diploma work

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1274 07558da8-63fa-0310-ba24-9fe276d99e06

diff --git a/conf/dipl/config.yml b/conf/dipl/config.yml
index 61e07db..fed5a27 100644 (file)
--- a/conf/dipl/config.yml
+++ b/conf/dipl/config.yml
@@ -77,6 +77,7 @@ webpac:
     excel: 'WebPAC::Input::Excel'
     dbf: 'WebPAC::Input::DBF'
     isi: 'WebPAC::Input::ISI'
+    csv: 'WebPAC::Input::CSV'
   # define delimiters for validation
   delimiters:
     - ' ; '
@@ -107,7 +108,7 @@ databases:
 #        path: '/data/FF/citati/proba.xls'
 #        encoding: 'windows-1250'
 #        normalize:
-#          path: 'conf/isi/isi2xls-citirani.pl'
+#          path: 'conf/dipl/isi2xls-citirani.pl'
 #    output:
 #      - module: 'Excel'
 #        path: '/data/FF/citiari/citirani.xls'
@@ -121,32 +122,32 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: arambasicl
         type: isi
         path: '/data/FF/citati/citing/arambasicl.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: bratkod
         type: isi
         path: '/data/FF/citati/citing/bratkod.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: buskov
         type: isi
         path: '/data/FF/citati/citing/buskov.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: butkovica.txt
         type: isi
@@ -154,8 +155,8 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: corkalobiruskid.txt
         type: isi
@@ -163,8 +164,8 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: hrend
         type: isi
@@ -172,16 +173,16 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: hromatkoi
         type: isi
         path: '/data/FF/citati/citing/hromatkoi.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: hudekknezevicj
         type: isi
@@ -189,32 +190,32 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: kaliternalipovcanlj
         type: isi
         path: '/data/FF/citati/citing/kaliternalipovcanlj.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: kardumi
         type: isi
         path: '/data/FF/citati/citing/kardumi.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: kerestesg
         type: isi
         path: '/data/FF/citati/citing/kerestesg.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: knezovicz
         type: isi
@@ -222,8 +223,8 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: kolesaricv
         type: isi
@@ -231,8 +232,8 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: kuterovacjagodicg
         type: isi
@@ -240,40 +241,40 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: lackovicgrgink
         type: isi
         path: '/data/FF/citati/citing/lackovicgrgink.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: lamzaposavecv
         type: isi
         path: '/data/FF/citati/citing/lamzaposavecv.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: manenicai 
         type: isi
         path: '/data/FF/citati/citing/manenicai.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: maslicsersicd
         type: isi
         path: '/data/FF/citati/citing/maslicsersicd.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: radosevicvidacekb
         type: isi
@@ -281,24 +282,24 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: sverkob
         type: isi
         path: '/data/FF/citati/citing/sverkob.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: galicz
         type: isi
         path: '/data/FF/citati/citing/galicz.txt'
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
         skip: 1
       - name: tadinacm
         type: isi
@@ -306,8 +307,8 @@ databases:
         encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/tables.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/tables.pl'
       - name: radovi 
         type: excel
         path: '/data/FF/citati/citirani_radovi.xls'
@@ -315,14 +316,27 @@ databases:
         # encoding: 'ISO-8859-1'
         normalize:
         # - path: 'conf/normalize/isi/isi2xls.pl'
-        # - path: 'conf/isi/isi2sorted.pl'
-          - path: 'conf/isi/citirani-radovi.pl'
+        # - path: 'conf/dipl/isi2sorted.pl'
+          - path: 'conf/dipl/citirani-radovi.pl'
       - name: cropsy
         type: isi
         path: '/data/FF/citati/croatia-psychol.txt'
         normalize:
-          - path: 'conf/isi/tables-cropsy.pl'
+          - path: 'conf/dipl/tables-cropsy.pl'
         skip: 0
+      - name: jcr
+        type: csv
+        path: "/data/FF/citati/jcr_recs.txt"
+        normalize:
+          - path: 'conf/dipl/tables-jcr.pl'
+      - name: izbori
+        type: excel
+        path: "/data/FF/citati/izbori.xls"
+        worksheet: podaci
+        from: 1
+        to: 886
+        normalize:
+          - path: 'conf/dipl/tables-izbori.pl'
     output:
 #      - module: 'Excel'
 #        path: '/data/FF/citati/data.xls'
@@ -332,5 +346,5 @@ databases:
 #        url: 'http://193.198.212.57:5984'
       - module: 'DBI'
         dsn: 'dbi:Pg:dbname=dipl'
-        schema: 'conf/isi/schema.sql'
+        schema: 'conf/dipl/schema.sql'
 #        table: 'citirani'

diff --git a/conf/dipl/schema.sql b/conf/dipl/schema.sql
index ac332d9..5dbb6f4 100644 (file)
--- a/conf/dipl/schema.sql
+++ b/conf/dipl/schema.sql
@@ -115,6 +115,29 @@ create table cropsy (
        py integer
 );
 
+drop table if exists jcr;
+create table jcr (
+       id serial,
+       abbt text,
+       issn text,
+       tc text,
+       impf text,
+       impf5 text,
+       ii text,
+       noart text,
+       hlife text,
+       eigen text,
+       infl text
+);     
+
+drop table if exists izbori;
+create table izbori (
+       id serial,
+       autor text,
+       casopis text,
+       godina text,
+       sd numeric
+);
 
 -- CREATE AGGREGATE array_accum (anyelement)
 -- (