Finish thp code DSL basic implementation

src/components/InteractiveCode.astro

@@ -1,5 +1,7 @@
 ---
 import { lex } from "../lexer/lexer";
+import type { Instruction } from "../thp_machine/machine_parser";
+import { parse_str } from "../thp_machine/machine_parser";
 const {code, steps} = Astro.props;
 
 /** 
@@ -97,6 +99,15 @@ function highlightCode(lines: Array<string>): string {
 }
 
 const codeHtml = highlightCode(trimAndDedent(code));
+let instructionSet: Array<Array<Instruction>>;
+try {
+    instructionSet = parse_str(steps);
+} catch (e) {
+    console.error(Astro.url);
+    throw e;
+}
+
+const serialized_inst = JSON.stringify(instructionSet);
 ---
 
 <div class="bg-black text-white rounded px-1"
@@ -104,7 +115,7 @@ const codeHtml = highlightCode(trimAndDedent(code));
         line: 0,
         stdout: "",
         ip: 0,
-        inst: ${steps},
+        inst: ${serialized_inst},
         done: false,
         state: {},
     }`}
@@ -120,7 +131,7 @@ const codeHtml = highlightCode(trimAndDedent(code));
         </div>
         <div>
             <div class="p-1 border-b border-white">state</div>
-            <div class="h-24 p-1">
+            <div class="h-24 p-1 overflow-y-scroll">
                 <template x-for="(value, key) in state">
                     <div x-text="key.replaceAll(' ', '&nbsp;') + ' = ' + value"></div>
                 </template>
@@ -128,22 +139,23 @@ const codeHtml = highlightCode(trimAndDedent(code));
         </div>
     </div>
     <div class="border-t border-white p-1">
-        <button class="font-mono px-1 rounded bg-pink-950 disabled:opacity-50 disabled:cursor-not-allowed" @click="alpineNext($data)" :disabled="done && 'true'">
+        <button class="font-mono px-1 rounded bg-pink-200 dark:bg-pink-950 text-black dark:text-white disabled:opacity-50 disabled:cursor-not-allowed" @click="alpineNext($data)" :disabled="done && 'true'">
             Step: <span x-text="ip"></span>
         </button>
-        <button class="font-mono px-1 rounded bg-pink-950" @click="alpineReset($data)">
+        <button class="font-mono px-1 rounded bg-pink-200 dark:bg-pink-950 text-black dark:text-white" @click="alpineReset($data)">
             Reset
         </button>
     </div>
 </div>
 
 <script>
-    type Instruction = "line" | "out" | "set" | "unset";
+    import { InstructionType, type Instruction } from "../thp_machine/machine_parser";
+
     type AlpineState = {
         line: number,
         stdout: string,
         ip: number,
-        inst: Array<Array<[Instruction, string, string | undefined]>>
+        inst: Array<Array<Instruction>>
         done: boolean,
         state: {[key: string]: string},
     }
@@ -156,22 +168,25 @@ const codeHtml = highlightCode(trimAndDedent(code));
 
         const instructions = data.inst[ip]!;
         for (const instructionSet of instructions) {
-            const instructionArr = instructionSet;
+            const i = instructionSet;
-            switch (instructionArr[0]) {
+
-                case "line": {
+            switch (i.t) {
-                    data.line = Number(instructionArr[1]);
+                case InstructionType.Line: {
+                    data.line = Number(i.v0);
                     break;
                 }
-                case "out": {
+                case InstructionType.Out: {
-                    data.stdout += String(instructionArr[1])
+                    data.stdout += i.v0
                     break;
                 }
-                case "set": {
+                case InstructionType.Set: {
-                    data.state[String(instructionArr[1])] = String(instructionArr[2]);
+                    const i_key = i.v0.slice(1, -1);
+                    const i_value = i.v1!.slice(1, -1);
+                    data.state[i_key] = i_value;
                     break;
                 }
-                case "unset": {
+                case InstructionType.Unset: {
-                    delete data.state[String(instructionArr[1])];
+                    delete data.state[i.v0];
                     break;
                 }
             }

src/lexer/identifier_lexer.ts

@@ -10,11 +10,11 @@ import { is_identifier_char } from "./utils";
  * @param is_datatype whether the identifier is a datatype
  */
 export function scan_identifier(input: string, starting_position: number, is_datatype = false): [Token, number] {
-    let value = input[starting_position];
+    let value = input[starting_position]!;
     let pos = starting_position + 1;
 
     while (pos < input.length) {
-        const c = input[pos];
+        const c = input[pos]!;
 
         if (is_identifier_char(c)) {
             pos += 1;

src/lexer/lexer.ts

@@ -34,7 +34,7 @@ export function lex(code: string): Array<Token> {
     let current_default_token = "";
 
     while (current_pos < code_len) {
-        const c = code[current_pos];
+        const c = code[current_pos]!;
 
         // try to scan a number
         if (is_digit(c)) {

src/pages/learn/error-handling/try.mdx

@@ -83,101 +83,36 @@ Use a naked `try` when you want to rethrow an error, if there is any.
     val res1 = run() // First, without error
     val res2 = run() // Then, an example with error
     `}
-    steps={`[
+    steps={`
-        [["line", 14]],
+    step { line 14 }
-        [["line", 7]],
+    step { line 7 }
-        [
+    step {
-            ["line", 8],
+        line 8
-            ["set", "= = =", "run() = = = ="]
+        set "= = =" "run() = = = ="
-        ],
+    }
-        [
+    step {
-            ["line", 10],
+        line 10
-            ["set", " dangerous()", "..."],
+        set " dangerous()" "..."
-        ],
+    }
-        [["line", 1]],
+    step {line 1}
-        [
+    step {
-            ["line", 2],
+        line 2
-            ["set", " = = =", "dangerous() = = = ="]
+        set " = = =" "dangerous() = = = ="
-        ],
+    }
-        [["line", 3]],
+    step {line 3}
-        [
+    step {set " Math.random()" "0.2"}
-            ["set", "  Math.random()", "0.2"]
+    step {
-        ],
+        unset " Math.random()"
-        [
+        set "  return" "Int 50"
-            ["unset", "  Math.random()"],
+    }
-            ["set", "  return", "Int 50"]
+    step {line 5}
-        ],
+    step {
-        [["line", 5]],
+        line 10
-        [
+        unset "  return"
-            ["line", 10],
+        unset " = = ="
-            ["unset", "  return"],
+        set " dangerous()" "Int 50"
-            ["unset", " = = ="],
+    }
-            ["set", " dangerous()", "Int 50"]
+`}
-        ],
-        [
-            ["unset", " dangerous()"],
-            ["set", " Int result", "Int 50"]
-        ],
-        [["line", 11]],
-        [
-            ["line", 12],
-            ["out", "The result is 50\\n"],
-        ],
-        [
-            ["line", 14],
-            ["unset", " Int result"],
-            ["unset", "= = ="],
-        ],
-        [
-            ["line", 15],
-            ["set", "!Exception res1", "<empty>"],
-        ],
-
-        [["line", 7]],
-        [
-            ["line", 8],
-            ["set", "= = =", "run() = = = ="]
-        ],
-        [
-            ["line", 10],
-            ["set", " dangerous()", "..."],
-        ],
-        [["line", 1]],
-        [
-            ["line", 2],
-            ["set", " = = =", "dangerous() = = = ="],
-        ],
-        [["line", 3]],
-        [
-            ["set", "  Math.random()", "0.9"],
-        ],
-        [
-            ["line", "4"],
-            ["unset", "  Math.random()"],
-            ["set", "  return", "Exception(\\"Unlucky\\")"],
-        ],
-        [["line", 5]],
-        [
-            ["line", 10],
-            ["unset", "  return"],
-            ["unset", " = = ="],
-            ["set", " dangerous()", "Exception \\"Unlucky\\""],
-        ],
-        [
-            ["line", 12],
-            ["unset", " dangerous()"],
-            ["set", " return", "Exception(\\"Unlucky\\")"],
-        ],
-        [
-            ["line", 15],
-            ["unset", " return"],
-            ["unset", "= = ="],
-        ],
-        [
-            ["line", 16],
-            ["set", "!Exception res2", "Exception(\\"Unlucky\\")"],
-        ],
-    ]`}
 ></InteractiveCode>
 
 

src/pages/learn/index.mdx

@@ -78,25 +78,25 @@ The documentation contains snippets of interactive THP code, like this:
 
     f(x, y)
     `}
-    steps={`[
+    steps={`
-        [["line", 1]],
+    step {line 1}
-        [
+    step {
-            ["line", 2],
+        line 2
-            ["set", "String x", "\\"android\\""]
+        set "String x" "\\"android\\""
-        ],
+    }
-        [
+    step {
-            ["line", 8],
+        line 8
-            ["set", "Int y", "17"]
+        set "Int y" "17"
-        ],
+    }
-        [["line", 4]],
+    step {line 4}
-        [["line", 5]],
+    step {line 5}
-        [
+    step {
-            ["out", "hello, android 17"],
+        out "hello, android 17"
-            ["line", 6]
+        line 6
-        ],
+    }
-        [["line", 8]],
+    step {line 8}
-        [["line", 0]]
+    step {line 0}
-    ]`}
+    `}
 ></InteractiveCode>
 
 Use the `Step` and `Reset` buttons to emulate the execution of a

src/thp_machine/machine_parser.ts

@@ -0,0 +1,218 @@
+/*
+step {
+    line 1
+    set "a" "b"
+    unset "a"
+}
+*/
+
+import { scan_number } from "../lexer/number_lexer";
+import { scan_string } from "../lexer/string_lexer";
+import { is_digit, is_lowercase, is_uppercase } from "../lexer/utils";
+
+enum TokenType {
+    Step,
+    Line,
+    Set,
+    Out,
+    Number,
+    String,
+    Unset,
+    BraceOpen,
+    BraceClose,
+};
+
+type Token = [TokenType, string | undefined];
+
+// Creates a stream of tokens
+function lex(input: string): Array<Token> {
+    const characters = input.split("");
+    const characters_len = characters.length;
+    let next_p = 0;
+
+    const tokens: Array<Token> = [];
+
+    while (next_p < characters_len)
+    {
+        const c = characters[next_p]!;
+
+        // word
+        if (is_lowercase(c) || is_uppercase(c))
+        {
+            const [token, next] = lex_word(characters, next_p);
+            tokens.push(token);
+            next_p = next;
+        }
+        // number
+        else if (is_digit(c))
+        {
+            const [token, next] = scan_number(input, next_p);
+            tokens.push([TokenType.Number, token.v]);
+            next_p = next;
+        }
+        // string
+        else if (c === "\"")
+        {
+            const [token, next] = scan_string(input, next_p);
+            tokens.push([TokenType.String, token.v]);
+            next_p = next;
+        }
+        else if (c === "{")
+        {
+            tokens.push([TokenType.BraceOpen, undefined]);
+            next_p += 1;
+        }
+        else if (c === "}")
+        {
+            tokens.push([TokenType.BraceClose, undefined]);
+            next_p += 1;
+        }
+        else if (c === " " || c === "\n")
+        {
+            next_p += 1;
+        }
+        else
+        {
+            throw new Error(`Invalid character: \`${c}\``);
+        }
+    }
+
+    return tokens;
+}
+
+function lex_word(input: Array<string>, pos: number): [Token, number] {
+    let next_p = pos;
+    let value = "";
+
+    let c = input[next_p];
+    while (c !== undefined && (is_lowercase(c) || is_uppercase(c) || is_digit(c) || c === "_"))
+    {
+        value += c;
+        next_p += 1;
+        c = input[next_p];
+    }
+
+    let token_type;
+    if (value === "step")      { token_type = TokenType.Step; }
+    else if (value === "line") { token_type = TokenType.Line; }
+    else if (value === "set")  { token_type = TokenType.Set; }
+    else if (value === "unset"){ token_type = TokenType.Unset; }
+    else if (value === "out")  { token_type = TokenType.Out; }
+    else
+    {
+        throw new Error(`Invalid word: ${value}`);
+    }
+
+    return [[token_type, value], next_p]
+}
+
+export enum InstructionType {
+    Line,
+    Set,
+    Unset,
+    Out,
+}
+
+export type Instruction = {
+    t: InstructionType,
+    v0: string,
+    v1?: string,
+}
+
+export function parse_str(input: string): Array<Array<Instruction>> {
+    return parse(lex(input));
+}
+
+// Parses the tokens into a instruction set
+function parse(tokens: Array<Token>): Array<Array<Instruction>> {
+    let pos = 0;
+    let max = tokens.length;
+
+    const ret = [];
+
+    while (pos < max) {
+        const [steps, next_pos] = parse_step(tokens, pos);
+        pos = next_pos;
+        ret.push(steps);
+    }
+
+    return ret;
+}
+
+function parse_step(tokens: Array<Token>, _pos: number): [Array<Instruction>, number] {
+    let pos = _pos;
+
+    expect(tokens, pos, TokenType.Step, "expected step");
+    pos += 1;
+    expect(tokens, pos, TokenType.BraceOpen, "expected opening brace");
+    pos += 1;
+
+    const instructions = [];
+
+    while (true) {
+        const [inst, next] = parse_instruction(tokens, pos);
+        if (inst === null) {
+            break;
+        }
+        instructions.push(inst);
+        pos = next;
+    }
+
+    expect(tokens, pos, TokenType.BraceClose, "expected closing brace");
+    pos += 1
+
+    return [instructions, pos];
+}
+
+function parse_instruction(tokens: Array<Token>, _pos: number): [Instruction|null, number] {
+    let pos = _pos;
+
+    let instruction_type = tokens[pos]![0];
+    if (instruction_type === TokenType.Line) {
+        pos += 1;
+        expect(tokens, pos, TokenType.Number, "expected a number after the `line` instruction");
+        return [{
+            t: InstructionType.Line,
+            v0: tokens[pos]![1]!,
+        }, pos + 1]
+    }
+    else if (instruction_type === TokenType.Set) {
+        pos += 1;
+        expect(tokens, pos, TokenType.String, "expected a string after the `set` instruction");
+        pos += 1;
+        expect(tokens, pos, TokenType.String, "expected a second string after the `set` instruction");
+
+        return [{
+            t: InstructionType.Set,
+            v0: tokens[pos - 1]![1]!,
+            v1: tokens[pos]![1]!,
+        }, pos + 1]
+    }
+    else if (instruction_type === TokenType.Unset) {
+        expect(tokens, pos + 1, TokenType.String, "expected a a string after the `unset` instruction");
+
+        return [{
+            t: InstructionType.Unset,
+            v0: tokens[pos + 1]![1]!,
+        }, pos + 2]
+    }
+    else if (instruction_type === TokenType.Out) {
+        expect(tokens, pos + 1, TokenType.String, "expected a a string after the `unset` instruction");
+
+        return [{
+            t: InstructionType.Out,
+            v0: tokens[pos + 1]![1]!,
+        }, pos + 2]
+    }
+
+    return [null, pos];
+}
+
+function expect(t: Array<Token>, pos: number, type: TokenType, err: string) {
+    const [t_type] = t[pos]!;
+    if (t_type !== type) {
+        console.error("`" + String(t[pos]) + "`");
+        throw new Error(err + " , got " + t[pos]);
+    }
+}
+