From 2c55e38822fcd2fc1a2e9cacaa85b6243de315b9 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 19:16:56 -0400
Subject: [PATCH 01/14] WIP replacement Notcl parser
---
src/parser2.ts | 155 +++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 155 insertions(+)
create mode 100644 src/parser2.ts
diff --git a/src/parser2.ts b/src/parser2.ts
new file mode 100644
index 0000000..75f9842
--- /dev/null
+++ b/src/parser2.ts
@@ -0,0 +1,155 @@
+import {
+ Command,
+ ErrorResult,
+ InterpolatedPiece,
+ Script,
+ SimplifyWord,
+ Word,
+} from "./words";
+
+/**
+ * Parse out a Notcl script into an easier-to-interpret representation.
+ * No script is actually executed yet.
+ *
+ * @param code code to parse
+ * @param offset source position of code, if embedded in a larger source document
+ * @returns parsed list of commands, or error message on failure
+ */
+export function parse(
+ code: string,
+ offset = 0
+): [true, Script] | [false, string] {
+ try {
+ const parser = new Parser(code);
+ const script = parser.parseScript();
+
+ // TODO: report error with error position
+
+ if (parser.lastIndex != code.length) {
+ return [false, "Couldn't parse full script"];
+ }
+
+ return [true, script];
+ } catch (ex) {
+ return [false, String(ex)];
+ }
+}
+
+// ---------------------------
+
+// Parser for evaluating Notcl scripts
+
+type TokenType =
+ | "newline"
+ | "whitespace"
+ | "semicolon"
+ | "{"
+ | "}"
+ | "["
+ | "]"
+ | "quote"
+ | "backslash"
+ | "comment"
+ | "text"
+ | "EOF"
+ | "ERROR";
+
+type Token = [TokenType, string, number];
+
+const Tokens: [TokenType, RegExp][] = [
+ ["newline", /(\n)/y],
+ ["whitespace", /([^\S\n]+)/y],
+ ["text", /([^\s\\;\[\]]+)/y],
+];
+
+class WipScript {
+ script: Command[] = [];
+ wipCommand: Word[] = [];
+ wipWord: InterpolatedPiece[] = [];
+ // TODO: thing to fail {}a & ""a
+
+ addWordPiece(piece: InterpolatedPiece) {
+ this.wipWord.push(piece);
+ }
+ finishWord() {
+ if (this.wipWord.length > 0) {
+ this.wipCommand.push(SimplifyWord(this.wipWord));
+ this.wipWord = [];
+ }
+ }
+ finishCommand() {
+ this.finishWord();
+ if (this.wipCommand.length > 0) {
+ this.script.push(this.wipCommand);
+ this.wipCommand = [];
+ }
+ }
+ finishScript(): Script {
+ this.finishCommand();
+ return this.script;
+ }
+}
+
+class Parser {
+ lastIndex: number = 0;
+ next: Token;
+
+ constructor(public text: string) {
+ this.next = this.advance();
+ }
+
+ advance(): Token {
+ const startPos = this.lastIndex;
+ if (startPos == this.text.length) {
+ return (this.next = ["EOF", "<EOF>", startPos]);
+ }
+
+ for (const [type, regex] of Tokens) {
+ regex.lastIndex = startPos;
+ const matches = regex.exec(this.text);
+ if (matches) {
+ this.lastIndex = regex.lastIndex;
+ return (this.next = [type, matches[1], startPos]);
+ }
+ }
+
+ return (this.next = ["ERROR", "Token not matched", startPos]);
+ }
+
+ parseScript(): Script {
+ const wip = new WipScript();
+
+ while (true) {
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "text":
+ wip.addWordPiece({ bare: chars, pos });
+ break;
+
+ case "whitespace":
+ wip.finishWord();
+ break;
+
+ case "newline":
+ case "semicolon":
+ wip.finishCommand();
+ break;
+
+ case "EOF":
+ case "]":
+ return wip.finishScript();
+
+ case "{":
+ case "}":
+ case "[":
+ case "quote":
+ case "backslash":
+ case "comment":
+ case "ERROR":
+ throw new Error(`Unhandled case: ${type} (${chars})`);
+ }
+
+ this.advance();
+ }
+ }
+}
From 618de2ac9957c6c12b7df271566e702db2a101fa Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 19:32:58 -0400
Subject: [PATCH 02/14] Add source position to words in new parser
---
src/parser2.ts | 13 ++++++++++---
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 75f9842..163ed25 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -66,15 +66,20 @@ class WipScript {
script: Command[] = [];
wipCommand: Word[] = [];
wipWord: InterpolatedPiece[] = [];
+ wordPos: number | undefined = undefined;
// TODO: thing to fail {}a & ""a
- addWordPiece(piece: InterpolatedPiece) {
+ addWordPiece(piece: InterpolatedPiece, pos?: number) {
+ if (this.wipWord.length == 0) {
+ this.wordPos = pos;
+ }
this.wipWord.push(piece);
}
finishWord() {
if (this.wipWord.length > 0) {
- this.wipCommand.push(SimplifyWord(this.wipWord));
+ this.wipCommand.push(SimplifyWord(this.wipWord, this.wordPos));
this.wipWord = [];
+ this.wordPos = undefined;
}
}
finishCommand() {
@@ -123,7 +128,7 @@ class Parser {
const [type, chars, pos] = this.next;
switch (type) {
case "text":
- wip.addWordPiece({ bare: chars, pos });
+ wip.addWordPiece({ bare: chars }, pos);
break;
case "whitespace":
@@ -147,6 +152,8 @@ class Parser {
case "comment":
case "ERROR":
throw new Error(`Unhandled case: ${type} (${chars})`);
+ default:
+ throw new Error(`Unhandled case: ${type satisfies never} (${chars})`);
}
this.advance();
From 9b81056d1d003be5c292e45faa7c432fc559bcee Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 19:43:19 -0400
Subject: [PATCH 03/14] Add more recognized tokens
---
src/parser2.ts | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/src/parser2.ts b/src/parser2.ts
index 163ed25..83ca5fe 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -59,6 +59,14 @@ type Token = [TokenType, string, number];
const Tokens: [TokenType, RegExp][] = [
["newline", /(\n)/y],
["whitespace", /([^\S\n]+)/y],
+ ["semicolon", /(;)/y],
+ ["{", /(\{)/y],
+ ["}", /(\})/y],
+ ["[", /(\[)/y],
+ ["]", /(\])/y],
+ ["quote", /(")/y],
+ ["backslash", /(\\)/y],
+ ["comment", /(\#)/y],
["text", /([^\s\\;\[\]]+)/y],
];
From 63d6fa836a7b9af60f6f4d60c30cc1896caa9f7f Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 20:02:20 -0400
Subject: [PATCH 04/14] Add expect() & startOfWord() helpers to new parser
---
src/parser2.ts | 22 +++++++++++++++-------
1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 83ca5fe..513f563 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -22,15 +22,11 @@ export function parse(
try {
const parser = new Parser(code);
const script = parser.parseScript();
-
- // TODO: report error with error position
-
- if (parser.lastIndex != code.length) {
- return [false, "Couldn't parse full script"];
- }
+ parser.expect("EOF");
return [true, script];
} catch (ex) {
+ // TODO: report error with error position
return [false, String(ex)];
}
}
@@ -77,8 +73,12 @@ class WipScript {
wordPos: number | undefined = undefined;
// TODO: thing to fail {}a & ""a
+ startOfWord(): boolean {
+ return this.wipWord.length == 0;
+ }
+
addWordPiece(piece: InterpolatedPiece, pos?: number) {
- if (this.wipWord.length == 0) {
+ if (this.startOfWord()) {
this.wordPos = pos;
}
this.wipWord.push(piece);
@@ -129,6 +129,14 @@ class Parser {
return (this.next = ["ERROR", "Token not matched", startPos]);
}
+ expect(type: TokenType) {
+ if (this.next[0] != type) {
+ throw new Error(
+ `Expected ${type}, found ${this.next[0]} (${this.next[1]})`
+ );
+ }
+ }
+
parseScript(): Script {
const wip = new WipScript();
From ab91d71170b29f363899a058ed16d96717a01c85 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 20:02:47 -0400
Subject: [PATCH 05/14] Support [] interpolation in new parser
---
src/parser2.ts | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 513f563..92c7233 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -147,6 +147,14 @@ class Parser {
wip.addWordPiece({ bare: chars }, pos);
break;
+ case "[": {
+ this.advance();
+ const script = this.parseScript();
+ wip.addWordPiece({ script });
+ this.expect("]");
+ break;
+ }
+
case "whitespace":
wip.finishWord();
break;
@@ -162,7 +170,6 @@ class Parser {
case "{":
case "}":
- case "[":
case "quote":
case "backslash":
case "comment":
From c1ce90fd63637326b2bba3ab5a28fa7fdd98fa16 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 20:26:09 -0400
Subject: [PATCH 06/14] WIP brace parsing for WIP parser
---
src/parser2.ts | 41 ++++++++++++++++++++++++++++++++++++++---
1 file changed, 38 insertions(+), 3 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 92c7233..9481d80 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -63,7 +63,7 @@ const Tokens: [TokenType, RegExp][] = [
["quote", /(")/y],
["backslash", /(\\)/y],
["comment", /(\#)/y],
- ["text", /([^\s\\;\[\]]+)/y],
+ ["text", /([^\s;\{\}\[\]"\\\#]+)/y],
];
class WipScript {
@@ -144,9 +144,18 @@ class Parser {
const [type, chars, pos] = this.next;
switch (type) {
case "text":
+ case "}":
wip.addWordPiece({ bare: chars }, pos);
break;
+ case "{": {
+ this.advance();
+ const text = this.parseBrace();
+ wip.addWordPiece({ text }, pos);
+ this.expect("}");
+ break;
+ }
+
case "[": {
this.advance();
const script = this.parseScript();
@@ -168,8 +177,6 @@ class Parser {
case "]":
return wip.finishScript();
- case "{":
- case "}":
case "quote":
case "backslash":
case "comment":
@@ -182,4 +189,32 @@ class Parser {
this.advance();
}
}
+
+ parseBrace(): string {
+ let wip = "";
+
+ while (true) {
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "{": {
+ wip += "{";
+ this.advance();
+ wip += this.parseBrace();
+ this.expect("}");
+ wip += "}";
+ break;
+ }
+ case "}":
+ return wip;
+ case "EOF":
+ throw new Error("Reached end of input while parsing a brace word");
+ case "ERROR":
+ throw new Error(chars);
+ default:
+ wip += chars;
+ }
+
+ this.advance();
+ }
+ }
}
From c61496fcc3ea203ea5d8de00df46befe6c147cb5 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Wed, 5 Jun 2024 20:44:13 -0400
Subject: [PATCH 07/14] Be strict about trailing characters after braces. Don't
give mid-word braces magic meaning.
---
src/parser2.ts | 26 +++++++++++++++++++-------
1 file changed, 19 insertions(+), 7 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 9481d80..be21529 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -71,24 +71,31 @@ class WipScript {
wipCommand: Word[] = [];
wipWord: InterpolatedPiece[] = [];
wordPos: number | undefined = undefined;
- // TODO: thing to fail {}a & ""a
+ endOfWordError: string | undefined = undefined;
startOfWord(): boolean {
return this.wipWord.length == 0;
}
addWordPiece(piece: InterpolatedPiece, pos?: number) {
+ if (this.endOfWordError) {
+ throw new Error(this.endOfWordError);
+ }
if (this.startOfWord()) {
this.wordPos = pos;
}
this.wipWord.push(piece);
}
+ freezeWord(error: string) {
+ this.endOfWordError = error;
+ }
finishWord() {
if (this.wipWord.length > 0) {
this.wipCommand.push(SimplifyWord(this.wipWord, this.wordPos));
- this.wipWord = [];
- this.wordPos = undefined;
}
+ this.wipWord = [];
+ this.wordPos = undefined;
+ this.endOfWordError = undefined;
}
finishCommand() {
this.finishWord();
@@ -149,10 +156,15 @@ class Parser {
break;
case "{": {
- this.advance();
- const text = this.parseBrace();
- wip.addWordPiece({ text }, pos);
- this.expect("}");
+ if (wip.startOfWord()) {
+ this.advance();
+ const text = this.parseBrace();
+ wip.addWordPiece({ text }, pos);
+ this.expect("}");
+ wip.freezeWord("Extra characters after closing brace");
+ } else {
+ wip.addWordPiece({ bare: chars }, pos);
+ }
break;
}
From b2c5f7ea904e37e301d6be5d37eb644cc8d38f76 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Fri, 7 Jun 2024 22:21:04 -0400
Subject: [PATCH 08/14] Handle backslashes in brace words correctly
---
src/parser2.ts | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/src/parser2.ts b/src/parser2.ts
index be21529..93829cd 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -208,6 +208,11 @@ class Parser {
while (true) {
const [type, chars, pos] = this.next;
switch (type) {
+ case "backslash":
+ wip += "\\";
+ this.advance();
+ wip += this.next[1];
+ break;
case "{": {
wip += "{";
this.advance();
From b536d304201afea5d77068ea19ca7d2fe22231bd Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 00:03:17 -0400
Subject: [PATCH 09/14] sketch out backslash escape handling
---
src/parser2.ts | 46 ++++++++++++++++++++++++++++++++++++++++++++--
1 file changed, 44 insertions(+), 2 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 93829cd..44796cc 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -189,11 +189,53 @@ class Parser {
case "]":
return wip.finishScript();
+ case "backslash": {
+ this.advance();
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "newline":
+ wip.finishWord();
+ break;
+
+ case "whitespace":
+ case "semicolon":
+ case "{":
+ case "}":
+ case "[":
+ case "]":
+ case "quote":
+ case "backslash":
+ case "comment":
+ wip.addWordPiece({ text: chars });
+ break;
+ case "text":
+ switch (chars) {
+ case "n":
+ wip.addWordPiece({ text: "\n" });
+ break;
+ default:
+ throw new Error(`Unknown backslash escape: ${chars}`);
+ }
+ break;
+ case "EOF":
+ throw new Error(
+ "Reached end of input while parsing a backslash escape"
+ );
+ case "ERROR":
+ throw new Error(chars);
+ default:
+ throw new Error(
+ `Unhandled case: ${type satisfies never} (${chars})`
+ );
+ }
+ break;
+ }
+
case "quote":
- case "backslash":
case "comment":
- case "ERROR":
throw new Error(`Unhandled case: ${type} (${chars})`);
+ case "ERROR":
+ throw new Error(chars);
default:
throw new Error(`Unhandled case: ${type satisfies never} (${chars})`);
}
From fa3be1e003aa7fbd3318518d071fd5e5a299f6b9 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 12:32:58 -0400
Subject: [PATCH 10/14] Support comments in new parser
---
src/parser2.ts | 31 ++++++++++++++++++++++++++++---
src/words.ts | 8 ++++++--
2 files changed, 34 insertions(+), 5 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 44796cc..fb7131d 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -77,6 +77,10 @@ class WipScript {
return this.wipWord.length == 0;
}
+ startOfCommand(): boolean {
+ return this.wipWord.length == 0 && this.wipCommand.length == 0;
+ }
+
addWordPiece(piece: InterpolatedPiece, pos?: number) {
if (this.endOfWordError) {
throw new Error(this.endOfWordError);
@@ -185,6 +189,28 @@ class Parser {
wip.finishCommand();
break;
+ case "comment":
+ if (wip.startOfCommand()) {
+ skipComment: while (this.advance()) {
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "newline":
+ case "EOF":
+ break skipComment;
+ case "backslash":
+ this.advance();
+ continue;
+ case "ERROR":
+ throw new Error(chars);
+ default:
+ continue;
+ }
+ }
+ } else {
+ wip.addWordPiece({ bare: chars }, pos);
+ }
+ break;
+
case "EOF":
case "]":
return wip.finishScript();
@@ -206,12 +232,12 @@ class Parser {
case "quote":
case "backslash":
case "comment":
- wip.addWordPiece({ text: chars });
+ wip.addWordPiece({ text: chars }, pos);
break;
case "text":
switch (chars) {
case "n":
- wip.addWordPiece({ text: "\n" });
+ wip.addWordPiece({ text: "\n" }, pos);
break;
default:
throw new Error(`Unknown backslash escape: ${chars}`);
@@ -232,7 +258,6 @@ class Parser {
}
case "quote":
- case "comment":
throw new Error(`Unhandled case: ${type} (${chars})`);
case "ERROR":
throw new Error(chars);
diff --git a/src/words.ts b/src/words.ts
index 5c8f5b7..7d70c6e 100644
--- a/src/words.ts
+++ b/src/words.ts
@@ -1,4 +1,4 @@
-import { escapeHtml } from './helpers';
+import { escapeHtml } from "./helpers";
export type SourcePos = number;
@@ -117,7 +117,11 @@ export function SimplifyWord(
if (consolidated.length == 0) {
return { text: "", pos: sourcePosition };
} else if (consolidated.length == 1 && IsTextPiece(consolidated[0])) {
- return { ...consolidated[0], pos: sourcePosition };
+ if (pieces.every((piece) => "bare" in piece)) {
+ return { bare: AsText(consolidated[0]), pos: sourcePosition };
+ } else {
+ return { ...consolidated[0], pos: sourcePosition };
+ }
} else {
return { pieces: consolidated };
}
From a2c8eb66b991c68aa41b031c554c9857c41ace17 Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 13:11:46 -0400
Subject: [PATCH 11/14] Factor out backslash parse helper
---
src/parser2.ts | 80 +++++++++++++++++++++++++++-----------------------
1 file changed, 43 insertions(+), 37 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index fb7131d..bb0a2ca 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -217,43 +217,7 @@ class Parser {
case "backslash": {
this.advance();
- const [type, chars, pos] = this.next;
- switch (type) {
- case "newline":
- wip.finishWord();
- break;
-
- case "whitespace":
- case "semicolon":
- case "{":
- case "}":
- case "[":
- case "]":
- case "quote":
- case "backslash":
- case "comment":
- wip.addWordPiece({ text: chars }, pos);
- break;
- case "text":
- switch (chars) {
- case "n":
- wip.addWordPiece({ text: "\n" }, pos);
- break;
- default:
- throw new Error(`Unknown backslash escape: ${chars}`);
- }
- break;
- case "EOF":
- throw new Error(
- "Reached end of input while parsing a backslash escape"
- );
- case "ERROR":
- throw new Error(chars);
- default:
- throw new Error(
- `Unhandled case: ${type satisfies never} (${chars})`
- );
- }
+ this.parseBackslashEscape(wip, "bare");
break;
}
@@ -269,6 +233,48 @@ class Parser {
}
}
+ parseBackslashEscape(wip: WipScript, wordType: "bare" | "quote") {
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "newline":
+ if (wordType == "bare") {
+ wip.finishWord();
+ } else {
+ // ignore newline
+ }
+ break;
+
+ case "whitespace":
+ case "semicolon":
+ case "{":
+ case "}":
+ case "[":
+ case "]":
+ case "quote":
+ case "backslash":
+ case "comment":
+ wip.addWordPiece({ text: chars }, pos);
+ break;
+ case "text":
+ switch (chars) {
+ case "n":
+ wip.addWordPiece({ text: "\n" }, pos);
+ break;
+ default:
+ throw new Error(`Unknown backslash escape: ${chars}`);
+ }
+ break;
+ case "EOF":
+ throw new Error(
+ "Reached end of input while parsing a backslash escape"
+ );
+ case "ERROR":
+ throw new Error(chars);
+ default:
+ throw new Error(`Unhandled case: ${type satisfies never} (${chars})`);
+ }
+ }
+
parseBrace(): string {
let wip = "";
From ac1a38e75fc40fa99143777097b4dedceca906cc Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 13:28:21 -0400
Subject: [PATCH 12/14] Tidy up error reporting
---
src/parser2.ts | 52 +++++++++++++++++++++++++++++---------------------
1 file changed, 30 insertions(+), 22 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index bb0a2ca..c9a65a8 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -35,6 +35,12 @@ export function parse(
// Parser for evaluating Notcl scripts
+export class ParseError extends Error {
+ constructor(message: string, public pos: number) {
+ super(message);
+ }
+}
+
type TokenType =
| "newline"
| "whitespace"
@@ -47,8 +53,7 @@ type TokenType =
| "backslash"
| "comment"
| "text"
- | "EOF"
- | "ERROR";
+ | "EOF";
type Token = [TokenType, string, number];
@@ -81,9 +86,9 @@ class WipScript {
return this.wipWord.length == 0 && this.wipCommand.length == 0;
}
- addWordPiece(piece: InterpolatedPiece, pos?: number) {
+ addWordPiece(piece: InterpolatedPiece, pos: number) {
if (this.endOfWordError) {
- throw new Error(this.endOfWordError);
+ throw new ParseError(this.endOfWordError, pos);
}
if (this.startOfWord()) {
this.wordPos = pos;
@@ -137,13 +142,14 @@ class Parser {
}
}
- return (this.next = ["ERROR", "Token not matched", startPos]);
+ throw new ParseError("Token not matched", startPos);
}
expect(type: TokenType) {
if (this.next[0] != type) {
- throw new Error(
- `Expected ${type}, found ${this.next[0]} (${this.next[1]})`
+ throw new ParseError(
+ `Expected ${type}, found ${this.next[0]} (${this.next[1]})`,
+ this.next[2]
);
}
}
@@ -175,7 +181,7 @@ class Parser {
case "[": {
this.advance();
const script = this.parseScript();
- wip.addWordPiece({ script });
+ wip.addWordPiece({ script }, pos);
this.expect("]");
break;
}
@@ -200,8 +206,6 @@ class Parser {
case "backslash":
this.advance();
continue;
- case "ERROR":
- throw new Error(chars);
default:
continue;
}
@@ -223,10 +227,11 @@ class Parser {
case "quote":
throw new Error(`Unhandled case: ${type} (${chars})`);
- case "ERROR":
- throw new Error(chars);
default:
- throw new Error(`Unhandled case: ${type satisfies never} (${chars})`);
+ throw new ParseError(
+ `Unhandled case: ${type satisfies never} (${chars})`,
+ pos
+ );
}
this.advance();
@@ -261,17 +266,19 @@ class Parser {
wip.addWordPiece({ text: "\n" }, pos);
break;
default:
- throw new Error(`Unknown backslash escape: ${chars}`);
+ throw new ParseError(`Unknown backslash escape: ${chars}`, pos);
}
break;
case "EOF":
- throw new Error(
- "Reached end of input while parsing a backslash escape"
+ throw new ParseError(
+ "Reached end of input while parsing a backslash escape",
+ pos
);
- case "ERROR":
- throw new Error(chars);
default:
- throw new Error(`Unhandled case: ${type satisfies never} (${chars})`);
+ throw new ParseError(
+ `Unhandled case: ${type satisfies never} (${chars})`,
+ pos
+ );
}
}
@@ -297,9 +304,10 @@ class Parser {
case "}":
return wip;
case "EOF":
- throw new Error("Reached end of input while parsing a brace word");
- case "ERROR":
- throw new Error(chars);
+ throw new ParseError(
+ "Reached end of input while parsing a brace word",
+ pos
+ );
default:
wip += chars;
}
From 6b1c2c48ef13f3130e8fff9e6810e3b74b7e1b3d Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 14:00:45 -0400
Subject: [PATCH 13/14] Support quoted words
---
src/parser2.ts | 62 +++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 61 insertions(+), 1 deletion(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index c9a65a8..3fc959c 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -178,6 +178,19 @@ class Parser {
break;
}
+ case "quote": {
+ if (wip.startOfWord()) {
+ wip.addWordPiece({ text: "" }, pos);
+ this.advance();
+ this.parseQuoteWord(wip);
+ this.expect("quote");
+ wip.freezeWord("Extra characters after quoted word");
+ } else {
+ wip.addWordPiece({ bare: chars }, pos);
+ }
+ break;
+ }
+
case "[": {
this.advance();
const script = this.parseScript();
@@ -225,8 +238,55 @@ class Parser {
break;
}
+ default:
+ throw new ParseError(
+ `Unhandled case: ${type satisfies never} (${chars})`,
+ pos
+ );
+ }
+
+ this.advance();
+ }
+ }
+
+ parseQuoteWord(wip: WipScript) {
+ while (true) {
+ const [type, chars, pos] = this.next;
+ switch (type) {
+ case "text":
+ case "{":
+ case "}":
+ case "]":
+ case "whitespace":
+ case "newline":
+ case "semicolon":
+ case "comment":
+ wip.addWordPiece({ text: chars }, pos);
+ break;
+
+ case "[": {
+ this.advance();
+ const script = this.parseScript();
+ wip.addWordPiece({ script }, pos);
+ this.expect("]");
+ break;
+ }
+
+ case "EOF":
+ throw new ParseError(
+ "Reached end of input while parsing a quoted word",
+ pos
+ );
+
+ case "backslash": {
+ this.advance();
+ this.parseBackslashEscape(wip, "quote");
+ break;
+ }
+
case "quote":
- throw new Error(`Unhandled case: ${type} (${chars})`);
+ return;
+
default:
throw new ParseError(
`Unhandled case: ${type satisfies never} (${chars})`,
From d22099323675f3a3dff8e3e4c8ac8dcf55861f6a Mon Sep 17 00:00:00 2001
From: Tangent Wantwight <tangent128@gmail.com>
Date: Sat, 8 Jun 2024 14:03:31 -0400
Subject: [PATCH 14/14] All original parser tests pass
---
src/parser2.ts | 12 ++++++++----
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/src/parser2.ts b/src/parser2.ts
index 3fc959c..47f9e9f 100644
--- a/src/parser2.ts
+++ b/src/parser2.ts
@@ -234,7 +234,7 @@ class Parser {
case "backslash": {
this.advance();
- this.parseBackslashEscape(wip, "bare");
+ this.parseBackslashEscape(wip, pos, "bare");
break;
}
@@ -280,7 +280,7 @@ class Parser {
case "backslash": {
this.advance();
- this.parseBackslashEscape(wip, "quote");
+ this.parseBackslashEscape(wip, pos, "quote");
break;
}
@@ -298,8 +298,12 @@ class Parser {
}
}
- parseBackslashEscape(wip: WipScript, wordType: "bare" | "quote") {
- const [type, chars, pos] = this.next;
+ parseBackslashEscape(
+ wip: WipScript,
+ pos: number,
+ wordType: "bare" | "quote"
+ ) {
+ const [type, chars] = this.next;
switch (type) {
case "newline":
if (wordType == "bare") {