Merge pull request #96 from HumanCellAtlas/analysis_validation

Revised analysis.json so it becomes valid json
HumanCellAtlas · Dec 13, 2017 · 1c26aab · 1c26aab
2 parents 14d8fd2 + e4d3762
commit 1c26aab
Show file tree

Hide file tree

Showing 2 changed files with 130 additions and 99 deletions.
diff --git a/json_schema/analysis.json b/json_schema/analysis.json
@@ -1,154 +1,185 @@
 {
-    "title": "analysis", 
+    "title": "analysis",
     "required": [
-        "timestamp_start_utc", 
-        "timestamp_stop_utc", 
-        "computational_method", 
-        "input_bundles", 
-        "reference_bundle", 
-        "analysis_id", 
-        "analysis_run_type", 
-        "metadata_schema", 
-        "tasks", 
-        "inputs", 
-        "outputs"
-    ], 
-    "additionalProperties": true, 
+        "timestamp_start_utc",
+        "timestamp_stop_utc",
+        "computational_method",
+        "input_bundles",
+        "reference_bundle",
+        "analysis_id",
+        "analysis_run_type",
+        "metadata_schema",
+        "tasks",
+        "inputs",
+        "outputs",
+        "core"
+    ],
+    "additionalProperties": true,
     "definitions": {
         "task": {
-            "additionalProperties": false, 
+            "additionalProperties": false,
             "required": [
-                "name", 
-                "start_time", 
-                "stop_time", 
-                "log", 
-                "disk_size", 
-                "docker_image", 
-                "cpus", 
-                "memory", 
+                "name",
+                "start_time",
+                "stop_time",
+                "log",
+                "disk_size",
+                "docker_image",
+                "cpus",
+                "memory",
                 "zone"
-            ], 
-            "type": "object", 
+            ],
+            "type": "object",
             "properties": {
-                "disk_size": "string", 
-                "name": "string", 
-                "zone": "string", 
-                "log_err": "uri", 
-                "start_time": "date-time", 
-                "cpus": "int", 
-                "log_out": "uri", 
-                "stop_time": "date-time", 
-                "memory": "string", 
-                "docker_image": "string"
+                "disk_size": {
+                    "type" : "string"
+                },
+                "name": {
+                    "type" : "string"
+                },
+                "zone": {
+                    "type" : "string"
+                },
+                "log_err": {
+                    "type" : "string"
+                },
+                "start_time":{
+                    "type": "string",
+                    "format": "date-time"
+                },
+                "cpus": {
+                    "type" : "integer"
+                },
+                "log_out":{
+                    "type" : "string"
+                },
+                "stop_time": {
+                    "type": "string",
+                    "format": "date-time"
+                },
+                "memory":{
+                    "type" : "string"
+                },
+                "docker_image":{
+                    "type" : "string"
+                }
             }
-        }, 
+        },
         "parameter": {
-            "additionalProperties": false, 
+            "additionalProperties": false,
             "required": [
-                "name", 
+                "name",
                 "value"
-            ], 
-            "type": "object", 
+            ],
+            "type": "object",
             "properties": {
-                "checksum": "string", 
-                "name": "string", 
-                "value": "string"
+                "checksum": {
+                    "type" : "string"
+                },
+                "name": {
+                    "type" : "string"
+                },
+                "value": {
+                    "type" : "string"
+                }
             }
-        }, 
+        },
         "file": {
-            "additionalProperties": false, 
+            "additionalProperties": false,
             "required": [
-                "checksum", 
-                "file_path", 
+                "checksum",
+                "file_path",
                 "format"
-            ], 
-            "type": "object", 
+            ],
+            "type": "object",
             "properties": {
-                "checksum": "string", 
-                "file_path": "uri", 
-                "format": "string"
+                "checksum": {
+                    "type" : "string"
+                },
+                "file_path": {
+                    "type" : "string"
+                },
+                "format": {
+                    "type" : "string"
+                }
             }
         }
-    }, 
-    "$schema": "http://json-schema.org/draft-04/schema#", 
-    "type": "object", 
+    },
+    "$schema": "http://json-schema.org/draft-04/schema#",
+    "type": "object",
     "properties": {
         "inputs": {
             "items": {
-                "type": {
-                    "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/parameter"
-                }
-            }, 
-            "type": "array", 
+                "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/parameter"
+            },
+            "type": "array",
             "description": "Input parameters used in the pipeline run, these can be files or string values (settings)."
-        }, 
+        },
         "reference_bundle": {
-            "type": "uri", 
+            "type": "string",
             "description": "Bundle containing the reference used in running the pipeline."
-        }, 
+        },
         "tasks": {
             "items": {
-                "type": {
-                    "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/task"
-                }
-            }, 
-            "type": "array", 
+                "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/task"
+            },
+            "type": "array",
             "description": "Descriptions of tasks in the workflow."
-        }, 
+        },
         "description": {
-            "type": "string", 
+            "type": "string",
             "description": "A general description of the analysis."
-        }, 
+        },
         "timestamp_stop_utc": {
-            "type": "date-time", 
+            "type": "string",
+            "format": "date-time",
             "description": "Terminal stop time of the full pipeline."
-        }, 
+        },
         "input_bundles": {
             "items": {
-                "type": "uri"
-            }, 
-            "type": "array", 
+                "type": "string"
+            },
+            "type": "array",
             "description": "The input bundles used in this analysis run."
-        }, 
+        },
         "outputs": {
             "items": {
-                "type": {
-                    "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/file"
-                }
-            }, 
-            "type": "array", 
+                "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/analysis.json#/definitions/file"
+            },
+            "type": "array",
             "description": "Output generated by the pipeline run."
-        }, 
+        },
         "analysis_id": {
-            "type": "number", 
+            "type": "string",
             "description": "A unique ID for this analysis."
-        }, 
+        },
         "computational_method": {
-            "type": "uri", 
+            "type": "string",
             "description": "A URI to a versioned workflow and versioned execution environment in a GA4GH-compliant repository."
-        }, 
+        },
         "timestamp_start_utc": {
-            "type": "date-time", 
+            "type": "string",
+            "format": "date-time",
             "description": "Initial start time of the full pipeline."
-        }, 
+        },
         "core": {
-            "description": "Type and schema for this object.", 
+            "description": "Type and schema for this object.",
             "$ref": "https://raw.githubusercontent.com/HumanCellAtlas/metadata-schema/4.4.0/json_schema/core.json"
-        }, 
+        },
         "analysis_run_type": {
             "enum": [
-                "run", 
+                "run",
                 "copy-forward"
-            ], 
-            "type": "string", 
+            ],
+            "type": "string",
             "description": "Indicator of whether the analysis actually ran or was just copied forward as an optimization."
-        }, 
+        },
         "metadata_schema": {
-            "type": "string", 
+            "type": "string",
             "description": "The version of the metadata schemas used for the json files."
-        }, 
+        },
         "name": {
-            "type": "string", 
+            "type": "string",
             "description": "A short, descriptive name for the analysis that need not be unique."
         }
     }

diff --git a/src/simple_test.py b/src/simple_test.py
@@ -8,7 +8,7 @@
 """
 
 # a*.json skipped for now
-schemas = glob.glob('../json_schema/' +"[b-z]*.json")
+schemas = glob.glob('../json_schema/' +"[a-z]*.json")
 
 for s in schemas:
     print("Checking whether %s is a valid json schema" % s)