1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
|
## Code Structure
Almost all of the rust code is either auto-generated or "glue" between the
swagger API spec on one side and the SQL schema on the other.
- `./migrations/*/up.sql`: SQL schema
- `./src/database_schema.rs`: autogenerated per-table Diesel schemas
- `./src/database_models.rs`: hand- and macro-generated Rust structs matching
Diesel schemas, and a small number of row-level helpers
- `./src/entity_crud.rs`: "struct-relational-mapping"; trait
implementations of CRUD (create, read, update, delete) actions for each
entity model, hitting the database (and building on `database_model` structs)
- `./src/endpoint_handlers.rs`: one function for each API endpoint, with rust-style
arguments and return types. mostly calls in to `api_entity_crud`.
- `./src/endpoints.rs`: hand- and macro-generated wrapper functions, one per
API endpoint, that map between API request and return types, and
rust-idiomatic request and return types (plus API models).
- `./fatcat-openapi`: autogenerated API models and endpoint types/signatures
- `../fatcat-openapi2.yaml`: OpenAPI 2 specification of API models and
endpoints
When deciding to use this structure, it wasn't expected that either
`api_wrappers.rs` or `database_models.rs` would need to hand-maintained; both
are verbose and implemented in a very mechanical fashion. The return type
mapping in `api_wrappers` might be necessary, but `database_models.rs` in
particular feels unnecessary; other projects have attempted to completely
automate generation of this file, but it doesn't sound reliable. In particular,
both regular "Row" (queryable) and "NewRow" (insertable) structs need to be
defined.
## Test Structure
- `./tests/test_api_server.rs`: Iron (web framework) level raw HTTP JSON
request/response tests of API endpoints.
## Updating Schemas
Regenerate API schemas after editing the fatcat-openapi2 schema. This will, as
a side-effect, also run `cargo fmt` on the whole project, so don't run it with
your editor open!
cargo install cargo-swagger # uses docker
./codegen_openapi2.sh
Update Rust database schema (after changing raw SQL schema):
diesel database reset
diesel print-schema > src/database_schema.rs
Debug SQL schema errors (if diesel commands fail):
psql fatcat_dev < migrations/2019-01-01-000000_init/up.sql
## Direct API Interaction
First setup an auth token and check that authentication is working
EDITOR_ID='aaaaaaaaaaaabkvkaaaaaaaaay'
AUTH_TOKEN=`./target/debug/fatcat-auth create-token $EDITOR_ID`
http get :9411/v0/auth/check "Authorization:Bearer $AUTH_TOKEN"
http get :9411/v0/auth/check?role=admin "Authorization:Bearer $AUTH_TOKEN"
You'll need to add the `$AUTH_TOKEN` bit to all requests below.
Creating entities via API:
http --json post localhost:9411/v0/container name=asdf issn=1234-5678
## Authentication
Uses macaroons. See `notes/auth.md` and maybe look in the guide.
## Codegen Updates
These are notes from another branch?
### OpenAPI Generator 5.0
export OPENAPI_GENERATOR_VERSION=5.0.0-SNAPSHOT
./openapi-generator-cli.sh version
### OpenAPI Standard 3.0
Should consider doing this conversion by hand, not using converter tool.
Motivation to make this change for CLI was that rust-server doesn't support
client Bearer authentication from OpenAPI 2.0.
Request bodies now don't have a "name", so the parameter ends up being infered
from the type. For auto_batch, this means "auto_batch_fileset" or whatever
instead of "auto_batch".
|
