Skip to content
This repository has been archived by the owner on Jul 15, 2021. It is now read-only.

Standard AST across all statement types (v1.0.0) #3

Closed
7 tasks done
nwronski opened this issue Jul 9, 2015 · 1 comment
Closed
7 tasks done

Standard AST across all statement types (v1.0.0) #3

nwronski opened this issue Jul 9, 2015 · 1 comment
Assignees
@nwronski
Milestone
Release v1.0.0 Final

Comments

@nwronski
Copy link
Contributor

nwronski commented Jul 9, 2015 •
edited
Loading

  • [In Progress] Finish standardizing AST format across all types of statements
    • ORDER BY
    • LIMIT
    • name property across node types
    • type and variant combine into single type property Move to v2.0.0 milestone
    • target
  • [DONE] Organize tests and SQL test queries by type and split out into different files/directories.
@nwronski nwronski added this to the Standardized AST milestone Jul 9, 2015
nwronski added a commit that referenced this issue Jul 10, 2015
@nwronski
lots of clean up to organize tests by category, split out to differen…
99a5e1f 
…t files and directories by type, define mocha.opts to run tests directory recursively. refs #3
nwronski added a commit that referenced this issue Jul 10, 2015
@nwronski
version bump to v0.10.2. refs #3
6a724ae
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Remove null and false values from AST. Refs #3
29a0074 
Taking steps to remove properties that are not required in
the AST such as default/null values and automatically
adding obscure properties by default. Especially properties such as autoIncrement and temporary.
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Use external json files in tests. Refs #3
f457c34 
Replaced grunt json with grunt rewrite to automatically
generate new test JSON files for all the specs.

Instead of parsing a JSON string in the spec JS file, the
helper now looks for a file in the "test/json/" folder that
matches the SQL file loaded from "test/sql/".
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Default optimization key removed. Refs #3
6f0ac6f 
Only include 'optimization' when it contains a value.
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Get grunt rewrite-json task working. Refs #3
4d00f9f 
Working version of "grunt rewrite-json" that also runs the
specs after generating each new JSON file.
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Test JSON files for specs. Refs #3
3af3f5d
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Remove null constraint names. Refs #3
146bb63 
Still had some null values for column constraint names.
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Update README.md for next version. Refs #3
085d31f
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Release v0.12.0
a50d7ea 
Refs #3
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Remove extra 'temporary' props. Refs #3
412b843
nwronski added a commit that referenced this issue Sep 29, 2015
@nwronski
Release v0.12.0-beta.1
5cfa812 
Previous 0.12.0 should have been tagged beta and not set
as default release.

Refs #3
@jdrew1303 jdrew1303 mentioned this issue Sep 28, 2016
Open
@jdrew1303
Copy link

Hi @nwronski

I was wondering if you were still working on standardising the node type and variant? I'm working on an AST walker based off the Estraverse code base and I've encountered some issue with nodes not having types on them (eg the on property). I can bypass it if necessary but it slows walking down. Every key has to be visited and checked to see if there is an object or array to walk into instead of knowing in advance where it needs to go.

If you're not working on it would it be ok for me to submit pull requests to add types to the nodes? They will be quite simplistic to allow the walker to get up and running but I wouldnt mind working on the full refactor of type and variant if you could do with some help?

nwronski added a commit that referenced this issue Oct 2, 2016
@nwronski
A lot of cleanup and refactoring utility funcs.
d641e6d 
Got rid of a bunch of unused code, cleaned up the utility
functions, moved the utilities back into the grammar.pegjs file to
avoid the extra magic of injecting the import statement into the
built parser.

Fixed the binary expression logic to properly handle cases such
as:

``` sql
SELECT * FROM t where -1 * (2 + 3);
SELECT * FROM t where 3 + 4 * 5 > 20;
SELECT * FROM t where v1 = ((v2 * 5) - v3);
```

Updated SQL in specs to contain correct ASTs for specs related
to expression grouping.

Should have broken this into several commits but the changes
are too intertwined to easily break apart now.

Refs #20
Refs #3
nwronski added a commit that referenced this issue Oct 3, 2016
@nwronski
Add a type and variant to root node. Refs #3
916ef4e 
The root node of the AST now has `type` and `variant`
properties:

``` json
{
  "type": "statement",
  "variant": "list",
  "statement": [{
    "type": "statement",
    "variant": "select",
    "statement": {}
  }]
}
```
nwronski added a commit that referenced this issue Oct 4, 2016
@nwronski
Add type and variant to on prop of CREATE INDEX.
647c09e 
Indexed columns (e.g., the column list in the `ON` part of a
`CREATE INDEX` statement) and ordering expressions (e.g., the
`ORDER BY` part of a `SELECT` statement) now have the
following format:

- When they are proceeded by an ordering term (e.g., `ASC`, `DESC`) and/or `COLLATE`, such as `ORDER BY nick ASC`
- But, when it is only an expression or column name without any ordering term or `COLLATE` then it will only be the expression itself, and the implicit `"direction": "asc"` will not be added to the AST, such as `ORDER BY nick`

Refs #3
nwronski added a commit that referenced this issue Oct 4, 2016
@nwronski
Allow qualified table name in CTE. Refs #3
6acff97 
Allow dbName.tableName format in a qualified table expression.
nwronski added a commit that referenced this issue Oct 4, 2016
@nwronski
Allow COLLATE to parse in indexed column list.
ad51730 
The definition for indexed columns and ordering expressions
allows for a COLLATE part, but that is already part of the
grammar for expression so it is redundant to have it explicitly
added to these other rules.

Refs #3
nwronski added a commit that referenced this issue Oct 4, 2016
@nwronski
Change format for indexed column and ordering term
9d279ad 
Indexed columns (e.g., the column list in the `ON` part of a
`CREATE INDEX` statement) and ordering expressions (e.g., the
`ORDER BY` part of a `SELECT` statement) now have the
following format:

- When they are proceeded by an ordering term (e.g., `ASC`, `DESC`) and/or `COLLATE`, such as `ORDER BY nick ASC`
- But, when it is only an expression or column name without any ordering term or `COLLATE` then it will only be the expression itself, and the implicit `"direction": "asc"` will not be added to the AST, such as `ORDER BY nick`

Refs #3
@nwronski nwronski closed this as completed Oct 9, 2016
@nwronski nwronski self-assigned this Oct 17, 2016
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees

@nwronski nwronski

Labels
None yet
Projects
None yet
Milestone
Release v1.0.0 Final
Development

No branches or pull requests
2 participants
@jdrew1303 @nwronski