Check provider-sourced pattern constraints

[jeskew]

Resolves #15325

Microsoft Reviewers: Open in CodeFlow

[github-actions]

Test this change out locally with the following install scripts (Action run 13509227187)

VSCode

• Mac/Linux

  bash <(curl -Ls https://aka.ms/bicep/nightly-vsix.sh) --run-id 13509227187

• Windows

  iex "& { $(irm https://aka.ms/bicep/nightly-vsix.ps1) } -RunId 13509227187"

Azure CLI

• Mac/Linux

  bash <(curl -Ls https://aka.ms/bicep/nightly-cli.sh) --run-id 13509227187

• Windows

  iex "& { $(irm https://aka.ms/bicep/nightly-cli.ps1) } -RunId 13509227187"

[github-actions]

Dotnet Test Results

    78 files   -     39      78 suites   - 39   31m 53s ⏱️ - 15m 32s
11 829 tests  -     11  11 829 ✅  -     11  0 💤 ±0  0 ❌ ±0 
27 416 runs   - 13 697  27 416 ✅  - 13 697  0 💤 ±0  0 ❌ ±0 

Results for commit 214110b. ± Comparison against base commit 3d03ba0.

This pull request removes 1790 and adds 613 tests. Note that renamed tests count towards both.

		nestedProp1: 1
		nestedProp2: 2
		prop1: true
		prop2: false
	1
	2
	\$'")
	prop1: true
	prop2: false
…

Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000
�ӽ\u000e�0\u0010\u0007��>\u0005OP�J�����+4J�G(\u0006J$1��e0q�� ����˥�k�2��~[�Cٴ�8S�\u0014s��h=z>@B��\u0002\u0002�H	���O2�k�m�(K���T�X���\u0002Y�Ό�LBnr�j\u0013j��\u001a&W�E�.��9ynk74TS
k?4\u001a����Zѧ�\u0003(�����"
�GM!��d������=A\u0014EQ��'� ��\u0000\u000c\u0000\u0000,"The path: index.json was not found in artifact contents")
Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000
�Խ\u000e�0\u0010\u0007��>E�\u0003�\u0016�+������\u0000\u0015Έ\u0011$�	���-�q���ab�c�^�����i�hR�j�t\u0018����[ e�zGHM�~�k�4!�\u001d}�\u001e��1�\u001de��~P\u0010Q�d9�BK.C\u0000\u0005\u000c8@\u0014ɕ�%C5���i�qV�زK}+��|�}�c:\u0003>�7�\u0015}�?\u000f%����B\u0007\u0001���R`�?K&�<�\u000f�\u001d
lKL\u001aLw�\u001f��6�d�5>���q\u001cǙ�\u000bS?:�\u0000\u000c\u0000\u0000,"Value cannot be null. (Parameter 'source')")
Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000
���\u000b�0\u0014\u0007��+�?`nso��C�!#,\u0008���A\u0006Z���?�<D\u0017��?��9n\u000f���}�۫z�U��\u0012����Q߈!8o]oP.\u0011�o\u0004\u0011	 \u0011r��;i�(+U�V�8�\u0007����4�!��p_\u0008\u0010\u0018�\u0004\u0002���޹k�\u0006�]U�0�\u0013]�ky˛�|�kZ\u001d>�7�\u0013}�?a\u001c�\u0019\u0000�JƄɿ\u0000\u0000��Q2�����Q|\n��is����1\u0013x�,˲��\u0000B��r\u0000\u000c\u0000\u0000,"'7' is an invalid end of a number. Expected a delimiter. Path: $.INVALID_JSON | LineNumber: 0 | BytePositionInLine: 20.")
Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000\u0003�ӽ\u000e�0\u0010\u0007��>\u0005OP��\u001eg\u0007vG_�Q\u0012?B1\u0002������\u0001�`b�c������t�M��ŵ�\u001aئRL
�̘��\u001e\u0012
�\u0017\u0010�D�v�M�ɀ�n�5�2Ǭ\u001f�։k�e�#\u001b0:# i	�U�
��h
��k\u000ey�Ͼ�yy�+�7�c
K?4\u001a���ߚ�)�\u0000J�abd\u0005\u001aC�QS��,�����\u001fKo\u0010EQ\u0014-�	\u0015\u000eK�\u0000\u000c\u0000\u0000,"The path: index.json was not found in artifact contents")
Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000\u0003�Խ
�0\u0010\u0007��>E�\u0003��6\u001fV� 8X�*\u0008�\u0012l�
��V(�\uda58\udcf8���C0�19�\u000b���ݫz�U���� \u0003����!\u0018k=oP&\u0011�o�@rA\u0011�ux��*�(c��A�\u001c�*�tH%\u0003�\u000b\u000e�\u0004�\u0007A��ܝ��h�諸�i��\�[ޔg]�S?������z|�?x\u000c�\u001d�%�\u001e�\u0014����(����?�(>.���9�bg��\u0006�z&˲,kx/硖W\u0000\u000c\u0000\u0000,"'7' is an invalid end of a number. Expected a delimiter. Path: $.INVALID_JSON | LineNumber: 0 | BytePositionInLine: 20.")
Bicep.Core.IntegrationTests.AzTypesViaRegistryTests ‑ Bicep_compiler_handles_corrupted_extension_package_gracefully (\u001f�\u0008\u0000\u0000\u0000\u0000\u0000\u0000\u0003��K
�0\u0010\u0006�=E�\u0001�ͣ\u0015�w#��\u0000�\u001d�bki+\u0014Ļ�.�M��>\u0004�-3\u00033!�a�i�hR�j\u0016��֌�
,%D�y�\u000bM�~\u0003\u0002Z*N\u0008mGߤǽnLeW�c�\u000f�Cj�,ǘk\u0001"P\u0012$�B�|%W��\u000c�xDKӜ�H�e��Vt��P���t\u0006|�o�\u0019��\u000f� �\u000f��k\u001f\u0002N@I���ϒ�?���z�\u0002�\u0012�\u0006�\u001d�G��
=�k�ϥws\u001c�q��\u0002�L�A\u0000\u000c\u0000\u0000,"Value cannot be null. (Parameter 'source')")
Bicep.Core.IntegrationTests.DecompilationTests ‑ Decompiler_handles_banned_function_replacement ("createArray(1, 2, 3)","array","[
  1
  2
  3
]")
Bicep.Core.IntegrationTests.DecompilationTests ‑ Decompiler_handles_banned_function_replacement ("createObject('key', 'value')","object","{
  key: 'value'
}")
Bicep.Core.IntegrationTests.DecompilationTests ‑ Decompiler_handles_strings_with_newlines ("
","\n")
Bicep.Core.IntegrationTests.DecompilationTests ‑ Decompiler_handles_strings_with_newlines ("
","\r\n")
…

♻️ This comment has been updated with latest results.

[jeskew]

I'm unsure if we should be special-casing patterns that come from the az provider. The behavioral differences between standard regexes and the ECMAScript dialect are subtle [0], and I would be surprised if Azure RPs are using the ECMAScript dialect in practice.

Needing to track whether a pattern came from the Az provider types (and thus needs to be interpreted per the ECMAScript dialect) or not (and therefore needs to be interpreted with the standard dialect and non-backtracking engine) adds a fair amount of complexity.

[0]: These would be:

• \w doesn't match alphabetic characters in languages that use non-Latin alphabets
• self-referencing capture groups are supported
• the \ means slightly different things when it's not referring to a capture group

Cf https://learn.microsoft.com/en-us/dotnet/standard/base-types/regular-expression-options#ecmascript-matching-behavior