Add flag to suppress validation errors to simplify codegen and back c…

…ompat checks (#300)

semantic-conventions/CHANGELOG.md

@@ -32,6 +32,10 @@ Please update the changelog as part of any significant pull request.
   ([#266](https://github.com/open-telemetry/build-tools/pull/266))
 - Sort attribute tables by requirement level and attribute name
   ([#260](https://github.com/open-telemetry/build-tools/pull/260))
+- Support suppressing all validation errors via flag that allows to
+  parse previous versions of semantic conventions for backward compatibility checks
+  and use code generation improvements on older semantic convention version.
+  ([#300](https://github.com/open-telemetry/build-tools/pull/300))
 
 ## v0.23.0
 

semantic-conventions/README.md

@@ -123,6 +123,12 @@ The following checks are performed:
     This check does not take into account opt-in attributes. Adding new attributes to metric is not always breaking,
     so it's considered non-critical and it's possible to suppress it with `--ignore-warnings`
 
+Previous versions of semantic conventions are not always compatible with newer versions of build-tools. You can suppress validation errors by adding `--continue-on-validation-errors` flag:
+
+```bash
+docker run --rm otel/semconvgen --yaml-root {yaml_folder} --continue-on-validation-errors compatibility --previous-version {semconv version}
+```
+
 ## Code Generator
 
 The image supports [Jinja](https://jinja.palletsprojects.com/en/2.11.x/) templates to generate code from the models.
@@ -171,13 +177,13 @@ Finally, additional value can be passed to the template in form of `key=value` p
 comma using the `--parameters [{key=value},]+` or `-D` flag.
 
 Generating code from older versions of semantic conventions with new tooling is, in general, not supported.
-However in some cases minor incompatibilities in semantic conventions can be ignored by setting `--strict-validation` flag to `false`
+However in some cases minor incompatibilities in semantic conventions can be suppressed by adding `--continue-on-validation-errors` flag:
 
 ```bash
 docker run --rm \
   otel/semconvgen:$GENERATOR_VERSION \
   --yaml-root /source \
-  `--strict-validation false`
+  --continue-on-validation-errors \
   code \
   ...other parameters...
 ```

semantic-conventions/src/opentelemetry/semconv/main.py

@@ -28,6 +28,7 @@
     CONVENTION_CLS_BY_GROUP_TYPE,
     SemanticConventionSet,
 )
+from opentelemetry.semconv.model.utils import ValidationContext
 from opentelemetry.semconv.templating.code import CodeRenderer
 from opentelemetry.semconv.templating.compatibility import CompatibilityChecker
 from opentelemetry.semconv.templating.markdown import MarkdownRenderer
@@ -42,7 +43,7 @@ def parse_semconv(
     for file in sorted(files):
         if not file.endswith(".yaml") and not file.endswith(".yml"):
             parser.error(f"{file} is not a yaml file.")
-        semconv.parse(file, strict_validation)
+        semconv.parse(file, ValidationContext(file, strict_validation))
     semconv.finish()
     if semconv.has_error():
         sys.exit(1)
@@ -73,7 +74,11 @@ def main():
     args = parser.parse_args()
     check_args(args, parser)
     semconv = parse_semconv(
-        args.yaml_root, args.exclude, args.debug, args.strict_validation, parser
+        args.yaml_root,
+        args.exclude,
+        args.debug,
+        not args.continue_on_validation_errors,
+        parser,
     )
     semconv_filter = parse_only_filter(args.only, parser)
     filter_semconv(semconv, semconv_filter)
@@ -93,9 +98,9 @@ def main():
 def process_markdown(semconv, args):
     options = MarkdownOptions(
         check_only=args.md_check,
-        disable_stable_badge=args.md_disable_stable,
-        disable_experimental_badge=args.md_disable_experimental,
-        disable_deprecated_badge=args.md_disable_deprecated,
+        disable_stable_badge=args.md_disable_stable_badge,
+        disable_experimental_badge=args.md_disable_experimental_badge,
+        disable_deprecated_badge=args.md_disable_deprecated_badge,
         break_count=args.md_break_conditional,
         exclude_files=exclude_file_list(args.markdown_root, args.exclude),
     )
@@ -106,7 +111,11 @@ def process_markdown(semconv, args):
 def check_compatibility(semconv, args, parser):
     prev_semconv_path = download_previous_version(args.previous_version)
     prev_semconv = parse_semconv(
-        prev_semconv_path, args.exclude, args.debug, args.strict_validation, parser
+        prev_semconv_path,
+        args.exclude,
+        args.debug,
+        not args.continue_on_validation_errors,
+        parser,
     )
     compatibility_checker = CompatibilityChecker(semconv, prev_semconv)
     problems = compatibility_checker.check()
@@ -220,12 +229,6 @@ def add_md_parser(subparsers):
         required=False,
         action="store_true",
     )
-    parser.add_argument(
-        "--check-compat",
-        help="Check backward compatibility with previous version of semantic conventions.",
-        type=str,
-        required=False,
-    )
     parser.add_argument(
         "--md-disable-stable-badge",
         help="Removes badges from attributes marked as stable.",
@@ -306,11 +309,14 @@ def setup_parser():
         help="YAML file containing a Semantic Convention",
     )
     parser.add_argument(
-        "--strict-validation",
-        help="Fail on non-critical yaml validation issues.",
+        "--continue-on-validation-errors",
+        help="""Continue parsing on yaml validation issues.
+        Should not be used to generate or validate semantic conventions.
+        Useful when running backward compatibility checks or using newer
+        tooling version to generate code for released semantic conventions.""",
         required=False,
-        default=True,
-        action="store_false",
+        default=False,
+        action="store_true",
     )
     subparsers = parser.add_subparsers(dest="flavor")
     add_code_parser(subparsers)

semantic-conventions/src/opentelemetry/semconv/model/constraints.py

@@ -17,7 +17,6 @@
 
 from ruamel.yaml.comments import CommentedSeq
 
-from opentelemetry.semconv.model.exceptions import ValidationError
 from opentelemetry.semconv.model.semantic_attribute import SemanticAttribute
 from opentelemetry.semconv.model.utils import validate_values
 
@@ -64,20 +63,20 @@ class Include:
     semconv_id: str
 
 
-def parse_constraints(yaml_constraints):
+def parse_constraints(yaml_constraints, validation_ctx):
     """This method parses the yaml representation for semantic convention attributes
     creating a list of Constraint objects.
     """
     constraints = ()
     allowed_keys = ("include", "any_of")
     for constraint in yaml_constraints:
-        validate_values(constraint, allowed_keys)
+        validate_values(constraint, allowed_keys, validation_ctx)
         if len(constraint.keys()) > 1:
             position = constraint.lc.data[list(constraint)[1]]
             msg = (
                 "Invalid entry in constraint array - multiple top-level keys in entry."
             )
-            raise ValidationError.from_yaml_pos(position, msg)
+            validation_ctx.raise_or_warn(position, msg, None)
         if "include" in constraint:
             constraints += (Include(constraint.get("include")),)
         elif "any_of" in constraint:

semantic-conventions/src/opentelemetry/semconv/model/exceptions.py

@@ -19,19 +19,15 @@ class ValidationError(Exception):
         line -- line in the file where the error occurred
         column -- column in the file where the error occurred
         message -- reason of the error
+        fqn -- identifier of the node that contains the error
     """
 
-    @classmethod
-    def from_yaml_pos(cls, pos, msg):
-        # the yaml parser starts counting from 0
-        # while in document is usually reported starting from 1
-        return cls(pos[0] + 1, pos[1] + 1, msg)
-
-    def __init__(self, line, column, message):
-        super().__init__(line, column, message)
+    def __init__(self, line, column, message, fqn):
+        super().__init__(line, column, message, fqn)
         self.message = message
         self.line = line
         self.column = column
+        self.fqn = fqn
 
     def __str__(self):
-        return f"{self.message} - @{self.line}:{self.column}"
+        return f"{self.message} - @{self.line}:{self.column} ('{self.fqn}')"