Repository navigation
Add native DynamoDB conversion and error mapping - #157
Draft
dkropachev wants to merge 7 commits into
Draft
dkropachev wants to merge 7 commits into
dkropachev wants to merge 7 commits into
Conversation
![github-code-quality[bot]](https://avatars.githubusercontent.com/u/9919?s=60&v=4){kind=small}
Convert the ten approved operation dictionaries directly into owned Rust SDK inputs before releasing the GIL. Generate typed converters from the compatibility manifest and locked SDK source, including legacy members, precise numeric strings and binary values. Project SDK outputs with optional members intact and use the final captured response for defaulted scalar presence and operation metadata. Decode compressed service responses in Rust while retaining original wire context. Map native validation, service failures and structured transport categories through cached botocore exception models without creating a Python client or invoking its serializers and transport. Keep interface version 1, the current Rust pin and staged operation boundary. Add embedded-Python differential checks and installed-wheel error coverage; renew source-bound startup evidence for the manifest changes. Refs #134 Part of #131.
Pop the Error member before checking it so result cleanup does not depend on Python assertion evaluation.
Project only required final-response presence paths, borrowing raw JSON branches so large attributes are not copied into a second value tree. Generate routes from the pinned model and SDK, retaining nested batch markers and typed output values. Keep empty and non-JSON HTTP failures as service errors with wire context. Preserve unknown integers, null error members and repeated headers while retaining runtime failures for successful-status SDK decoding and malformed HTTP. Leave modeled request maximums to the service, matching Python. Renew the native manifest startup anchor for the existing serde_json precision and raw-value features. Keep the Rust pin, Tokio features and interface version unchanged. Refs #134
Match the pinned SDK's single-member gzip and per-layer expansion limits. Keep TLS handshake I/O retryable and wrap transport failures only once. Preserve proxy service messages, explicit null messages and overflowing unknown JSON floats while retaining wire context and exact integers. Cover error-code precedence and these boundaries through the existing converter and local-socket SDK tests. Refs #134
Accept the pinned SDK's compression aliases and parameterized tokens. Preserve expansion-limit diagnostics when considering raw-deflate fallback, so oversized responses retain their actual failure reason. Match botocore's generic server-error handling for empty and HTML bodies, including HTML bytes outside UTF-8. Retain status/reason fields and wire context while keeping non-generic decode failures as protocol errors. Abort the TLS EOF test's peer task when its owner unwinds or times out, so a failed assertion cannot leave its listener on the shared runtime. Refs #134
dkropachev
force-pushed
the
feat/native-conversion-errors
branch
from
October 27, 2026 06:18
d255012 to
f82adcf
Compare
Retain unrelated fields in dictionary Error and ResponseMetadata payloads while canonical exception, HTTP and retry fields remain authoritative. Reserve optional metadata names even when genuine values are absent, and keep overwritten originals in the raw body. Document these exceptions. Normalize malformed modeled error shapes to protocol failures, preserve supported null variants and sanitize error-code suffixes before namespaces. Retain TLS protocol retries separately from their Python category and preserve the original request URL in transport failures. Keep test peers owned on unwind and isolate the metadata retry barrier from the SDK upload observer's periodic timers. Preserve its timeout and attempt assertions while qualifying the combined source. Refs #134
|
Important Draft PR not reviewedDraft PRs are not automatically reviewed by default.
To automatically review draft PRs, update your CodeRabbit configuration: reviews:
auto_review:
drafts: true
Comment |
Borrow path fragments through standard formatting instead of retaining an allocated prefix at every recursive conversion frame. Render the useful path only for a validation error, preserving the depth guard and owned SDK values without adding input limits or path context state. Report length bounds in plain language. Reject a second active member in modeled error AttributeValues while preserving supported null members, ignored protocol tags and single unknown variants. Refs #134
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #134. Part of #131.
The staged native backend has construction and transport but needs a conversion boundary for its ten approved DynamoDB operations. Convert low-level Python dictionaries directly into owned Rust SDK inputs, including legacy parameters, nested attributes, exact numeric strings, binary values and list-based sets. Reject missing or unknown members, invalid types and malformed AttributeValue unions with parameter paths before native execution.
Project typed SDK responses into fresh Python dictionaries, retaining absent and empty fields. Use each operation's captured final response for defaulted scalar presence and metadata, including retries that had no HTTP response. The scope includes compact parsing of required field-presence paths while skipping large attribute values and preserving absent, null and empty distinctions, including nested batch fields. JSON scanning and a second decompression of compressed responses remain. Decode compressed bodies in Rust while preserving original wire bytes and headers.
Export the reserved validation/service exceptions and structured transport categories. A private Python adapter constructs modeled or unknown botocore-compatible errors using cached service models without creating an admin client. Native service errors retain unknown body members and message casing, with decoded binary Items and cancellation reasons.
The scope includes preserving unrelated nested fields in dictionary-valued
ErrorandResponseMetadatabody members. Canonical exception, HTTP and retry field names remain reserved, even when their genuine values are absent. Non-dictionary values are replaced by generated containers; overwritten originals remain in raw.body.Decisions
ErrorandResponseMetadatapayloads. Exception, HTTP and retry field names remain reserved, even when their genuine values are absent. Generated values take precedence; replace non-dictionary payload values at these two names. Overwritten originals remain in raw.body. These are explicit exceptions to decoded-payload retention. Reason: retain useful payload data while keeping exception identity and transport/retry details authoritative. Source: the user selected option 1 for collision handling.Draft status
Native execute methods and factories remain staged; this PR does not claim end-to-end operation support. Exact-head native core qualification and the existing 20 interpreter/platform wheel cells are required before readiness. Final dependency-pin qualification remains separate #133 work.