* conditional serialization add only those property which user wants to configure.
* fixed the samples
* fixed the sample String to string issue.
* fixed the sample String to string issue.
* updated the sample
* Added unit test for conditional serialization
* update samples
* remove trailing spaces, update samples
* remove files
* add back files
* build the project in the ci
Co-authored-by: William Cheng <wing328hk@gmail.com>
* first commit: add cli option for saga and records. Added dummy sagas.mustache test file.
* More progress with default values. First prototype for isEntity and isUniqueId.
* record generation complete
* record generation complete
* progress with saga generation
* progress with saga generation
* first fully working saga generation
* merge with latest master
* removed unneeded "items" properties.
* moved global CodegenModel modifications into subclass ExtendedCodegenModel used exclusively by TypescriptFetchClient. Adding missing samples files.
* moved global CodegenOperation modifications into subclass ExtendedCodegenOperation used exclusively by TypescriptFetchClient.
* moved global CodegenProperty modifications into subclass ExtendedCodegenProperty used exclusively by TypescriptFetchClient.
* moved global CodegenParameter modifications into subclass ExtendedCodegenParameter used exclusively by TypescriptFetchClient.
* added the missing "allSagas" export.
* renamed & reworked "meta data response" flags to a more useful general concept of "operation return passthrough"
* added vendor flag keepAsJSObject as escape hatch to support circular dependencies in models and other special cases. Also fixed issues with default values for some records properties.
* added autodetection for passthrough to simplify standardised specs.
* fix small issue with passthrough void
* fix small issues with passthrough void and missing passthrough imports in some cases. Fix issues with enum default values.
* fix small issues with passthrough void and missing passthrough imports in some cases. Fix issues with enum default values.
* Added "reservedRecordField" feature to support remapping fields names that cannot be used in Records. Added missing export to record: toApi().
* added uniqueId inference. Fix small generation when uniqueId property is an array.
* removed feature "reservedRecordField" and replaced it with existing built-in "reserved words" feature. Fix minor issues with typings in generated files.
* Changed api recType names to make them less likely to cause name conflicts. Added generated ApiEntities (record, reducer & selector) files.
* Moved location of ApiEntities related files and fix issues with exports.
* - merge latest master
- renamed fake test apis to better fit the "pet theme"
- added mode for "SourceOnlyLibrary" (same as used in codegen typescript jquery)
* - missing ganarate sampless
* - Modified way to export apiEntitiesSelectpr to reduce typescript analysis time for consuming project. Removed tab characters in mustache files. Reformat code for TypeScriptFetchClientCodegen to try to remove false positive for tabs vs spaces.
* - added markErrorsAsHandled property to api sagas. Increased typescript version to address some typing errors on library build.
* - fix bug in saga interfaces. Upgraded to typescript "strict" mode to ensure proper typechecking info is generated.
* - added optional id for apiEntity selectors. Added toInlined() support to convert an entity to an inlined model recursively.
* - minor tweak for apiEntitySelector to accept null id
* - minor tweak for apiEntitySelector
* - runned ensure up to date.
* Revert "- runned ensure up to date."
This reverts commit ea9b4aed
* - runned ensure up to date.
* - runned ensure up to date.
Co-authored-by: Bruno Flamand <bflamand@stingray.com>
According to the [documentation](https://openapi-generator.tech/docs/generators/go-server) the go-server generator should support an `addResponseHeaders` option but a template uses an `addResponseHeader` option in some places which this PR fixes.
Closes#9795
Signed-off-by: Max Goltzsche <max.goltzsche@gmail.com>
* Continuing from #1317 and its PRs for pistache server string enum code generation;
* A class that has an `anyOf` specification, in cpp side will have no members: in stead it should have a member having the type `classname_anyOf`
* Thus, Its `==` operator is not present or wrongly filled
* An string enum, should have a better usage, hence the `setEnumValue`
* this PR, is a brigde between `stringenumclassname_anyOf` and `stringenumclassname`
* `anyOf` specification is not just about `Enums`, so a better handling is needed from mustache templates, hence new template model parameter `isStringEnumContainer`
* PR fix: muttleyxd: `double semicolon`
* PR fix: wing328: `I think std::string is C++ only. What about adding x-is-string-enum-container instead in the postProcessModel operation in the C++ pistache server generator?`
* PR fix: wing328: `I think std::string is C++ only...` after fix get latest codes and then generate samples
Co-authored-by: Mehmet Fatih <mfyuce@netas.com.tr>
Use Gradle 6.9 and Kotlin 1.4.32. Generate Paths for other HTTP verbs
(#828) and fix imports (#5640). Use 'object' when no parameters are
used. Introduce 'featureMetrics' to control metrics plugin usage. Remove
HOCON configuration parsing. This is provided by
`Application.environment.config already` and removes a dependency.
Resolves#9087, resolves#828, resolves#5640
Relates-To #5346
* Revert a change to the Go Client code that broke
the ability to download files. This reverts the change
from 14d41310b9
* Missed import for "io/ioutil"
* update samples
Co-authored-by: icubbon <icubbon@gmail.com>
* Add imports for inner items of arrays/maps in composedSchemas
* Add test case for typescript-angular
* tests
* tests
* disable generate alias
* Create different config
* [python][client] allow passing floats for _request_timeout
While several method docstrings suggested that you could pass floats for
_request_timeout and the underlying urllib3.Timeout can deal with them,
the code only allowed integer values so far when passed as a single
number.
* [python][samples] update with fix for _request_timeout type
* fix compilation break with validate function
* fix error handling in handleParsingException
bug caused all errors to be regarded as an internal server error
* generate samples
* map free-form objects to Dict[str, Any]
* support Forward Type References
Forward type references will be supported by default from Python 3.10 on only. Until then (and starting with Python 3.7), we can opt in by a __future__ import, cf. https://docs.python.org/3.9/whatsnew/3.7.html?highlight=forward#pep-563-postponed-evaluation-of-annotations
* re-created pet-store sample
* bump required Python version to 3.7 for generated FastAPI projects
* make pydantic modell classes process forward type references
* [haskell][server]: Set "AnyType" to `Aeson.Value`
`AnyType` would not be translated to the json value type.
This fixes that.
* [haskell][server] Fix haddock error in `API.hs`
Haddock (the Haskell documentation generator) would choke on the
misplaced docstring. The heading has to appear inside the list.
Add new configuration flags: REMOVE_OPERATION_ID_PREFIX_DELIMITER and REMOVE_OPERATION_ID_PREFIX_COUNT
Add the corresponding fields in the DefaultCodegen
Sanitize the operationId according to the new fields
Add delimiter to list of nonNameElementPattern
Add unit test
* pass init and url as returned by the premiddleware to the postmiddleware
Current implementation will always pass the initial url and init object. However if a pre middleware changes returns an object whose `init` or `url` property refer to a different url string or object, any following post middleware wont have access to it. Effectively providing the post middleware with a completely different request information then what was actually used to make the fetch call.
* update samples and docs
Model naming conforms to the property: modelPropertyNaming, however the
parameter naming scheme is currently hardcoded to camelcase. Since the
parameter names are closely tied to the model properties names, is maybe
desirable to use a similar naming approach.
The toParamName cannot directly use the paramPropertyNaming since that
will cause issues with current generated code, since the
modelPropertyNaming currently defaults to "original"
A new property has been created to override this behavior, named
"paramPropertyNaming", and its default is set to "camelcase".
