Updated ElasticSearch mapping scripts

gbhl · Jul 13, 2018 · 3c9f455 · 3c9f455
1 parent d88fb68
commit 3c9f455
Show file tree

Hide file tree

Showing 13 changed files with 321 additions and 119 deletions.
diff --git a/ElasticSearch/AuthorsMappings.json → ElasticSearch/Authors.json b/ElasticSearch/AuthorsMappings.json → ElasticSearch/Authors.json
@@ -1,4 +1,14 @@
 {
+	"settings": {
+		"analysis": {
+			"analyzer": {
+				"default": {
+					"tokenizer": "icu_tokenizer",
+					"filter":  [ "lowercase", "icu_folding" ]
+				}
+			}
+		}
+	},
 	"author" : {
 		"properties" : {
 			"authorNames" : {

diff --git a/ElasticSearch/AuthorsSettings.json b/ElasticSearch/AuthorsSettings.json
diff --git a/ElasticSearch/BuildESIndex.sh b/ElasticSearch/BuildESIndex.sh
@@ -2,34 +2,26 @@
 
 sServer=$1
 sIndex=$2
-sDocType=$3
-sSettings=$4
-sMappings=$5
+sSettings=$3
 
-if [ $# -eq 5 ]
+if [ $# -eq 3 ]
 then
   echo
   echo Building index $sServer/$sIndex with settings in $sSettings
   curl -X PUT $sServer/$sIndex -d @$sSettings --header "Content-Type:application/json"
   echo
-  echo
-  echo Adding mappings for $sServer/$sIndex/_mapping/$sDocType with mappings in $sMappings
-  curl -X POST $sServer/$sIndex/_mapping/$sDocType -d @$sMappings --header "Content-Type:application/json"
-  echo
   echo Done
 else
   echo
   echo Usage
   echo
-  echo BuildESIndex SERVER-NAME INDEX-NAME DOC-TYPE SETTINGS-FILENAME MAPPINGS-FILENAME
+  echo BuildESIndex SERVER-NAME INDEX-NAME SETTINGS-FILENAME
   echo
   echo SERVER-NAME is the name of the search server, including the port number
   echo INDEX-NAME is the name of the index to be created
-  echo DOC-TYPE is the type of documents to be added to the index
-  echo SETTINGS-FILENAME is the name of a file containing index settings
-  echo MAPPINGS-FILENAME is the name of a file containing index mappings
+  echo SETTINGS-FILENAME is the name of a file containing index settings and mappings
   echo
-  echo Example: BuildESIndex http://localhost:9200 items item ItemsSettings.json ItemsMappings.json
+  echo Example: BuildESIndex http://localhost:9200 items Items.json
   echo
 fi
 echo
diff --git a/ElasticSearch/Catalog.json b/ElasticSearch/Catalog.json
@@ -0,0 +1,239 @@
+{
+	"settings": {
+		"analysis": {
+			"analyzer": {
+				"default": {
+					"tokenizer": "icu_tokenizer",
+					"filter":  [ "lowercase", "icu_folding" ]
+				}
+			}
+		}
+	},
+	"mappings" : {
+		"catalogitem" : {
+			"properties" : {
+				"associations" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"authors" : {
+					"type" : "keyword",
+					"index" : false
+				},
+				"collections" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"container" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"contributors" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					},
+					"include_in_all" : false
+				},
+				"dateRanges" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"dates" : {
+					"type" : "keyword"
+				},
+				"doi" : {
+					"type" : "text",
+					"analyzer" : "standard",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"facetAuthors" : {
+					"type" : "keyword",
+					"index" : false
+				},			
+				"genre" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"id" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"isbn" : {
+					"type" : "text",
+					"analyzer" : "standard",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"issn" : {
+					"type" : "text",
+					"analyzer" : "standard",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"issue" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"itemId" : {
+					"type" : "long",
+					"include_in_all" : false
+				},
+				"keywords" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"language" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"materialType" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"oclc" : {
+					"type" : "text",
+					"analyzer" : "standard",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"pageRange" : {
+					"type" : "keyword",
+					"index" : false
+				},
+				"publicationPlace" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					},
+					"include_in_all" : false
+				},
+				"publisher" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					},
+					"include_in_all" : false
+				},
+				"searchAuthors" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"segmentId" : {
+					"type" : "long",
+					"include_in_all" : false
+				},
+				"series" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"sortTitle" : {
+					"type" : "keyword",
+					"include_in_all" : false
+				},
+				"startPageId" : {
+					"type" : "long",
+					"include_in_all" : false
+				},
+				"title" : {
+					"type" : "text",
+					"analyzer" : "default",
+					"fields" : {
+						"raw" : {
+							"type" : "keyword"
+						}
+					}
+				},
+				"titleId" : {
+					"type" : "long",
+					"include_in_all" : false
+				},
+				"translatedTitle" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"uniformTitle" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"url" : {
+					"type" : "keyword",
+					"index" : false
+				},
+				"variants" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"volume" : {
+					"type" : "text",
+					"analyzer" : "default"
+				},
+				"volumes": { 
+					"type": "nested", 
+					"properties": {
+						"id" : { 
+							"type" : "keyword", 
+							"include_in_all" : false 
+						},
+						"itemId" : { 
+							"type" : "long", 
+							"include_in_all" : false 
+						},
+						"volume" : { 
+							"type" : "text",  
+							"analyzer" : "default"
+						},
+						"date" : { 
+							"type" : "keyword",
+							"include_in_all" : false
+						},
+						"hasExternalContent" : {
+							"type" : "boolean"
+						},
+						"hasIllustrations" : {
+							"type" : "boolean"
+						},
+						"hasLocalContent" : {
+							"type" : "boolean"
+						},
+						"hasSegments" : {
+							"type" : "boolean"
+						}
+					}
+				}
+			}
+		}
+	}
+}
diff --git a/ElasticSearch/ItemsMappings.json → ElasticSearch/Items.json b/ElasticSearch/ItemsMappings.json → ElasticSearch/Items.json
@@ -1,4 +1,14 @@
 {
+	"settings": {
+		"analysis": {
+			"analyzer": {
+				"default": {
+					"tokenizer": "icu_tokenizer",
+					"filter":  [ "lowercase", "icu_folding" ]
+				}
+			}
+		}
+	},
 	"item" : {
 		"properties" : {
 			"associations" : {

diff --git a/ElasticSearch/ItemsSettings.json b/ElasticSearch/ItemsSettings.json
diff --git a/ElasticSearch/KeywordsMappings.json → ElasticSearch/Keywords.json b/ElasticSearch/KeywordsMappings.json → ElasticSearch/Keywords.json
@@ -1,4 +1,14 @@
 {
+	"settings": {
+		"analysis": {
+			"analyzer": {
+				"default": {
+					"tokenizer": "icu_tokenizer",
+					"filter":  [ "lowercase", "icu_folding" ]
+				}
+			}
+		}
+	},
 	"keyword" : {
 		"properties" : {
 			"id" : {

diff --git a/ElasticSearch/KeywordsSettings.json b/ElasticSearch/KeywordsSettings.json
diff --git a/ElasticSearch/NamesMappings.json → ElasticSearch/Names.json b/ElasticSearch/NamesMappings.json → ElasticSearch/Names.json
@@ -1,4 +1,14 @@
 {
+	"settings": {
+		"analysis": {
+			"analyzer": {
+				"default": {
+					"tokenizer": "icu_tokenizer",
+					"filter":  [ "lowercase", "icu_folding" ]
+				}
+			}
+		}
+	},
 	"name" : {
 		"properties" : {
 			"count" : {

diff --git a/ElasticSearch/NamesSettings.json b/ElasticSearch/NamesSettings.json