convex-schema-translator.ts249 lines · main
| 1 | import { ValidationError } from '@briven/shared'; |
| 2 | |
| 3 | /** |
| 4 | * Convex schema → briven schema DSL translator (offline; no convex |
| 5 | * deployment required). Accepts the literal TypeScript source of the |
| 6 | * customer's convex/schema.ts (paste-in flow from the admin row) and |
| 7 | * emits a draft briven/schema.ts the operator reviews before shipping. |
| 8 | * |
| 9 | * Scope today: defineTable() with v.string/number/int64/boolean/id/ |
| 10 | * optional. union-of-literal types fall through to text() + a TODO |
| 11 | * comment so the operator picks the validation approach. Indexes parse |
| 12 | * but become inline TODOs in the briven schema (we don't have a |
| 13 | * stable enough indexes-on-table DSL output yet). |
| 14 | * |
| 15 | * Errors: ValidationError on completely unparseable input. Warnings |
| 16 | * collected per-table for the operator to read. |
| 17 | */ |
| 18 | |
| 19 | const INPUT_CAP = 100_000; |
| 20 | |
| 21 | export interface TranslationResult { |
| 22 | brivenSchema: string; |
| 23 | warnings: readonly string[]; |
| 24 | tables: ReadonlyArray<{ name: string; columns: number }>; |
| 25 | } |
| 26 | |
| 27 | export function translateConvexSchema(source: string): TranslationResult { |
| 28 | if (!source || typeof source !== 'string') { |
| 29 | throw new ValidationError('schema source is required'); |
| 30 | } |
| 31 | if (source.length > INPUT_CAP) { |
| 32 | throw new ValidationError(`schema source exceeds ${INPUT_CAP}-character cap`); |
| 33 | } |
| 34 | |
| 35 | const warnings: string[] = []; |
| 36 | const tables: { name: string; columns: number }[] = []; |
| 37 | |
| 38 | // Strip TS comments (line + block) so the parser doesn't trip on |
| 39 | // // defineTable references inside comments. Crude but reliable. |
| 40 | const cleaned = source |
| 41 | .replace(/\/\/[^\n]*/g, '') |
| 42 | .replace(/\/\*[\s\S]*?\*\//g, ''); |
| 43 | |
| 44 | // Find every `<key>: defineTable({ ... })` block. We don't try to |
| 45 | // parse a full TS AST — a balanced-brace walk after the open is |
| 46 | // enough for the patterns convex users actually write. |
| 47 | const tableRegex = /(\w+)\s*:\s*defineTable\s*\(\s*{/g; |
| 48 | const lines: string[] = []; |
| 49 | let match: RegExpExecArray | null; |
| 50 | |
| 51 | while ((match = tableRegex.exec(cleaned)) !== null) { |
| 52 | const tableName = match[1]!; |
| 53 | const bodyStart = match.index + match[0].length; |
| 54 | const bodyEnd = findMatchingBrace(cleaned, bodyStart); |
| 55 | if (bodyEnd === -1) { |
| 56 | warnings.push( |
| 57 | `${tableName}: unbalanced braces in defineTable body — skipped. operator should hand-port this table.`, |
| 58 | ); |
| 59 | continue; |
| 60 | } |
| 61 | const body = cleaned.slice(bodyStart, bodyEnd); |
| 62 | const columns = parseColumns(body, tableName, warnings); |
| 63 | tables.push({ name: tableName, columns: columns.length }); |
| 64 | lines.push(emitTable(tableName, columns)); |
| 65 | } |
| 66 | |
| 67 | if (tables.length === 0) { |
| 68 | warnings.push( |
| 69 | 'no defineTable() calls found. paste only the schema.ts content (defineSchema + every defineTable).', |
| 70 | ); |
| 71 | } |
| 72 | |
| 73 | const brivenSchema = renderSchemaFile(lines, tables); |
| 74 | return { brivenSchema, warnings, tables }; |
| 75 | } |
| 76 | |
| 77 | interface ColumnDecl { |
| 78 | name: string; |
| 79 | brivenType: string; |
| 80 | notNull: boolean; |
| 81 | references?: string; |
| 82 | todoComment?: string; |
| 83 | } |
| 84 | |
| 85 | function findMatchingBrace(s: string, start: number): number { |
| 86 | let depth = 1; |
| 87 | for (let i = start; i < s.length; i++) { |
| 88 | const c = s[i]; |
| 89 | if (c === '{') depth++; |
| 90 | else if (c === '}') { |
| 91 | depth--; |
| 92 | if (depth === 0) return i; |
| 93 | } |
| 94 | } |
| 95 | return -1; |
| 96 | } |
| 97 | |
| 98 | const SIMPLE_TYPE_MAP: Record<string, string> = { |
| 99 | 'v.string()': 'text()', |
| 100 | 'v.number()': 'bigint()', |
| 101 | 'v.int64()': 'bigint()', |
| 102 | 'v.float64()': 'bigint()', |
| 103 | 'v.boolean()': 'boolean()', |
| 104 | 'v.bytes()': 'text()', |
| 105 | 'v.null()': 'text()', |
| 106 | 'v.any()': 'jsonb()', |
| 107 | }; |
| 108 | |
| 109 | function parseColumns( |
| 110 | body: string, |
| 111 | tableName: string, |
| 112 | warnings: string[], |
| 113 | ): ColumnDecl[] { |
| 114 | // Match `name: <expr>` pairs at the top level of the object literal. |
| 115 | // We walk the body char-by-char to handle nested parens/braces inside |
| 116 | // each value expression. |
| 117 | const cols: ColumnDecl[] = []; |
| 118 | let i = 0; |
| 119 | while (i < body.length) { |
| 120 | // Skip whitespace and commas. |
| 121 | while (i < body.length && /[\s,]/.test(body[i]!)) i++; |
| 122 | if (i >= body.length) break; |
| 123 | // Read name. |
| 124 | const nameStart = i; |
| 125 | while (i < body.length && /\w/.test(body[i]!)) i++; |
| 126 | const name = body.slice(nameStart, i); |
| 127 | if (!name) { |
| 128 | i++; |
| 129 | continue; |
| 130 | } |
| 131 | // Expect colon. |
| 132 | while (i < body.length && /\s/.test(body[i]!)) i++; |
| 133 | if (body[i] !== ':') { |
| 134 | // Not a key:value — probably a `.index(...)` call hanging off |
| 135 | // the closing brace. Bail. |
| 136 | break; |
| 137 | } |
| 138 | i++; |
| 139 | // Read value until top-level comma or end. |
| 140 | const valueStart = i; |
| 141 | let depth = 0; |
| 142 | while (i < body.length) { |
| 143 | const c = body[i]!; |
| 144 | if (c === '(' || c === '{' || c === '[') depth++; |
| 145 | else if (c === ')' || c === '}' || c === ']') depth--; |
| 146 | else if (c === ',' && depth === 0) break; |
| 147 | i++; |
| 148 | } |
| 149 | const valueExpr = body.slice(valueStart, i).trim(); |
| 150 | const col = translateColumn(name, valueExpr, tableName, warnings); |
| 151 | if (col) cols.push(col); |
| 152 | } |
| 153 | return cols; |
| 154 | } |
| 155 | |
| 156 | function translateColumn( |
| 157 | name: string, |
| 158 | expr: string, |
| 159 | tableName: string, |
| 160 | warnings: string[], |
| 161 | ): ColumnDecl | null { |
| 162 | let notNull = true; |
| 163 | let working = expr; |
| 164 | // Peel one layer of v.optional(...). |
| 165 | const optMatch = /^v\.optional\s*\(\s*([\s\S]+)\s*\)$/.exec(working); |
| 166 | if (optMatch) { |
| 167 | notNull = false; |
| 168 | working = optMatch[1]!.trim(); |
| 169 | } |
| 170 | // v.id("users") → text().references("users", "id") |
| 171 | const idMatch = /^v\.id\s*\(\s*['"]([^'"]+)['"]\s*\)$/.exec(working); |
| 172 | if (idMatch) { |
| 173 | return { |
| 174 | name, |
| 175 | brivenType: `text().references('${idMatch[1]}', 'id')`, |
| 176 | notNull, |
| 177 | references: idMatch[1], |
| 178 | }; |
| 179 | } |
| 180 | // Simple types. |
| 181 | if (SIMPLE_TYPE_MAP[working]) { |
| 182 | return { name, brivenType: SIMPLE_TYPE_MAP[working]!, notNull }; |
| 183 | } |
| 184 | // Array / object — store as jsonb. |
| 185 | if (working.startsWith('v.array(') || working.startsWith('v.object(')) { |
| 186 | return { |
| 187 | name, |
| 188 | brivenType: 'jsonb()', |
| 189 | notNull, |
| 190 | todoComment: 'convex v.array/v.object → jsonb; refine to nested table if you query into it', |
| 191 | }; |
| 192 | } |
| 193 | // Union — usually v.union(v.literal("a"), v.literal("b")). Treat as |
| 194 | // text() + a TODO. The narrower validation gets done in the function |
| 195 | // layer with zod, mirroring the docs guidance. |
| 196 | if (working.startsWith('v.union(')) { |
| 197 | return { |
| 198 | name, |
| 199 | brivenType: 'text()', |
| 200 | notNull, |
| 201 | todoComment: 'convex v.union — validate this enum at the function layer with zod', |
| 202 | }; |
| 203 | } |
| 204 | warnings.push( |
| 205 | `${tableName}.${name}: couldn't translate "${expr.slice(0, 60)}". emitting text() — review manually.`, |
| 206 | ); |
| 207 | return { |
| 208 | name, |
| 209 | brivenType: 'text()', |
| 210 | notNull, |
| 211 | todoComment: `unrecognised convex type: ${expr.slice(0, 80)}`, |
| 212 | }; |
| 213 | } |
| 214 | |
| 215 | function emitTable(name: string, columns: ColumnDecl[]): string { |
| 216 | const colLines = columns |
| 217 | .map((c) => { |
| 218 | const todo = c.todoComment ? ` // TODO: ${c.todoComment}\n` : ''; |
| 219 | const notNull = c.notNull ? '.notNull()' : ''; |
| 220 | return `${todo} ${c.name}: ${c.brivenType}${notNull},`; |
| 221 | }) |
| 222 | .join('\n'); |
| 223 | return ` ${name}: table({ |
| 224 | columns: { |
| 225 | ${colLines} |
| 226 | }, |
| 227 | }),`; |
| 228 | } |
| 229 | |
| 230 | function renderSchemaFile( |
| 231 | tableLines: string[], |
| 232 | tables: { name: string; columns: number }[], |
| 233 | ): string { |
| 234 | const header = `import { bigint, boolean, jsonb, schema, table, text } from '@briven/cli/schema'; |
| 235 | |
| 236 | // Generated from convex/schema.ts by the briven schema translator. |
| 237 | // Review every TODO before deploying. Convex's _creationTime is implicit |
| 238 | // in convex but explicit in briven — add createdAt: bigint().notNull() |
| 239 | // to tables that need it. Convex indexes (.index("by_x", ["x"])) move |
| 240 | // to a table-level \`indexes: [{ columns: ['x'] }]\` array; not yet |
| 241 | // auto-emitted by the translator. |
| 242 | |
| 243 | `; |
| 244 | const summary = `// translator summary: ${tables.length} table${tables.length === 1 ? '' : 's'} — ${tables.map((t) => `${t.name}(${t.columns})`).join(', ') || 'none'}.\n\n`; |
| 245 | return `${header}${summary}export default schema({ |
| 246 | ${tableLines.join('\n\n')} |
| 247 | }); |
| 248 | `; |
| 249 | } |