From 10453f117247eddc92ae5326b857f8e83ee2da6d Mon Sep 17 00:00:00 2001 From: David Ball Date: Sun, 26 May 2024 04:47:53 -0400 Subject: [PATCH] Updated solr and tika configurations. --- .env.nm3clol => .env.example | 0 .gitignore | 5 +- fullstack/docker-compose.yml | 412 ++++++++++++++++- solr-log4j2.xml | 86 ++++ solr-tika/docker-compose.yml | 414 +++++++++++++++++- solr/docker-compose.yml | 389 +++++++++++++++- ...ties => tika-TesseractOCRConfig.properties | 0 tika-log4j2.xml | 13 + tika/docker-compose.yml | 25 +- 9 files changed, 1303 insertions(+), 41 deletions(-) rename .env.nm3clol => .env.example (100%) create mode 100644 solr-log4j2.xml rename TesseractOCRConfig.properties => tika-TesseractOCRConfig.properties (100%) create mode 100644 tika-log4j2.xml diff --git a/.env.nm3clol b/.env.example similarity index 100% rename from .env.nm3clol rename to .env.example diff --git a/.gitignore b/.gitignore index fa3b7f0..80783ee 100644 --- a/.gitignore +++ b/.gitignore @@ -1,6 +1,5 @@ # Production .env .env - # Docker container data -docker/solr-data -docker/tika-data +solr-data +tika-data diff --git a/fullstack/docker-compose.yml b/fullstack/docker-compose.yml index ad7a669..aaeafd6 100644 --- a/fullstack/docker-compose.yml +++ b/fullstack/docker-compose.yml @@ -1,36 +1,232 @@ name: nm3clol services: + solr-init: + image: alpine + container_name: solr-init + restart: "no" + volumes: + - solr-data:/var/solr + entrypoint: | + /bin/sh -c "chown -R 8983:8983 /var/solr" solr: + depends_on: + - solr-init image: solr:latest container_name: solr restart: unless-stopped ports: - "${SOLR_DOCS_PORT-8983}:8983" volumes: - - ../solr-data:/var/solr + - solr-data:/var/solr # environment: # - SOLR_CORE=my_core command: + # - /bin/sh -c "chown -R 8983:8983 /var/solr" - solr-precreate -# - gettingstarted - user: "1000:995" - + # - gettingstarted + user: "8983:8983" + configs: + - source: solr_log4j2 + target: /var/solr/log4j2.xml + - source: solr_nm3clol_core_properties + target: /var/solr/data/nm3clol_core/core.properties + - source: solr_nm3clol_core_solrconfig + target: /var/solr/data/nm3clol_core/conf/solrconfig.xml + - source: solr_nm3clol_core_managed_schema + target: /var/solr/data/nm3clol_core/conf/managed-schema.xml + - source: solr_nm3clol_core_lang_contractions_ca + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ca.txt + - source: solr_nm3clol_core_lang_contractions_fr + target: /var/solr/data/nm3clol_core/conf/lang/contractions_fr.txt + - source: solr_nm3clol_core_lang_contractions_ga + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ga.txt + - source: solr_nm3clol_core_lang_contractions_it + target: /var/solr/data/nm3clol_core/conf/lang/contractions_it.txt + - source: solr_nm3clol_core_lang_hyphenations_ga + target: /var/solr/data/nm3clol_core/conf/lang/hyphenations_ga.txt + - source: solr_nm3clol_core_lang_stemdict_nl + target: /var/solr/data/nm3clol_core/conf/lang/stemdict_nl.txt + - source: solr_nm3clol_core_lang_stoptags_ja + target: /var/solr/data/nm3clol_core/conf/lang/stoptags_ja.txt + - source: solr_nm3clol_core_lang_stopwords_ar + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ar.txt + - source: solr_nm3clol_core_lang_stopwords_bg + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_bg.txt + - source: solr_nm3clol_core_lang_stopwords_ca + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ca.txt + - source: solr_nm3clol_core_lang_stopwords_cz + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_cz.txt + - source: solr_nm3clol_core_lang_stopwords_da + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_da.txt + - source: solr_nm3clol_core_lang_stopwords_de + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_de.txt + - source: solr_nm3clol_core_lang_stopwords_el + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_el.txt + - source: solr_nm3clol_core_lang_stopwords_en + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_en.txt + - source: solr_nm3clol_core_lang_stopwords_es + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_es.txt + - source: solr_nm3clol_core_lang_stopwords_et + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_et.txt + - source: solr_nm3clol_core_lang_stopwords_eu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_eu.txt + - source: solr_nm3clol_core_lang_stopwords_fa + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fa.txt + - source: solr_nm3clol_core_lang_stopwords_fi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fi.txt + - source: solr_nm3clol_core_lang_stopwords_fr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fr.txt + - source: solr_nm3clol_core_lang_stopwords_ga + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ga.txt + - source: solr_nm3clol_core_lang_stopwords_gl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_gl.txt + - source: solr_nm3clol_core_lang_stopwords_hi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hi.txt + - source: solr_nm3clol_core_lang_stopwords_hu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hu.txt + - source: solr_nm3clol_core_lang_stopwords_hy + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hy.txt + - source: solr_nm3clol_core_lang_stopwords_id + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_id.txt + - source: solr_nm3clol_core_lang_stopwords_it + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_it.txt + - source: solr_nm3clol_core_lang_stopwords_ja + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ja.txt + - source: solr_nm3clol_core_lang_stopwords_lv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_lv.txt + - source: solr_nm3clol_core_lang_stopwords_nl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_nl.txt + - source: solr_nm3clol_core_lang_stopwords_no + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_no.txt + - source: solr_nm3clol_core_lang_stopwords_pt + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_pt.txt + - source: solr_nm3clol_core_lang_stopwords_ro + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ro.txt + - source: solr_nm3clol_core_lang_stopwords_ru + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ru.txt + - source: solr_nm3clol_core_lang_stopwords_sv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_sv.txt + - source: solr_nm3clol_core_lang_stopwords_th + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_th.txt + - source: solr_nm3clol_core_lang_stopwords_tr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_tr.txt + - source: solr_nm3clol_core_lang_userdict_ja + target: /var/solr/data/nm3clol_core/conf/lang/userdict_ja.txt + - source: solr_nm3clol_core_protwords + target: /var/solr/data/nm3clol_core/conf/protwords.txt + - source: solr_nm3clol_core_stopwords + target: /var/solr/data/nm3clol_core/conf/stopwords.txt + - source: solr_nm3clol_core_synonyms + target: /var/solr/data/nm3clol_core/conf/synonyms.txt + - source: solr_vacode_core_properties + target: /var/solr/data/vacode_core/core.properties + - source: solr_vacode_core_solrconfig + target: /var/solr/data/vacode_core/conf/solrconfig.xml + - source: solr_vacode_core_managed_schema + target: /var/solr/data/vacode_core/conf/managed-schema.xml + - source: solr_vacode_core_lang_contractions_ca + target: /var/solr/data/vacode_core/conf/lang/contractions_ca.txt + - source: solr_vacode_core_lang_contractions_fr + target: /var/solr/data/vacode_core/conf/lang/contractions_fr.txt + - source: solr_vacode_core_lang_contractions_ga + target: /var/solr/data/vacode_core/conf/lang/contractions_ga.txt + - source: solr_vacode_core_lang_contractions_it + target: /var/solr/data/vacode_core/conf/lang/contractions_it.txt + - source: solr_vacode_core_lang_hyphenations_ga + target: /var/solr/data/vacode_core/conf/lang/hyphenations_ga.txt + - source: solr_vacode_core_lang_stemdict_nl + target: /var/solr/data/vacode_core/conf/lang/stemdict_nl.txt + - source: solr_vacode_core_lang_stoptags_ja + target: /var/solr/data/vacode_core/conf/lang/stoptags_ja.txt + - source: solr_vacode_core_lang_stopwords_ar + target: /var/solr/data/vacode_core/conf/lang/stopwords_ar.txt + - source: solr_vacode_core_lang_stopwords_bg + target: /var/solr/data/vacode_core/conf/lang/stopwords_bg.txt + - source: solr_vacode_core_lang_stopwords_ca + target: /var/solr/data/vacode_core/conf/lang/stopwords_ca.txt + - source: solr_vacode_core_lang_stopwords_cz + target: /var/solr/data/vacode_core/conf/lang/stopwords_cz.txt + - source: solr_vacode_core_lang_stopwords_da + target: /var/solr/data/vacode_core/conf/lang/stopwords_da.txt + - source: solr_vacode_core_lang_stopwords_de + target: /var/solr/data/vacode_core/conf/lang/stopwords_de.txt + - source: solr_vacode_core_lang_stopwords_el + target: /var/solr/data/vacode_core/conf/lang/stopwords_el.txt + - source: solr_vacode_core_lang_stopwords_en + target: /var/solr/data/vacode_core/conf/lang/stopwords_en.txt + - source: solr_vacode_core_lang_stopwords_es + target: /var/solr/data/vacode_core/conf/lang/stopwords_es.txt + - source: solr_vacode_core_lang_stopwords_et + target: /var/solr/data/vacode_core/conf/lang/stopwords_et.txt + - source: solr_vacode_core_lang_stopwords_eu + target: /var/solr/data/vacode_core/conf/lang/stopwords_eu.txt + - source: solr_vacode_core_lang_stopwords_fa + target: /var/solr/data/vacode_core/conf/lang/stopwords_fa.txt + - source: solr_vacode_core_lang_stopwords_fi + target: /var/solr/data/vacode_core/conf/lang/stopwords_fi.txt + - source: solr_vacode_core_lang_stopwords_fr + target: /var/solr/data/vacode_core/conf/lang/stopwords_fr.txt + - source: solr_vacode_core_lang_stopwords_ga + target: /var/solr/data/vacode_core/conf/lang/stopwords_ga.txt + - source: solr_vacode_core_lang_stopwords_gl + target: /var/solr/data/vacode_core/conf/lang/stopwords_gl.txt + - source: solr_vacode_core_lang_stopwords_hi + target: /var/solr/data/vacode_core/conf/lang/stopwords_hi.txt + - source: solr_vacode_core_lang_stopwords_hu + target: /var/solr/data/vacode_core/conf/lang/stopwords_hu.txt + - source: solr_vacode_core_lang_stopwords_hy + target: /var/solr/data/vacode_core/conf/lang/stopwords_hy.txt + - source: solr_vacode_core_lang_stopwords_id + target: /var/solr/data/vacode_core/conf/lang/stopwords_id.txt + - source: solr_vacode_core_lang_stopwords_it + target: /var/solr/data/vacode_core/conf/lang/stopwords_it.txt + - source: solr_vacode_core_lang_stopwords_ja + target: /var/solr/data/vacode_core/conf/lang/stopwords_ja.txt + - source: solr_vacode_core_lang_stopwords_lv + target: /var/solr/data/vacode_core/conf/lang/stopwords_lv.txt + - source: solr_vacode_core_lang_stopwords_nl + target: /var/solr/data/vacode_core/conf/lang/stopwords_nl.txt + - source: solr_vacode_core_lang_stopwords_no + target: /var/solr/data/vacode_core/conf/lang/stopwords_no.txt + - source: solr_vacode_core_lang_stopwords_pt + target: /var/solr/data/vacode_core/conf/lang/stopwords_pt.txt + - source: solr_vacode_core_lang_stopwords_ro + target: /var/solr/data/vacode_core/conf/lang/stopwords_ro.txt + - source: solr_vacode_core_lang_stopwords_ru + target: /var/solr/data/vacode_core/conf/lang/stopwords_ru.txt + - source: solr_vacode_core_lang_stopwords_sv + target: /var/solr/data/vacode_core/conf/lang/stopwords_sv.txt + - source: solr_vacode_core_lang_stopwords_th + target: /var/solr/data/vacode_core/conf/lang/stopwords_th.txt + - source: solr_vacode_core_lang_stopwords_tr + target: /var/solr/data/vacode_core/conf/lang/stopwords_tr.txt + - source: solr_vacode_core_lang_userdict_ja + target: /var/solr/data/vacode_core/conf/lang/userdict_ja.txt + - source: solr_vacode_core_protwords + target: /var/solr/data/vacode_core/conf/protwords.txt + - source: solr_vacode_core_stopwords + target: /var/solr/data/vacode_core/conf/stopwords.txt + - source: solr_vacode_core_synonyms + target: /var/solr/data/vacode_core/conf/synonyms.txt tika: image: apache/tika:latest-full container_name: tika restart: unless-stopped + configs: + - source: tika_config + target: /tika-config.xml + - source: tika_log4j2 + target: /log4j2.xml + - source: tika_tesseract_ocr_config_properties + target: /TesseractOCRConfig.properties + command: -c /tika-config.xml environment: - TIKA_LOG_PATH=/tika-logs - command: -c /tika-config.xml entrypoint: [ "/bin/sh", "-c", "exec java -cp \"/customocr:/tika-server-standard-2.9.2.jar:/tika-extras/*\" org.apache.tika.server.core.TikaServerCli -h 0.0.0.0 $$0 $$@"] volumes: - - ../tika-config.xml:/tika-config.xml - - ../tika-data/logs:/tika-logs - - ../tika-data/log4j2.xml:/log4j2.xml - - ../TesseractOCRConfig.properties:/TesseractOCRConfig.properties + - tika-logs:/tika-logs ports: - "${TIKA_PORT-9998}:9998" - nm3clol: build: ../nm3clol.Dockerfile container_name: nm3clol @@ -53,8 +249,200 @@ services: - $PWD/..:/srv - caddy_data:/data - caddy_config:/config - +configs: + solr_log4j2: + file: ../solr-log4j2.xml + solr_nm3clol_core_properties: + content: | + name=nm3clol_core + solr_nm3clol_core_solrconfig: + file: ../solr-data/nm3clol_core/conf/solrconfig.xml + solr_nm3clol_core_managed_schema: + file: ../solr-data/nm3clol_core/conf/managed-schema.xml + solr_nm3clol_core_lang_contractions_ca: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ca.txt + solr_nm3clol_core_lang_contractions_fr: + file: ../solr-data/nm3clol_core/conf/lang/contractions_fr.txt + solr_nm3clol_core_lang_contractions_ga: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ga.txt + solr_nm3clol_core_lang_contractions_it: + file: ../solr-data/nm3clol_core/conf/lang/contractions_it.txt + solr_nm3clol_core_lang_hyphenations_ga: + file: ../solr-data/nm3clol_core/conf/lang/hyphenations_ga.txt + solr_nm3clol_core_lang_stemdict_nl: + file: ../solr-data/nm3clol_core/conf/lang/stemdict_nl.txt + solr_nm3clol_core_lang_stoptags_ja: + file: ../solr-data/nm3clol_core/conf/lang/stoptags_ja.txt + solr_nm3clol_core_lang_stopwords_ar: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ar.txt + solr_nm3clol_core_lang_stopwords_bg: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_bg.txt + solr_nm3clol_core_lang_stopwords_ca: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ca.txt + solr_nm3clol_core_lang_stopwords_cz: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_cz.txt + solr_nm3clol_core_lang_stopwords_da: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_da.txt + solr_nm3clol_core_lang_stopwords_de: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_de.txt + solr_nm3clol_core_lang_stopwords_el: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_el.txt + solr_nm3clol_core_lang_stopwords_en: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_en.txt + solr_nm3clol_core_lang_stopwords_es: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_es.txt + solr_nm3clol_core_lang_stopwords_et: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_et.txt + solr_nm3clol_core_lang_stopwords_eu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_eu.txt + solr_nm3clol_core_lang_stopwords_fa: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fa.txt + solr_nm3clol_core_lang_stopwords_fi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fi.txt + solr_nm3clol_core_lang_stopwords_fr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fr.txt + solr_nm3clol_core_lang_stopwords_ga: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ga.txt + solr_nm3clol_core_lang_stopwords_gl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_gl.txt + solr_nm3clol_core_lang_stopwords_hi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hi.txt + solr_nm3clol_core_lang_stopwords_hu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hu.txt + solr_nm3clol_core_lang_stopwords_hy: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hy.txt + solr_nm3clol_core_lang_stopwords_id: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_id.txt + solr_nm3clol_core_lang_stopwords_it: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_it.txt + solr_nm3clol_core_lang_stopwords_ja: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ja.txt + solr_nm3clol_core_lang_stopwords_lv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_lv.txt + solr_nm3clol_core_lang_stopwords_nl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_nl.txt + solr_nm3clol_core_lang_stopwords_no: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_no.txt + solr_nm3clol_core_lang_stopwords_pt: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_pt.txt + solr_nm3clol_core_lang_stopwords_ro: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ro.txt + solr_nm3clol_core_lang_stopwords_ru: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ru.txt + solr_nm3clol_core_lang_stopwords_sv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_sv.txt + solr_nm3clol_core_lang_stopwords_th: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_th.txt + solr_nm3clol_core_lang_stopwords_tr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_tr.txt + solr_nm3clol_core_lang_userdict_ja: + file: ../solr-data/nm3clol_core/conf/lang/userdict_ja.txt + solr_nm3clol_core_protwords: + file: ../solr-data/nm3clol_core/conf/protwords.txt + solr_nm3clol_core_stopwords: + file: ../solr-data/nm3clol_core/conf/stopwords.txt + solr_nm3clol_core_synonyms: + file: ../solr-data/nm3clol_core/conf/synonyms.txt + solr_vacode_core_properties: + content: | + name=vacode_core + solr_vacode_core_solrconfig: + file: ../solr-data/vacode_core/conf/solrconfig.xml + solr_vacode_core_managed_schema: + file: ../solr-data/vacode_core/conf/managed-schema.xml + solr_vacode_core_lang_contractions_ca: + file: ../solr-data/vacode_core/conf/lang/contractions_ca.txt + solr_vacode_core_lang_contractions_fr: + file: ../solr-data/vacode_core/conf/lang/contractions_fr.txt + solr_vacode_core_lang_contractions_ga: + file: ../solr-data/vacode_core/conf/lang/contractions_ga.txt + solr_vacode_core_lang_contractions_it: + file: ../solr-data/vacode_core/conf/lang/contractions_it.txt + solr_vacode_core_lang_hyphenations_ga: + file: ../solr-data/vacode_core/conf/lang/hyphenations_ga.txt + solr_vacode_core_lang_stemdict_nl: + file: ../solr-data/vacode_core/conf/lang/stemdict_nl.txt + solr_vacode_core_lang_stoptags_ja: + file: ../solr-data/vacode_core/conf/lang/stoptags_ja.txt + solr_vacode_core_lang_stopwords_ar: + file: ../solr-data/vacode_core/conf/lang/stopwords_ar.txt + solr_vacode_core_lang_stopwords_bg: + file: ../solr-data/vacode_core/conf/lang/stopwords_bg.txt + solr_vacode_core_lang_stopwords_ca: + file: ../solr-data/vacode_core/conf/lang/stopwords_ca.txt + solr_vacode_core_lang_stopwords_cz: + file: ../solr-data/vacode_core/conf/lang/stopwords_cz.txt + solr_vacode_core_lang_stopwords_da: + file: ../solr-data/vacode_core/conf/lang/stopwords_da.txt + solr_vacode_core_lang_stopwords_de: + file: ../solr-data/vacode_core/conf/lang/stopwords_de.txt + solr_vacode_core_lang_stopwords_el: + file: ../solr-data/vacode_core/conf/lang/stopwords_el.txt + solr_vacode_core_lang_stopwords_en: + file: ../solr-data/vacode_core/conf/lang/stopwords_en.txt + solr_vacode_core_lang_stopwords_es: + file: ../solr-data/vacode_core/conf/lang/stopwords_es.txt + solr_vacode_core_lang_stopwords_et: + file: ../solr-data/vacode_core/conf/lang/stopwords_et.txt + solr_vacode_core_lang_stopwords_eu: + file: ../solr-data/vacode_core/conf/lang/stopwords_eu.txt + solr_vacode_core_lang_stopwords_fa: + file: ../solr-data/vacode_core/conf/lang/stopwords_fa.txt + solr_vacode_core_lang_stopwords_fi: + file: ../solr-data/vacode_core/conf/lang/stopwords_fi.txt + solr_vacode_core_lang_stopwords_fr: + file: ../solr-data/vacode_core/conf/lang/stopwords_fr.txt + solr_vacode_core_lang_stopwords_ga: + file: ../solr-data/vacode_core/conf/lang/stopwords_ga.txt + solr_vacode_core_lang_stopwords_gl: + file: ../solr-data/vacode_core/conf/lang/stopwords_gl.txt + solr_vacode_core_lang_stopwords_hi: + file: ../solr-data/vacode_core/conf/lang/stopwords_hi.txt + solr_vacode_core_lang_stopwords_hu: + file: ../solr-data/vacode_core/conf/lang/stopwords_hu.txt + solr_vacode_core_lang_stopwords_hy: + file: ../solr-data/vacode_core/conf/lang/stopwords_hy.txt + solr_vacode_core_lang_stopwords_id: + file: ../solr-data/vacode_core/conf/lang/stopwords_id.txt + solr_vacode_core_lang_stopwords_it: + file: ../solr-data/vacode_core/conf/lang/stopwords_it.txt + solr_vacode_core_lang_stopwords_ja: + file: ../solr-data/vacode_core/conf/lang/stopwords_ja.txt + solr_vacode_core_lang_stopwords_lv: + file: ../solr-data/vacode_core/conf/lang/stopwords_lv.txt + solr_vacode_core_lang_stopwords_nl: + file: ../solr-data/vacode_core/conf/lang/stopwords_nl.txt + solr_vacode_core_lang_stopwords_no: + file: ../solr-data/vacode_core/conf/lang/stopwords_no.txt + solr_vacode_core_lang_stopwords_pt: + file: ../solr-data/vacode_core/conf/lang/stopwords_pt.txt + solr_vacode_core_lang_stopwords_ro: + file: ../solr-data/vacode_core/conf/lang/stopwords_ro.txt + solr_vacode_core_lang_stopwords_ru: + file: ../solr-data/vacode_core/conf/lang/stopwords_ru.txt + solr_vacode_core_lang_stopwords_sv: + file: ../solr-data/vacode_core/conf/lang/stopwords_sv.txt + solr_vacode_core_lang_stopwords_th: + file: ../solr-data/vacode_core/conf/lang/stopwords_th.txt + solr_vacode_core_lang_stopwords_tr: + file: ../solr-data/vacode_core/conf/lang/stopwords_tr.txt + solr_vacode_core_lang_userdict_ja: + file: ../solr-data/vacode_core/conf/lang/userdict_ja.txt + solr_vacode_core_protwords: + file: ../solr-data/vacode_core/conf/protwords.txt + solr_vacode_core_stopwords: + file: ../solr-data/vacode_core/conf/stopwords.txt + solr_vacode_core_synonyms: + file: ../solr-data/vacode_core/conf/synonyms.txt + tika_config: + file: ../tika-config.xml + tika_log4j2: + file: ../tika-log4j2.xml + tika_tesseract_ocr_config_properties: + file: ../tika-TesseractOCRConfig.properties volumes: caddy_data: external: true - caddy_config: \ No newline at end of file + caddy_config: + solr-data: + tika-logs: diff --git a/solr-log4j2.xml b/solr-log4j2.xml new file mode 100644 index 0000000..c4dd7c8 --- /dev/null +++ b/solr-log4j2.xml @@ -0,0 +1,86 @@ + + + + + + + + + + + %maxLen{%d{yyyy-MM-dd HH:mm:ss.SSS} %-5p (%t) [%notEmpty{c:%X{collection}}%notEmpty{ s:%X{shard}}%notEmpty{ r:%X{replica}}%notEmpty{ x:%X{core}}%notEmpty{ t:%X{trace_id}}] %c{1.} %m%notEmpty{ =>%ex{short}}}{10240}%n + + + + + + + + %maxLen{%d{yyyy-MM-dd HH:mm:ss.SSS} %-5p (%t) [%notEmpty{c:%X{collection}}%notEmpty{ s:%X{shard}}%notEmpty{ r:%X{replica}}%notEmpty{ x:%X{core}}%notEmpty{ t:%X{trace_id}}] %c{1.} %m%notEmpty{ =>%ex{short}}}{10240}%n + + + + + + + + + + + + + %maxLen{%d{yyyy-MM-dd HH:mm:ss.SSS} %-5p (%t) [%notEmpty{c:%X{collection}}%notEmpty{ s:%X{shard}}%notEmpty{ r:%X{replica}}%notEmpty{ x:%X{core}}%notEmpty{ t:%X{trace_id}}] %c{1.} %m%notEmpty{ =>%ex{short}}}{10240}%n + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/solr-tika/docker-compose.yml b/solr-tika/docker-compose.yml index cf13887..6f95dfd 100644 --- a/solr-tika/docker-compose.yml +++ b/solr-tika/docker-compose.yml @@ -1,36 +1,422 @@ -version: '3' - services: + solr-init: + image: alpine + container_name: solr-init + restart: "no" + volumes: + - solr-data:/var/solr + entrypoint: | + /bin/sh -c "chown -R 8983:8983 /var/solr" solr: + depends_on: + - solr-init image: solr:latest container_name: solr restart: unless-stopped ports: - "${SOLR_DOCS_PORT-8983}:8983" volumes: - - ../solr-data:/var/solr + - solr-data:/var/solr # environment: # - SOLR_CORE=my_core command: + # - /bin/sh -c "chown -R 8983:8983 /var/solr" - solr-precreate -# - gettingstarted - user: "1000:995" - + # - gettingstarted + user: "8983:8983" + configs: + - source: solr_log4j2 + target: /var/solr/log4j2.xml + - source: solr_nm3clol_core_properties + target: /var/solr/data/nm3clol_core/core.properties + - source: solr_nm3clol_core_solrconfig + target: /var/solr/data/nm3clol_core/conf/solrconfig.xml + - source: solr_nm3clol_core_managed_schema + target: /var/solr/data/nm3clol_core/conf/managed-schema.xml + - source: solr_nm3clol_core_lang_contractions_ca + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ca.txt + - source: solr_nm3clol_core_lang_contractions_fr + target: /var/solr/data/nm3clol_core/conf/lang/contractions_fr.txt + - source: solr_nm3clol_core_lang_contractions_ga + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ga.txt + - source: solr_nm3clol_core_lang_contractions_it + target: /var/solr/data/nm3clol_core/conf/lang/contractions_it.txt + - source: solr_nm3clol_core_lang_hyphenations_ga + target: /var/solr/data/nm3clol_core/conf/lang/hyphenations_ga.txt + - source: solr_nm3clol_core_lang_stemdict_nl + target: /var/solr/data/nm3clol_core/conf/lang/stemdict_nl.txt + - source: solr_nm3clol_core_lang_stoptags_ja + target: /var/solr/data/nm3clol_core/conf/lang/stoptags_ja.txt + - source: solr_nm3clol_core_lang_stopwords_ar + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ar.txt + - source: solr_nm3clol_core_lang_stopwords_bg + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_bg.txt + - source: solr_nm3clol_core_lang_stopwords_ca + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ca.txt + - source: solr_nm3clol_core_lang_stopwords_cz + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_cz.txt + - source: solr_nm3clol_core_lang_stopwords_da + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_da.txt + - source: solr_nm3clol_core_lang_stopwords_de + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_de.txt + - source: solr_nm3clol_core_lang_stopwords_el + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_el.txt + - source: solr_nm3clol_core_lang_stopwords_en + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_en.txt + - source: solr_nm3clol_core_lang_stopwords_es + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_es.txt + - source: solr_nm3clol_core_lang_stopwords_et + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_et.txt + - source: solr_nm3clol_core_lang_stopwords_eu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_eu.txt + - source: solr_nm3clol_core_lang_stopwords_fa + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fa.txt + - source: solr_nm3clol_core_lang_stopwords_fi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fi.txt + - source: solr_nm3clol_core_lang_stopwords_fr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fr.txt + - source: solr_nm3clol_core_lang_stopwords_ga + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ga.txt + - source: solr_nm3clol_core_lang_stopwords_gl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_gl.txt + - source: solr_nm3clol_core_lang_stopwords_hi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hi.txt + - source: solr_nm3clol_core_lang_stopwords_hu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hu.txt + - source: solr_nm3clol_core_lang_stopwords_hy + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hy.txt + - source: solr_nm3clol_core_lang_stopwords_id + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_id.txt + - source: solr_nm3clol_core_lang_stopwords_it + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_it.txt + - source: solr_nm3clol_core_lang_stopwords_ja + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ja.txt + - source: solr_nm3clol_core_lang_stopwords_lv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_lv.txt + - source: solr_nm3clol_core_lang_stopwords_nl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_nl.txt + - source: solr_nm3clol_core_lang_stopwords_no + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_no.txt + - source: solr_nm3clol_core_lang_stopwords_pt + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_pt.txt + - source: solr_nm3clol_core_lang_stopwords_ro + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ro.txt + - source: solr_nm3clol_core_lang_stopwords_ru + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ru.txt + - source: solr_nm3clol_core_lang_stopwords_sv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_sv.txt + - source: solr_nm3clol_core_lang_stopwords_th + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_th.txt + - source: solr_nm3clol_core_lang_stopwords_tr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_tr.txt + - source: solr_nm3clol_core_lang_userdict_ja + target: /var/solr/data/nm3clol_core/conf/lang/userdict_ja.txt + - source: solr_nm3clol_core_protwords + target: /var/solr/data/nm3clol_core/conf/protwords.txt + - source: solr_nm3clol_core_stopwords + target: /var/solr/data/nm3clol_core/conf/stopwords.txt + - source: solr_nm3clol_core_synonyms + target: /var/solr/data/nm3clol_core/conf/synonyms.txt + - source: solr_vacode_core_properties + target: /var/solr/data/vacode_core/core.properties + - source: solr_vacode_core_solrconfig + target: /var/solr/data/vacode_core/conf/solrconfig.xml + - source: solr_vacode_core_managed_schema + target: /var/solr/data/vacode_core/conf/managed-schema.xml + - source: solr_vacode_core_lang_contractions_ca + target: /var/solr/data/vacode_core/conf/lang/contractions_ca.txt + - source: solr_vacode_core_lang_contractions_fr + target: /var/solr/data/vacode_core/conf/lang/contractions_fr.txt + - source: solr_vacode_core_lang_contractions_ga + target: /var/solr/data/vacode_core/conf/lang/contractions_ga.txt + - source: solr_vacode_core_lang_contractions_it + target: /var/solr/data/vacode_core/conf/lang/contractions_it.txt + - source: solr_vacode_core_lang_hyphenations_ga + target: /var/solr/data/vacode_core/conf/lang/hyphenations_ga.txt + - source: solr_vacode_core_lang_stemdict_nl + target: /var/solr/data/vacode_core/conf/lang/stemdict_nl.txt + - source: solr_vacode_core_lang_stoptags_ja + target: /var/solr/data/vacode_core/conf/lang/stoptags_ja.txt + - source: solr_vacode_core_lang_stopwords_ar + target: /var/solr/data/vacode_core/conf/lang/stopwords_ar.txt + - source: solr_vacode_core_lang_stopwords_bg + target: /var/solr/data/vacode_core/conf/lang/stopwords_bg.txt + - source: solr_vacode_core_lang_stopwords_ca + target: /var/solr/data/vacode_core/conf/lang/stopwords_ca.txt + - source: solr_vacode_core_lang_stopwords_cz + target: /var/solr/data/vacode_core/conf/lang/stopwords_cz.txt + - source: solr_vacode_core_lang_stopwords_da + target: /var/solr/data/vacode_core/conf/lang/stopwords_da.txt + - source: solr_vacode_core_lang_stopwords_de + target: /var/solr/data/vacode_core/conf/lang/stopwords_de.txt + - source: solr_vacode_core_lang_stopwords_el + target: /var/solr/data/vacode_core/conf/lang/stopwords_el.txt + - source: solr_vacode_core_lang_stopwords_en + target: /var/solr/data/vacode_core/conf/lang/stopwords_en.txt + - source: solr_vacode_core_lang_stopwords_es + target: /var/solr/data/vacode_core/conf/lang/stopwords_es.txt + - source: solr_vacode_core_lang_stopwords_et + target: /var/solr/data/vacode_core/conf/lang/stopwords_et.txt + - source: solr_vacode_core_lang_stopwords_eu + target: /var/solr/data/vacode_core/conf/lang/stopwords_eu.txt + - source: solr_vacode_core_lang_stopwords_fa + target: /var/solr/data/vacode_core/conf/lang/stopwords_fa.txt + - source: solr_vacode_core_lang_stopwords_fi + target: /var/solr/data/vacode_core/conf/lang/stopwords_fi.txt + - source: solr_vacode_core_lang_stopwords_fr + target: /var/solr/data/vacode_core/conf/lang/stopwords_fr.txt + - source: solr_vacode_core_lang_stopwords_ga + target: /var/solr/data/vacode_core/conf/lang/stopwords_ga.txt + - source: solr_vacode_core_lang_stopwords_gl + target: /var/solr/data/vacode_core/conf/lang/stopwords_gl.txt + - source: solr_vacode_core_lang_stopwords_hi + target: /var/solr/data/vacode_core/conf/lang/stopwords_hi.txt + - source: solr_vacode_core_lang_stopwords_hu + target: /var/solr/data/vacode_core/conf/lang/stopwords_hu.txt + - source: solr_vacode_core_lang_stopwords_hy + target: /var/solr/data/vacode_core/conf/lang/stopwords_hy.txt + - source: solr_vacode_core_lang_stopwords_id + target: /var/solr/data/vacode_core/conf/lang/stopwords_id.txt + - source: solr_vacode_core_lang_stopwords_it + target: /var/solr/data/vacode_core/conf/lang/stopwords_it.txt + - source: solr_vacode_core_lang_stopwords_ja + target: /var/solr/data/vacode_core/conf/lang/stopwords_ja.txt + - source: solr_vacode_core_lang_stopwords_lv + target: /var/solr/data/vacode_core/conf/lang/stopwords_lv.txt + - source: solr_vacode_core_lang_stopwords_nl + target: /var/solr/data/vacode_core/conf/lang/stopwords_nl.txt + - source: solr_vacode_core_lang_stopwords_no + target: /var/solr/data/vacode_core/conf/lang/stopwords_no.txt + - source: solr_vacode_core_lang_stopwords_pt + target: /var/solr/data/vacode_core/conf/lang/stopwords_pt.txt + - source: solr_vacode_core_lang_stopwords_ro + target: /var/solr/data/vacode_core/conf/lang/stopwords_ro.txt + - source: solr_vacode_core_lang_stopwords_ru + target: /var/solr/data/vacode_core/conf/lang/stopwords_ru.txt + - source: solr_vacode_core_lang_stopwords_sv + target: /var/solr/data/vacode_core/conf/lang/stopwords_sv.txt + - source: solr_vacode_core_lang_stopwords_th + target: /var/solr/data/vacode_core/conf/lang/stopwords_th.txt + - source: solr_vacode_core_lang_stopwords_tr + target: /var/solr/data/vacode_core/conf/lang/stopwords_tr.txt + - source: solr_vacode_core_lang_userdict_ja + target: /var/solr/data/vacode_core/conf/lang/userdict_ja.txt + - source: solr_vacode_core_protwords + target: /var/solr/data/vacode_core/conf/protwords.txt + - source: solr_vacode_core_stopwords + target: /var/solr/data/vacode_core/conf/stopwords.txt + - source: solr_vacode_core_synonyms + target: /var/solr/data/vacode_core/conf/synonyms.txt tika: image: apache/tika:latest-full container_name: tika restart: unless-stopped + configs: + - source: tika_config + target: /tika-config.xml + - source: tika_log4j2 + target: /log4j2.xml + - source: tika_tesseract_ocr_config_properties + target: /TesseractOCRConfig.properties + command: -c /tika-config.xml environment: - TIKA_LOG_PATH=/tika-logs - command: -c /tika-config.xml entrypoint: [ "/bin/sh", "-c", "exec java -cp \"/customocr:/tika-server-standard-2.9.2.jar:/tika-extras/*\" org.apache.tika.server.core.TikaServerCli -h 0.0.0.0 $$0 $$@"] volumes: - - ../tika-config.xml:/tika-config.xml - - ../tika-data/logs:/tika-logs - - ../tika-data/log4j2.xml:/log4j2.xml - - ../TesseractOCRConfig.properties:/TesseractOCRConfig.properties + - tika-logs:/tika-logs ports: - "${TIKA_PORT-9998}:9998" - -#volumes: -# data: +configs: + solr_log4j2: + file: ../solr-log4j2.xml + solr_nm3clol_core_properties: + content: | + name=nm3clol_core + solr_nm3clol_core_solrconfig: + file: ../solr-data/nm3clol_core/conf/solrconfig.xml + solr_nm3clol_core_managed_schema: + file: ../solr-data/nm3clol_core/conf/managed-schema.xml + solr_nm3clol_core_lang_contractions_ca: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ca.txt + solr_nm3clol_core_lang_contractions_fr: + file: ../solr-data/nm3clol_core/conf/lang/contractions_fr.txt + solr_nm3clol_core_lang_contractions_ga: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ga.txt + solr_nm3clol_core_lang_contractions_it: + file: ../solr-data/nm3clol_core/conf/lang/contractions_it.txt + solr_nm3clol_core_lang_hyphenations_ga: + file: ../solr-data/nm3clol_core/conf/lang/hyphenations_ga.txt + solr_nm3clol_core_lang_stemdict_nl: + file: ../solr-data/nm3clol_core/conf/lang/stemdict_nl.txt + solr_nm3clol_core_lang_stoptags_ja: + file: ../solr-data/nm3clol_core/conf/lang/stoptags_ja.txt + solr_nm3clol_core_lang_stopwords_ar: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ar.txt + solr_nm3clol_core_lang_stopwords_bg: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_bg.txt + solr_nm3clol_core_lang_stopwords_ca: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ca.txt + solr_nm3clol_core_lang_stopwords_cz: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_cz.txt + solr_nm3clol_core_lang_stopwords_da: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_da.txt + solr_nm3clol_core_lang_stopwords_de: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_de.txt + solr_nm3clol_core_lang_stopwords_el: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_el.txt + solr_nm3clol_core_lang_stopwords_en: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_en.txt + solr_nm3clol_core_lang_stopwords_es: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_es.txt + solr_nm3clol_core_lang_stopwords_et: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_et.txt + solr_nm3clol_core_lang_stopwords_eu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_eu.txt + solr_nm3clol_core_lang_stopwords_fa: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fa.txt + solr_nm3clol_core_lang_stopwords_fi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fi.txt + solr_nm3clol_core_lang_stopwords_fr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fr.txt + solr_nm3clol_core_lang_stopwords_ga: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ga.txt + solr_nm3clol_core_lang_stopwords_gl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_gl.txt + solr_nm3clol_core_lang_stopwords_hi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hi.txt + solr_nm3clol_core_lang_stopwords_hu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hu.txt + solr_nm3clol_core_lang_stopwords_hy: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hy.txt + solr_nm3clol_core_lang_stopwords_id: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_id.txt + solr_nm3clol_core_lang_stopwords_it: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_it.txt + solr_nm3clol_core_lang_stopwords_ja: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ja.txt + solr_nm3clol_core_lang_stopwords_lv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_lv.txt + solr_nm3clol_core_lang_stopwords_nl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_nl.txt + solr_nm3clol_core_lang_stopwords_no: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_no.txt + solr_nm3clol_core_lang_stopwords_pt: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_pt.txt + solr_nm3clol_core_lang_stopwords_ro: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ro.txt + solr_nm3clol_core_lang_stopwords_ru: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ru.txt + solr_nm3clol_core_lang_stopwords_sv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_sv.txt + solr_nm3clol_core_lang_stopwords_th: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_th.txt + solr_nm3clol_core_lang_stopwords_tr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_tr.txt + solr_nm3clol_core_lang_userdict_ja: + file: ../solr-data/nm3clol_core/conf/lang/userdict_ja.txt + solr_nm3clol_core_protwords: + file: ../solr-data/nm3clol_core/conf/protwords.txt + solr_nm3clol_core_stopwords: + file: ../solr-data/nm3clol_core/conf/stopwords.txt + solr_nm3clol_core_synonyms: + file: ../solr-data/nm3clol_core/conf/synonyms.txt + solr_vacode_core_properties: + content: | + name=vacode_core + solr_vacode_core_solrconfig: + file: ../solr-data/vacode_core/conf/solrconfig.xml + solr_vacode_core_managed_schema: + file: ../solr-data/vacode_core/conf/managed-schema.xml + solr_vacode_core_lang_contractions_ca: + file: ../solr-data/vacode_core/conf/lang/contractions_ca.txt + solr_vacode_core_lang_contractions_fr: + file: ../solr-data/vacode_core/conf/lang/contractions_fr.txt + solr_vacode_core_lang_contractions_ga: + file: ../solr-data/vacode_core/conf/lang/contractions_ga.txt + solr_vacode_core_lang_contractions_it: + file: ../solr-data/vacode_core/conf/lang/contractions_it.txt + solr_vacode_core_lang_hyphenations_ga: + file: ../solr-data/vacode_core/conf/lang/hyphenations_ga.txt + solr_vacode_core_lang_stemdict_nl: + file: ../solr-data/vacode_core/conf/lang/stemdict_nl.txt + solr_vacode_core_lang_stoptags_ja: + file: ../solr-data/vacode_core/conf/lang/stoptags_ja.txt + solr_vacode_core_lang_stopwords_ar: + file: ../solr-data/vacode_core/conf/lang/stopwords_ar.txt + solr_vacode_core_lang_stopwords_bg: + file: ../solr-data/vacode_core/conf/lang/stopwords_bg.txt + solr_vacode_core_lang_stopwords_ca: + file: ../solr-data/vacode_core/conf/lang/stopwords_ca.txt + solr_vacode_core_lang_stopwords_cz: + file: ../solr-data/vacode_core/conf/lang/stopwords_cz.txt + solr_vacode_core_lang_stopwords_da: + file: ../solr-data/vacode_core/conf/lang/stopwords_da.txt + solr_vacode_core_lang_stopwords_de: + file: ../solr-data/vacode_core/conf/lang/stopwords_de.txt + solr_vacode_core_lang_stopwords_el: + file: ../solr-data/vacode_core/conf/lang/stopwords_el.txt + solr_vacode_core_lang_stopwords_en: + file: ../solr-data/vacode_core/conf/lang/stopwords_en.txt + solr_vacode_core_lang_stopwords_es: + file: ../solr-data/vacode_core/conf/lang/stopwords_es.txt + solr_vacode_core_lang_stopwords_et: + file: ../solr-data/vacode_core/conf/lang/stopwords_et.txt + solr_vacode_core_lang_stopwords_eu: + file: ../solr-data/vacode_core/conf/lang/stopwords_eu.txt + solr_vacode_core_lang_stopwords_fa: + file: ../solr-data/vacode_core/conf/lang/stopwords_fa.txt + solr_vacode_core_lang_stopwords_fi: + file: ../solr-data/vacode_core/conf/lang/stopwords_fi.txt + solr_vacode_core_lang_stopwords_fr: + file: ../solr-data/vacode_core/conf/lang/stopwords_fr.txt + solr_vacode_core_lang_stopwords_ga: + file: ../solr-data/vacode_core/conf/lang/stopwords_ga.txt + solr_vacode_core_lang_stopwords_gl: + file: ../solr-data/vacode_core/conf/lang/stopwords_gl.txt + solr_vacode_core_lang_stopwords_hi: + file: ../solr-data/vacode_core/conf/lang/stopwords_hi.txt + solr_vacode_core_lang_stopwords_hu: + file: ../solr-data/vacode_core/conf/lang/stopwords_hu.txt + solr_vacode_core_lang_stopwords_hy: + file: ../solr-data/vacode_core/conf/lang/stopwords_hy.txt + solr_vacode_core_lang_stopwords_id: + file: ../solr-data/vacode_core/conf/lang/stopwords_id.txt + solr_vacode_core_lang_stopwords_it: + file: ../solr-data/vacode_core/conf/lang/stopwords_it.txt + solr_vacode_core_lang_stopwords_ja: + file: ../solr-data/vacode_core/conf/lang/stopwords_ja.txt + solr_vacode_core_lang_stopwords_lv: + file: ../solr-data/vacode_core/conf/lang/stopwords_lv.txt + solr_vacode_core_lang_stopwords_nl: + file: ../solr-data/vacode_core/conf/lang/stopwords_nl.txt + solr_vacode_core_lang_stopwords_no: + file: ../solr-data/vacode_core/conf/lang/stopwords_no.txt + solr_vacode_core_lang_stopwords_pt: + file: ../solr-data/vacode_core/conf/lang/stopwords_pt.txt + solr_vacode_core_lang_stopwords_ro: + file: ../solr-data/vacode_core/conf/lang/stopwords_ro.txt + solr_vacode_core_lang_stopwords_ru: + file: ../solr-data/vacode_core/conf/lang/stopwords_ru.txt + solr_vacode_core_lang_stopwords_sv: + file: ../solr-data/vacode_core/conf/lang/stopwords_sv.txt + solr_vacode_core_lang_stopwords_th: + file: ../solr-data/vacode_core/conf/lang/stopwords_th.txt + solr_vacode_core_lang_stopwords_tr: + file: ../solr-data/vacode_core/conf/lang/stopwords_tr.txt + solr_vacode_core_lang_userdict_ja: + file: ../solr-data/vacode_core/conf/lang/userdict_ja.txt + solr_vacode_core_protwords: + file: ../solr-data/vacode_core/conf/protwords.txt + solr_vacode_core_stopwords: + file: ../solr-data/vacode_core/conf/stopwords.txt + solr_vacode_core_synonyms: + file: ../solr-data/vacode_core/conf/synonyms.txt + tika_config: + file: ../tika-config.xml + tika_log4j2: + file: ../tika-log4j2.xml + tika_tesseract_ocr_config_properties: + file: ../tika-TesseractOCRConfig.properties +volumes: + solr-data: + tika-logs: diff --git a/solr/docker-compose.yml b/solr/docker-compose.yml index 4d08e5d..a7bd42e 100644 --- a/solr/docker-compose.yml +++ b/solr/docker-compose.yml @@ -1,17 +1,396 @@ -version: '3' - services: + solr-init: + image: alpine + container_name: solr-init + restart: "no" + volumes: + - solr-data:/var/solr + entrypoint: | + /bin/sh -c "chown -R 8983:8983 /var/solr" solr: + depends_on: + - solr-init image: solr:latest container_name: solr restart: unless-stopped ports: - "${SOLR_DOCS_PORT-8983}:8983" volumes: - - ../solr-data:/var/solr + - solr-data:/var/solr # environment: # - SOLR_CORE=my_core command: + # - /bin/sh -c "chown -R 8983:8983 /var/solr" - solr-precreate -# - gettingstarted - user: "1000:995" + # - gettingstarted + user: "8983:8983" + configs: + - source: solr_log4j2 + target: /var/solr/log4j2.xml + - source: solr_nm3clol_core_properties + target: /var/solr/data/nm3clol_core/core.properties + - source: solr_nm3clol_core_solrconfig + target: /var/solr/data/nm3clol_core/conf/solrconfig.xml + - source: solr_nm3clol_core_managed_schema + target: /var/solr/data/nm3clol_core/conf/managed-schema.xml + - source: solr_nm3clol_core_lang_contractions_ca + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ca.txt + - source: solr_nm3clol_core_lang_contractions_fr + target: /var/solr/data/nm3clol_core/conf/lang/contractions_fr.txt + - source: solr_nm3clol_core_lang_contractions_ga + target: /var/solr/data/nm3clol_core/conf/lang/contractions_ga.txt + - source: solr_nm3clol_core_lang_contractions_it + target: /var/solr/data/nm3clol_core/conf/lang/contractions_it.txt + - source: solr_nm3clol_core_lang_hyphenations_ga + target: /var/solr/data/nm3clol_core/conf/lang/hyphenations_ga.txt + - source: solr_nm3clol_core_lang_stemdict_nl + target: /var/solr/data/nm3clol_core/conf/lang/stemdict_nl.txt + - source: solr_nm3clol_core_lang_stoptags_ja + target: /var/solr/data/nm3clol_core/conf/lang/stoptags_ja.txt + - source: solr_nm3clol_core_lang_stopwords_ar + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ar.txt + - source: solr_nm3clol_core_lang_stopwords_bg + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_bg.txt + - source: solr_nm3clol_core_lang_stopwords_ca + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ca.txt + - source: solr_nm3clol_core_lang_stopwords_cz + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_cz.txt + - source: solr_nm3clol_core_lang_stopwords_da + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_da.txt + - source: solr_nm3clol_core_lang_stopwords_de + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_de.txt + - source: solr_nm3clol_core_lang_stopwords_el + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_el.txt + - source: solr_nm3clol_core_lang_stopwords_en + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_en.txt + - source: solr_nm3clol_core_lang_stopwords_es + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_es.txt + - source: solr_nm3clol_core_lang_stopwords_et + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_et.txt + - source: solr_nm3clol_core_lang_stopwords_eu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_eu.txt + - source: solr_nm3clol_core_lang_stopwords_fa + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fa.txt + - source: solr_nm3clol_core_lang_stopwords_fi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fi.txt + - source: solr_nm3clol_core_lang_stopwords_fr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_fr.txt + - source: solr_nm3clol_core_lang_stopwords_ga + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ga.txt + - source: solr_nm3clol_core_lang_stopwords_gl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_gl.txt + - source: solr_nm3clol_core_lang_stopwords_hi + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hi.txt + - source: solr_nm3clol_core_lang_stopwords_hu + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hu.txt + - source: solr_nm3clol_core_lang_stopwords_hy + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_hy.txt + - source: solr_nm3clol_core_lang_stopwords_id + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_id.txt + - source: solr_nm3clol_core_lang_stopwords_it + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_it.txt + - source: solr_nm3clol_core_lang_stopwords_ja + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ja.txt + - source: solr_nm3clol_core_lang_stopwords_lv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_lv.txt + - source: solr_nm3clol_core_lang_stopwords_nl + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_nl.txt + - source: solr_nm3clol_core_lang_stopwords_no + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_no.txt + - source: solr_nm3clol_core_lang_stopwords_pt + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_pt.txt + - source: solr_nm3clol_core_lang_stopwords_ro + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ro.txt + - source: solr_nm3clol_core_lang_stopwords_ru + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_ru.txt + - source: solr_nm3clol_core_lang_stopwords_sv + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_sv.txt + - source: solr_nm3clol_core_lang_stopwords_th + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_th.txt + - source: solr_nm3clol_core_lang_stopwords_tr + target: /var/solr/data/nm3clol_core/conf/lang/stopwords_tr.txt + - source: solr_nm3clol_core_lang_userdict_ja + target: /var/solr/data/nm3clol_core/conf/lang/userdict_ja.txt + - source: solr_nm3clol_core_protwords + target: /var/solr/data/nm3clol_core/conf/protwords.txt + - source: solr_nm3clol_core_stopwords + target: /var/solr/data/nm3clol_core/conf/stopwords.txt + - source: solr_nm3clol_core_synonyms + target: /var/solr/data/nm3clol_core/conf/synonyms.txt + - source: solr_vacode_core_properties + target: /var/solr/data/vacode_core/core.properties + - source: solr_vacode_core_solrconfig + target: /var/solr/data/vacode_core/conf/solrconfig.xml + - source: solr_vacode_core_managed_schema + target: /var/solr/data/vacode_core/conf/managed-schema.xml + - source: solr_vacode_core_lang_contractions_ca + target: /var/solr/data/vacode_core/conf/lang/contractions_ca.txt + - source: solr_vacode_core_lang_contractions_fr + target: /var/solr/data/vacode_core/conf/lang/contractions_fr.txt + - source: solr_vacode_core_lang_contractions_ga + target: /var/solr/data/vacode_core/conf/lang/contractions_ga.txt + - source: solr_vacode_core_lang_contractions_it + target: /var/solr/data/vacode_core/conf/lang/contractions_it.txt + - source: solr_vacode_core_lang_hyphenations_ga + target: /var/solr/data/vacode_core/conf/lang/hyphenations_ga.txt + - source: solr_vacode_core_lang_stemdict_nl + target: /var/solr/data/vacode_core/conf/lang/stemdict_nl.txt + - source: solr_vacode_core_lang_stoptags_ja + target: /var/solr/data/vacode_core/conf/lang/stoptags_ja.txt + - source: solr_vacode_core_lang_stopwords_ar + target: /var/solr/data/vacode_core/conf/lang/stopwords_ar.txt + - source: solr_vacode_core_lang_stopwords_bg + target: /var/solr/data/vacode_core/conf/lang/stopwords_bg.txt + - source: solr_vacode_core_lang_stopwords_ca + target: /var/solr/data/vacode_core/conf/lang/stopwords_ca.txt + - source: solr_vacode_core_lang_stopwords_cz + target: /var/solr/data/vacode_core/conf/lang/stopwords_cz.txt + - source: solr_vacode_core_lang_stopwords_da + target: /var/solr/data/vacode_core/conf/lang/stopwords_da.txt + - source: solr_vacode_core_lang_stopwords_de + target: /var/solr/data/vacode_core/conf/lang/stopwords_de.txt + - source: solr_vacode_core_lang_stopwords_el + target: /var/solr/data/vacode_core/conf/lang/stopwords_el.txt + - source: solr_vacode_core_lang_stopwords_en + target: /var/solr/data/vacode_core/conf/lang/stopwords_en.txt + - source: solr_vacode_core_lang_stopwords_es + target: /var/solr/data/vacode_core/conf/lang/stopwords_es.txt + - source: solr_vacode_core_lang_stopwords_et + target: /var/solr/data/vacode_core/conf/lang/stopwords_et.txt + - source: solr_vacode_core_lang_stopwords_eu + target: /var/solr/data/vacode_core/conf/lang/stopwords_eu.txt + - source: solr_vacode_core_lang_stopwords_fa + target: /var/solr/data/vacode_core/conf/lang/stopwords_fa.txt + - source: solr_vacode_core_lang_stopwords_fi + target: /var/solr/data/vacode_core/conf/lang/stopwords_fi.txt + - source: solr_vacode_core_lang_stopwords_fr + target: /var/solr/data/vacode_core/conf/lang/stopwords_fr.txt + - source: solr_vacode_core_lang_stopwords_ga + target: /var/solr/data/vacode_core/conf/lang/stopwords_ga.txt + - source: solr_vacode_core_lang_stopwords_gl + target: /var/solr/data/vacode_core/conf/lang/stopwords_gl.txt + - source: solr_vacode_core_lang_stopwords_hi + target: /var/solr/data/vacode_core/conf/lang/stopwords_hi.txt + - source: solr_vacode_core_lang_stopwords_hu + target: /var/solr/data/vacode_core/conf/lang/stopwords_hu.txt + - source: solr_vacode_core_lang_stopwords_hy + target: /var/solr/data/vacode_core/conf/lang/stopwords_hy.txt + - source: solr_vacode_core_lang_stopwords_id + target: /var/solr/data/vacode_core/conf/lang/stopwords_id.txt + - source: solr_vacode_core_lang_stopwords_it + target: /var/solr/data/vacode_core/conf/lang/stopwords_it.txt + - source: solr_vacode_core_lang_stopwords_ja + target: /var/solr/data/vacode_core/conf/lang/stopwords_ja.txt + - source: solr_vacode_core_lang_stopwords_lv + target: /var/solr/data/vacode_core/conf/lang/stopwords_lv.txt + - source: solr_vacode_core_lang_stopwords_nl + target: /var/solr/data/vacode_core/conf/lang/stopwords_nl.txt + - source: solr_vacode_core_lang_stopwords_no + target: /var/solr/data/vacode_core/conf/lang/stopwords_no.txt + - source: solr_vacode_core_lang_stopwords_pt + target: /var/solr/data/vacode_core/conf/lang/stopwords_pt.txt + - source: solr_vacode_core_lang_stopwords_ro + target: /var/solr/data/vacode_core/conf/lang/stopwords_ro.txt + - source: solr_vacode_core_lang_stopwords_ru + target: /var/solr/data/vacode_core/conf/lang/stopwords_ru.txt + - source: solr_vacode_core_lang_stopwords_sv + target: /var/solr/data/vacode_core/conf/lang/stopwords_sv.txt + - source: solr_vacode_core_lang_stopwords_th + target: /var/solr/data/vacode_core/conf/lang/stopwords_th.txt + - source: solr_vacode_core_lang_stopwords_tr + target: /var/solr/data/vacode_core/conf/lang/stopwords_tr.txt + - source: solr_vacode_core_lang_userdict_ja + target: /var/solr/data/vacode_core/conf/lang/userdict_ja.txt + - source: solr_vacode_core_protwords + target: /var/solr/data/vacode_core/conf/protwords.txt + - source: solr_vacode_core_stopwords + target: /var/solr/data/vacode_core/conf/stopwords.txt + - source: solr_vacode_core_synonyms + target: /var/solr/data/vacode_core/conf/synonyms.txt +configs: + solr_log4j2: + file: ../solr-log4j2.xml + solr_nm3clol_core_properties: + content: | + name=nm3clol_core + solr_nm3clol_core_solrconfig: + file: ../solr-data/nm3clol_core/conf/solrconfig.xml + solr_nm3clol_core_managed_schema: + file: ../solr-data/nm3clol_core/conf/managed-schema.xml + solr_nm3clol_core_lang_contractions_ca: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ca.txt + solr_nm3clol_core_lang_contractions_fr: + file: ../solr-data/nm3clol_core/conf/lang/contractions_fr.txt + solr_nm3clol_core_lang_contractions_ga: + file: ../solr-data/nm3clol_core/conf/lang/contractions_ga.txt + solr_nm3clol_core_lang_contractions_it: + file: ../solr-data/nm3clol_core/conf/lang/contractions_it.txt + solr_nm3clol_core_lang_hyphenations_ga: + file: ../solr-data/nm3clol_core/conf/lang/hyphenations_ga.txt + solr_nm3clol_core_lang_stemdict_nl: + file: ../solr-data/nm3clol_core/conf/lang/stemdict_nl.txt + solr_nm3clol_core_lang_stoptags_ja: + file: ../solr-data/nm3clol_core/conf/lang/stoptags_ja.txt + solr_nm3clol_core_lang_stopwords_ar: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ar.txt + solr_nm3clol_core_lang_stopwords_bg: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_bg.txt + solr_nm3clol_core_lang_stopwords_ca: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ca.txt + solr_nm3clol_core_lang_stopwords_cz: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_cz.txt + solr_nm3clol_core_lang_stopwords_da: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_da.txt + solr_nm3clol_core_lang_stopwords_de: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_de.txt + solr_nm3clol_core_lang_stopwords_el: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_el.txt + solr_nm3clol_core_lang_stopwords_en: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_en.txt + solr_nm3clol_core_lang_stopwords_es: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_es.txt + solr_nm3clol_core_lang_stopwords_et: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_et.txt + solr_nm3clol_core_lang_stopwords_eu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_eu.txt + solr_nm3clol_core_lang_stopwords_fa: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fa.txt + solr_nm3clol_core_lang_stopwords_fi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fi.txt + solr_nm3clol_core_lang_stopwords_fr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_fr.txt + solr_nm3clol_core_lang_stopwords_ga: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ga.txt + solr_nm3clol_core_lang_stopwords_gl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_gl.txt + solr_nm3clol_core_lang_stopwords_hi: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hi.txt + solr_nm3clol_core_lang_stopwords_hu: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hu.txt + solr_nm3clol_core_lang_stopwords_hy: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_hy.txt + solr_nm3clol_core_lang_stopwords_id: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_id.txt + solr_nm3clol_core_lang_stopwords_it: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_it.txt + solr_nm3clol_core_lang_stopwords_ja: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ja.txt + solr_nm3clol_core_lang_stopwords_lv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_lv.txt + solr_nm3clol_core_lang_stopwords_nl: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_nl.txt + solr_nm3clol_core_lang_stopwords_no: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_no.txt + solr_nm3clol_core_lang_stopwords_pt: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_pt.txt + solr_nm3clol_core_lang_stopwords_ro: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ro.txt + solr_nm3clol_core_lang_stopwords_ru: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_ru.txt + solr_nm3clol_core_lang_stopwords_sv: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_sv.txt + solr_nm3clol_core_lang_stopwords_th: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_th.txt + solr_nm3clol_core_lang_stopwords_tr: + file: ../solr-data/nm3clol_core/conf/lang/stopwords_tr.txt + solr_nm3clol_core_lang_userdict_ja: + file: ../solr-data/nm3clol_core/conf/lang/userdict_ja.txt + solr_nm3clol_core_protwords: + file: ../solr-data/nm3clol_core/conf/protwords.txt + solr_nm3clol_core_stopwords: + file: ../solr-data/nm3clol_core/conf/stopwords.txt + solr_nm3clol_core_synonyms: + file: ../solr-data/nm3clol_core/conf/synonyms.txt + solr_vacode_core_properties: + content: | + name=vacode_core + solr_vacode_core_solrconfig: + file: ../solr-data/vacode_core/conf/solrconfig.xml + solr_vacode_core_managed_schema: + file: ../solr-data/vacode_core/conf/managed-schema.xml + solr_vacode_core_lang_contractions_ca: + file: ../solr-data/vacode_core/conf/lang/contractions_ca.txt + solr_vacode_core_lang_contractions_fr: + file: ../solr-data/vacode_core/conf/lang/contractions_fr.txt + solr_vacode_core_lang_contractions_ga: + file: ../solr-data/vacode_core/conf/lang/contractions_ga.txt + solr_vacode_core_lang_contractions_it: + file: ../solr-data/vacode_core/conf/lang/contractions_it.txt + solr_vacode_core_lang_hyphenations_ga: + file: ../solr-data/vacode_core/conf/lang/hyphenations_ga.txt + solr_vacode_core_lang_stemdict_nl: + file: ../solr-data/vacode_core/conf/lang/stemdict_nl.txt + solr_vacode_core_lang_stoptags_ja: + file: ../solr-data/vacode_core/conf/lang/stoptags_ja.txt + solr_vacode_core_lang_stopwords_ar: + file: ../solr-data/vacode_core/conf/lang/stopwords_ar.txt + solr_vacode_core_lang_stopwords_bg: + file: ../solr-data/vacode_core/conf/lang/stopwords_bg.txt + solr_vacode_core_lang_stopwords_ca: + file: ../solr-data/vacode_core/conf/lang/stopwords_ca.txt + solr_vacode_core_lang_stopwords_cz: + file: ../solr-data/vacode_core/conf/lang/stopwords_cz.txt + solr_vacode_core_lang_stopwords_da: + file: ../solr-data/vacode_core/conf/lang/stopwords_da.txt + solr_vacode_core_lang_stopwords_de: + file: ../solr-data/vacode_core/conf/lang/stopwords_de.txt + solr_vacode_core_lang_stopwords_el: + file: ../solr-data/vacode_core/conf/lang/stopwords_el.txt + solr_vacode_core_lang_stopwords_en: + file: ../solr-data/vacode_core/conf/lang/stopwords_en.txt + solr_vacode_core_lang_stopwords_es: + file: ../solr-data/vacode_core/conf/lang/stopwords_es.txt + solr_vacode_core_lang_stopwords_et: + file: ../solr-data/vacode_core/conf/lang/stopwords_et.txt + solr_vacode_core_lang_stopwords_eu: + file: ../solr-data/vacode_core/conf/lang/stopwords_eu.txt + solr_vacode_core_lang_stopwords_fa: + file: ../solr-data/vacode_core/conf/lang/stopwords_fa.txt + solr_vacode_core_lang_stopwords_fi: + file: ../solr-data/vacode_core/conf/lang/stopwords_fi.txt + solr_vacode_core_lang_stopwords_fr: + file: ../solr-data/vacode_core/conf/lang/stopwords_fr.txt + solr_vacode_core_lang_stopwords_ga: + file: ../solr-data/vacode_core/conf/lang/stopwords_ga.txt + solr_vacode_core_lang_stopwords_gl: + file: ../solr-data/vacode_core/conf/lang/stopwords_gl.txt + solr_vacode_core_lang_stopwords_hi: + file: ../solr-data/vacode_core/conf/lang/stopwords_hi.txt + solr_vacode_core_lang_stopwords_hu: + file: ../solr-data/vacode_core/conf/lang/stopwords_hu.txt + solr_vacode_core_lang_stopwords_hy: + file: ../solr-data/vacode_core/conf/lang/stopwords_hy.txt + solr_vacode_core_lang_stopwords_id: + file: ../solr-data/vacode_core/conf/lang/stopwords_id.txt + solr_vacode_core_lang_stopwords_it: + file: ../solr-data/vacode_core/conf/lang/stopwords_it.txt + solr_vacode_core_lang_stopwords_ja: + file: ../solr-data/vacode_core/conf/lang/stopwords_ja.txt + solr_vacode_core_lang_stopwords_lv: + file: ../solr-data/vacode_core/conf/lang/stopwords_lv.txt + solr_vacode_core_lang_stopwords_nl: + file: ../solr-data/vacode_core/conf/lang/stopwords_nl.txt + solr_vacode_core_lang_stopwords_no: + file: ../solr-data/vacode_core/conf/lang/stopwords_no.txt + solr_vacode_core_lang_stopwords_pt: + file: ../solr-data/vacode_core/conf/lang/stopwords_pt.txt + solr_vacode_core_lang_stopwords_ro: + file: ../solr-data/vacode_core/conf/lang/stopwords_ro.txt + solr_vacode_core_lang_stopwords_ru: + file: ../solr-data/vacode_core/conf/lang/stopwords_ru.txt + solr_vacode_core_lang_stopwords_sv: + file: ../solr-data/vacode_core/conf/lang/stopwords_sv.txt + solr_vacode_core_lang_stopwords_th: + file: ../solr-data/vacode_core/conf/lang/stopwords_th.txt + solr_vacode_core_lang_stopwords_tr: + file: ../solr-data/vacode_core/conf/lang/stopwords_tr.txt + solr_vacode_core_lang_userdict_ja: + file: ../solr-data/vacode_core/conf/lang/userdict_ja.txt + solr_vacode_core_protwords: + file: ../solr-data/vacode_core/conf/protwords.txt + solr_vacode_core_stopwords: + file: ../solr-data/vacode_core/conf/stopwords.txt + solr_vacode_core_synonyms: + file: ../solr-data/vacode_core/conf/synonyms.txt +volumes: + solr-data: diff --git a/TesseractOCRConfig.properties b/tika-TesseractOCRConfig.properties similarity index 100% rename from TesseractOCRConfig.properties rename to tika-TesseractOCRConfig.properties diff --git a/tika-log4j2.xml b/tika-log4j2.xml new file mode 100644 index 0000000..245e974 --- /dev/null +++ b/tika-log4j2.xml @@ -0,0 +1,13 @@ + + + + + + + + + + + + + diff --git a/tika/docker-compose.yml b/tika/docker-compose.yml index 31927b5..cd2c372 100644 --- a/tika/docker-compose.yml +++ b/tika/docker-compose.yml @@ -1,18 +1,29 @@ -version: '3' - services: tika: image: apache/tika:latest-full container_name: tika restart: unless-stopped + configs: + - source: tika_config + target: /tika-config.xml + - source: tika_log4j2 + target: /log4j2.xml + - source: tika_tesseract_ocr_config_properties + target: /TesseractOCRConfig.properties + command: -c /tika-config.xml environment: - TIKA_LOG_PATH=/tika-logs - command: -c /tika-config.xml entrypoint: [ "/bin/sh", "-c", "exec java -cp \"/customocr:/tika-server-standard-2.9.2.jar:/tika-extras/*\" org.apache.tika.server.core.TikaServerCli -h 0.0.0.0 $$0 $$@"] volumes: - - ../tika-config.xml:/tika-config.xml - - ../tika-data/logs:/tika-logs - - ../tika-data/log4j2.xml:/log4j2.xml - - ../TesseractOCRConfig.properties:/TesseractOCRConfig.properties + - tika-logs:/tika-logs ports: - "${TIKA_PORT-9998}:9998" +configs: + tika_config: + file: ../tika-config.xml + tika_log4j2: + file: ../tika-log4j2.xml + tika_tesseract_ocr_config_properties: + file: ../tika-TesseractOCRConfig.properties +volumes: + tika-logs: