First steps in integrating COMHIS cleanups (#94):

- add date substitution CSV - add script for making a simplified version of it for Catmandu processing (with unit test) - add a test record with a broken data field that will be used later
NatLibFi · Jul 3, 2019 · b0ced3c · b0ced3c
1 parent b955f9c
commit b0ced3c
Show file tree

Hide file tree

Showing 5 changed files with 33 additions and 0 deletions.
diff --git a/Makefile b/Makefile
@@ -28,6 +28,9 @@ split-input/%.md5: input/%.alephseq
 slices/%.md5: split-input/%.md5
 	scripts/update-slices.sh $^ $@
 
+refdata/subst-260c.csv: refdata/fennica-dates.csv.gz
+	zcat $^ | scripts/extract-subst-260c.py >$@
+
 refdata/iso639-2-fi.csv: sparql/extract-iso639-2-fi.rq
 	$(RSPARQL) --service $(FINTOSPARQL) --query $^ --results=CSV >$@
 

diff --git a/refdata/fennica-dates.csv.gz b/refdata/fennica-dates.csv.gz
diff --git a/test/00_refdata.bats b/test/00_refdata.bats
@@ -51,3 +51,7 @@ setup () {
   [ -s refdata/RDAMediaType.nt ]
 }
 
+@test "Reference data: 260c date value substitutions" {
+  make refdata/subst-260c.csv
+  [ -s refdata/subst-260c.csv ]
+}
diff --git a/test/input/suoja-pirtti.alephseq b/test/input/suoja-pirtti.alephseq
@@ -0,0 +1,25 @@
+000005646 FMT   L BK
+000005646 LDR   L 00000cam^a2200577zi^4500
+000005646 001   L 000005646
+000005646 005   L 20180104164905.0
+000005646 008   L 820419s1981^^^^fi^|||||||||||||||||fin||
+000005646 015   L $$af820558$$2skl
+000005646 035   L $$a(FI-MELINDA)000005646
+000005646 040   L $$aFI-NL
+000005646 0410  L $$afin
+000005646 042   L $$afinb
+000005646 080   L $$a361/362
+000005646 080   L $$a364$$21974/fin/fennica$$9FENNI<KEEP>
+000005646 080   L $$a613$$21974/fin/fennica$$9FENNI<KEEP>
+000005646 080   L $$a613.81$$21974/fin/fennica$$9FENNI<KEEP>
+000005646 080   L $$a364.65$$x-05$$x(091)$$21974/fin/fennica$$9FENNI<KEEP>
+000005646 24500 L $$a20 vuotta päihdehuoltoa 1961-1981 /$$c[julk.] Suoja-pirti r.y.
+000005646 260   L $$a[Hki] :$$bSuoja-pirtti,$$c[1981] f [Merkur]
+000005646 300   L $$a[2], 33, [1] s. :$$bkuvitettu ;$$c25 cm
+000005646 336   L $$ateksti$$btxt$$2rdacontent
+000005646 337   L $$akäytettävissä ilman laitetta$$bn$$2rdamedia
+000005646 338   L $$anide$$bnc$$2rdacarrier
+000005646 7102  L $$aSuoja-pirtti.
+000005646 7400  L $$aKaksikymmentä vuotta päihdehuoltoa 1961-1981.
+000005646 903   L $$ab$$5FENNI
+000005646 904   L $$ap$$5FENNI
diff --git a/test/refdata/fennica-dates.csv.gz b/test/refdata/fennica-dates.csv.gz
@@ -0,0 +1 @@
+../../refdata/fennica-dates.csv.gz