Add --oss flag to the ECS generator script (elastic#991)

CHANGELOG.next.md

@@ -43,6 +43,7 @@ Thanks, you're awesome :-) -->
 * Introduced `--strict` flag to perform stricter schema validation when running the generator script. #937
 * Added check under `--strict` that ensures composite types in example fields are quoted. #966
 * Added `ignore_above` and `normalizer` support for keyword multi-fields. #971
+* Added `--oss` flag for users who want to generate ECS templates for use on OSS clusters. #991
 
 #### Improvements
 

USAGE.md

@@ -29,6 +29,7 @@ relevant artifacts for their unique set of data sources.
     + [Subset](#subset)
     + [Ref](#ref)
     + [Mapping & Template Settings](#mapping--template-settings)
+    + [OSS](#oss)
     + [Strict Mode](#strict-mode)
     + [Intermediate-Only](#intermediate-only)
 
@@ -295,14 +296,38 @@ The `--template-settings` argument defines [index level settings](https://www.el
 
 For `template.json`, the `mappings` object is left empty: `{}`. Likewise the `properties` object remains empty in the `mapping.json` example. This will be filled in automatically by the script.
 
+#### OSS
+
+**IMPORTANT**: This feature is unnecessary for most users. Our default free distribution
+comes with the Elastic Basic license, and supports all data types used by ECS.
+Learn more about our licenses [here](https://www.elastic.co/subscriptions).
+
+Users that want to use the open source version of Elasticsearch do not have access to the basic data types.
+However some of these types have an OSS replacement that can be used instead, without too much loss of functionality.
+
+This flag performs a best effort fallback, replacing basic data types with their OSS replacement.
+
+Indices using purely OSS types will benefit from the normalization of ECS, but may be missing on some of the added functionality of these basic types.
+
+Current fallbacks applied by this flag are:
+
+- `wildcard` => `keyword`
+- `version` => `keyword`
+
+Usage:
+
+```
+$ python scripts/generator.py --oss
+```
+
 #### Strict Mode
 
 The `--strict` argument enables "strict mode". Strict mode performs a stricter validation step against the schema's contents.
 
 Basic usage:
 
 ```
-$ python/generator.py --strict
+$ python scripts/generator.py --strict
 ```
 
 Strict mode requires the following conditions, else the script exits on an exception:

scripts/generator.py

@@ -12,6 +12,7 @@
 from generators import intermediate_files
 
 from schema import loader
+from schema import oss
 from schema import cleaner
 from schema import finalizer
 from schema import subset_filter
@@ -41,6 +42,8 @@ def main():
     # ecs_helpers.yaml_dump('ecs.yml', fields)
 
     fields = loader.load_schemas(ref=args.ref, included_files=args.include)
+    if args.oss:
+        oss.fallback(fields)
     cleaner.clean(fields, strict=args.strict)
     finalizer.finalize(fields)
     fields = subset_filter.filter(fields, args.subset, out_dir)
@@ -60,20 +63,21 @@ def main():
 
 def argument_parser():
     parser = argparse.ArgumentParser()
-    parser.add_argument('--intermediate-only', action='store_true',
-                        help='generate intermediary files only')
+    parser.add_argument('--ref', action='store', help='git reference to use when building schemas')
     parser.add_argument('--include', nargs='+',
                         help='include user specified directory of custom field definitions')
     parser.add_argument('--subset', nargs='+',
                         help='render a subset of the schema')
-    parser.add_argument('--out', action='store', help='directory to store the generated files')
-    parser.add_argument('--ref', action='store', help='git reference to use when building schemas')
+    parser.add_argument('--out', action='store', help='directory to output the generated files')
     parser.add_argument('--template-settings', action='store',
                         help='index template settings to use when generating elasticsearch template')
     parser.add_argument('--mapping-settings', action='store',
                         help='mapping settings to use when generating elasticsearch template')
+    parser.add_argument('--oss', action='store_true', help='replace basic data types with oss ones where possible')
     parser.add_argument('--strict', action='store_true',
-                        help='enforce stricter checking at schema cleanup')
+                        help='enforce strict checking at schema cleanup')
+    parser.add_argument('--intermediate-only', action='store_true',
+                        help='generate intermediary files only')
     args = parser.parse_args()
     # Clean up empty include of the Makefile
     if args.include and [''] == args.include:

scripts/schema/oss.py

@@ -0,0 +1,29 @@
+# This script performs a best effort fallback of basic data types to equivalent
+# OSS data types.
+# Note however that not all basic data types have an OSS replacement.
+#
+# The way this script is currently written, it has to be run on the fields *before*
+# the cleaner script applies defaults, as there's no concept of defaults here.
+# But since it navigates using the visitor script, it can easily be moved around
+# in the chain, provided we add support for defaults as well.
+#
+# For now, no warning is output on basic fields that don't have a fallback.
+# This could be improved if ECS starts using such types.
+
+from schema import visitor
+
+TYPE_FALLBACKS = {
+    'wildcard': 'keyword',
+    'version': 'keyword'
+}
+
+
+def fallback(fields):
+    """Verify all fields for basic data type usage, and fallback to an OSS equivalent if appropriate."""
+    visitor.visit_fields(fields, field_func=perform_fallback)
+
+
+def perform_fallback(field):
+    """Performs a best effort fallback of basic data types to equivalent OSS data types."""
+    if field['field_details']['type'] in TYPE_FALLBACKS.keys():
+        field['field_details']['type'] = TYPE_FALLBACKS[field['field_details']['type']]

scripts/tests/unit/test_schema_oss.py

@@ -0,0 +1,35 @@
+import os
+import pprint
+import sys
+import unittest
+
+sys.path.append(os.path.join(os.path.dirname(__file__), '../..'))
+
+from schema import oss
+from schema import visitor
+
+
+class TestSchemaOss(unittest.TestCase):
+
+    def setUp(self):
+        self.maxDiff = None
+
+    def test_wildcard_fallback(self):
+        field = {'field_details': {'name': 'myfield', 'type': 'wildcard'}}
+        oss.perform_fallback(field)
+        self.assertEqual('keyword', field['field_details']['type'])
+
+    def test_version_fallback(self):
+        field = {'field_details': {'name': 'myfield', 'type': 'version'}}
+        oss.perform_fallback(field)
+        self.assertEqual('keyword', field['field_details']['type'])
+
+    def test_basic_without_fallback(self):
+        field = {'field_details': {'name': 'myfield', 'type': 'histogram'}}
+        oss.perform_fallback(field)
+        self.assertEqual('histogram', field['field_details']['type'])
+
+    def test_oss_no_fallback(self):
+        field = {'field_details': {'name': 'myfield', 'type': 'keyword'}}
+        oss.perform_fallback(field)
+        self.assertEqual('keyword', field['field_details']['type'])