Prototype it with simple functions

Author: chbrandt

validate.py

@@ -4,13 +4,14 @@
 _here = os.path.abspath(__file__)
 sys.path.insert(0, _here)
 
+from validator import Validator
+
 
 config = {
     'geo_path': 'tmp/PM-MAR-MS-Arsinoes_01/vector/PM-MAR-MS-Arsinoes_01.gpkg',
     'geopackage_tables': ['units', 'contacts']
 }
 
-from validator import Validator
 
 def run(config, log_path):
     validator = Validator(config, log_path)

validator/validator.py

@@ -2,6 +2,54 @@
 
 from tasks import *
 
+
+def check_exist(filename):
+    import os
+    return os.path.exists(filename)
+
+def check_is_gpkg(filename):
+    import geopandas
+    try:
+        geopandas.read_file(filename, driver='GPKG')
+    except:
+        return False
+    return True
+
+def check_layer(pkg, layer):
+    out = layer in pkg
+    return out
+
+def check_gpkg_layers(filename, layers=[]):
+    pkg = load_gpkg(filename)
+    for layer in layers:
+        out += check_layer(pkg, layer)
+    return out
+
+def check_layer_columns(gdf, columns):
+    out = []
+    for column in columns:
+        out += column in gdf
+    return out
+
+def check_layers_columns(filename, layers={}):
+    pkg = load_gpkg(filename)
+    gdf = pkg[layer]
+    for layer, columns in layers.items():
+        out += check_layer_columns(gdf, columns)
+    return out
+
+def check_column_content(gdf, column, rules):
+    for rule in rules:
+        out += gdf[column].apply(rule)
+    return out
+
+def check_columns_content(filename, layer, columns={}):
+    pkg = load_gpkg(filename)
+    gdf = pkg[layer]
+    for column, rules in columns.items():
+        out += check_column_content(gdf, column, rules)
+    return out
+
 class Validator(object):
     tasks = None
 
@@ -18,7 +66,12 @@ def __init__(self, config, log_path):
 
         self.config = config
 
-        tasks = [TaskVector, TaskVectorTables]
+        # tasks = [TaskVector, TaskVectorTables]
+        tasks = [check_exist(filename),
+                 check_is_gpkg(filename),
+                 check_gpkg_layers(filename),
+                 check_layers_columns(filename),
+                 check_columns_content(filename)]
         self.tasks = self.set_tasks(tasks)
 
     def set_tasks(self, tasks):

validator/vector/geopackage_layers.py

@@ -0,0 +1,16 @@
+
+_component = __file__.split('.')[0]
+_schemafile = '.'.join([_component, 'schema', 'json'])
+
+with open(_schemafile) as f:
+    import json
+    schema = json.load(f)
+
+def validate(gpkg_path):
+    import fiona
+    data = fiona.listlayers(gpkg_path)
+
+    import jsonschema
+    res = jsonschema.validate(data, schema)
+
+    return res

validator/vector/geopackage_layers.schema.json

@@ -1,8 +1,13 @@
 {
-    "$schema": "http://json-schema.org/draft-07/schema#",
+"$schema": "http://json-schema.org/draft-07/schema#",
+
+"title": "Match Geopackage layers list",
+
+"$comment": "Works for, e.g., ['units','contacts','bla'], but not for ['bla','units','contacts']",
+"$example": ["units","contacts","another-column"],
 
     "definitions": {
-        "columns": {
+        "layers": {
             "type": "string",
             "enum": ["units", "contacts"]
         }
@@ -11,13 +16,11 @@
     "type": "array",
     "items": [
         {
-            "$ref": "#/definitions/columns"
-        },
-        {
-            "$ref": "#/definitions/columns"
+            "$ref": "#/definitions/layers"
         }
     ],
     "minItems": 2,
+
     "uniqueItems": true,
     "additionalItems": { "type": "string" }
 }