1603_45_16.lib.sh (#45): bootstrap_1603_16_1__lsf works

EticaAI · Jul 23, 2022 · e901a44 · e901a44
1 parent b764406
commit e901a44
Show file tree

Hide file tree

Showing 2 changed files with 54 additions and 24 deletions.
diff --git a/officina/999999999/0/L999999999_0.py b/officina/999999999/0/L999999999_0.py
@@ -840,7 +840,7 @@ def _expand_hxl_ad_rdf():
     # 'p': 'http://www.wikidata.org/prop/',
     'wdt': 'http://www.wikidata.org/prop/direct/',
     'wdv': 'http://www.wikidata.org/value/',
-    'p': 'http://www.wikidata.org/prop/',  # NOTE: preffer wdata for datasets
+    'p': 'http://www.wikidata.org/prop/',  # NOTE: preffer use wdata for datasets
 }
 # For "Base OWL" of Wikidata, download link: http://wikiba.se/ontology
 
@@ -1060,7 +1060,7 @@ def _expand_hxl_ad_rdf():
     'obo:locationofatsometime': 'obo:BFO_0000124',
     'obo:bfo124': 'obo:BFO_0000124',
 }
-# Note: prefixes that already are lower case do not be here
+# Note: prefixes that already are lower case should not be here
 RDF_SPATIA_NOMINALIBUS_PREFIX_EXTRAS = {
 }
 
@@ -1071,6 +1071,7 @@ def _expand_hxl_ad_rdf():
 
 def _rdf_spatia_nominalibus_prefix_normali(rem: str) -> dict:
     # exemplum: obo:bfo29 -> obo:BFO_0000029
+    # exemplum: wdata:p2082 -> wdata:P2082
     rem_l = rem.lower()
     if rem_l in RDF_SPATIA_NOMINALIBUS_PREFIX:
         return RDF_SPATIA_NOMINALIBUS_PREFIX[rem_l]
@@ -1084,6 +1085,10 @@ def _rdf_spatia_nominalibus_prefix_normali(rem: str) -> dict:
         rem_item_alpha = rem_ls.replace(rem_digits, '').upper()
         rem_digits_full = rem_digits.zfill(7)
         return 'obo:{0}_{1}'.format(rem_item_alpha, rem_digits_full)
+    if rem_l.startswith('wdata:'):
+        rem_ls = rem_l.replace('wdata:', '')
+        rem_item_alpha = rem_ls.upper()
+        return 'wdata:{0}'.format(rem_item_alpha)
 
     # Worst case: assume input already is normalized
     return rem
@@ -2589,6 +2594,10 @@ def bcp47_rdf_extension_poc(
         '_error': [],
     }
 
+    # Enforce normalization for Wikidata predicates, we apply
+    # _rdf_spatia_nominalibus_prefix_normali()
+    _to_upper = ['wdata']
+
     if not rdf_sine_spatia_nominalibus or len(rdf_sine_spatia_nominalibus) == 0:
         rdf_sine_spatia_nominalibus = None
 
@@ -2751,6 +2760,10 @@ def _helper_aux_triple(
 
             _temp1, _temp2 = predicate_and_subject.split('||')
             predicate = _temp1
+            for _item in _to_upper:
+                if predicate.startswith(_item + ':'):
+                    predicate = _rdf_spatia_nominalibus_prefix_normali(
+                        predicate)
 
             if value_separator is not None and \
                 object_literal.find(value_separator) > -1 and \

diff --git a/officina/999999999/1603_45_16.lib.sh b/officina/999999999/1603_45_16.lib.sh
@@ -96,6 +96,7 @@ bootstrap_1603_16_1__lsf() {
   objetivum_archivum_no11="${ROOTDIR}/1603/16/1/0/$_nomen.no11.tm.hxl.csv"
   objetivum_archivum_no11_bcp47min="${ROOTDIR}/1603/16/1/0/$_nomen.no11.bcp47.csv"
   objetivum_archivum_no11_skos="${ROOTDIR}/1603/16/1/0/$_nomen.no11.skos.ttl"
+  objetivum_archivum_datapackage="${ROOTDIR}/1603/16/1/0/datapackage.json"
 
   opus_temporibus_temporarium="${DESTDIR}/999999/0/${_nomen}~TEMP~1.csv"
   opus_temporibus_temporarium_2="${DESTDIR}/999999/0/${_nomen}~TEMP~2.csv"
@@ -187,37 +188,39 @@ bootstrap_1603_16_1__lsf() {
   file_update_if_necessary "csv" "${fontem_archivum_temporarium_no11}" "${objetivum_archivum_no11}"
   file_update_if_necessary "csv" "${fontem_archivum_temporarium_no1}" "${objetivum_archivum_no1}"
 
+  ##  Computational-like RDF serialization, "OWL version" --------------------
+  # @TODO fix generation of invalid format if
+  #       --rdf-sine-spatia-nominalibus=skos,devnull is enabled
+  rdf_ontologia_ordinibus='4'
+  rdf_trivio='5000'
   set -x
 
-  ##  Computational-like RDF serialization, "OWL version" --------------------
-    # @TODO fix generation of invalid format if
-    #       --rdf-sine-spatia-nominalibus=skos,devnull is enabled
+  "${ROOTDIR}/999999999/0/999999999_54872.py" \
+    --methodus=_temp_no1 \
+    --numerordinatio-cum-antecessoribus \
+    --rdf-sine-spatia-nominalibus=devnull \
+    --rdf-ontologia-ordinibus="${rdf_ontologia_ordinibus}" \
+    --rdf-trivio="${rdf_trivio}" \
+    <"${objetivum_archivum_no1}" >"${opus_temporibus_temporarium_ttl_1}"
 
-  # "${ROOTDIR}/999999999/0/999999999_54872.py" \
-  #   --methodus=_temp_no1 \
-  #   --numerordinatio-cum-antecessoribus \
-  #   --rdf-sine-spatia-nominalibus=devnull \
-  #   --rdf-ontologia-ordinibus="${rdf_ontologia_ordinibus}" \
-  #   --rdf-trivio="${rdf_trivio}" \
-  #   <"${objectivum_archivum_no1}" >"${opus_temporibus_temporarium}"
+  rdfpipe --input-format=turtle --output-format=longturtle \
+    "${opus_temporibus_temporarium_ttl_1}" \
+    >"${opus_temporibus_temporarium_ttl_2}"
 
-  # # rapper --quiet --input=turtle --output=turtle \
-  # #   "${opus_temporibus_temporarium}" \
-  # #   >"${objectivum_archivum_no1_owl_ttl}"
+  riot --validate "${opus_temporibus_temporarium_ttl_2}"
 
-  # rdfpipe --input-format=turtle --output-format=longturtle \
-  #   "${opus_temporibus_temporarium}" \
-  #   >"${objectivum_archivum_no1_owl_ttl}"
+  # sleep 10
+  set +x
 
-  # riot --validate "${objectivum_archivum_no1_owl_ttl}"
+  file_update_if_necessary "ttl" "${opus_temporibus_temporarium_ttl_2}" "${objetivum_archivum_no1_owl}"
 
-  ##  Linguistic-like RDF serialization, "SKOS version" ----------------------
+  ##  Linguistic-like RDF serialization, "SKOS version" ------------------------
   # @TODO fix invalid generation if disabling OWL with
   #        --rdf-sine-spatia-nominalibus=owl
-
   rdf_ontologia_ordinibus='4'
   rdf_trivio='5000'
 
+  set -x
   "${ROOTDIR}/999999999/0/999999999_54872.py" \
     --methodus=_temp_no1 \
     --numerordinatio-cum-antecessoribus \
@@ -226,16 +229,30 @@ bootstrap_1603_16_1__lsf() {
     --rdf-trivio="${rdf_trivio}" \
     <"${objetivum_archivum_no11}" >"${opus_temporibus_temporarium_ttl_1}"
 
-    rdfpipe --input-format=turtle --output-format=longturtle \
-      "${opus_temporibus_temporarium_ttl_1}" \
-      >"${opus_temporibus_temporarium_ttl_2}"
+  rdfpipe --input-format=turtle --output-format=longturtle \
+    "${opus_temporibus_temporarium_ttl_1}" \
+    >"${opus_temporibus_temporarium_ttl_2}"
 
   riot --validate "${opus_temporibus_temporarium_ttl_2}"
 
   set +x
 
   # file_update_if_necessary "ttl" "${opus_temporibus_temporarium_ttl_2}" "${objetivum_archivum_no1_owl}"
   file_update_if_necessary "ttl" "${opus_temporibus_temporarium_ttl_2}" "${objetivum_archivum_no11_skos}"
+
+  ## Now create the packages ---------------------------------------------------
+
+  set -x
+  "${ROOTDIR}/999999999/0/1603_1.py" \
+    --methodus='data-apothecae-unicae' \
+    --data-apothecae-ex='1603_16_1_0' \
+    --data-apothecae-ad-stdout \
+    --data-apothecae-formato='datapackage' \
+    >"$objetivum_archivum_datapackage"
+
+  frictionless validate "$objetivum_archivum_datapackage"
+
+  set +x
 }
 
 #######################################