and input)
89	default_record_uri: 'ps/peri/1'
90
91	# directives after this are used when indexing using core WebPAC modules
92
93	databases:
94	# This is empty database created only in Hyper Estraier to merge
95	# all three databases
96	'webpac2':
97	name: 'Search all'
98	links:
99	- to: isi
100	credit: 10000
101
102	citirani:
103	name: 'CITIRANI'
104	input:
105	- name: radovi
106	type: excel
107	path: '/data/citiranje/proba.xls'
108	encoding: 'windows-1250'
109	# modify_file: 'conf/modify/common.pl'
110	normalize:
111	path: 'conf/normalize/isi2xls-citirani.pl'
112	output:
113	- module: 'Excel'
114	path: '/data/citiranje/citirani.xls'
115
116	isi:
117	name: 'ISI'
118	input:
119	- name: isi
120	type: isi
121	path: '/home/mglavica/citiranje/citirajuci_clanci/sverkob.txt'
122	encoding: 'ISO-8859-1'
123	# modify_file: 'conf/modify/common.pl'
124	normalize:
125	path: 'conf/normalize/isi2xls.pl'
126	output:
127	- module: 'Excel'
128	path: '/data/citiranje/sverko.xls'