999999999_7200235.py (#45): --methodus=cod_ab_et_wdata uses no11.tm.h…

…xl.csv (near 250+ languages for each adm0) if cached file exist on disk
EticaAI · Jul 21, 2022 · 2005f32 · 2005f32
1 parent 7ef079a
commit 2005f32
Show file tree

Hide file tree

Showing 4 changed files with 81 additions and 5 deletions.
diff --git a/officina/999999999/0/999999999_7200235.py b/officina/999999999/0/999999999_7200235.py
@@ -710,7 +710,14 @@ def execute_cli(self, pyargs, stdin=STDIN, _stdout=sys.stdout,
                     numerordinatio_praefixo=numerordinatio_praefixo,
                     no1_simplici=True, cod_ab_level__inline=True)
 
-                caput_wdata, data_wdata = hxltm_carricato(WDATA_ADM0)
+                if exists(WDATA_ADM0_NO11):
+                    caput_wdata, data_wdata = hxltm_carricato(WDATA_ADM0_NO11)
+                elif exists(WDATA_ADM0):
+                    caput_wdata, data_wdata = hxltm_carricato(WDATA_ADM0)
+                else:
+                    raise FileNotFoundError('[{0}, {1}] ?'.format(
+                        WDATA_ADM0_NO11, WDATA_ADM0
+                    ))
                 # print(data_wdata[0])
 
                 caput, data = hxltm_carricato__cod_ab_et_wdata(

diff --git a/officina/999999999/1603_45_16.lib.sh b/officina/999999999/1603_45_16.lib.sh
@@ -13,6 +13,7 @@
 #  REQUIREMENTS:  - Bash shell (or better)
 #                 - wget
 #                 - csvkit (https://github.com/wireservice/csvkit)
+#                 - rapper (sudo apt install raptor2-utils)
 #          BUGS:  ---
 #         NOTES:  ---
 #        AUTHOR:  Emerson Rocha <rocha[at]ieee.org>

diff --git a/officina/999999999/999999999.lib.sh b/officina/999999999/999999999.lib.sh
@@ -17,6 +17,7 @@
 #                 - ssconvert (sudo apt install gnumeric)
 #                 - wget
 #                 - unzip
+#                 - rapper (sudo apt install raptor2-utils)
 #
 #          BUGS:  ---
 #         NOTES:  ---
@@ -1017,6 +1018,64 @@ file_convert_rdf_skos_ttl_de_numerordinatio11() {
   file_update_if_necessary 'rdf_skos_ttl' "$objectivum_archivum_temporarium" "$objectivum_archivum"
 }
 
+#######################################
+# Convert a "full" .no11.tm.hxl.csv to .no11.skos.ttl, method optimized for
+# large datasets.
+#
+# @see https://www.w3.org/2015/03/ShExValidata/
+#
+# Globals:
+#   ROOTDIR
+#   DESTDIR
+# Arguments:
+#   numerordinatio
+# Outputs:
+#   Convert files
+#######################################
+file_convert_rdf_skos_ttl_de_numerordinatio11__v2() {
+  numerordinatio="$1"
+
+  _path=$(numerordinatio_neo_separatum "$numerordinatio" "/")
+  _nomen=$(numerordinatio_neo_separatum "$numerordinatio" "_")
+  _prefix=$(numerordinatio_neo_separatum "$numerordinatio" ":")
+
+  _basim_fontem="${ROOTDIR}"
+  _basim_objectivum="${DESTDIR}"
+
+  fontem_archivum="${_basim_fontem}/$_path/$_nomen.no11.tm.hxl.csv"
+  objectivum_archivum="${_basim_objectivum}/$_path/$_nomen.no11.skos.ttl"
+  objectivum_archivum_temporarium="${ROOTDIR}/999999/0/$_nomen.no11.skos.ttl"
+  # tmeta="${ROOTDIR}/999999999/0/hxltm-exemplum.tmeta.yml"
+
+  echo "@TODO finish this version. See also function bootstrap_1603_45_16__item_rdf()"
+  return 0
+
+  # echo "${FUNCNAME[0]}: [$fontem_archivum] --> [$objectivum_archivum]"
+
+  # "${ROOTDIR}/999999999/0/1603_1.py" \
+  #   --methodus='status-quo' \
+  #   --status-quo-in-rdf-skos-turtle \
+  #   --codex-de "$_nomen" \
+  #   >"$objectivum_archivum_temporarium"
+
+  # file_update_if_necessary 'rdf_skos_ttl' "$objectivum_archivum_temporarium" "$objectivum_archivum"
+
+  # "${ROOTDIR}/999999999/0/999999999_54872.py" \
+  #   --objectivum-formato=_temp_no1 \
+  #   --numerordinatio-cum-antecessoribus \
+  #   --rdf-sine-spatia-nominalibus=devnull \
+  #   --rdf-ontologia-ordinibus="${rdf_ontologia_ordinibus}" \
+  #   --rdf-trivio="${rdf_trivio}" \
+  #   <"${objectivum_archivum_no1}" >"${opus_temporibus_temporarium}"
+
+  # rapper --quiet --input=turtle --output=turtle \
+  #   "${opus_temporibus_temporarium}" \
+  #   >"${objectivum_archivum_no1_owl_ttl}"
+
+  # riot --validate "${objectivum_archivum_no1_owl_ttl}"
+
+}
+
 #######################################
 # Hotfix to remove duplicated merge keys in files (in place change)
 # See file_merge_numerordinatio_de_wiki_q() for reasoning
@@ -1702,8 +1761,8 @@ file_translate_csv_de_numerordinatio_q__v2() {
   # wikiq=$(file_extract_ix_wikiq "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.tm.hxl.csv")
   wikiq=$(file_extract_ix_wikiq "$fontem_archivum")
 
-  # wikidata_q_ex_totalibus "$wikiq" "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.wikiq.tm.hxl.csv" 
-  wikidata_q_ex_totalibus "$wikiq" "$objectivum_archivum" 
+  # wikidata_q_ex_totalibus "$wikiq" "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.wikiq.tm.hxl.csv"
+  wikidata_q_ex_totalibus "$wikiq" "$objectivum_archivum"
   printf "\t%40s\n" "${tty_green}${FUNCNAME[0]} FINISHED OKAY ${tty_normal}"
 }
 

diff --git a/officina/999999999/999999_17.sh b/officina/999999999/999999_17.sh
@@ -63,9 +63,18 @@ ROOTDIR="$(pwd)"
 ### Quick tests
 
 # file_extract_ix_wikiq "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.tm.hxl.csv" "999999/0/1603_3_45_16_1_1.uniq.q.txt"
-wikiq=$(file_extract_ix_wikiq "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.tm.hxl.csv")
+# wikiq=$(file_extract_ix_wikiq "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.tm.hxl.csv")
 
-wikidata_q_ex_totalibus "$wikiq" "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.wikiq.tm.hxl.csv" 
+# wikidata_q_ex_totalibus "$wikiq" "999999/1603/3/45/16/1/1/1603_3_45_16_1_1.wikiq.tm.hxl.csv" 
+set -x
+./999999999/0/999999999_7200235.py --methodus='cod_ab_et_wdata' --numerordinatio-praefixo='1603_16' > 999999/0/teste.csv
+ls -lha 999999/0/teste.csv
+
+./999999999/0/999999999_54872.py --objectivum-formato=_temp_no1 --numerordinatio-cum-antecessoribus --rdf-trivio=1603 999999/0/teste.csv > 999999/0/teste.ttl-simplici.ttl
+
+rapper --quiet --input=turtle --output=turtle 999999/0/teste.ttl-simplici.ttl > 999999/0/teste.ttl
+
+set +x
 exit 0
 ### Really boostrapping downloads, end _________________________________________