diff --git a/packages/opencode/src/altimate/tools/cte-detect.ts b/packages/opencode/src/altimate/tools/cte-detect.ts new file mode 100644 index 0000000000..816aa6405f --- /dev/null +++ b/packages/opencode/src/altimate/tools/cte-detect.ts @@ -0,0 +1,382 @@ +/** + * CTE detection and count-query building for the CTE profiler. + * + * Works on the raw model text (Jinja included) with string offsets. The scanner + * skips SQL line/block comments, Jinja comments and string literals, and + * understands quoted CTE names and CTE column lists. + */ + +export interface CteInfo { + /** Identifier as written, including a trailing column list such as `my_cte (id, name)`. */ + name: string + /** Offset where the name starts. */ + nameStart: number + /** Offset just after the name (and its column list). */ + nameEnd: number + /** Offset of the first character after the CTE body's opening parenthesis. */ + queryStart: number + /** Offset of the CTE body's closing parenthesis. */ + queryEnd: number + /** Order of the CTE within its WITH clause. */ + index: number + /** Offset of the WITH keyword that opens this CTE's clause. */ + withClauseStart: number +} + +/** Position just past the string literal, quoted identifier or comment (`--`, block, Jinja) that starts at `pos`; `pos` itself when none does. An unterminated one runs to the end of the text. */ +function skipLiteral(text: string, pos: number): number { + const char = text[pos] + const next = text[pos + 1] + if (char === "'" || char === '"' || char === "`") { + let i = pos + 1 + while (i < text.length) { + if (text[i] === char) { + // A doubled quote is an escaped quote. + if (text[i + 1] === char) i += 2 + else return i + 1 + } else i++ + } + return text.length + } + if (char === "$" && !/[\p{L}\p{N}_$]/u.test(text[pos - 1] ?? " ")) { + // A dollar-quoted string (`$$ … $$`, `$tag$ … $tag$`) runs to its matching tag. + const tag = /^\$(?:[A-Za-z_][A-Za-z0-9_]*)?\$/.exec(text.substring(pos, pos + 70))?.[0] + if (tag) { + const end = text.indexOf(tag, pos + tag.length) + return end === -1 ? text.length : end + tag.length + } + } + const close = char === "-" && next === "-" ? "\n" : char === "/" && next === "*" ? "*/" : char === "{" && next === "#" ? "#}" : null + if (close === null) return pos + const end = text.indexOf(close, pos + 2) + if (end === -1) return text.length + return close === "\n" ? end : end + close.length +} + +/** Position of the first character at or after `pos` that is not whitespace, a comment, or (when `jinja`) a Jinja `{% %}` tag. */ +function skipTrivia(text: string, pos: number, jinja: boolean): number { + while (pos < text.length) { + if (/\s/.test(text[pos])) { + pos++ + continue + } + const end = skipLiteral(text, pos) + if (end !== pos && !/['"`]/.test(text[pos])) { + pos = end + continue + } + if (jinja && text[pos] === "{" && text[pos + 1] === "%") { + pos = skipJinjaBlock(text, pos) + continue + } + return pos + } + return pos +} + +/** Offset of the `)` that closes the `(` at `openParenPos`, or -1 when it is never closed. */ +function findMatchingClosingParen(text: string, openParenPos: number): number { + let depth = 1 + let pos = openParenPos + 1 + while (pos < text.length) { + const end = skipLiteral(text, pos) + if (end !== pos) { + pos = end + continue + } + // A captured block (`{% set x %}…{% endset %}`) is not the CTE's SQL: a `)` written there does not close it. + if (text[pos] === "{" && text[pos + 1] === "%") { + pos = skipJinjaBlock(text, pos) + continue + } + if (text[pos] === "(") depth++ + else if (text[pos] === ")" && --depth === 0) return pos + pos++ + } + return -1 +} + +interface JinjaTag { + /** Tag name (`if`, `set`, `macro`…), without the `end` prefix. */ + name: string + /** The tag closes a block (`{% endif %}`). */ + closes: boolean + /** The tag opens a block: not an `end` tag, `else`/`elif`, or an inline `{% set x = … %}`. */ + opens: boolean + /** Offset just past the tag. */ + end: number +} + +const BLOCK_TAGS = new Set(["if", "for", "macro", "call", "filter", "with", "set", "raw", "block"]) + +/** The block tag of the `{% … %}` that spans `at` to `close` (the offset of its `%}`); undefined for any other tag. */ +function parseTag(text: string, at: number, close: number): JinjaTag | undefined { + const head = /^\{%[-+]?\s*(end)?(\w+)/.exec(text.substring(at, Math.min(close + 2, at + 200))) + if (!head || !BLOCK_TAGS.has(head[2])) return undefined + const closes = head[1] !== undefined + // `{% set x = … %}` is an inline set; `{% set x %}` and `{% set x | filter %}` open a block. + const inline = head[2] === "set" && /^\{%[-+]?\s*set\s+[\w.,\s()]+=(?!=)/.test(text.substring(at, close + 2)) + return { name: head[2], closes, opens: !closes && !inline, end: close + 2 } +} + +/** + * The Jinja statement tags of `text` from `from`, in order, in one linear pass. Comments (`{# #}`) and the + * inside of `{% raw %}` are not tags; an unterminated comment or tag ends the text. + */ +function* jinjaTags(text: string, from: number): Generator { + let pos = from + for (;;) { + const at = text.indexOf("{", pos) + if (at === -1 || at + 1 >= text.length) return + const kind = text[at + 1] + if (kind !== "#" && kind !== "%") { + pos = at + 1 + continue + } + const close = text.indexOf(kind === "#" ? "#}" : "%}", at + 2) + if (close === -1) return + pos = close + 2 + if (kind === "#") continue + const tag = parseTag(text, at, close) + if (!tag) continue + if (tag.name === "raw" && tag.opens) { + // Nothing inside `{% raw %}` is Jinja: the block runs to its end tag. + const end = /\{%[-+]?\s*endraw\b/g + end.lastIndex = pos + const found = end.exec(text) + if (!found) { + yield { name: "raw", closes: false, opens: true, end: text.length } + return + } + const endClose = text.indexOf("%}", found.index) + pos = endClose === -1 ? text.length : endClose + 2 + yield { name: "raw", closes: false, opens: false, end: pos } + continue + } + yield tag + } +} + +/** + * Position just past the Jinja tag at `pos`, and past its whole block when it opens one whose text + * is not the model's own SQL (`{% set x %}…{% endset %}`, `{% call %}`, `{% macro %}`): a query + * written there is a string for dbt to run, not a CTE of the model. + */ +function skipJinjaBlock(text: string, pos: number): number { + const close = text.indexOf("%}", pos + 2) + if (close === -1) return text.length + const first = parseTag(text, pos, close) + if (!first || !first.opens || (first.name !== "set" && first.name !== "call" && first.name !== "macro")) return close + 2 + // Blocks of the same kind nest: the block ends at the end tag that matches this one. + let depth = 1 + for (const tag of jinjaTags(text, close + 2)) { + if (tag.name !== first.name) continue + if (tag.closes) depth-- + else if (tag.opens) depth++ + if (depth === 0) return tag.end + } + return text.length +} + +/** Every WITH keyword outside comments, strings and parentheses: a WITH inside a CTE body or a subquery belongs to that query, not to the model. */ +function findTopLevelWithKeywords(text: string): number[] { + const positions: number[] = [] + const keyword = /with\b/iy + let depth = 0 + let pos = 0 + while (pos < text.length) { + const end = skipLiteral(text, pos) + if (end !== pos) { + pos = end + continue + } + const char = text[pos] + if (char === "{" && text[pos + 1] === "%") { + pos = skipJinjaBlock(text, pos) + continue + } + if (char === "(") depth++ + else if (char === ")") depth = Math.max(0, depth - 1) + else if (depth === 0 && (char === "w" || char === "W") && !/[A-Za-z0-9_$]/.test(text[pos - 1] ?? " ")) { + keyword.lastIndex = pos + if (keyword.test(text)) { + positions.push(pos) + pos += 4 + continue + } + } + pos++ + } + return positions +} + +/** Offset just past the identifier (`a`, `"a""b"`, `` `a` ``, `[a]`, `db.schema.a`) that starts at `pos`; -1 when none does. */ +function readIdentifier(text: string, pos: number): number { + let end = pos + for (;;) { + const char = text[end] + if (char === '"' || char === "`") { + const close = skipLiteral(text, end) + if (close === end || text[close - 1] !== char || close - end < 2) return -1 + end = close + } else if (char === "[") { + const bracketed = /\[(?:[^\]\n]|\]\]){1,256}\]/y + bracketed.lastIndex = end + if (!bracketed.test(text)) return -1 + end = bracketed.lastIndex + } else { + const bare = /[\p{L}_][\p{L}\p{N}_$]*/uy + bare.lastIndex = end + if (!bare.test(text)) return -1 + end = bare.lastIndex + } + if (text[end] !== ".") return end + end++ + } +} + +/** Offset of the `(` that opens a CTE body after `AS [NOT] [MATERIALIZED]` starting at `pos`; -1 when the text there is not such a header. */ +function skipHeader(text: string, pos: number): number { + let at = pos + const word = (w: string) => { + const re = new RegExp(`${w}\\b`, "iy") + re.lastIndex = at + if (!re.test(text)) return false + at = skipTrivia(text, at + w.length, true) + return true + } + if (!word("as")) return -1 + if (word("not")) { + if (!word("materialized")) return -1 + } else word("materialized") + return text[at] === "(" ? at : -1 +} + +/** The CTEs of the WITH clause that opens at `withPos`: each `name [(columns)] AS (body)`, separated by commas, up to the first thing that is not one. */ +function parseWithClause(text: string, withPos: number, ctes: CteInfo[]): void { + let pos = skipTrivia(text, withPos + 4, true) + const keyword = /recursive\b/iy + keyword.lastIndex = pos + if (keyword.test(text)) pos = skipTrivia(text, pos + "recursive".length, true) + + for (let index = 0; ; index++) { + const nameStart = pos + const identifierEnd = readIdentifier(text, nameStart) + if (identifierEnd === -1) return + let nameEnd = identifierEnd + let name = text.substring(nameStart, identifierEnd) + pos = skipTrivia(text, identifierEnd, false) + if (text[pos] === "(") { + const close = findMatchingClosingParen(text, pos) + if (close === -1) return + // The name as written plus its column list; a comment between them is left out. + name += ` ${text.substring(pos, close + 1)}` + nameEnd = close + 1 + pos = skipTrivia(text, nameEnd, false) + } + // `AS [NOT] [MATERIALIZED] (`, with comments or Jinja tags allowed between the words. + const open = skipHeader(text, pos) + if (open === -1) return + const close = findMatchingClosingParen(text, open) + if (close === -1) return + ctes.push({ + name, + nameStart, + nameEnd, + queryStart: open + 1, + queryEnd: close, + index, + withClauseStart: withPos, + }) + pos = skipTrivia(text, close + 1, true) + if (text[pos] !== ",") return + pos = skipTrivia(text, pos + 1, true) + } +} + +/** Find every CTE of the model's own WITH clauses, in source order. A WITH inside a CTE body or a subquery is not listed: its CTEs cannot be counted without the query around them. */ +export function detectCtes(text: string): CteInfo[] { + const ctes: CteInfo[] = [] + for (const withPos of findTopLevelWithKeywords(text)) parseWithClause(text, withPos, ctes) + return ctes +} + +/** + * Offset from which the CTEs could not be read, or undefined when every CTE was read: the + * comma after the last CTE read of a WITH clause that goes on, or the start of a clause where none was read + * (its first name is built by a Jinja expression or loop). The list is short from there on. + */ +export function unreadCteAfter(text: string, ctes: CteInfo[]): number | undefined { + const found: number[] = [] + for (const withPos of findTopLevelWithKeywords(text)) { + const own = ctes.filter((c) => c.withClauseStart === withPos) + if (own.length === 0) { + // `START WITH` (a hierarchical query) opens no CTE list. + if (/\bstart\s*$/i.test(text.substring(Math.max(0, withPos - 24), withPos))) continue + let first = skipTrivia(text, withPos + 4, true) + if (/recursive\b/iy.test(text.substring(first, first + 10))) first = skipTrivia(text, first + "recursive".length, true) + const unreadName = /^[\p{L}\p{N}_$]*\{\{/u.exec(text.substring(first, first + 130)) + if (unreadName) found.push(first) + continue + } + const last = own[own.length - 1] + const next = skipTrivia(text, last.queryEnd + 1, true) + if (text[next] === ",") found.push(next) + } + return found.length ? Math.min(...found) : undefined +} + +/** Drop a trailing column list: `my_cte (id, name)` -> `my_cte`. */ +export function stripColumnList(rawName: string): string { + const end = readIdentifier(rawName, 0) + return end === -1 || !rawName.substring(end).trimStart().startsWith("(") ? rawName.trim() : rawName.substring(0, end) +} + +export function quoteSqlIdentifier(identifier: string): string { + if (identifier.match(/^["'`[]/) || identifier.includes(".")) return identifier + if (!identifier.match(/^[\p{L}_][\p{L}\p{N}_$]*$/u)) return `"${identifier}"` + return identifier +} + +/** Name of the first Jinja block (`if`, `for`, `macro`, `call`, `filter`, `with`, `raw`, or `set` as a block) opened in `text` and never closed there; undefined when every block is closed. Tags inside Jinja comments and `{% raw %}` do not count. */ +export function unclosedJinjaBlock(text: string): string | undefined { + const open: string[] = [] + const counts = new Map() + for (const tag of jinjaTags(text, 0)) { + if (tag.closes) { + // An end tag closes the nearest open block of its kind, and any block opened inside it. + if (!counts.get(tag.name)) continue + for (let top = open.pop(); top !== undefined; top = open.pop()) { + counts.set(top, counts.get(top)! - 1) + if (top === tag.name) break + } + } else if (tag.opens) { + open.push(tag.name) + counts.set(tag.name, (counts.get(tag.name) ?? 0) + 1) + } + } + return open[0] +} + +/** + * `SELECT COUNT(*)` over one CTE: the model's text up to the end of that CTE (dbt configs, variables, + * Jinja between CTEs and every earlier CTE of the WITH clause, as written), so the query compiles on its own. + * Undefined for a CTE of a second WITH clause in the same file, which has no standalone query. + */ +export function buildCountQuery(text: string, ctes: CteInfo[], target: CteInfo): string | undefined { + // The text before a later WITH holds the earlier clause's whole query: nothing valid stands alone there. + if (ctes.some((c) => c.withClauseStart < target.withClauseStart)) return undefined + if (target.queryStart <= target.nameStart) return undefined + const head = text.substring(0, target.queryEnd + 1).trim() + return `${head}\nSELECT COUNT(*) AS _profile_count FROM ${quoteSqlIdentifier(stripColumnList(target.name))}` +} + +/** + * Read the single COUNT(*) value by position: Snowflake and Oracle upper-case + * the alias, so a lookup by name would report 0 rows. + */ +export function extractRowCount(data: Record[] | undefined): number { + if (!data || data.length === 0) return 0 + const first = Object.values(data[0])[0] + return typeof first === "number" ? first : Number(first) || 0 +} diff --git a/packages/opencode/src/altimate/tools/dbt-cte-profile.ts b/packages/opencode/src/altimate/tools/dbt-cte-profile.ts index 0e4913f06d..512cb3851b 100644 --- a/packages/opencode/src/altimate/tools/dbt-cte-profile.ts +++ b/packages/opencode/src/altimate/tools/dbt-cte-profile.ts @@ -1,17 +1,273 @@ +import fs from "fs" +import path from "path" import z from "zod" import { Tool } from "../../tool/tool" +import { loadRawManifest, findModel } from "../native/dbt/helpers" +import { projectRoot } from "../native/connections/registry" +import { isQueryTimeoutError } from "@altimateai/drivers/types" +import { executeQuery } from "../native/query-execution" +import { dbtAdapterFor } from "./dbt-adapter-access" +import { buildCountQuery, detectCtes, extractRowCount, stripColumnList, unclosedJinjaBlock, unreadCteAfter, type CteInfo } from "./cte-detect" + +/** Folders a model search never enters. */ +const SKIPPED_DIRS = new Set(["target", "dbt_packages", "dbt_modules", "node_modules", "logs", "venv", "site-packages"]) + +/** Deepest folder level the model search follows. */ +const MAX_SEARCH_DEPTH = 12 + +/** Largest model file the tool reads. */ +const MAX_MODEL_BYTES = 4 * 1024 * 1024 + +/** A CTE name without its column list or surrounding quotes, quote escapes undone. */ +function bareName(name: string): string { + const plain = stripColumnList(name) + const quote = plain[0] + if ((quote === '"' || quote === "`") && plain.length > 1 && plain.endsWith(quote)) return plain.slice(1, -1).replaceAll(quote + quote, quote) + if (quote === "[" && plain.endsWith("]")) return plain.slice(1, -1).replaceAll("]]", "]") + return plain +} + +function inside(root: string, file: string): boolean { + const rel = path.relative(root, file) + return rel !== "" && rel !== ".." && !rel.startsWith(".." + path.sep) && !path.isAbsolute(rel) +} + +function realpath(p: string): string { + try { + return fs.realpathSync(p) + } catch { + return path.resolve(p) + } +} + +/** Every `.sql` under the project, outside the build folders. */ +function findSqlFiles(root: string, name: string): string[] { + const found: string[] = [] + const walk = (dir: string, depth: number) => { + let entries: fs.Dirent[] + try { + entries = fs.readdirSync(dir, { withFileTypes: true }) + } catch { + return + } + for (const entry of entries) { + if (entry.name.startsWith(".")) continue + const full = path.join(dir, entry.name) + if (entry.isDirectory()) { + if (depth < MAX_SEARCH_DEPTH && !SKIPPED_DIRS.has(entry.name)) walk(full, depth + 1) + } else if (entry.isFile() && entry.name === `${name}.sql`) { + found.push(full) + } + } + } + walk(root, 0) + return found.sort() +} + +/** The model's source file: a project-relative or absolute `.sql` path, the manifest's record of a name or unique_id, or a search by name. */ +function modelFile(root: string, model: string): { file?: string; error?: string } { + const realRoot = realpath(root) + if (/\.sql$/i.test(model) || /[\\/]/.test(model)) { + const candidate = path.resolve(realRoot, model) + if (!/\.sql$/i.test(candidate)) return { error: `${model} is not a .sql file` } + if (!inside(realRoot, realpath(candidate))) return { error: `${model} is outside the dbt project` } + return fs.existsSync(candidate) && fs.statSync(candidate).isFile() ? { file: realpath(candidate) } : { error: `no file ${model} in the dbt project` } + } + try { + const manifest = loadRawManifest(path.join(realRoot, "target", "manifest.json")) + const own = /^name:\s*['"]?([\w-]+)/m.exec(fs.readFileSync(path.join(realRoot, "dbt_project.yml"), "utf-8"))?.[1] + let node = manifest?.nodes ? findModel(manifest.nodes, model) : null + if (node && !model.startsWith("model.")) { + // The project's own model of that name comes before a package's; versions of one model share a name and the manifest can't say which is meant. + const named = Object.values(manifest.nodes).filter((n) => n.resource_type === "model" && n.name === model) + const mine = named.filter((n) => !own || !n.package_name || n.package_name === own) + const pool = mine.length > 0 ? mine : named + if (pool.length > 1) return { error: `${model} matches ${pool.length} models (${pool.map((n) => n.unique_id).join(", ")}): pass the path of the one to profile` } + node = pool[0] + } + if (node?.original_file_path) { + // dbt records the path relative to the model's own package: only the project's models can be read here. + if (node.package_name && own && node.package_name !== own) return { error: `${model} belongs to the package ${node.package_name}: pass the path of its .sql file` } + const candidate = path.resolve(realRoot, node.original_file_path) + if (inside(realRoot, realpath(candidate)) && fs.existsSync(candidate)) return { file: realpath(candidate) } + } + } catch { + // A manifest that can't be read doesn't stop the search by name. + } + // A unique_id names one model of one package: only the manifest can say which file it is. + if (model.startsWith("model.")) return { error: `${model} is not in target/manifest.json: run dbt compile, or pass the model's name or path` } + const matches = findSqlFiles(realRoot, model) + if (matches.length === 1) return { file: realpath(matches[0]) } + if (matches.length === 0) return { error: `no model named ${model} in the dbt project` } + const rel = matches.map((m) => path.relative(realRoot, m)) + return { error: `${model} matches ${matches.length} files (${rel.join(", ")}): pass the path` } +} + +/** The dbt project of the session's directory; else the one the dbt adapter is set up for. */ +async function projectDir(): Promise { + const { findProjectRoot } = await import("../../../../dbt-tools/src/config") + const here = findProjectRoot(projectRoot()) + if (here) return here + const { getDbtProjectRoot } = await import("../native/connections/register") + return getDbtProjectRoot() +} + +// The Jinja tags of a query, found by a scan that steps over quoted strings so a `}}` inside one does not end the tag early. +function jinjaTags(sql: string): string[] { + const text = sql.replace(/\{#[\s\S]*?#\}/g, "").replace(/\{%-?\s*raw\s*-?%\}[\s\S]*?\{%-?\s*endraw\s*-?%\}/g, "") + const tags: string[] = [] + for (let i = 0; i < text.length - 1; i++) { + const open = text.substr(i, 2) + if (open !== "{{" && open !== "{%") continue + const close = open === "{{" ? "}}" : "%}" + let j = i + 2 + while (j < text.length && !text.startsWith(close, j)) { + const ch = text[j] + if (ch === "'" || ch === '"') { + j++ + while (j < text.length && text[j] !== ch) j += text[j] === "\\" ? 2 : 1 + } + j++ + } + tags.push(text.substring(i, j)) + i = j + 1 + } + return tags +} + +// True when a Jinja tag (quoted strings inside it left out) calls is_incremental() or names this. +function readsModelContext(sql: string): boolean { + return jinjaTags(sql).some((tag) => /\bis_incremental\s*\(|\bthis\b/.test(tag.replace(/'(?:[^'\\]|\\.)*'|"(?:[^"\\]|\\.)*"/g, "''"))) +} + +// The CTEs a typed name picks: those spelled exactly so, else those that differ only in letter case. +function nameMatches(list: CteInfo[], typed: string): CteInfo[] { + const exact = list.filter((c) => bareName(c.name) === typed) + return exact.length > 0 ? exact : list.filter((c) => bareName(c.name).toLowerCase() === typed.toLowerCase()) +} + +function failure(title: string, output: string, extra: Record = {}) { + return { title: `dbt_cte_profile: ${title}`, metadata: { success: false, error: output, ...extra }, output } +} export const DbtCteProfileTool = Tool.define("dbt_cte_profile", { - description: "List a dbt model's CTEs, or time one CTE with a row-count query.", + description: + "List a dbt model's CTEs, or time one CTE with a row-count query. Without `cte` or `position` it returns the CTE names in source order (name, line, position) and runs nothing, so a read-only call works. With one of them it runs `SELECT COUNT(*)` over that CTE and the CTEs before it in the same WITH clause on the warehouse, and returns the elapsed milliseconds and the row count; that run needs write permission (`--allow-write`) because it starts the dbt adapter and compiles the model. Each run recomputes the earlier CTEs, so a model of N CTEs costs N queries. Only the model's own top-level WITH clause is read: a WITH inside a CTE body is part of that CTE. The list stops at a CTE whose name Jinja builds, and `incomplete` is set in the metadata then; a CTE whose count query reads is_incremental() or this, in its own SQL, an earlier CTE or the model header, is not run, because the count runs outside the model. Macros the query calls are not opened: one that reads is_incremental() inside is not caught, so treat the count of such a model as a full-history count.", parameters: z.object({ - model: z.string().describe("Model name"), - cte: z.string().optional().describe("CTE to time; without it, the CTE list is returned"), + model: z + .string() + .describe("Model name, unique_id, or project-relative or absolute path to the model's .sql file (e.g. 'customers' or 'models/marts/customers.sql')"), + cte: z.string().optional().describe("CTE to time, by name; without it (and without `position`), the CTE list is returned"), + position: z + .number() + .int() + .min(0) + .optional() + .describe("The CTE's position in the list (0-based); needed when several CTEs share a name"), }), - async execute() { - return { - title: "dbt_cte_profile: not implemented", - metadata: { success: false, error: "not implemented" }, - output: "dbt_cte_profile is not implemented in this build of altimate-code yet.", + async execute(args, ctx) { + const root = await projectDir() + if (!root) return failure("no dbt project", "No dbt project here: run from a dbt project folder.") + const located = modelFile(root, args.model) + if (!located.file) return failure("model not found", located.error ?? `no model named ${args.model}`) + const modelPath = located.file + const size = fs.statSync(modelPath).size + if (size > MAX_MODEL_BYTES) return failure("model too large", `${args.model} is ${size} bytes; the largest file read is ${MAX_MODEL_BYTES}.`) + const text = fs.readFileSync(located.file, "utf-8") + const ctes = detectCtes(text) + const file = path.relative(realpath(root), located.file) + + if (args.cte === undefined && args.position === undefined) { + // The CTEs come in source order: count the lines once, not once per CTE. + let line = 1 + let counted = 0 + const lineAt = (offset: number) => { + for (; counted < offset; counted++) if (text[counted] === "\n") line++ + return line + } + const list = ctes.map((c, position) => ({ + name: stripColumnList(c.name), + line: lineAt(c.nameStart), + index: c.index, + clause: c.withClauseStart, + position, + })) + const stop = unreadCteAfter(text, ctes) + const note = stop === undefined ? "" : `\nThe list is incomplete: the CTEs from line ${text.substring(0, stop).split("\n").length} on could not be read (its name is built by a Jinja expression or loop).` + const output = (list.length + ? [`${list.length} CTE${list.length === 1 ? "" : "s"} in ${file}:`, ...list.map((c) => `position ${c.position}: ${c.name} (line ${c.line})`)].join("\n") + : `No CTEs in ${file}.`) + note + return { + title: `dbt_cte_profile: ${list.length} CTE${list.length === 1 ? "" : "s"} in ${args.model}`, + metadata: { success: true, model: args.model, file, ctes: list, ...(stop === undefined ? {} : { incomplete: true }) }, + output, + } + } + + const unread = unreadCteAfter(text, ctes) !== undefined + const incompleteNote = unread ? " The list is incomplete: a CTE built by a Jinja expression or loop could not be read." : "" + const names = ctes.map((c) => stripColumnList(c.name)).join(", ") + let target: CteInfo | undefined + if (args.position !== undefined) { + target = ctes[args.position] + if (!target) return failure("CTE not found", `${args.model} has ${ctes.length} CTEs: there is none at position ${args.position}.${incompleteNote}`, { model: args.model }) + // The name as typed wins here too: another CTE spelled exactly so is the one meant, whatever the case of this one. + const typed = args.cte === undefined ? undefined : bareName(args.cte) + if (typed !== undefined && !nameMatches(ctes, typed).includes(target)) + return failure("CTE not found", `The CTE at position ${args.position} of ${args.model} is ${stripColumnList(target.name)}, not ${args.cte}.`, { model: args.model }) + } else { + // The name as typed wins; letter case only decides when nothing matches it exactly. + const wanted = bareName(args.cte ?? "") + const matches = nameMatches(ctes, wanted) + if (matches.length === 0) + return failure("CTE not found", `${args.model} has no CTE named ${args.cte}${names ? `; it has: ${names}` : ""}.${incompleteNote}`, { model: args.model }) + if (matches.length > 1) { + const at = matches.map((c) => ctes.indexOf(c)).join(", ") + return failure("CTE ambiguous", `${args.model} has ${matches.length} CTEs that match ${args.cte} (positions ${at}): pass position.`, { model: args.model }) + } + target = matches[0] + } + const query = buildCountQuery(text, ctes, target) + const name = stripColumnList(target.name) + if (!query) { + const block = unclosedJinjaBlock(text.substring(0, target.withClauseStart)) + return failure( + "no query", + block + ? `${name} sits in a WITH clause of ${args.model} that is one branch of a Jinja {% ${block} %} block, so it has no query of its own to count.` + : `${name} is in a second WITH clause of ${args.model}, which has no query of its own to count.`, + { model: args.model, cte: name }, + ) + } + const block = unclosedJinjaBlock(query) + if (block) return failure("no query", `${name} is inside a Jinja {% ${block} %} block of ${args.model}, which cannot be cut off at that CTE and stay valid: compile the model and profile the compiled SQL, or move the WITH out of the block.`, { model: args.model, cte: name }) + + // The adapter runs the query as an inline node, not as the model, so is_incremental() and this would resolve for that node instead. + if (readsModelContext(query)) return failure("not run", `${name} depends on is_incremental() or this in ${args.model}. Where the count runs as an inline node (the CLI and Fusion modes), those resolve for that node, not the model, so the count would not match what the model reads. Compile the model and profile the compiled SQL instead.`, { model: args.model, cte: name }) + + const access = await dbtAdapterFor(ctx) + if (!access.adapter) return failure("not run", access.reason ?? "the dbt adapter is not available", { model: args.model, cte: name, refusal: access.refusal }) + const adapter = access.adapter + ctx.abort.throwIfAborted() + const started = performance.now() + try { + // The adapter has no cancel of its own: the run's timeout, budget and abort end the wait, not the query. + const result: { data?: Record[] } = await executeQuery({ type: "dbt" }, {}, { signal: ctx.abort, sessionID: ctx.sessionID }, () => + adapter.immediatelyExecuteSQLWithLimit(query, path.basename(modelPath, ".sql"), 1), + ) + const queryMs = Math.round(performance.now() - started) + const rows = extractRowCount(result?.data) + return { + title: `dbt_cte_profile: ${name} ${queryMs} ms`, + metadata: { success: true, model: args.model, cte: name, index: target.index, position: ctes.indexOf(target), queryMs, rows }, + output: `${name}: ${queryMs} ms, ${rows} row${rows === 1 ? "" : "s"}`, + } + } catch (e) { + if (ctx.abort.aborted) throw e + const message = e instanceof Error ? e.message : String(e) + // A timeout gives up the wait, not the query: it may still be running on the warehouse. + const timedOut = isQueryTimeoutError(e) + return failure(timedOut ? "timed out" : "query failed", `The count query for ${name} ${timedOut ? "timed out and may still be running on the warehouse" : "failed"}: ${message}`, { model: args.model, cte: name, ...(timedOut && { timedOut }) }) } }, }) diff --git a/packages/opencode/test/altimate/cte-detect.test.ts b/packages/opencode/test/altimate/cte-detect.test.ts new file mode 100644 index 0000000000..f4adbd2346 --- /dev/null +++ b/packages/opencode/test/altimate/cte-detect.test.ts @@ -0,0 +1,1641 @@ +import { describe, expect, it } from "bun:test" +import { + buildCountQuery, + detectCtes as detectCtesInText, + extractRowCount, + quoteSqlIdentifier, + stripColumnList, + unclosedJinjaBlock, + type CteInfo, +} from "../../src/altimate/tools/cte-detect" + +// Detector cases ported from the dbt Power User CTE detector suite. A "document" is the model text. +describe("detectCtes", () => { + const detectCtes = (document: string): CteInfo[] => detectCtesInText(document) + + const assertCTE = (cte: CteInfo, expectedName: string, expectedIndex: number, _description?: string) => { + expect(cte.name).toBe(expectedName) + expect(cte.index).toBe(expectedIndex) + expect(typeof cte.nameStart).toBe("number") + expect(typeof cte.queryStart).toBe("number") + expect(typeof cte.withClauseStart).toBe("number") + } + + describe("Basic CTE Detection", () => { + it("should detect single CTE correctly with proper ranges", () => { + const sql = "with my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with simple select and ID column") + }) + + it("should detect multiple CTEs correctly with proper indexing", () => { + const sql = `with + first_cte as (select 1 as id), + second_cte as (select 2 as id), + third_cte as (select 3 as id) + select * from third_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "first_cte", 0, "CTE with simple select") + assertCTE(ctes[1], "second_cte", 1, "CTE with different ID value") + assertCTE(ctes[2], "third_cte", 2, "CTE with third ID value") + }) + + it("should handle CTEs with column lists", () => { + const sql = "with my_cte (id, name) as (select 1, 'test') select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (id, name)", 0, "CTE with explicit column list specification") + }) + + it("should handle quoted CTE names", () => { + const sql = `with "my cte" as (select 1 as id) select * from "my cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], '"my cte"', 0, "CTE with double-quoted name containing space") + }) + }) + + describe("Comments in CTE Body", () => { + it("should handle line comments with unmatched quotes in CTE body", () => { + const sql = `with + cte_without_lens as ( + select * + from {{ ref('parent') }} + where + condition = true -- CTE doesn't have a lens + ), + cte_with_lens as ( + select * + from {{ ref('parent') }} + where + condition = true -- CTE has a 'lens' + ) + select * from cte_with_lens` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(2) + assertCTE(ctes[0], "cte_without_lens", 0, "CTE with line comment containing unmatched single quote") + assertCTE(ctes[1], "cte_with_lens", 1, "CTE with line comment containing matched single quotes") + }) + + it("should handle block comments with unmatched quotes in CTE body", () => { + const sql = `with + cte_with_block_comment as ( + select * + from {{ ref('parent') }} + where + condition = true /* This is a block comment with unmatched quote ' */ + ) + select * from cte_with_block_comment` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "cte_with_block_comment", 0, "CTE with block comment containing unmatched single quote") + }) + + it("should handle Jinja comments with unmatched quotes in CTE body", () => { + const sql = `with + cte_with_jinja_comment as ( + select * + from {{ ref('parent') }} + where + condition = true {# This is a Jinja comment with unmatched quote ' #} + ) + select * from cte_with_jinja_comment` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "cte_with_jinja_comment", 0, "CTE with Jinja comment containing unmatched single quote") + }) + }) + + describe("Comments Between CTE Name and AS", () => { + it("should handle single-line block comment between name and AS", () => { + const sql = "with source /* test */ as (select 1 as id) select * from source" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "source", 0, "CTE with single-line block comment between name and AS keyword") + // The important part is that CTE detection works correctly + }) + + it("should handle multi-line block comment between name and AS", () => { + const sql = `with source +/* this is +a multi-line comment +that should be ignored */ +as ( + select * from {{ ref('raw_orders') }} +) +select * from source` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "source", 0, "CTE with multi-line block comment between name and AS keyword") + // The important part is that CTE detection works correctly + }) + + it("should handle Jinja comment between name and AS", () => { + const sql = "with target {# some comment #} as (select 1 as id) select * from target" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "target", 0, "CTE with Jinja comment between name and AS keyword") + }) + + it("should handle line comment between name and AS (multiline)", () => { + const sql = `with multiline_cte -- this is a comment + as ( + select 4 as id + ) + select * from multiline_cte` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "multiline_cte", 0, "CTE with line comment on separate line before AS keyword") + }) + + it("should handle complex comment with quotes between name and AS", () => { + const sql = `with complex_comment /* more complex comment with 'quotes' */ as ( + select 5 as id + ) select * from complex_comment` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "complex_comment", 0, "CTE with block comment containing both single and double quotes") + }) + }) + + describe("Comments Before and Around CTE Names", () => { + it("should handle block comment before CTE name", () => { + const sql = "with /* comment before name */ my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with block comment before name") + }) + + it("should handle line comment before CTE name on separate line", () => { + const sql = `with + -- This is a comment before the CTE name + my_cte as ( + select 1 as id + ) + select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with line comment before name on separate line") + }) + + it("should handle Jinja comment before CTE name", () => { + const sql = "with {# comment before name #} my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with Jinja comment before name") + }) + + it("should handle comment immediately after WITH keyword", () => { + const sql = "with /* comment after WITH */ my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with block comment immediately after WITH keyword") + }) + + it("should handle line comment after WITH keyword", () => { + const sql = `with -- comment after WITH + my_cte as ( + select 1 as id + ) + select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with line comment after WITH keyword") + }) + + it("should handle multiple comments around CTE name", () => { + const sql = "with /* before */ my_cte /* after */ as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with block comments before and after name") + }) + + it("should handle mixed comment types around CTE name", () => { + const sql = `with -- line comment + /* block comment */ my_cte {# jinja comment #} as ( + select 1 as id + ) + select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with mixed comment types around name") + }) + + it("should handle comments in multi-CTE scenarios", () => { + const sql = `with + /* comment */ first_cte as ( + select 1 as id + ), + -- another comment + second_cte /* inline comment */ as ( + select 2 as id + ), + {# jinja comment #} third_cte as ( + select 3 as id + ) + select * from third_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "first_cte", 0, "First CTE with block comment before name") + assertCTE(ctes[1], "second_cte", 1, "Second CTE with line and inline comments") + assertCTE(ctes[2], "third_cte", 2, "Third CTE with Jinja comment before name") + }) + + it("should handle comments with special characters around CTE name", () => { + const sql = "with /* comment with $pecial ch@rs & symbols! */ my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with comment containing special characters") + }) + + it("should handle unterminated block comment before CTE name", () => { + const sql = "with /* unterminated comment my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + // The comment swallows the rest of the text, so there is no CTE to find. + expect(ctes).toEqual([]) + }) + + it("should handle sequential line and block comments between name and AS", () => { + const sql = `with source -- test +/* this is +a multi-line comment +that should be ignored */ +as ( + select * from {{ ref('raw_orders') }} +) +select * from renamed` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "source", 0, "CTE with sequential line and block comments between name and AS") + }) + }) + + describe("Complex Comment Scenarios", () => { + it("should handle CTE with mixed block and Jinja comments", () => { + const sql = `with +source /* test */ as( + {#- + Normally we would select from the table here, but we are using seeds to load + our data in this project + #} + select * from {{ ref('raw_orders') }} +), + +renamed as ( + select + id as order_id, + user_id as customer_id, + order_date, + status + from source +) + +select * from renamed` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(2) + assertCTE(ctes[0], "source", 0, "CTE with block comment after name and Jinja comment in body") + assertCTE(ctes[1], "renamed", 1, "CTE that references another CTE with column aliasing") + }) + + it("should handle CTE with multi-line block comment before AS keyword", () => { + const sql = `with source +/* this is +a multi-line comment +that should be ignored */ +as ( + select * from {{ ref('raw_orders') }} +), + +renamed as ( + select + id as order_id, + user_id as customer_id, + order_date, + status + from source +) + +select * from renamed` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(2) + assertCTE(ctes[0], "source", 0, "CTE with multi-line block comment spanning multiple lines before AS") + assertCTE(ctes[1], "renamed", 1, "CTE that selects from another CTE with column renaming") + }) + }) + + describe("Very Lengthy Comments with Quoted Names", () => { + it("should handle extremely long block comment before quoted CTE name", () => { + // Create a 1500+ character comment to test real-world lengthy comment scenarios + const longComment = + "/* " + + "This is an extremely long comment that might appear in real-world SQL files. ".repeat(25) + + "It contains multiple sentences and potentially problematic content like 'quoted strings', " + + "special characters !@#$%^&*()_+-=[]{}|;:,.<>?, and even problematic keywords " + + "that might confuse the parser. This comment also contains backslashes \\ and forward slashes // " + + "and other regex-problematic characters like .*+?^${}()|[] that could cause catastrophic backtracking " + + "in poorly designed regular expressions. The comment goes on and on longer. ".repeat(10) + + " */" + + const sql = `with ${longComment} "quoted_cte_name" as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"quoted_cte_name"', + 0, + "CTE with extremely long block comment (1500+ chars) before quoted name", + ) + }) + + it("should handle extremely long line comment before quoted CTE name", () => { + // Create a 1000+ character line comment + const longLineComment = + "-- " + + "This is an extremely long line comment that might appear in real-world SQL files. ".repeat(12) + + "It contains 'quoted strings' and special characters that could cause issues. " + + const sql = `with\n${longLineComment}\n"quoted_cte_name" as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"quoted_cte_name"', + 0, + "CTE with extremely long line comment (1000+ chars) before quoted name", + ) + }) + + it("should handle extremely long Jinja comment before quoted CTE name", () => { + // Create a 1200+ character Jinja comment + const longJinjaComment = + "{# " + + "This is an extremely long Jinja comment that might appear in real-world dbt files. ".repeat(15) + + "It contains 'quoted strings', template variables like {{ var('my_var') }}, and complex logic. " + + "The comment includes special characters and database keywords that could confuse parsers. " + + " #}" + + const sql = `with ${longJinjaComment} "quoted_cte_name" as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"quoted_cte_name"', + 0, + "CTE with extremely long Jinja comment (1200+ chars) before quoted name", + ) + }) + + it("should handle extremely long block comment after quoted CTE name", () => { + // Create a 2000+ character comment after the name + const longComment = + "/* " + + "This is an extremely long comment that appears after the table name but before AS. ".repeat(30) + + "It contains multiple paragraphs, 'quoted strings', and complex descriptions that might " + + "explain the business logic behind the transformation. The comment includes database keywords, special " + + "characters, and other potentially problematic content for regex parsing. ".repeat(5) + + " */" + + const sql = `with "quoted_cte_name" ${longComment} as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"quoted_cte_name"', + 0, + "CTE with extremely long block comment (2000+ chars) after quoted name", + ) + }) + + it("should handle multiple very long comments around quoted CTE name", () => { + // Create multiple long comments around the CTE name + const longCommentBefore = + "/* " + "Long comment before the CTE name. ".repeat(20) + "Contains 'quotes' and special characters. " + " */" + + const longCommentAfter = + "/* " + + "Long comment after the CTE name but before AS. ".repeat(20) + + "Also contains 'quotes' and special characters. " + + " */" + + const sql = `with ${longCommentBefore} "quoted_cte_name" ${longCommentAfter} as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], '"quoted_cte_name"', 0, "CTE with multiple very long comments around quoted name") + }) + + it("should handle extremely long mixed comment types around quoted CTE name", () => { + // Test with line comment, block comment, and Jinja comment (avoid SQL keywords in comments) + const longLineComment = "-- " + "Very long line comment containing 'quotes' and special characters. ".repeat(10) + + const longBlockComment = + "/* " + "Very long block comment containing 'quotes' and special characters. ".repeat(15) + " */" + + const longJinjaComment = + "{# " + "Very long Jinja comment containing 'quotes' and template logic. ".repeat(10) + " #}" + + const sql = `${longLineComment}\n${longBlockComment}\nwith "quoted_cte_name" ${longJinjaComment} as (select 1 as id) select * from "quoted_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], '"quoted_cte_name"', 0, "CTE with extremely long mixed comment types around quoted name") + }) + + it("should handle lengthy comments with quoted names in multi-CTE scenario", () => { + // Test performance with multiple CTEs each having lengthy comments + const longComment1 = + "/* " + + "First data source containing very long description. ".repeat(25) + + "Contains 'quotes' and special characters for testing. " + + " */" + + const longComment2 = + "/* " + + "Second data source containing very long description. ".repeat(25) + + "Also contains 'quotes' and special characters for testing. " + + " */" + + const sql = `with + ${longComment1} "first_quoted_cte" as ( + select 1 as id, 'first' as name + ), + ${longComment2} "second_quoted_cte" as ( + select 2 as id, 'second' as name + ) + select * from "first_quoted_cte" union all select * from "second_quoted_cte"` + + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(2) + assertCTE(ctes[0], '"first_quoted_cte"', 0, "First CTE with very long comment before quoted name") + assertCTE(ctes[1], '"second_quoted_cte"', 1, "Second CTE with very long comment before quoted name") + }) + + it("should handle extremely long comments with regex-problematic content", () => { + // Create a comment with content that could cause regex issues + const problemComment = + "/* " + + "This comment contains regex-problematic content: .*+?^${}()|[] ".repeat(20) + + "It also has nested quotes like 'text with \"nested quotes\" inside' and " + + "backslashes \\ and forward slashes // that could break poorly designed regex. " + + "The comment includes SQL injection attempts like '; DROP TABLE users; -- " + + "and other malicious content that should be handled safely. ".repeat(10) + + " */" + + const sql = `with ${problemComment} "secure_cte_name" as (select 1 as id) select * from "secure_cte_name"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], '"secure_cte_name"', 0, "CTE with extremely long comment containing regex-problematic content") + }) + }) + + describe("Long quoted strings inside comments", () => { + it("should ignore a long quoted string inside a comment", () => { + // Create a quoted string in comment that's longer than 500 chars + const veryLongQuotedString = '"' + "a".repeat(600) + '"' + const commentWithLongQuotedString = `/* This comment contains a very long quoted string: ${veryLongQuotedString} that is long */` + + const sql = `with ${commentWithLongQuotedString} "real_cte" as (select 1 as id) select * from "real_cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"real_cte"', + 0, + "CTE with comment containing quoted string longer than 500 characters", + ) + }) + + it("should handle quoted CTE name approaching 500 characters", () => { + // Create a quoted CTE name that's close to but under the 500 char limit + const longCTEName = '"' + "very_long_cte_name_" + "x".repeat(450) + '"' + + const sql = `with ${longCTEName} as (select 1 as id) select * from ${longCTEName}` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], longCTEName, 0, "CTE with quoted name approaching 500 characters") + }) + + it("should handle quoted CTE name exactly at 500 characters", () => { + // Create a quoted CTE name that's exactly at the 500 char limit (498 chars inside quotes + 2 quotes = 500) + const exactLimitCTEName = '"' + "x".repeat(498) + '"' + + const sql = `with ${exactLimitCTEName} as (select 1 as id) select * from ${exactLimitCTEName}` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], exactLimitCTEName, 0, "CTE with quoted name exactly at 500 characters") + }) + + it("should handle comment with multiple long quoted strings and real CTE", () => { + // Test multiple long quoted strings in comments + const longQuotedString1 = '"' + "fake_identifier_1_" + "a".repeat(600) + '"' + const longQuotedString2 = '"' + "fake_identifier_2_" + "b".repeat(700) + '"' + const commentWithMultipleLongQuotes = `/* Comment with ${longQuotedString1} and also ${longQuotedString2} */` + + const sql = `with ${commentWithMultipleLongQuotes} "actual_cte" as (select 1 as id) select * from "actual_cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"actual_cte"', + 0, + "CTE with comment containing multiple quoted strings longer than 500 characters", + ) + }) + + it("should handle extremely long comment with quoted content and real long CTE name", () => { + // Combine very long comment with quoted content AND a long CTE name + const longQuotedInComment = '"' + "fake_cte_name_in_comment_" + "z".repeat(800) + '"' + const longComment = `/* ${"Very long comment text. ".repeat(50)} Contains fake quoted identifier: ${longQuotedInComment} ${"More comment text. ".repeat(30)} */` + const longRealCTEName = '"' + "real_cte_name_" + "r".repeat(400) + '"' + + const sql = `with ${longComment} ${longRealCTEName} as (select 1 as id) select * from ${longRealCTEName}` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + longRealCTEName, + 0, + "CTE with extremely long comment containing long quoted string and long real CTE name", + ) + }) + + it("should handle line comment with long quoted string before quoted CTE", () => { + // Test line comment with long quoted content + const longQuotedInLineComment = '"' + "fake_line_comment_identifier_" + "l".repeat(600) + '"' + const lineComment = `-- Line comment with long quoted string: ${longQuotedInLineComment}` + + const sql = `with\n${lineComment}\n"real_cte" as (select 1 as id) select * from "real_cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"real_cte"', + 0, + "CTE with line comment containing quoted string longer than 500 characters", + ) + }) + + it("should handle Jinja comment with long quoted string before quoted CTE", () => { + // Test Jinja comment with long quoted content + const longQuotedInJinjaComment = '"' + "fake_jinja_identifier_" + "j".repeat(650) + '"' + const jinjaComment = `{# Jinja comment with long quoted string: ${longQuotedInJinjaComment} #}` + + const sql = `with ${jinjaComment} "real_cte" as (select 1 as id) select * from "real_cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"real_cte"', + 0, + "CTE with Jinja comment containing quoted string longer than 500 characters", + ) + }) + + it("should handle mixed comment types with various length quoted strings", () => { + // Test combination of different comment types with various length quoted strings + const shortQuoted = '"short"' + const mediumQuoted = '"' + "medium_" + "m".repeat(200) + '"' + const longQuoted = '"' + "very_long_" + "v".repeat(800) + '"' + + const lineComment = `-- Line comment with ${shortQuoted}` + const blockComment = `/* Block comment with ${mediumQuoted} and ${longQuoted} */` + const jinjaComment = `{# Jinja comment with ${longQuoted} #}` + + const sql = `${lineComment}\n${blockComment}\nwith ${jinjaComment} "mixed_test_cte" as (select 1 as id) select * from "mixed_test_cte"` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], '"mixed_test_cte"', 0, "CTE with mixed comment types containing various length quoted strings") + }) + }) + + describe("CTE Column List Test Cases", () => { + describe("Basic Column List Variations", () => { + it("should handle simple column list with multiple columns", () => { + const sql = + "with my_cte (id, name, email, created_at) as (select 1, 'John', 'john@example.com', now()) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (id, name, email, created_at)", 0, "CTE with multiple column names in list") + }) + + it("should handle column list with quoted column names", () => { + const sql = + 'with my_cte ("user id", "full name", "email address") as (select 1, \'John Doe\', \'john@example.com\') select * from my_cte' + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + 'my_cte ("user id", "full name", "email address")', + 0, + "CTE with quoted column names containing spaces", + ) + }) + + it("should handle column list with mixed quoted and unquoted names", () => { + const sql = + "with my_cte (id, \"user name\", email, \"created at\") as (select 1, 'John', 'john@example.com', now()) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + 'my_cte (id, "user name", email, "created at")', + 0, + "CTE with mixed quoted and unquoted column names", + ) + }) + + it("should handle column list with simple data types", () => { + const sql = + "with my_cte (id integer, name varchar, amount decimal) as (select 1, 'John', 100.50) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + "my_cte (id integer, name varchar, amount decimal)", + 0, + "CTE with simple column names and data types", + ) + }) + + it("should handle column list with single column", () => { + const sql = "with my_cte (single_column) as (select 'value') select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (single_column)", 0, "CTE with single column in list") + }) + }) + + describe("Length and Performance Edge Cases", () => { + it("should handle column list approaching MAX_COLUMN_LIST_LENGTH limit", () => { + // Create a column list that's close to but under the 1000 char limit + const longColumnList = Array.from({ length: 20 }, (_, i) => `column_${i}_with_long_name`).join(", ") + const sql = `with my_cte (${longColumnList}) as (select ${Array.from({ length: 20 }, (_, i) => i).join(", ")}) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte (${longColumnList})`, + 0, + "CTE with column list approaching MAX_COLUMN_LIST_LENGTH limit", + ) + }) + + it("should handle column list near MAX_COLUMN_LIST_LENGTH limit", () => { + // Create a column list that's close to the 1000 char limit + const nearLimitColumnList = "x".repeat(900) // 900 chars - well within limit + const sql = `with my_cte (${nearLimitColumnList}) as (select 1) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte (${nearLimitColumnList})`, + 0, + "CTE with column list near MAX_COLUMN_LIST_LENGTH limit", + ) + }) + + it("should handle very long individual column names", () => { + const longColumnName1 = '"' + "very_long_column_name_" + "a".repeat(200) + '"' + const longColumnName2 = '"' + "another_long_column_name_" + "b".repeat(200) + '"' + const sql = `with my_cte (${longColumnName1}, ${longColumnName2}) as (select 1, 2) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte (${longColumnName1}, ${longColumnName2})`, + 0, + "CTE with very long individual column names", + ) + }) + }) + + describe("Complex Column List Scenarios", () => { + it("should handle column list with nested parentheses (function calls)", () => { + const sql = + "with my_cte (id, calculated_value, formatted_date) as (select 1, round(avg(amount), 2), to_char(created_at, 'YYYY-MM-DD')) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + "my_cte (id, calculated_value, formatted_date)", + 0, + "CTE with column list containing nested parentheses from function calls", + ) + }) + + it("should handle column list with simple constraints", () => { + const sql = + "with my_cte (id not null, name, amount default 0) as (select 1, 'John', 100.0) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + "my_cte (id not null, name, amount default 0)", + 0, + "CTE with column list containing simple constraints", + ) + }) + + it("should handle column list with line breaks and formatting", () => { + const sql = `with my_cte ( + id, + name, + email, + created_at + ) as ( + select 1, 'John', 'john@example.com', now() + ) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte ( + id, + name, + email, + created_at + )`, + 0, + "CTE with multi-line formatted column list", + ) + }) + + it("should handle column list with special characters and escaped quotes", () => { + const sql = `with my_cte ("col_with_'quote", "col with spaces", "col@with#symbols") as (select 1, 2, 3) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte ("col_with_'quote", "col with spaces", "col@with#symbols")`, + 0, + "CTE with column list containing special characters and quotes", + ) + }) + }) + + describe("Column List + Comment Combinations", () => { + it("should handle comments before the opening parenthesis", () => { + const sql = "with my_cte /* comment before column list */ (id, name) as (select 1, 'John') select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (id, name)", 0) + }) + + it("should handle comments after the closing parenthesis", () => { + const sql = "with my_cte (id, name) /* comment after column list */ as (select 1, 'John') select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (id, name)", 0, "CTE with comment after column list closing parenthesis") + }) + + it("should handle comments between column list and AS keyword", () => { + const sql = `with my_cte (id, name, email) + -- This is a comment before AS + /* Another comment */ + as (select 1, 'John', 'john@example.com') select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte (id, name, email)", 0, "CTE with comments between column list and AS keyword") + }) + + it("should handle mixed comment types around column lists", () => { + const sql = `with /* before */ my_cte ( + id, -- line comment + name /* inline comment */, + email + ) {# jinja comment #} as (select 1, 'John', 'john@example.com') select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + // Comments within column lists are currently included in the CTE name + // This is expected behavior given the current parsing approach + assertCTE( + ctes[0], + `my_cte ( + id, -- line comment + name /* inline comment */, + email + )`, + 0, + "CTE with mixed comment types around column list", + ) + }) + }) + + describe("Error Handling and Edge Cases", () => { + it("should handle empty column lists", () => { + const sql = "with my_cte () as (select 1) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte ()", 0, "CTE with empty column list") + }) + + it("should handle column lists with only whitespace", () => { + const sql = "with my_cte ( ) as (select 1) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte ( )", 0, "CTE with whitespace-only column list") + }) + + it("should handle column lists in combination with quoted CTE names", () => { + const sql = + 'with "my cte" (id, "user name", email) as (select 1, \'John\', \'john@example.com\') select * from "my cte"' + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + '"my cte" (id, "user name", email)', + 0, + "CTE with quoted name and column list with quoted columns", + ) + }) + + it("should handle deeply nested parentheses in column definitions", () => { + const sql = + "with my_cte (calculated_field, nested_calc) as (select sum(case when (x > 0 and (y < 10 or z = (a + b))) then 1 else 0 end), avg(coalesce(field1, field2, 0))) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + "my_cte (calculated_field, nested_calc)", + 0, + "CTE with column list and deeply nested parentheses in query", + ) + }) + }) + + describe("Real-World Complex Scenarios", () => { + it("should handle simple BigQuery-style column definitions", () => { + const sql = `with my_cte ( + user_id INT64, + user_name STRING, + created_at TIMESTAMP + ) as ( + select 1, 'John', current_timestamp() + ) select * from my_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE( + ctes[0], + `my_cte ( + user_id INT64, + user_name STRING, + created_at TIMESTAMP + )`, + 0, + "CTE with simple BigQuery-style column definitions", + ) + }) + + it("should handle multi-CTE scenarios with varying column list complexity", () => { + const sql = `with + simple_cte (id, name) as ( + select 1, 'John' + ), + complex_cte ( + user_id, + "full name", + calculated_field, + metadata_json + ) as ( + select id, name, id * 100, '{"key": "value"}' + from simple_cte + ), + no_columns_cte as ( + select * from complex_cte + ) + select * from no_columns_cte` + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "simple_cte (id, name)", 0, "First CTE with simple column list") + assertCTE( + ctes[1], + `complex_cte ( + user_id, + "full name", + calculated_field, + metadata_json + )`, + 1, + "Second CTE with complex multi-line column list", + ) + assertCTE(ctes[2], "no_columns_cte", 2, "Third CTE without column list") + }) + }) + }) + + describe("CTE Cross-References", () => { + it("should handle simple sequential CTE references", () => { + const sql = `with + base_data as ( + select 1 as id, 'test' as name + ), + filtered_data as ( + select * from base_data where id > 0 + ), + final_data as ( + select id, upper(name) as name from filtered_data + ) + select * from final_data` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "base_data", 0, "CTE with hardcoded test data") + assertCTE(ctes[1], "filtered_data", 1, "CTE filtering data from base_data CTE") + assertCTE(ctes[2], "final_data", 2, "CTE transforming data from filtered_data CTE") + }) + + it("should handle CTE references with joins", () => { + const sql = `with + orders as ( + select 1 as order_id, 100 as customer_id + ), + customers as ( + select 100 as id, 'John' as name + ), + order_details as ( + select o.order_id, c.name + from orders o + join customers c on o.customer_id = c.id + ) + select * from order_details` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "orders", 0, "CTE with order and customer ID data") + assertCTE(ctes[1], "customers", 1, "CTE with customer ID and name data") + assertCTE(ctes[2], "order_details", 2, "CTE joining orders and customers CTEs") + }) + + it("should handle CTE references with comments and quotes", () => { + const sql = `with + source_data as ( + select id, name -- 'source' data extraction + from raw_table + ), + cleaned_data as ( + /* Reference to 'source_data' CTE */ + select * from source_data + where name is not null + ) + select * from cleaned_data` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(2) + assertCTE(ctes[0], "source_data", 0, "CTE with line comment containing single quotes") + assertCTE(ctes[1], "cleaned_data", 1, "CTE referencing source_data with block comment containing quotes") + }) + + it("should handle multiple CTE references in one query", () => { + const sql = `with + cte_a as ( + select 1 as id, 'A' as source + ), + cte_b as ( + select 2 as id, 'B' as source + ), + cte_combined as ( + select * from cte_a + union all + select * from cte_b + ) + select * from cte_combined` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "cte_a", 0, "CTE with ID 1 and source A") + assertCTE(ctes[1], "cte_b", 1, "CTE with ID 2 and source B") + assertCTE(ctes[2], "cte_combined", 2, "CTE combining cte_a and cte_b with UNION ALL") + }) + + it("should handle nested CTE references with subqueries", () => { + const sql = `with + base as ( + select id, amount from transactions + ), + aggregated as ( + select id, sum(amount) as total + from base + group by id + ), + ranked as ( + select *, + (select count(*) from aggregated a2 where a2.total > aggregated.total) as rank + from aggregated + ) + select * from ranked` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(3) + assertCTE(ctes[0], "base", 0, "CTE selecting ID and amount from transactions table") + assertCTE(ctes[1], "aggregated", 1, "CTE grouping base CTE data by ID with sum") + assertCTE(ctes[2], "ranked", 2, "CTE adding rank using correlated subquery on aggregated CTE") + }) + }) + + describe("Edge Cases", () => { + it("should handle no CTEs", () => { + const sql = "select * from my_table" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(0) + }) + + it("should handle empty WITH clause", () => { + const sql = "with select * from my_table" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(0) + }) + + it("should handle complex nested queries", () => { + const sql = `with my_cte as ( + select id, ( + select count(*) + from other_table + where other_table.id = main_table.id + ) as count + from main_table + ) select * from my_cte` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with correlated subquery counting from other table") + + }) + + it("should handle unterminated comments gracefully", () => { + const sql = "with source /* unterminated comment as (select 1) select * from source" + const document = sql + + const ctes = detectCtes(document) + + // The comment swallows the rest of the text, so there is no CTE to find. + expect(ctes).toEqual([]) + }) + }) + + describe("CTE Range and Content Validation", () => { + it("should provide correct ranges for CTE name and query", () => { + const sql = "with my_cte as (select 1 as id) select * from my_cte" + const document = sql + + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + const cte = ctes[0] + + // Verify basic range structure is defined + expect(typeof cte.nameStart).toBe("number") + expect(typeof cte.queryStart).toBe("number") + expect(cte.nameStart).toBeDefined() + expect(cte.nameEnd).toBeDefined() + expect(cte.queryStart).toBeDefined() + expect(cte.queryEnd).toBeDefined() + + // Verify the CTE name was extracted correctly + expect(cte.name).toBe("my_cte") + expect(cte.index).toBe(0) + expect(typeof cte.withClauseStart).toBe("number") + }) + + it("should detect complex CTEs correctly", () => { + const sql = `with + orders_summary as ( + select + customer_id, + count(*) as order_count, + sum(total_amount) as total_spent + from orders + where status = 'completed' + group by customer_id + ) + select * from orders_summary` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "orders_summary", 0, "CTE with aggregated customer order data") + + // Verify range structure is correct + expect(typeof ctes[0].queryStart).toBe("number") + }) + + it("should handle CTE detection with comments correctly", () => { + const sql = `with my_cte as ( + select + id, + name, -- This is a comment with quotes 'inside' + /* Another comment */ + status + from users + {# Jinja comment with 'quotes' #} + where active = true + ) select * from my_cte` + + const document = sql + const ctes = detectCtes(document) + + expect(ctes).toHaveLength(1) + assertCTE(ctes[0], "my_cte", 0, "CTE with line, block, and Jinja comments containing quotes") + + // Verify that CTE was detected despite comments with quotes + expect(ctes[0].name).toBe("my_cte") + expect(typeof ctes[0].queryStart).toBe("number") + }) + }) +}) + +describe("detectCtes offsets", () => { + it("returns offsets that slice the name and the body", () => { + const sql = "with my_cte as (select 1 as id) select * from my_cte" + const [cte] = detectCtesInText(sql) + expect(sql.slice(cte.nameStart, cte.nameEnd)).toBe("my_cte") + expect(sql.slice(cte.queryStart, cte.queryEnd)).toBe("select 1 as id") + expect(cte.withClauseStart).toBe(0) + }) + + it("keeps the column list in the name", () => { + const sql = "with my_cte (id, name) as (select 1, 'test') select * from my_cte" + const [cte] = detectCtesInText(sql) + expect(cte.name).toBe("my_cte (id, name)") + expect(stripColumnList(cte.name)).toBe("my_cte") + }) + + it("does not detect a WITH inside a comment or string", () => { + const sql = "-- with fake as (select 1)\nselect 'with x as (select 2)' as s from t" + expect(detectCtesInText(sql)).toEqual([]) + }) + + it("lists the CTEs of each top-level WITH, each with its own clause", () => { + const sql = "with a as (select 1) with b as (select 2) select * from b" + const ctes = detectCtesInText(sql) + expect(ctes.map((c) => [c.name, c.index])).toEqual([["a", 0], ["b", 0]]) + expect(ctes[0].withClauseStart).not.toBe(ctes[1].withClauseStart) + }) + + it("lists only the model's own CTEs: a WITH inside a CTE body is part of that body", () => { + const sql = `with outer_cte as ( + with inner_cte as (select 1 as x) + select x from inner_cte +), +last as (select * from outer_cte) +select * from last` + const ctes = detectCtesInText(sql) + expect(ctes.map((c) => [c.name, c.index])).toEqual([["outer_cte", 0], ["last", 1]]) + // The outer CTE's count query carries its nested WITH whole and closes every parenthesis. + const query = buildCountQuery(sql, ctes, ctes[1])! + expect(query).toContain("inner_cte as (select 1 as x)") + expect(query.match(/\(/g)!.length).toBe(query.match(/\)/g)!.length) + expect(query).toEndWith("SELECT COUNT(*) AS _profile_count FROM last") + }) + + it("a WITH inside a subquery or a window definition is not a model CTE", () => { + const sql = "select * from (with s as (select 1) select * from s) t join u on true window w as (partition by a)" + expect(detectCtesInText(sql)).toEqual([]) + }) + + it("comment markers and CTE-like text inside string literals neither hide nor add CTEs", () => { + expect(detectCtesInText("with a as (select '--' as s), b as (select 2) select * from b").map((c) => c.name)).toEqual(["a", "b"]) + expect(detectCtesInText("with a as (select '/*' as s), b as (select '*/' as t) select * from b").map((c) => c.name)).toEqual(["a", "b"]) + expect(detectCtesInText("with a as (select '(x)' as s, 'fake as (select 2)' as t) select * from a").map((c) => c.name)).toEqual(["a"]) + expect(detectCtesInText("with a as (select 'it''s ) as (' as s), b as (select 2) select * from b").map((c) => c.name)).toEqual(["a", "b"]) + }) + + it("a quoted name keeps its escaped quote", () => { + const sql = 'with "a""b" as (select 1), `c``d` as (select 2) select * from "a""b"' + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(['"a""b"', "`c``d`"]) + }) + + it("skips a captured {% set %} block between CTEs and reads a bracket name with an escaped ]]", () => { + const sql = "with a as (select 1),\n{% set helper %}select 2{% endset %}\nb as (select 2)\nselect * from b" + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["a", "b"]) + expect(detectCtesInText("with [we]]ird] as (select 1) select * from [we]]ird]").map((c) => c.name)).toEqual(["[we]]ird]"]) + }) + + it("a ) inside a captured {% set %} block does not close the CTE body around it", () => { + const sql = "with a as (\n{% set helper %}){% endset %}\nselect 1), b as (select 2)\nselect * from b" + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["a", "b"]) + }) + + it("keeps CTEs that follow a Jinja tag between them, and reads WITH RECURSIVE and MATERIALIZED", () => { + const jinja = "with a as (select 1),\n{% if x %}\nb as (select 2),\n{% endif %}\nc as (select 3)\nselect * from c" + expect(detectCtesInText(jinja).map((c) => c.name)).toEqual(["a", "b", "c"]) + expect(detectCtesInText("with recursive r as (select 1) select * from r").map((c) => c.name)).toEqual(["r"]) + expect(detectCtesInText("with m as materialized (select 1), n as not materialized (select 2) select 1").map((c) => c.name)).toEqual(["m", "n"]) + }) + + it("scans in linear time: a 2 MB model is read in well under a second", () => { + const body = "select a, b, c from t where x = 1 and y = 'it''s' -- note\n" + const sql = `with big as (\n${body.repeat(Math.floor(2_000_000 / body.length))}) select * from big` + const started = performance.now() + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["big"]) + expect(performance.now() - started).toBeLessThan(1000) + }) + + it("keeps WITH RECURSIVE in the count query, and a WITH written inside a Jinja block is not the model's", () => { + const recursive = "with recursive r as (select 1 as n union all select n + 1 from r where n < 3) select * from r" + const ctes = detectCtesInText(recursive) + expect(buildCountQuery(recursive, ctes, ctes[0])).toStartWith("with recursive r as (") + const block = "{% set q %}with z as (select 1) select * from z{% endset %}\nwith a as (select 1), b as (select 2) select * from b" + const found = detectCtesInText(block) + expect(found.map((c) => c.name)).toEqual(["a", "b"]) + expect(buildCountQuery(block, found, found[1])).toContain("b as (select 2)") + const inline = "{% set q = 'with z as (select 1)' %}\nwith a as (select 1) select * from a" + expect(detectCtesInText(inline).map((c) => c.name)).toEqual(["a"]) + const macro = "{% macro m() %}with z as (select 1) select * from z{% endmacro %}\nwith a as (select 1) select * from a" + expect(detectCtesInText(macro).map((c) => c.name)).toEqual(["a"]) + }) + + it("reads $ and non-ASCII letters in a name, and survives crafted input in linear time", () => { + expect(detectCtesInText("with my$cte as (select 1), é_t as (select 2) select * from é_t").map((c) => c.name)).toEqual(["my$cte", "é_t"]) + for (const unit of ["with [", "with {%", "with ((", "with a as (", "with '", "with /*", "with \"x"]) { + const started = performance.now() + detectCtesInText(unit.repeat(200_000)) + expect(performance.now() - started).toBeLessThan(2000) + } + }) +}) + +describe("buildCountQuery", () => { + const sql = [ + "{{ config(materialized='table') }}", + "", + "with orders as (select * from {{ ref('stg_orders') }}),", + "customers as (select * from {{ ref('stg_customers') }}),", + "joined as (select o.id from orders o join customers c on o.cid = c.id)", + "select * from joined", + ].join("\n") + + it("keeps the preamble and every earlier CTE up to the target", () => { + const ctes = detectCtesInText(sql) + expect(ctes.map((c) => c.name)).toEqual(["orders", "customers", "joined"]) + const query = buildCountQuery(sql, ctes, ctes[1])! + expect(query.startsWith("{{ config(materialized='table') }}")).toBe(true) + expect(query).toContain("with orders as (") + expect(query).toContain("customers as (") + expect(query).not.toContain("joined") + expect(query.endsWith("SELECT COUNT(*) AS _profile_count FROM customers")).toBe(true) + }) + + it("counts the last CTE with all of its inputs", () => { + const ctes = detectCtesInText(sql) + const query = buildCountQuery(sql, ctes, ctes[2])! + expect(query).toContain("joined as (") + expect(query).toContain("from orders o join customers c") + expect(query.endsWith("FROM joined")).toBe(true) + }) + + it("keeps a column list in the definition but not in the FROM clause", () => { + const text = "with a (x, y) as (select 1, 2) select * from a" + const ctes = detectCtesInText(text) + const query = buildCountQuery(text, ctes, ctes[0])! + expect(query).toContain("with a (x, y) as (") + expect(query.endsWith("FROM a")).toBe(true) + }) + + it("builds no query for a CTE of a second WITH clause, whose preamble would hold the first clause's query", () => { + const text = "with a as (select 1) select * from a;\nwith b as (select 2) select * from b" + const ctes = detectCtesInText(text) + expect(buildCountQuery(text, ctes, ctes.find((c) => c.name === "a")!)).toContain("with a as (") + expect(buildCountQuery(text, ctes, ctes.find((c) => c.name === "b")!)).toBeUndefined() + }) + + it("returns undefined for a CTE that is not in the list", () => { + expect( + buildCountQuery("select 1", [], { + name: "x", + nameStart: 0, + nameEnd: 1, + queryStart: 0, + queryEnd: 0, + index: 0, + withClauseStart: 0, + }), + ).toBeUndefined() + }) +}) + +describe("count query edge cases", () => { + it("keeps Jinja written between CTEs, and reports a block left open by the cut", () => { + const kept = "{% set n = 2 %}\nwith a as (select {{ n }}),\n{% if true %}b as (select 2),{% endif %}\nc as (select 3) select * from c" + const ctes = detectCtesInText(kept) + expect(buildCountQuery(kept, ctes, ctes[2])).toContain("{% if true %}b as (select 2),{% endif %}") + expect(unclosedJinjaBlock(buildCountQuery(kept, ctes, ctes[2])!)).toBeUndefined() + const open = "{% if x %}\nwith a as (select 1) select * from a\n{% endif %}" + const found = detectCtesInText(open) + expect(unclosedJinjaBlock(buildCountQuery(open, found, found[0])!)).toBe("if") + expect(unclosedJinjaBlock("{% set q %}x")).toBe("set") + expect(unclosedJinjaBlock("{% raw %}{% if x %}{% endraw %}")).toBeUndefined() + expect(unclosedJinjaBlock("{% block body %}with a as (select 1)")).toBe("block") + expect(unclosedJinjaBlock("{% if x %}a{%+ endif %}")).toBeUndefined() + expect(unclosedJinjaBlock("{% set (a, b) = (1, 2) %}x")).toBeUndefined() + }) + + it("matches nested set blocks to their own end tag", () => { + const sql = "{% set outer %}{% set inner %}with fake as (select 1){% endset %}x{% endset %}\nwith real as (select 2) select * from real" + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["real"]) + }) + + it("reads a comment or Jinja tag between AS, NOT and MATERIALIZED and the body", () => { + expect(detectCtesInText("with a as /* c */ (select 1), b as not /* c */ materialized (select 2) select 1").map((c) => c.name)).toEqual(["a", "b"]) + expect(detectCtesInText("with a /* c */ as {# j #} materialized -- x\n(select 1) select 1").map((c) => c.name)).toEqual(["a"]) + }) + + it("strips a column list that holds a quoted ')' or a comment", () => { + expect(stripColumnList('a ("x)")')).toBe("a") + expect(stripColumnList("a (x /* ) */)")).toBe("a") + expect(stripColumnList("Big Spenders")).toBe("Big Spenders") + const text = 'with a ("x)", y) as (select 1, 2) select * from a' + const ctes = detectCtesInText(text) + expect(buildCountQuery(text, ctes, ctes[0])).toEndWith("FROM a") + }) + + it("a dollar-quoted string does not end the CTE body early", () => { + const text = "with a as (select $$ ) $$ as s), b as (select $tag$ ( $tag$ as t) select 1" + expect(detectCtesInText(text).map((c) => c.name)).toEqual(["a", "b"]) + }) + + it("tags inside a Jinja comment are not blocks, and a % inside a tag does not hide it", () => { + expect(unclosedJinjaBlock("{# {% if x %} #}with a as (select 1)")).toBeUndefined() + expect(unclosedJinjaBlock("{% if x %}{# {% endif %} #}with a as (select 1)")).toBe("if") + expect(unclosedJinjaBlock("{% if x % 2 == 0 %}with a as (select 1)")).toBe("if") + const sql = "{% set q %}{# {% endset %} #}with z as (select 1){% endset %}\nwith a as (select 2) select * from a" + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["a"]) + }) + + it("counts open Jinja blocks in linear time: 100k tags in a second", () => { + const sql = "{% if x %}".repeat(100_000) + "{% endfor %}".repeat(100_000) + const started = performance.now() + expect(unclosedJinjaBlock(sql)).toBe("if") + expect(performance.now() - started).toBeLessThan(1000) + }) + + it("unterminated tags, comments and many block tags are scanned in linear time", () => { + for (const unit of ["{% if x ", "{# c ", "{% set q %}", "{% set q | upper %}"]) { + const sql = unit.repeat(100_000) + "with a as (select 1) select 1" + const started = performance.now() + unclosedJinjaBlock(sql) + detectCtesInText(sql) + expect(performance.now() - started).toBeLessThan(1500) + } + }) + + it("a filtered set block is a block, and raw text holds no tags", () => { + expect(unclosedJinjaBlock("{% set q | upper %}x")).toBe("set") + expect(unclosedJinjaBlock("{% set q = 1 | upper %}x")).toBeUndefined() + const sql = "{% set q %}{% raw %}{% endset %}{% endraw %}with z as (select 1){% endset %}\nwith a as (select 2) select * from a" + expect(detectCtesInText(sql).map((c) => c.name)).toEqual(["a"]) + }) + + it("a dollar-quote tag may hold digits", () => { + expect(detectCtesInText("with a as (select $f1$ ) $f1$ as s), b as (select 2) select 1").map((c) => c.name)).toEqual(["a", "b"]) + }) + + it("writes a name with $ or a non-ASCII letter as it was declared", () => { + for (const name of ["my$cte", "Ünits"]) { + const text = `with ${name} as (select 1) select * from ${name}` + const ctes = detectCtesInText(text) + expect(buildCountQuery(text, ctes, ctes[0])).toEndWith(`FROM ${name}`) + } + }) +}) + +describe("quoteSqlIdentifier", () => { + it("leaves plain, quoted and dotted names alone and quotes the rest", () => { + expect(quoteSqlIdentifier("my_cte")).toBe("my_cte") + expect(quoteSqlIdentifier('"My Cte"')).toBe('"My Cte"') + expect(quoteSqlIdentifier("`x`")).toBe("`x`") + expect(quoteSqlIdentifier("[x]")).toBe("[x]") + expect(quoteSqlIdentifier("a.b")).toBe("a.b") + expect(quoteSqlIdentifier("my-cte")).toBe('"my-cte"') + }) + + it("is applied to a quoted CTE name in the count query", () => { + const text = 'with "My Cte" as (select 1) select * from "My Cte"' + const ctes = detectCtesInText(text) + expect(buildCountQuery(text, ctes, ctes[0])!.endsWith('FROM "My Cte"')).toBe(true) + }) +}) + +describe("extractRowCount", () => { + it("reads the count by position for upper- and lower-case aliases", () => { + expect(extractRowCount([{ _PROFILE_COUNT: 99 }])).toBe(99) + expect(extractRowCount([{ _profile_count: 42 }])).toBe(42) + }) + + it("coerces string counts", () => { + expect(extractRowCount([{ _PROFILE_COUNT: "1234" }])).toBe(1234) + }) + + it("returns 0 for no rows, a missing value or a non-numeric value", () => { + expect(extractRowCount([])).toBe(0) + expect(extractRowCount(undefined)).toBe(0) + expect(extractRowCount([{ _PROFILE_COUNT: "not-a-number" }])).toBe(0) + }) +}) diff --git a/packages/opencode/test/altimate/dbt-cte-profile.test.ts b/packages/opencode/test/altimate/dbt-cte-profile.test.ts new file mode 100644 index 0000000000..002b1193fc --- /dev/null +++ b/packages/opencode/test/altimate/dbt-cte-profile.test.ts @@ -0,0 +1,507 @@ +// dbt_cte_profile: the CTE list, and one timed count query per CTE. The dbt adapter is replaced through +// `dbtAdapterInternals`, so no warehouse, Python or dbt is needed. +import { afterAll, afterEach, beforeAll, beforeEach, describe, expect, test } from "bun:test" +import { mkdir, mkdtemp, rm, symlink, writeFile } from "node:fs/promises" +import os from "node:os" +import path from "node:path" +import { Effect } from "effect" +import type { Tool } from "@/tool/tool" +import { SessionID, MessageID } from "@/session/schema" +import { dbtAdapterInternals, resetDbtAdapter } from "@/altimate/native/connections/register" +import { DbtCteProfileTool } from "@/altimate/tools/dbt-cte-profile" +import { initTool } from "./tool-fixture" + +const ctx: Tool.Context = { + sessionID: SessionID.make("ses_dbt_cte_profile"), + messageID: MessageID.make("msg_dbt_cte_profile"), + agent: "analyst", + abort: new AbortController().signal, + messages: [], + metadata: () => Effect.void, + ask: () => Effect.void, +} + +const CUSTOMERS = `{{ config(materialized='table') }} + +-- orders first +with orders as ( + select * from {{ ref('stg_orders') }} +), + +"Big Spenders" (customer_id, total) as ( + select customer_id, sum(amount) from orders group by 1 +), + +final as ( + select * from "Big Spenders" +) + +select * from final +` + +let project: string +let queries: { sql: string; name: string; limit: number }[] +let respond: (sql: string) => Promise + +function useAdapter() { + dbtAdapterInternals.readConfig = async () => ({ + projectRoot: project, + pythonPath: "python3", + dbtIntegration: "corecommand", + queryLimit: 500, + }) + dbtAdapterInternals.create = async () => ({ + isInitialized: () => true, + dispose: async () => {}, + immediatelyExecuteSQLWithLimit: async (sql: string, name: string, limit: number) => { + queries.push({ sql, name, limit }) + return respond(sql) + }, + }) +} + +beforeAll(async () => { + project = await mkdtemp(path.join(os.tmpdir(), "dbt-cte-profile-")) + await writeFile(path.join(project, "dbt_project.yml"), "name: jaffle_shop\n") + await mkdir(path.join(project, "models", "marts"), { recursive: true }) + await mkdir(path.join(project, "models", "staging"), { recursive: true }) + await mkdir(path.join(project, "dbt_packages", "pkg", "models"), { recursive: true }) + await writeFile(path.join(project, "models", "marts", "customers.sql"), CUSTOMERS) + await writeFile(path.join(project, "models", "staging", "plain.sql"), "select 1 as id\n") + await writeFile(path.join(project, "models", "marts", "dup.sql"), "with a as (select 1) select * from a") + await writeFile(path.join(project, "models", "staging", "dup.sql"), "with b as (select 1) select * from b") + await writeFile( + path.join(project, "models", "marts", "twins.sql"), + 'with "Foo" as (select 1 as a), foo as (select 2 as a), keep as (select 3 as a)\nselect * from keep\n', + ) + await writeFile(path.join(project, "models", "marts", "bracketed.sql"), "with [we]]ird] as (select 1 as a) select * from [we]]ird]\n") + await writeFile( + path.join(project, "models", "marts", "wrapped.sql"), + "{% if is_incremental() %}\nwith a as (select 1 as x) select * from a\n{% else %}\nwith b as (select 2 as x) select * from b\n{% endif %}\n", + ) + await writeFile( + path.join(project, "models", "marts", "second.sql"), + "with a as (select 1 as x) select * from a;\nwith b as (select 2 as x) select * from b\n", + ) + await mkdir(path.join(project, "target"), { recursive: true }) + await writeFile( + path.join(project, "target", "manifest.json"), + JSON.stringify({ + nodes: { + "model.jaffle_shop.customers": { resource_type: "model", name: "customers", original_file_path: "models/marts/customers.sql" }, + "model.jaffle_shop.gone": { resource_type: "model", name: "gone", original_file_path: "models/marts/gone.sql" }, + "model.jaffle_shop.dup.v1": { resource_type: "model", name: "sales", unique_id: "model.jaffle_shop.sales.v1", original_file_path: "models/marts/dup.sql" }, + "model.jaffle_shop.sales.v2": { resource_type: "model", name: "sales", unique_id: "model.jaffle_shop.sales.v2", original_file_path: "models/staging/dup.sql" }, + "model.pkg.customers": { resource_type: "model", name: "customers", package_name: "pkg", original_file_path: "models/marts/customers.sql" }, + }, + }), + ) + await writeFile(path.join(project, "dbt_packages", "pkg", "models", "ignored.sql"), "with z as (select 1) select * from z") +}) + +afterAll(async () => { + await rm(project, { recursive: true, force: true }) +}) + +beforeEach(() => { + queries = [] + respond = async () => ({ columnNames: ["_PROFILE_COUNT"], data: [{ _PROFILE_COUNT: 42 }] }) + resetDbtAdapter() + useAdapter() +}) + +afterEach(() => { + resetDbtAdapter() + delete dbtAdapterInternals.readConfig + delete dbtAdapterInternals.create +}) + +describe("dbt_cte_profile list", () => { + test("lists the CTEs in source order with their lines, without the adapter, on a read-only call", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers" }, { ...ctx, extra: { forceReadOnly: true } }) + expect(result.metadata.success).toBe(true) + expect(result.metadata.file).toBe(path.join("models", "marts", "customers.sql")) + expect(result.metadata.ctes).toEqual([ + { name: "orders", line: 4, index: 0, clause: expect.any(Number), position: 0 }, + { name: '"Big Spenders"', line: 8, index: 1, clause: expect.any(Number), position: 1 }, + { name: "final", line: 12, index: 2, clause: expect.any(Number), position: 2 }, + ]) + expect(result.output).toContain("3 CTEs") + // The numbers shown are the ones `position` takes: it counts from 0. + expect(result.output).toContain("position 0: orders (line 4)") + expect(result.output).toContain("position 2: final (line 12)") + expect(queries).toEqual([]) + }) + + test("a CTE whose name a Jinja expression builds ends the list, and the list says so", async () => { + await writeFile( + path.join(project, "models", "marts", "looped.sql"), + "with base as (select 1 as a),\n{% for t in ['x'] %}\n{{ t }}_c as (select 1 as a),\n{% endfor %}\nfinal as (select 1 as a) select * from final\n", + ) + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "looped" }, ctx) + expect(result.metadata.ctes.map((c: { name: string }) => c.name)).toEqual(["base"]) + expect(result.metadata.incomplete).toBe(true) + expect(result.output).toContain("The list is incomplete: the CTEs from line 1 on could not be read") + const absent = await tool.execute({ model: "looped", cte: "final" }, ctx) + expect(absent.metadata.success).toBe(false) + expect(absent.output).toContain("The list is incomplete") + await writeFile( + path.join(project, "models", "marts", "firstloop.sql"), + "{% if true %}\nwith a as (select 1 as x), {{ y }}_c as (select 1 as x) select * from a\n{% else %}\nwith b as (select 2 as x) select * from b\n{% endif %}\n", + ) + const multi = await tool.execute({ model: "firstloop" }, ctx) + expect(multi.output).toContain("the CTEs from line 2 on could not be read") + await writeFile(path.join(project, "models", "marts", "loopfirst.sql"), "with {{ x }}_c as (select 1 as a), final as (select 1 as a) select * from final\n") + const first = await tool.execute({ model: "loopfirst" }, ctx) + expect(first.metadata.incomplete).toBe(true) + expect(first.metadata.ctes).toEqual([]) + expect(first.output).toContain("No CTEs in") + expect(first.output).toContain("the CTEs from line 1 on could not be read") + for (const [model, sql] of [ + ["prefixed", "with recursive stg_{{ var('s') }} as (select 1 as a) select * from stg_x\n"], + ["hierarchy", "select * from emp start with {{ root_filter() }} connect by prior id = parent_id\n"], + ]) { + await writeFile(path.join(project, "models", "marts", `${model}.sql`), sql) + } + expect((await tool.execute({ model: "prefixed" }, ctx)).metadata.incomplete).toBe(true) + const hierarchy = await tool.execute({ model: "hierarchy" }, ctx) + expect(hierarchy.metadata.incomplete).toBeUndefined() + expect(hierarchy.output).toBe("No CTEs in " + path.join("models", "marts", "hierarchy.sql") + ".") + const byPosition = await tool.execute({ model: "looped", position: 5 }, ctx) + expect(byPosition.output).toContain("The list is incomplete") + const whole = await tool.execute({ model: "customers" }, ctx) + expect(whole.metadata.incomplete).toBeUndefined() + }) + + test("the query runs against the model's own name", async () => { + const tool = await initTool(DbtCteProfileTool) + await tool.execute({ model: "customers", cte: "orders" }, ctx) + expect(queries[0].name).toBe("customers") + }) + + test("a CTE chain that reads is_incremental() or this is not run, since the count query would not see the model's context", async () => { + const tool = await initTool(DbtCteProfileTool) + for (const [model, sql] of [ + ["inc_filter", "with a as (select * from {{ ref('x') }} {% if is_incremental() %} where id > 1 {% endif %}), b as (select * from a) select * from b"], + ["inc_this", "with a as (select max(id) m from {{ this }}), b as (select * from a) select * from b"], + ["inc_set", "{% set rel = this %} with a as (select * from {{ rel }}), b as (select * from a) select * from b"], + ] as const) { + await writeFile(path.join(project, "models", "marts", `${model}.sql`), sql) + const before = queries.length + const result = await tool.execute({ model, cte: "b" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("is_incremental() or this") + expect(queries.length).toBe(before) + } + }) + + test("is_incremental() or this in a comment, a string or a ref name does not stop the run", async () => { + const tool = await initTool(DbtCteProfileTool) + await writeFile( + path.join(project, "models", "marts", "quiet.sql"), + "{# is_incremental() is unused, {{ this }} #}\n-- is_incremental()\nwith a as (select 'is_incremental()' as label from {{ ref('this_model') }}), b as (select * from a) select * from b", + ) + const result = await tool.execute({ model: "quiet", cte: "b" }, ctx) + expect(result.metadata.success).toBe(true) + }) + + test("a {{ this }} inside a quoted config() hook does not stop the run, a {% raw %} block is skipped, and a real this after them still does", async () => { + const tool = await initTool(DbtCteProfileTool) + const body = "with a as (select 1 as x), b as (select * from a) select * from b" + for (const [model, sql, ok] of [ + ["hook_this", `{{ config(post_hook="grant select on {{ this }} to role reporter") }}\n${body}`, true], + ["raw_this", `{% raw %}{{ this }}{% endraw %}\n${body}`, true], + ["hook_then_this", `{{ config(post_hook="grant on {{ this }}") }}\nwith a as (select * from {{ this }}), b as (select * from a) select * from b`, false], + ] as const) { + await writeFile(path.join(project, "models", "marts", `${model}.sql`), sql) + const result = await tool.execute({ model, cte: "b" }, ctx) + expect(result.metadata.success).toBe(ok) + if (!ok) expect(result.output).toContain("is_incremental() or this") + } + }) + + test("a model without CTEs lists none", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "plain" }, ctx) + expect(result.metadata.success).toBe(true) + expect(result.metadata.ctes).toEqual([]) + }) + + test("takes a project-relative path and ignores dbt_packages", async () => { + const tool = await initTool(DbtCteProfileTool) + const byPath = await tool.execute({ model: "models/marts/customers.sql" }, ctx) + expect(byPath.metadata.success).toBe(true) + const packaged = await tool.execute({ model: "ignored" }, ctx) + expect(packaged.metadata.success).toBe(false) + expect(packaged.output).toContain("no model named ignored") + }) + + test("an ambiguous name asks for the path; a path outside the project is refused", async () => { + const tool = await initTool(DbtCteProfileTool) + const dup = await tool.execute({ model: "dup" }, ctx) + expect(dup.metadata.success).toBe(false) + expect(dup.output).toContain("matches 2 files") + const outside = await tool.execute({ model: "../../etc/passwd.sql" }, ctx) + expect(outside.metadata.success).toBe(false) + expect(outside.output).toContain("outside the dbt project") + }) +}) + +describe("dbt_cte_profile model lookup", () => { + test("finds a model by unique_id through the manifest, and takes an absolute path", async () => { + const tool = await initTool(DbtCteProfileTool) + const byId = await tool.execute({ model: "model.jaffle_shop.customers" }, ctx) + expect(byId.metadata.success).toBe(true) + const absolute = await tool.execute({ model: path.join(project, "models", "marts", "customers.sql") }, ctx) + expect(absolute.metadata.success).toBe(true) + expect(absolute.metadata.file).toBe(path.join("models", "marts", "customers.sql")) + }) + + test("a unique_id the manifest does not hold is refused, never read as a name", async () => { + const tool = await initTool(DbtCteProfileTool) + const other = await tool.execute({ model: "model.other_package.customers" }, ctx) + expect(other.metadata.success).toBe(false) + expect(other.output).toContain("is not in target/manifest.json") + // The manifest knows this model but its file is gone: the name search must not answer for another package's. + const gone = await tool.execute({ model: "model.jaffle_shop.gone" }, ctx) + expect(gone.metadata.success).toBe(false) + // A package's model is recorded relative to its own package: never read as the project's file of that path. + const packaged = await tool.execute({ model: "model.pkg.customers" }, ctx) + expect(packaged.metadata.success).toBe(false) + expect(packaged.output).toContain("belongs to the package pkg") + }) + + test("a folder, or a path that is not a .sql file, is a failure, not a crash", async () => { + const tool = await initTool(DbtCteProfileTool) + const folder = await tool.execute({ model: "models/marts/" }, ctx) + expect(folder.metadata.success).toBe(false) + expect(folder.output).toContain("not a .sql file") + const config = await tool.execute({ model: "models/../dbt_project.yml" }, ctx) + expect(config.metadata.success).toBe(false) + expect(config.output).toContain("not a .sql file") + const dirNamedSql = path.join(project, "models", "dir.sql") + await mkdir(dirNamedSql, { recursive: true }) + const named = await tool.execute({ model: "models/dir.sql" }, ctx) + expect(named.metadata.success).toBe(false) + await rm(dirNamedSql, { recursive: true }) + }) + + test("a symlink out of the project is refused", async () => { + const outside = await mkdtemp(path.join(os.tmpdir(), "dbt-cte-outside-")) + await writeFile(path.join(outside, "secret.sql"), "with s as (select 1) select * from s") + await symlink(path.join(outside, "secret.sql"), path.join(project, "models", "link.sql")) + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "models/link.sql" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("outside the dbt project") + await rm(path.join(project, "models", "link.sql")) + await rm(outside, { recursive: true, force: true }) + }) + + test("the list is read from the session's project, not the one the adapter was set up for", async () => { + const other = await mkdtemp(path.join(os.tmpdir(), "dbt-cte-other-")) + await writeFile(path.join(other, "dbt_project.yml"), "name: other\n") + await mkdir(path.join(other, "models"), { recursive: true }) + await writeFile(path.join(other, "models", "customers.sql"), "with only_here as (select 1) select * from only_here") + const before = process.cwd() + process.chdir(other) + try { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers" }, ctx) + expect((result.metadata.ctes as { name: string }[]).map((c) => c.name)).toEqual(["only_here"]) + const run = await tool.execute({ model: "customers", cte: "only_here" }, ctx) + expect(run.metadata.success).toBe(false) + expect(run.metadata.refusal).toBe("other_project") + } finally { + process.chdir(before) + await rm(other, { recursive: true, force: true }) + } + }) + + test("an oversize model is refused", async () => { + await writeFile(path.join(project, "models", "huge.sql"), "-- " + "x".repeat(4 * 1024 * 1024 + 10)) + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "huge" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("largest file read") + await rm(path.join(project, "models", "huge.sql")) + }) +}) + +describe("dbt_cte_profile bracket-quoted names", () => { + test("a ]] inside a bracketed name is one ], so the CTE is reachable by its plain name", async () => { + const tool = await initTool(DbtCteProfileTool) + for (const cte of ["we]ird", "[we]]ird]"]) { + const run = await tool.execute({ model: "bracketed", cte }, ctx) + expect(run.metadata.success).toBe(true) + expect(run.metadata.position).toBe(0) + } + }) +}) + +describe("dbt_cte_profile same-named CTEs", () => { + test("two CTEs whose names differ only in case are listed apart, and each is reachable by its exact name", async () => { + const tool = await initTool(DbtCteProfileTool) + const list = await tool.execute({ model: "twins" }, ctx) + expect((list.metadata.ctes as { name: string; position: number }[]).map((c) => [c.name, c.position])).toEqual([['"Foo"', 0], ["foo", 1], ["keep", 2]]) + const exact = await tool.execute({ model: "twins", cte: "foo" }, ctx) + expect(exact.metadata.position).toBe(1) + expect(queries[0].sql).toEndWith("FROM foo") + const quoted = await tool.execute({ model: "twins", cte: '"Foo"' }, ctx) + expect(quoted.metadata.position).toBe(0) + expect(queries[1].sql).toEndWith('FROM "Foo"') + // Letter case alone does not pick between two matches. + const loose = await tool.execute({ model: "twins", cte: "FOO" }, ctx) + expect(loose.metadata.success).toBe(false) + expect(loose.output).toContain("positions 0, 1") + expect(queries).toHaveLength(2) + }) + + test("position picks one CTE, and a name that disagrees with it is refused", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "twins", position: 2 }, ctx) + expect(result.metadata.cte).toBe("keep") + const cased = await tool.execute({ model: "twins", cte: "KEEP", position: 2 }, ctx) + expect(cased.metadata.success).toBe(true) + const wrong = await tool.execute({ model: "twins", cte: "foo", position: 2 }, ctx) + expect(wrong.metadata.success).toBe(false) + const shadowed = await tool.execute({ model: "twins", cte: "foo", position: 0 }, ctx) + expect(shadowed.metadata.success).toBe(false) + const exactAt = await tool.execute({ model: "twins", cte: "foo", position: 1 }, ctx) + expect(exactAt.metadata.success).toBe(true) + const missing = await tool.execute({ model: "twins", position: 9 }, ctx) + expect(missing.metadata.success).toBe(false) + expect(queries).toHaveLength(3) + }) + + test("a CTE of a second WITH clause has no query of its own and runs nothing", async () => { + const tool = await initTool(DbtCteProfileTool) + const first = await tool.execute({ model: "second", cte: "a" }, ctx) + expect(first.metadata.success).toBe(true) + const later = await tool.execute({ model: "second", cte: "b" }, ctx) + expect(later.metadata.success).toBe(false) + expect(later.output).toContain("second WITH clause") + expect(queries).toHaveLength(1) + }) +}) + +describe("dbt_cte_profile run", () => { + test("every failure carries an error, so the CLI and MCP report it as one", async () => { + const tool = await initTool(DbtCteProfileTool) + const results = [ + await tool.execute({ model: "nope" }, ctx), + await tool.execute({ model: "customers", cte: "nope" }, ctx), + await tool.execute({ model: "customers", cte: "orders" }, { ...ctx, extra: { forceReadOnly: true } }), + ] + for (const result of results) { + expect(result.metadata.success).toBe(false) + expect(typeof result.metadata.error).toBe("string") + } + }) + + test("a query that outlives the call's abort ends the wait at the abort", async () => { + respond = () => new Promise(() => {}) + const controller = new AbortController() + const tool = await initTool(DbtCteProfileTool) + const run = tool.execute({ model: "customers", cte: "orders" }, { ...ctx, abort: controller.signal }) + setTimeout(() => controller.abort(new Error("stopped")), 50) + await expect(run).rejects.toThrow("stopped") + }) + + test("a name shared by several versions of one model is refused, not guessed", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "sales" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("matches 2 models") + }) + + test("a query that timed out is reported as timed out, since it may still be running", async () => { + respond = async () => { + throw Object.assign(new Error("exceeded 300000 ms"), { code: "QUERY_TIMEOUT" }) + } + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "orders" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.metadata.timedOut).toBe(true) + expect(result.output).toContain("may still be running") + }) + + test("a CTE inside a Jinja block that the query would cut open is refused, and nothing runs", async () => { + const tool = await initTool(DbtCteProfileTool) + for (const cte of ["a", "b"]) { + const result = await tool.execute({ model: "wrapped", cte }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("Jinja") + } + expect(queries).toHaveLength(0) + }) + + test("runs one count query over the CTE and the earlier ones, limit 1, and reads the count by position", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "Big Spenders" }, ctx) + expect(result.metadata.success).toBe(true) + expect(result.metadata.cte).toBe('"Big Spenders"') + expect(result.metadata.rows).toBe(42) + expect(typeof result.metadata.queryMs).toBe("number") + expect(queries).toHaveLength(1) + expect(queries[0].name).toBe("customers") + expect(queries[0].limit).toBe(1) + expect(queries[0].sql).toContain("{{ config(materialized='table') }}") + expect(queries[0].sql).toContain("orders as (") + expect(queries[0].sql).toContain('"Big Spenders" (customer_id, total) as (') + expect(queries[0].sql).not.toContain("final as (") + expect(queries[0].sql).toEndWith('SELECT COUNT(*) AS _profile_count FROM "Big Spenders"') + }) + + test("matches the CTE name ignoring case and quotes", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "ORDERS" }, ctx) + expect(result.metadata.success).toBe(true) + expect(result.metadata.cte).toBe("orders") + expect(queries[0].sql).toEndWith("FROM orders") + }) + + test("an unknown CTE names the ones the model has and runs nothing", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "nope" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("orders") + expect(queries).toEqual([]) + }) + + test("a read-only call is refused and never starts the adapter", async () => { + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "orders" }, { ...ctx, extra: { forceReadOnly: true } }) + expect(result.metadata.success).toBe(false) + expect(result.metadata.refusal).toBe("read_only") + expect(result.output).toContain("--allow-write") + expect(queries).toEqual([]) + }) + + test("a failing query is reported as a failure, not thrown", async () => { + respond = async () => { + throw new Error("Object 'STG_ORDERS' does not exist") + } + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "orders" }, ctx) + expect(result.metadata.success).toBe(false) + expect(result.output).toContain("STG_ORDERS") + }) + + test("an empty result counts as zero rows", async () => { + respond = async () => ({ columnNames: ["_PROFILE_COUNT"], data: [] }) + const tool = await initTool(DbtCteProfileTool) + const result = await tool.execute({ model: "customers", cte: "final" }, ctx) + expect(result.metadata.rows).toBe(0) + }) + + test("a cancelled call throws instead of answering", async () => { + const controller = new AbortController() + controller.abort() + const tool = await initTool(DbtCteProfileTool) + await expect(tool.execute({ model: "customers", cte: "orders" }, { ...ctx, abort: controller.signal })).rejects.toMatchObject({ name: "AbortError" }) + expect(queries).toEqual([]) + }) +})