Allow spoken forms generator to use custom (cursorless-dev#1947)

This PR generalizes our spoken form generator to be able to handle
custom spoken forms. Note that this PR doesn't contain the machinery to
read custom spoken forms from talon; that happens in cursorless-dev#1940. This PR just
prepares the api surface of the spoken form generator to handle those
custom spoken forms

## Checklist

- [ ] I have added
[tests](https://www.cursorless.org/docs/contributing/test-case-recorder/)
- [ ] I have updated the
[docs](https://github.com/cursorless-dev/cursorless/tree/main/docs) and
[cheatsheet](https://github.com/cursorless-dev/cursorless/tree/main/cursorless-talon/src/cheatsheet)
- [ ] I have not broken the cheatsheet

Author: pokey

packages/common/src/index.ts

@@ -11,6 +11,7 @@ export { getKey, splitKey } from "./util/splitKey";
 export { hrtimeBigintToSeconds } from "./util/timeUtils";
 export * from "./util/walkSync";
 export * from "./util/walkAsync";
+export * from "./util/camelCaseToAllDown";
 export { Notifier } from "./util/Notifier";
 export type { Listener } from "./util/Notifier";
 export type { TokenHatSplittingMode } from "./ide/types/Configuration";
@@ -42,6 +43,7 @@ export * from "./types/TextEditorOptions";
 export * from "./types/TextLine";
 export * from "./types/Token";
 export * from "./types/HatTokenMap";
+export * from "./types/SpokenForm";
 export * from "./util/textFormatters";
 export * from "./types/snippet.types";
 export * from "./testUtil/fromPlainObject";

packages/common/src/types/SpokenForm.ts

@@ -0,0 +1,47 @@
+/**
+ * The spoken form of a command, scope type, etc, that can be spoken
+ * using a given set of custom or default spoken forms.
+ */
+export interface SpokenFormSuccess {
+  type: "success";
+
+  /**
+   * The spoken forms for this entity. These could either be a user's custom
+   * spoken forms, if we have access to them, or the default spoken forms, if we
+   * don't, or if we're testing. There will often only be a single entry in this
+   * array, but there can be multiple if the user has used the `|` syntax in their
+   * spoken form csv's to define aliases for a single spoken form.
+   */
+  spokenForms: string[];
+}
+
+/**
+ * An error spoken form, which indicates that the given entity (command, scope
+ * type, etc) cannot be spoken, and the reason why.
+ */
+export interface SpokenFormError {
+  type: "error";
+
+  /**
+   * The reason why the entity cannot be spoken.
+   */
+  reason: string;
+
+  /**
+   * If `true`, indicates that the entity wasn't found in the user's Talon spoken
+   * forms json, and so they need to update their cursorless-talon to get the
+   * given entity.
+   */
+  requiresTalonUpdate: boolean;
+
+  /**
+   * If `true`, indicates that the entity is only for internal experimentation,
+   * and should not be exposed to users except within a targeted working group.
+   */
+  isPrivate: boolean;
+}
+
+/**
+ * A spoken form, which can either be a success or an error.
+ */
+export type SpokenForm = SpokenFormSuccess | SpokenFormError;

packages/common/src/util/camelCaseToAllDown.ts

@@ -0,0 +1,17 @@
+/**
+ * Converts a camelCase string to a string with spaces between each word, and
+ * all words in lowercase.
+ *
+ * Example: `camelCaseToAllDown("fooBarBaz")` returns `"foo bar baz"`.
+ *
+ * @param input A camelCase string
+ * @returns The same string, but with spaces between each word, and all words
+ * in lowercase
+ */
+export function camelCaseToAllDown(input: string): string {
+  return input
+    .replace(/([A-Z])/g, " $1")
+    .split(" ")
+    .map((word) => word.toLowerCase())
+    .join(" ");
+}

packages/cursorless-engine/src/generateSpokenForm/NoSpokenFormError.ts

@@ -1,5 +1,9 @@
 export class NoSpokenFormError extends Error {
-  constructor(public reason: string) {
+  constructor(
+    public reason: string,
+    public requiresTalonUpdate: boolean = false,
+    public isPrivate: boolean = false,
+  ) {
     super(`No spoken form for: ${reason}`);
   }
 }

packages/cursorless-engine/src/generateSpokenForm/SpokenFormComponent.ts

@@ -0,0 +1,40 @@
+import { SpokenFormMapEntry } from "../spokenForms/SpokenFormMap";
+import {
+  SpokenFormMapKeyTypes,
+  SpokenFormType,
+} from "../spokenForms/SpokenFormType";
+
+/**
+ * A component of a spoken form used internally during spoken form generation.
+ * This is a recursive type, so it can contain other spoken form components.
+ * During the final step of spoken form generation, it is flattened.
+ *
+ * FIXME: In the future, we want to replace `string` with something like
+ * `LiteralSpokenFormComponent` and `SpokenFormComponent[]` with something like
+ * `SequenceSpokenFormComponent`. We'd also like to avoid throwing
+ * `NoSpokenFormError` and instead return a `SpokenFormComponent` that
+ * represents an error. This would allow us to localize errors and still render
+ * the remainder of the spoken form component.
+ */
+export type SpokenFormComponent =
+  | CustomizableSpokenFormComponent
+  | string
+  | SpokenFormComponent[];
+
+export interface CustomizableSpokenFormComponentForType<
+  T extends SpokenFormType,
+> {
+  type: "customizable";
+  spokenForms: SpokenFormMapEntry;
+  spokenFormType: T;
+  id: SpokenFormMapKeyTypes[T];
+}
+
+/**
+ * A customizable spoken form component. This is a spoken form component that
+ * can be customized by the user. It is used internally during spoken form
+ * generation.
+ */
+export type CustomizableSpokenFormComponent = {
+  [K in SpokenFormType]: CustomizableSpokenFormComponentForType<K>;
+}[SpokenFormType];

packages/cursorless-engine/src/generateSpokenForm/defaultSpokenForms/modifiers.ts

@@ -1,136 +1,10 @@
-import {
-  ModifierType,
-  SimpleScopeTypeType,
-  SurroundingPairName,
-  CompositeKeyMap,
-} from "@cursorless/common";
-
-export const modifiers = {
-  excludeInterior: "bounds",
-  toRawSelection: "just",
-  leading: "leading",
-  trailing: "trailing",
-  keepContentFilter: "content",
-  keepEmptyFilter: "empty",
-  inferPreviousMark: "its",
-  startOf: "start of",
-  endOf: "end of",
-  interiorOnly: "inside",
-  extendThroughStartOf: "head",
-  extendThroughEndOf: "tail",
-  everyScope: "every",
-
-  containingScope: null,
-  ordinalScope: null,
-  relativeScope: null,
-  modifyIfUntyped: null,
-  cascading: null,
-  range: null,
-} as const satisfies Record<ModifierType, string | null>;
-
-export const modifiersExtra = {
-  first: "first",
-  last: "last",
-  previous: "previous",
-  next: "next",
-  forward: "forward",
-  backward: "backward",
-};
-
-export const scopeSpokenForms = {
-  argumentOrParameter: "arg",
-  attribute: "attribute",
-  functionCall: "call",
-  functionCallee: "callee",
-  className: "class name",
-  class: "class",
-  comment: "comment",
-  functionName: "funk name",
-  namedFunction: "funk",
-  ifStatement: "if state",
-  instance: "instance",
-  collectionItem: "item",
-  collectionKey: "key",
-  anonymousFunction: "lambda",
-  list: "list",
-  map: "map",
-  name: "name",
-  regularExpression: "regex",
-  section: "section",
-  sectionLevelOne: "one section",
-  sectionLevelTwo: "two section",
-  sectionLevelThree: "three section",
-  sectionLevelFour: "four section",
-  sectionLevelFive: "five section",
-  sectionLevelSix: "six section",
-  selector: "selector",
-  statement: "state",
-  string: "string",
-  branch: "branch",
-  type: "type",
-  value: "value",
-  condition: "condition",
-  unit: "unit",
-  //  XML, JSX
-  xmlElement: "element",
-  xmlBothTags: "tags",
-  xmlStartTag: "start tag",
-  xmlEndTag: "end tag",
-  // LaTeX
-  part: "part",
-  chapter: "chapter",
-  subSection: "subsection",
-  subSubSection: "subsubsection",
-  namedParagraph: "paragraph",
-  subParagraph: "subparagraph",
-  environment: "environment",
-  // Talon
-  command: "command",
-  // Text-based scope types
-  character: "char",
-  word: "word",
-  token: "token",
-  identifier: "identifier",
-  line: "line",
-  sentence: "sentence",
-  paragraph: "block",
-  document: "file",
-  nonWhitespaceSequence: "paint",
-  boundedNonWhitespaceSequence: "short paint",
-  url: "link",
-  notebookCell: "cell",
-
-  switchStatementSubject: null,
-  ["private.fieldAccess"]: null,
-} as const satisfies Record<SimpleScopeTypeType, string | null>;
-
-type ExtendedSurroundingPairName = SurroundingPairName | "whitespace";
-
-const surroundingPairsSpoken: Record<
-  ExtendedSurroundingPairName,
-  string | null
-> = {
-  curlyBrackets: "curly",
-  angleBrackets: "diamond",
-  escapedDoubleQuotes: "escaped quad",
-  escapedSingleQuotes: "escaped twin",
-  escapedParentheses: "escaped round",
-  escapedSquareBrackets: "escaped box",
-  doubleQuotes: "quad",
-  parentheses: "round",
-  backtickQuotes: "skis",
-  squareBrackets: "box",
-  singleQuotes: "twin",
-  any: "pair",
-  string: "string",
-  whitespace: "void",
-
-  // Used internally by the "item" scope type
-  collectionBoundary: null,
-};
+import { CompositeKeyMap } from "@cursorless/common";
+import { SpeakableSurroundingPairName } from "../../spokenForms/SpokenFormType";
+import { SpokenFormComponentMap } from "../getSpokenFormComponentMap";
+import { CustomizableSpokenFormComponentForType } from "../SpokenFormComponent";
 
 const surroundingPairsDelimiters: Record<
-  ExtendedSurroundingPairName,
+  SpeakableSurroundingPairName,
   [string, string] | null
 > = {
   curlyBrackets: ["{", "}"],
@@ -150,36 +24,19 @@ const surroundingPairsDelimiters: Record<
   string: null,
   collectionBoundary: null,
 };
+
 const surroundingPairDelimiterToName = new CompositeKeyMap<
   [string, string],
-  SurroundingPairName
+  SpeakableSurroundingPairName
 >((pair) => pair);
 
 for (const [name, pair] of Object.entries(surroundingPairsDelimiters)) {
   if (pair != null) {
-    surroundingPairDelimiterToName.set(pair, name as SurroundingPairName);
-  }
-}
-
-export const surroundingPairForceDirections = {
-  left: "left",
-  right: "right",
-};
-
-/**
- * Given a pair name (eg `parentheses`), returns the spoken form of the
- * surrounding pair.
- * @param surroundingPair The name of the surrounding pair
- * @returns The spoken form of the surrounding pair
- */
-export function surroundingPairNameToSpokenForm(
-  surroundingPair: SurroundingPairName,
-): string {
-  const result = surroundingPairsSpoken[surroundingPair];
-  if (result == null) {
-    throw Error(`Unknown surrounding pair '${surroundingPair}'`);
+    surroundingPairDelimiterToName.set(
+      pair,
+      name as SpeakableSurroundingPairName,
+    );
   }
-  return result;
 }
 
 /**
@@ -191,12 +48,13 @@ export function surroundingPairNameToSpokenForm(
  * @returns The spoken form of the surrounding pair
  */
 export function surroundingPairDelimitersToSpokenForm(
+  spokenFormMap: SpokenFormComponentMap,
   left: string,
   right: string,
-): string {
+): CustomizableSpokenFormComponentForType<"pairedDelimiter"> {
   const pairName = surroundingPairDelimiterToName.get([left, right]);
   if (pairName == null) {
     throw Error(`Unknown surrounding pair delimiters '${left} ${right}'`);
   }
-  return surroundingPairNameToSpokenForm(pairName);
+  return spokenFormMap.pairedDelimiter[pairName];
 }