Add H5 & H4: Listen to Cursor/Line Context (#7)

* Update parser

Implied nodes caused ambiguity when querying for context. Now every line
is considered, even blank lines, to make the process much, *much* easier.

* Update pylex tests

* Add runCursorContext
This commit is contained in:
Jake Grossman 2021-11-16 11:35:38 -06:00 committed by GitHub
parent 6904b5dacf
commit 887f88eba6
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
10 changed files with 321 additions and 171 deletions

View File

@ -46,7 +46,18 @@
{
"command": "mind-reader.selectTheme",
"title": "Select Theme"
},
{
"command": "mind-reader.runLineContext",
"title": "Run Line Context"
},
{
"command": "mind-reader.runCursorContext",
"title": "Run Cursor Context"
}
],
"keybindings": [
{
@ -145,7 +156,7 @@
"LEGO® Education SPIKE™ Prime Set (45678)"
]
},
"mindreader.screenReader": {
"mindreader.reader.screenReader": {
"type": "string",
"description": "Specifies which screen reader to optimize for.",
"default": "NVDA",
@ -160,6 +171,11 @@
"Apple VoiceOver (macOS)"
]
},
"mindreader.reader.contextWindow": {
"type": "number",
"description": "The number of words around the cursor to use when reading the cursor context",
"default": 1
},
"mindreader.connection.connectAutomatically": {
"type": "boolean",
"description": "Specifies whether to try to automatically detect and communicate with a connected Hub.",

View File

@ -1,4 +1,5 @@
import * as vscode from 'vscode';
import * as pl from './pylex';
/**
* @type {Object} Command // Command to register with the VS Code Extension API
@ -48,6 +49,16 @@ const commands: Command[] = [
name: 'mind-reader.resetEditorScale',
callback: resetEditorScale,
},
{
name: 'mind-reader.runLineContext',
callback: runLineContext,
},
{
name: 'mind-reader.runCursorContext',
callback: runCursorContext
}
];
// COMMAND CALLBACK IMPLEMENTATIONS
@ -76,4 +87,123 @@ function resetEditorScale(): void {
vscode.commands.executeCommand('workbench.action.zoomReset');
}
function runLineContext(): void {
let editor = vscode.window.activeTextEditor;
if (editor) {
// current text and line number
let editorText = editor.document.getText();
let line = editor.selection.active.line;
// get tab info settings
let size = parseInt(editor.options.tabSize as string);
let hard = !editor.options.insertSpaces;
// initialize parser
let parser = new pl.Parser(editorText, {size, hard});
parser.parse();
let context = parser.context(line);
// build text
let contentString = createContextString(context, line);
vscode.window.showInformationMessage(contentString);
} else {
vscode.window.showErrorMessage('No document currently active');
}
}
function createContextString(context: pl.LexNode[], line: number): string {
if (context.length < 1) {
throw new Error('Cannot create context string for empty context');
}
let contextString = 'Line ' + (line+1); // 1 based
if (context[0].token && context[0].token.attr) {
contextString += ': ' + context[0].token.type.toString() + ' ' + context[0].token.attr.toString();
}
for (let i = 1; i < context.length; i++) {
let node = context[i];
if (node.label === 'root') {
// root
contextString += ' in the Document Root';
continue;
}
if (node.token!.type !== pl.PylexSymbol.EMPTY &&
node.token!.type !== pl.PylexSymbol.INDENT) {
contextString += ' inside ' + node.token!.type.toString();
if (node.token!.attr) {
contextString += ' ' + node.token!.attr.toString();
}
}
}
return contextString;
}
// find up to `n` words around the cursor, where `n` is
// the value of `#mindreader.reader.contextWindow`
function runCursorContext(): void {
let editor = vscode.window.activeTextEditor;
if (!editor) {
vscode.window.showErrorMessage("RunCursorContext: No Active Editor");
return;
}
const cursorPos: vscode.Position = editor.selection.active;
const text: string = editor.document.lineAt(cursorPos).text;
const windowSize: number = vscode.workspace.getConfiguration('mindreader').get('reader.contextWindow')!;
let trimmedText = text.trimStart(); // trim leading whitespace
let leadingWS = text.length - trimmedText.length; // # of characters of leading whitespace
trimmedText = trimmedText.trimEnd(); // trim trailing whitespace
let pos = leadingWS;
let maxPos = text.length;
// clamp cursor start/end to new range
let col = cursorPos.character; // effective column of the cursor position
if (col < leadingWS) {
// move effective start to first non-whitespace character in the line
col = leadingWS;
} else if (col > leadingWS + trimmedText.length - 1) {
// move effective end to last non-whitespace character in the line
col = leadingWS + trimmedText.length - 1;
}
// generate list of space separate words with range data (start, end)
// TODO: can find user position to be done in one pass
let spaceWords: {word: string, start: number, end: number}[] = [];
while (pos < maxPos && trimmedText.length > 0) {
let word = trimmedText.replace(/ .*/, '');
spaceWords.push({word, start: pos, end: pos+word.length});
// remove processed word from trimmed text
const oldText = trimmedText;
trimmedText = trimmedText.replace(/[^ ]+/, '').trimStart();
// update pos to start of next word
pos += oldText.length - trimmedText.length;
}
// find word the user is in
let contextStart: number = -1, contextEnd: number = -1;
for (let i = 0; i < spaceWords.length; i++) {
if (col >= spaceWords[i].start && col <= spaceWords[i].end) {
// found the word
contextStart = Math.max(0, i - windowSize); // clamp start index
contextEnd = Math.min(spaceWords.length, i + windowSize + 1); // clamp end index
// construct cursor context string
let contextString = '';
for (let i = contextStart; i < contextEnd; i++) {
contextString += spaceWords[i].word + ' ';
}
// output cursor context string
vscode.window.showInformationMessage(contextString);
return;
}
}
}
export default commands;

View File

@ -4,4 +4,4 @@ export {default as LineToken} from './token';
export {default as Lexer} from './lexer';
export {default as LexNode} from './node';
export {TabInfo as TabInfo} from './token';
export {Symbol as PylexSymbol} from './token';

View File

@ -159,14 +159,15 @@ export default class Lexer {
}
// No rules matched
// Skip this line if it is whitespace, comment, or empty
// TODO: move to rules
if (/^\s*(#.*)?$/.test(line)) {
this.pos++;
continue;
// "empty" line
token = new LineToken(Symbol.EMPTY, this.pos, 999999);
} else {
// This is an INDENT token
token = new LineToken(Symbol.INDENT, this.pos, indent);
}
// This is an INDENT token
token = new LineToken(Symbol.INDENT, this.pos, indent);
this._currToken = token;
this.pos++;
return this.currToken();

View File

@ -2,6 +2,8 @@ import * as vscode from 'vscode';
import LineToken from './token';
/* TODO: make accessing children and parent less tedious */
/* TODO: 'root.children()![i])' */
/**
* A node in a Parse tree.
*/
@ -46,8 +48,9 @@ export default class LexNode extends vscode.TreeItem {
* Adopt child nodes.
*
* @param `child` Array of nodes to adopt.
* @returns an updated version of itself
*/
adopt(children: LexNode[]): void {
adopt(children: LexNode[]): LexNode {
let parentedChildren = children.map(c => new LexNode(
c.label,
c.collapsibleState,
@ -64,6 +67,7 @@ export default class LexNode extends vscode.TreeItem {
// No....
this._children = parentedChildren;
}
return this;
}
/**

View File

@ -3,6 +3,7 @@ import * as vscode from 'vscode';
import { EOFTOKEN, Symbol, TabInfo } from './token';
import Lexer from './lexer';
import LexNode from './node';
/* TODO: update design doc */
/**
* A parse tree generator
@ -71,8 +72,20 @@ export default class Parser {
// go up 1 level of recursion at a time to unravel properly
this.currIndent--;
return children;
} else if (this.lexer.currToken().type === Symbol.INDENT) {
}
if (this.lexer.currToken().type === Symbol.INDENT ||
this.lexer.currToken().type === Symbol.EMPTY) {
const label = this.lexer.currToken().type;
// regular code, advance and stay in same block
children.push(new LexNode(
label,
vscode.TreeItemCollapsibleState.None,
this.lexer.currToken(),
null,
parent)
);
this.lexer.next();
continue;
} else {
@ -105,29 +118,28 @@ export default class Parser {
* @param `lineNumber` The line number to query context for.
* @return An array of LexNodes for the root path containing `lineNumber`
*/
context(lineNumber: number): LexNode[] {
if (!this.root.children()) {
return [];
context(lineNumber: number, root?: LexNode): LexNode[] {
if (!root) {
root = this.root;
}
// Returns the LexNode that is the parent
// of the queried line number
let find = (root: LexNode): LexNode | undefined => {
let prevChild: LexNode;
for (var child of root.children()!) {
if (lineNumber < child.token!.linenr) {
if (prevChild!.children()) {
return find(prevChild!);
} else {
return prevChild!;
}
} else {
prevChild = child;
// is this the target?
if (root.token && root.token!.linenr === lineNumber) {
// match
return root.rootPath();
}
if (root.children()) {
// recursive call
for (let i = 0; i < root.children()!.length; i++) {
let ctx = this.context(lineNumber, root.children()![i]);
if (ctx.length > 0) {
// a rootpath was returned
return ctx;
}
}
};
let target = find(this.root);
return target!.rootPath();
}
// no matches
return [];
}
}

View File

@ -17,6 +17,7 @@ export enum Symbol {
FINALLY = "finally",
WITH = "with",
INDENT = "INDENT", // Indent token, default if not EOF, only contains indent information
EMPTY = "EMPTY", // empty line, used only to associate with the previous line
EOF = "EOF"
}

View File

@ -17,13 +17,18 @@ suite('Lexer Test Suite', () => {
});
test('Undefined', () => {
let l: Lexer = new Lexer('');
let l: Lexer = new Lexer(undefined);
assert.deepStrictEqual(l.currToken(), EOFTOKEN);
});
test('Whitespace', () => {
let l: Lexer = new Lexer(' \t\t'.repeat(4).repeat(4));
assert.deepStrictEqual(l.currToken(), EOFTOKEN);
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.EMPTY, 0, 999999));
});
test('Comment', () => {
let l: Lexer = new Lexer('# ur mom eats toes');
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.EMPTY, 0, 999999));
});
test('Non-Whitespace with no construct', () => {
@ -120,98 +125,6 @@ suite('Lexer Test Suite', () => {
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.IF, 0, 0, 'is_old()'));
});
test('next() ignores empty lines', () => {
let lines: string[] = [
'if wurst_available():',
'',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
l.next();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.INDENT, 2, 1));
});
test('retract() ignores empty lines', () => {
let lines: string[] = [
'if wurst_available():',
'',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
l.next();
l.retract();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.IF, 0, 0, 'wurst_available()'));
});
test('next() ignores whitespace lines', () => {
let lines: string[] = [
'if wurst_available():',
' \t \t ',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
l.next();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.INDENT, 2, 1));
});
test('retract() ignores whitespace lines', () => {
let lines: string[] = [
'if wurst_available():',
' \t \t ',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
// Advance to end of input
// Eliminates dependence on next()
// skipping whitespace
do {} while (l.next() !== EOFTOKEN);
l.retract(); // retract past EOFTOKEn
l.retract();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.IF, 0, 0, 'wurst_available()'));
});
test('next() ignores comment lines', () => {
let lines: string[] = [
'if wurst_available():',
' \t # I hate testing \t',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
l.next();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.INDENT, 2, 1));
});
test('retract() ignores comment lines', () => {
let lines: string[] = [
'if wurst_available():',
' \t # \t',
' eat_wurst()'
];
let l: Lexer = new Lexer(lines.join('\n'));
// Advance to end of input
// Eliminates dependence on next()
// skipping comment
do {} while (l.next() !== EOFTOKEN);
l.retract(); // retract past EOFTOKEn
l.retract();
assert.deepStrictEqual(l.currToken(), new LineToken(Symbol.IF, 0, 0, 'wurst_available()'));
});
test('next() out of range', () => {
let l: Lexer = new Lexer('foo = zaboomafoo');
l.next();

View File

@ -1,51 +1,61 @@
import * as assert from 'assert';
import * as vscode from 'vscode';
import { after } from 'mocha';
import { deparent, root } from '../../util';
import Parser from '../../../pylex/parser';
import LexNode from '../../../pylex/node';
import LineToken from '../../../pylex/token';
import { Symbol } from '../../../pylex/token';
import { root,indent,empty } from '../../util';
/**
* Test Descriptor
*/
type ParserTest = {
name: string,
input: string[],
output: LexNode,
name: string, // short name for the test
input: string[], // input lines for the test
output: LexNode // expected output. outputs are compared for token equality *only*
};
const tests: ParserTest[] = [
{ name: 'No Input', input: [], output: root(null) },
{ name: 'Single Empty Line', input: [''], output: root(null) },
{
name: 'No Input',
input: [ ],
output: root(null),
},
{
name: 'Single line without construct',
input: [ 'foo = "Yellow M&Ms make me angry >:(' ],
output: root(null),
},
{
name: 'Single line with construct',
input: [ 'for x of y:' ],
name: 'Single Whitespace Only Line',
input: [' '],
output: root([
new LexNode(
'for x of y',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.FOR, 0, 0, 'x of y')
)
]),
empty(0)
])
},
{
name: 'Single Comment Only Line',
input: ['# ur mom likes peas'],
output: root([
empty(0)
])
},
{
name: 'Single Non-Control Line',
input: ['my_age = 42'],
output: root([
indent(0, 0)
])
},
{
name: 'Single Control Line',
input: ['while True:'],
output: root([
new LexNode('', 0, new LineToken(Symbol.WHILE, 0, 0, 'True'))
])
},
{
name: 'Sequential lines, without construct',
input: [
'bar = "Blue M&Ms make me happy <:)"',
'reba = "A hard working gal"'
],
output: root(null),
output: root([indent(0,0), indent(1,0)]),
},
{
@ -58,7 +68,9 @@ const tests: ParserTest[] = [
output: root([
new LexNode('if radioshack',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 0, 0, 'radioshack'))
new LineToken(Symbol.IF, 0, 0, 'radioshack'),
[indent(1, 1)]),
indent(2, 0)
])
},
@ -70,9 +82,11 @@ const tests: ParserTest[] = [
' print radioshack.hours'
],
output: root([
indent(0, 0),
new LexNode('if radioshack',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 1, 0, 'radioshack'))
new LineToken(Symbol.IF, 1, 0, 'radioshack'),
[indent(2, 1)])
])
},
@ -89,13 +103,13 @@ const tests: ParserTest[] = [
output: root([
new LexNode('if yummy',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 0, 0, 'yummy')),
new LineToken(Symbol.IF, 0, 0, 'yummy'), [indent(1, 1)]),
new LexNode('elif just_ok',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELIF, 2, 0, 'just_ok')),
new LineToken(Symbol.ELIF, 2, 0, 'just_ok'), [indent(3, 1)]),
new LexNode('else',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELSE, 4, 0)),
new LineToken(Symbol.ELSE, 4, 0), [indent(5,1)]),
])
},
@ -113,8 +127,9 @@ const tests: ParserTest[] = [
[
new LexNode('if in_my_tummy',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 1, 1, 'in_my_tummy'))
]
new LineToken(Symbol.IF, 1, 1, 'in_my_tummy'),
[indent(2, 2)])
],
)
])
},
@ -135,12 +150,14 @@ const tests: ParserTest[] = [
[
new LexNode('if in_my_tummy',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 1, 1, 'in_my_tummy'))
new LineToken(Symbol.IF, 1, 1, 'in_my_tummy'),
[indent(2, 2)])
]
),
new LexNode('else',
new LexNode('else',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELSE, 3, 0),
[indent(4, 1)]
)
])
},
@ -166,7 +183,8 @@ const tests: ParserTest[] = [
[
new LexNode('if looks_like_a_mummy',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 2, 2, 'looks_like_a_mummy'))
new LineToken(Symbol.IF, 2, 2, 'looks_like_a_mummy'),
[indent(3, 3)])
]
)
]
@ -174,12 +192,13 @@ const tests: ParserTest[] = [
new LexNode('else',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELSE, 4, 0),
[indent(5, 1)]
)
])
},
{
name: 'Doubly Nested Block, with multiple indent resets',
name: 'Doubly Nested Block, with multiple indent resets > 1',
input: [
'if yummy:',
' if in_my_tummy:',
@ -188,7 +207,7 @@ const tests: ParserTest[] = [
' else:',
' print("eek! a zombie!)',
' elif in_my_mouth:',
' print("ill be in my tummy soon!"',
' print("itll be in my tummy soon!"',
'else:',
' print("Food is food...")'
],
@ -203,25 +222,70 @@ const tests: ParserTest[] = [
[
new LexNode('if looks_like_a_mummy',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.IF, 2, 2, 'looks_like_a_mummy')),
new LineToken(Symbol.IF, 2, 2, 'looks_like_a_mummy'),
[indent(3, 3)]),
new LexNode('else',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELSE, 4, 2))
new LineToken(Symbol.ELSE, 4, 2),
[indent(5, 3)])
]
),
new LexNode('elif in_my_mouth',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELIF, 6, 1, 'in_my_mouth'))
new LineToken(Symbol.ELIF, 6, 1, 'in_my_mouth'),
[indent(7, 2)]
)
]
),
new LexNode('else',
vscode.TreeItemCollapsibleState.None,
new LineToken(Symbol.ELSE, 8, 0)
new LineToken(Symbol.ELSE, 8, 0),
[indent(9, 1)]
)
])
},
{
name: 'Multiline Block',
input: [
'if yummy:',
' print("you have a spot on your tummy"',
' print("eek! a zombie!)',
' print("itll be in my tummy soon!"',
'else:',
' print("Food is food...")'
],
output: root([
new LexNode('if yummy', 0, new LineToken(Symbol.IF, 0, 0, 'yummy'),
[
indent(1, 1),
indent(2, 1),
indent(3, 1),
]
),
new LexNode('else', 0, new LineToken(Symbol.ELSE, 4, 0), [indent(5, 1)])
])
}
];
/* Checks for strict equality between the tokens of a lex node tree */
const checkEq = (reference: LexNode, subject: LexNode) => {
if (!reference.children()) {
// subject should also have no children
assert.deepStrictEqual(subject.children(), null);
return;
}
assert.notStrictEqual(subject.children(), null);
assert.deepStrictEqual(reference.children()!.length, subject.children()!.length);
for (let i = 0; i < subject.children()!.length; i++) {
// compare top level nodes
assert.deepStrictEqual(reference.children()![i].token, subject.children()![i].token);
// compare all children
checkEq(reference.children()![i], subject.children()![i]);
}
};
suite('Parser Test Suite', () => {
after(() => {
vscode.window.showInformationMessage('All tests passed!');
@ -230,10 +294,8 @@ suite('Parser Test Suite', () => {
for (var t of tests) {
let currTest = t; // without this, all test calls get the last test
test(currTest.name, () => {
let result: LexNode = deparent(new Parser(currTest.input.join('\n')).parse());
process.stdout.write(Object.entries(result).toString());
assert.deepStrictEqual(result, currTest.output);
let result = new Parser(currTest.input.join('\n')).parse();
checkEq(currTest.output, result);
});
}
});

View File

@ -1,6 +1,5 @@
import * as vscode from 'vscode';
import LexNode from '../pylex/node';
import { LineToken, PylexSymbol, LexNode} from '../pylex';
/**
* TODO: Eliminate need for me.
@ -50,8 +49,20 @@ function root(nodes: LexNode[] | null): LexNode {
);
}
/* short hand for returning an indentation token for a certain line and indentation */
function indent(linenr: number, indentLevel: number): LexNode {
return new LexNode('INDENT', 0, new LineToken(PylexSymbol.INDENT, linenr, indentLevel));
}
/* short hand for returning an empty token for a certain line*/
function empty(linenr: number): LexNode {
return new LexNode('EMPTY', 0, new LineToken(PylexSymbol.EMPTY, linenr, 999999));
}
export {
deparent,
root,
indent,
empty
};