Separate validation into two phases

 - snippet contents (always) and full-repo (optional; includes sample files, bad words, and SPDX checks) (#22)

Author: DavidSouther

action.yaml

@@ -4,9 +4,9 @@ inputs:
   root:
     description: "Root of the repository checkout to validate metadata within"
     default: ${{ github.workspace }}
-  check_spdx:
-    description: "Check for two-line SDPX header conformance"
-    default: "False"
+  doc_gen_only:
+    description: "Only perform extended validation on doc_gen snippets"
+    default: "True"
 runs:
   using: "composite"
   steps:
@@ -18,5 +18,5 @@ runs:
       run: pip install -e "${{ github.action_path }}"
       shell: bash
     - name: Run validator
-      run: python3 -m "aws_doc_sdk_examples_tools.validate" --root "${{ inputs.root }}" --check-spdx "${{ inputs.check_spdx }}"
+      run: python3 -m "aws_doc_sdk_examples_tools.validate" --root "${{ inputs.root }}" --doc_gen_only "${{ inputs.doc_gen_only }}"
       shell: bash

aws_doc_sdk_examples_tools/doc_gen.py

@@ -16,12 +16,13 @@
 )
 from .metadata_errors import MetadataErrors, MetadataError
 from .metadata_validator import validate_metadata
-from .project_validator import check_files, verify_sample_files, ValidationConfig
+from .project_validator import ValidationConfig
 from .sdks import Sdk, parse as parse_sdks
 from .services import Service, parse as parse_services
 from .snippets import (
     Snippet,
     collect_snippets,
+    collect_snippet_files,
     validate_snippets,
 )
 
@@ -53,6 +54,9 @@ def collect_snippets(
         if snippets_root is None:
             snippets_root = self.root
         snippets, errs = collect_snippets(snippets_root, prefix)
+        collect_snippet_files(
+            self.examples.values(), snippets=snippets, errors=errs, root=self.root
+        )
         self.snippets = snippets
         self.errors.extend(errs)
 
@@ -193,19 +197,17 @@ def for_root(self, root: Path, config: Optional[Path] = None) -> "DocGen":
     def from_root(cls, root: Path, config: Optional[Path] = None) -> "DocGen":
         return DocGen.empty().for_root(root, config)
 
-    def validate(self, check_spdx: bool):
+    def validate(self):
         for sdk in self.sdks.values():
             sdk.validate(self.errors)
         for service in self.services.values():
             service.validate(self.errors)
-        check_files(self.root, check_spdx, self.validation, self.errors)
-        verify_sample_files(self.root, self.validation, self.errors)
         validate_metadata(self.root, self.errors)
         validate_no_duplicate_api_examples(self.examples.values(), self.errors)
         validate_snippets(
             [*self.examples.values()],
             self.snippets,
-            self.snippet_files,
+            self.validation,
             self.errors,
             self.root,
         )

aws_doc_sdk_examples_tools/project_validator.py

@@ -52,7 +52,6 @@ def clone(self):
 
 def check_files(
     root: Path,
-    do_check_spdx: bool,
     validation: ValidationConfig,
     errors: MetadataErrors,
 ):
@@ -78,8 +77,7 @@ def check_files(
 
         verify_no_deny_list_words(file_contents, file_path, errors)
         verify_no_secret_keys(file_contents, file_path, validation, errors)
-        if do_check_spdx:
-            verify_spdx(file_contents, file_path, errors)
+        verify_spdx(file_contents, file_path, errors)
 
     print(f"{file_count} files scanned in {root}.\n")
 

aws_doc_sdk_examples_tools/snippets.py

@@ -3,14 +3,19 @@
 
 from dataclasses import dataclass
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Set, Tuple
+from typing import Any, Dict, Iterable, List, Optional, Set, Tuple
 from shutil import copyfile
 import re
 
 from .validator_config import skip
 from .file_utils import get_files, clear
 from .metadata import Example
 from .metadata_errors import MetadataErrors, MetadataError
+from .project_validator import (
+    verify_no_deny_list_words,
+    verify_no_secret_keys,
+    ValidationConfig,
+)
 
 SNIPPET_START = "snippet-start:["
 SNIPPET_END = "snippet-end:["
@@ -171,6 +176,49 @@ def collect_snippets(
     return snippets, errors
 
 
+def collect_snippet_files(
+    examples: Iterable[Example],
+    snippets: Dict[str, Snippet],
+    errors: MetadataErrors,
+    root: Path,
+):
+    for example in examples:
+        for lang in example.languages:
+            language = example.languages[lang]
+            for version in language.versions:
+                for excerpt in version.excerpts:
+                    for snippet_file in excerpt.snippet_files:
+                        if not (root / snippet_file).exists():
+                            # Ensure all snippet_files exist
+                            errors.append(
+                                MissingSnippetFile(
+                                    file=example.file,
+                                    snippet_file=snippet_file,
+                                    id=f"{lang}:{version.sdk_version}",
+                                )
+                            )
+                            continue
+                        if re.search(win_unsafe_re, str(snippet_file)):
+                            errors.append(
+                                WindowsUnsafeSnippetFile(
+                                    file=example.file,
+                                    snippet_file=snippet_file,
+                                    id=f"{lang}:{version.sdk_version}",
+                                )
+                            )
+                            continue
+                        name = str(snippet_file).replace("/", ".")
+                        with open(root / snippet_file, encoding="utf8") as file:
+                            code = file.readlines()
+                            snippets[name] = Snippet(
+                                id=name,
+                                file=str(snippet_file),
+                                line_start=0,
+                                line_end=len(code),
+                                code="\n".join(file),
+                            )
+
+
 @dataclass
 class MissingSnippet(MetadataError):
     tag: Optional[str] = None
@@ -200,9 +248,9 @@ def message(self):
 
 
 def validate_snippets(
-    examples: List[Example],
+    examples: Iterable[Example],
     snippets: Dict[str, Snippet],
-    snippet_files: Set[str],
+    validation: ValidationConfig,
     errors: MetadataErrors,
     root: Path,
 ):
@@ -221,24 +269,10 @@ def validate_snippets(
                                     tag=snippet_tag,
                                 )
                             )
-    for snippet_file in excerpt.snippet_files:
-        if not (root / snippet_file).exists():
-            # Ensure all snippet_files exist
-            errors.append(
-                MissingSnippetFile(
-                    file=example.file,
-                    snippet_file=snippet_file,
-                    id=f"{lang}:{version.sdk_version}",
-                )
-            )
-        if re.search(win_unsafe_re, str(snippet_file)):
-            errors.append(
-                WindowsUnsafeSnippetFile(
-                    file=example.file,
-                    snippet_file=snippet_file,
-                    id=f"{lang}:{version.sdk_version}",
-                )
-            )
+
+    for snippet in snippets.values():
+        verify_no_deny_list_words(snippet.code, root / snippet.file, errors)
+        verify_no_secret_keys(snippet.code, root / snippet.file, validation, errors)
 
 
 def write_snippets(root: Path, snippets: Dict[str, Snippet], check: bool = False):

aws_doc_sdk_examples_tools/validate.py

@@ -7,6 +7,7 @@
 from sys import exit
 
 from .doc_gen import DocGen
+from .project_validator import check_files, verify_sample_files
 
 
 def main():
@@ -16,25 +17,22 @@ def main():
         default=f"{Path(__file__).parent.parent.parent}",
         help="The root path from which to search for files to check. The default is the root of the git repo (two up from this file).",
     )
-    # parser.add_argument(
-    #     "--doc-gen",
-    #     default=f"{Path(__file__).parent.parent.parent / '.doc_gen'}",
-    #     help="The folder that contains schema and metadata files. The default is .doc_gen in the root of this repo.",
-    #     required=False,
-    # )
     parser.add_argument(
-        "--check-spdx",
+        "--doc_gen_only",
         type=literal_eval,
-        default=False,
-        help="Verify all files start with SPDX header",
+        default=True,
+        help="Only perform extended validation on snippet contents",
         required=False,
     )
     args = parser.parse_args()
     root_path = Path(args.root).resolve()
 
     doc_gen = DocGen.from_root(root=root_path)
     doc_gen.collect_snippets(snippets_root=root_path)
-    doc_gen.validate(args.check_spdx)
+    doc_gen.validate()
+    if not args.doc_gen_only:
+        check_files(doc_gen.root, doc_gen.validation, doc_gen.errors)
+        verify_sample_files(doc_gen.root, doc_gen.validation, doc_gen.errors)
 
     error_count = len(doc_gen.errors)
     if error_count > 0: