add maestro + platform mappings (#18)

* add maestro + platform  mappings

* typos

* add settings/autocomplete/types

* brackets

* add s to mapping

* Update maestro_file_centric.json

set shards to 3

* Update platform_file_centric.json

set shards to 3

* correct date type on dates

* correct md5_sum

* fix types

* remove dynamic - do group by workflow

* update analyzers
move workflow under analysis

* add analyzer to platform file mapping

fix edge ngram min gram

* remove sequencing fields from mapping

* remove/add workflow field

* actually commit removed workflow

* rename workflow fields

* add additional autocompletes

* Update maestro_file_centric.json

min gram to 2

* Update platform_file_centric.json

min gram to 2

* - remove PD id as it one snot have one _yet_
- add submitter id to follow + treatment

Co-authored-by: blabadi <bashar.allabadi@oicr.on.ca>

mappings/maestro_file_centric.json

@@ -0,0 +1,225 @@
+{
+  "aliases": {
+    "file_centric": {}
+  },
+  "settings": {
+    "index.number_of_shards": 3,
+    "index.max_results_window": 300000,
+    "analysis": {
+      "analyzer": {
+        "autocomplete_analyzer": {
+          "filter": ["lowercase", "edge_ngram"],
+          "tokenizer": "standard"
+        },
+        "autocomplete_prefix": {
+          "filter": ["lowercase", "edge_ngram"],
+          "tokenizer": "keyword"
+        },
+        "lowercase_keyword": {
+          "filter": ["lowercase"],
+          "tokenizer": "keyword"
+        }
+      },
+      "filter": {
+        "edge_ngram": {
+          "max_gram": 20,
+          "min_gram": 2,
+          "side": "front",
+          "type": "edge_ngram"
+        }
+      }
+    }
+  },
+  "mappings": {
+    "dynamic": false,
+    "date_detection": false,
+    "properties": {
+      "file_autocomplete": {
+        "fields": {
+          "analyzed": {
+            "analyzer": "autocomplete_analyzer",
+            "search_analyzer": "lowercase_keyword",
+            "type": "text"
+          },
+          "lowercase": {
+            "analyzer": "lowercase_keyword",
+            "type": "text"
+          },
+          "prefix": {
+            "analyzer": "autocomplete_prefix",
+            "search_analyzer": "lowercase_keyword",
+            "type": "text"
+          }
+        },
+        "type": "keyword"
+      },
+      "study_id": {
+        "type": "keyword",
+        "copy_to": ["file_autocomplete"]
+      },
+      "object_id": {
+        "type": "keyword",
+        "copy_to": ["file_autocomplete"]
+      },
+      "file_type": {
+        "type": "keyword"
+      },
+      "data_type": {
+        "type": "keyword",
+        "copy_to": ["file_autocomplete"]
+      },
+      "file_access": {
+        "type": "keyword"
+      },
+      "analysis": {
+        "properties": {
+          "analysis_id": {
+            "type": "keyword",
+            "copy_to": ["file_autocomplete"]
+          },
+          "analysis_type": {
+            "type": "keyword"
+          },
+          "analysis_version": {
+            "type": "integer"
+          },
+          "workflow": {
+            "properties": {
+              "name": {
+                "type": "keyword"
+              },
+              "version": {
+                "type": "keyword"
+              }
+            }
+          },
+          "experiment": {
+            "type": "object",
+            "properties": {
+              "platform": {
+                "type": "keyword"
+              },
+              "library_strategy": {
+                "type": "keyword"
+              }
+            }
+          }
+        }
+      },
+      "file": {
+        "properties": {
+          "id": {
+            "type": "keyword"
+          },
+          "size": {
+            "type": "integer"
+          },
+          "md5sum": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "keyword",
+            "copy_to": ["file_autocomplete"]
+          }
+        }
+      },
+      "index_file": {
+        "properties": {
+          "object_id": {
+            "type": "keyword"
+          },
+          "format": {
+            "type": "keyword"
+          },
+          "md5sum": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "keyword"
+          },
+          "size": {
+            "type": "integer"
+          }
+        }
+      },
+      "donors": {
+        "type": "nested",
+        "properties": {
+          "id": {
+            "type": "keyword",
+            "copy_to": ["file_autocomplete"]
+          },
+          "submitter_donor_id": {
+            "type": "keyword",
+            "copy_to": ["file_autocomplete"]
+          },
+          "gender": {
+            "type": "keyword"
+          },
+          "specimens": {
+            "type": "nested",
+            "properties": {
+              "id": {
+                "type": "keyword",
+                "copy_to": ["file_autocomplete"]
+              },
+              "submitter_specimen_id": {
+                "type": "keyword",
+                "copy_to": ["file_autocomplete"]
+              },
+              "tumour_normal_designation": {
+                "type": "keyword"
+              },
+              "specimen_tissue_source": {
+                "type": "keyword"
+              },
+              "specimen_type": {
+                "type": "keyword"
+              },
+              "samples": {
+                "type": "nested",
+                "properties": {
+                  "id": {
+                    "type": "keyword",
+                    "copy_to": ["file_autocomplete"]
+                  },
+                  "submitter_sample_id": {
+                    "type": "keyword",
+                    "copy_to": ["file_autocomplete"]
+                  },
+                  "sample_type": {
+                    "type": "keyword"
+                  },
+                  "matched_normal_submitter_sample_id": {
+                    "type": "keyword",
+                    "copy_to": ["file_autocomplete"]
+                  }
+                }
+              }
+            }
+          }
+        }
+      },
+      "repositories": {
+        "type": "nested",
+        "properties": {
+          "code": {
+            "type": "keyword"
+          },
+          "name": {
+            "type": "keyword"
+          },
+          "organization": {
+            "type": "keyword"
+          },
+          "country": {
+            "type": "keyword"
+          },
+          "url": {
+            "type": "keyword"
+          }
+        }
+      }
+    }
+  }
+}