@thecodecodex/tools 0.0.0-stage → 0.1.0

package/src/table-convert/serialize.ts

@@ -0,0 +1,435 @@
+import { inferValue, uniqueKeys, type Decimal, type Scalar, type Table } from "./model"
+import { toGo, toPandas, toPhp, toPython, toTypeScript } from "./code"
+import { DELIMITERS, type Delimiter } from "./parse"
+
+// Formatos agrupados como aparecem na tela
+export const OUTPUT_GROUPS = {
+  data: ["csv", "xlsx", "json", "jsonl", "sql", "yaml", "toml", "ini", "xml"],
+  docs: ["markdown", "html", "ascii", "latex", "jira", "mediawiki", "asciidoc", "rst"],
+  code: ["python", "pandas", "typescript", "go", "php"],
+} as const
+export type OutputGroup = keyof typeof OUTPUT_GROUPS
+export const OUTPUT_FORMATS = [...OUTPUT_GROUPS.data, ...OUTPUT_GROUPS.docs, ...OUTPUT_GROUPS.code] as const
+export type OutputFormat = (typeof OUTPUT_FORMATS)[number]
+export type TextFormat = Exclude<OutputFormat, "xlsx">
+
+export const JSON_SHAPES = ["objects", "arrays", "columns", "keyed"] as const
+export type JsonShape = (typeof JSON_SHAPES)[number]
+export const INDENTS = ["2", "4", "tab", "min"] as const
+export type Indent = (typeof INDENTS)[number]
+export const SQL_DIALECTS = ["mysql", "postgres", "sqlite", "mssql"] as const
+export type SqlDialect = (typeof SQL_DIALECTS)[number]
+export const ALIGNS = ["none", "left", "center", "right"] as const
+export type Align = (typeof ALIGNS)[number]
+export const ASCII_STYLES = ["unicode", "ascii"] as const
+export type AsciiStyle = (typeof ASCII_STYLES)[number]
+
+export interface SerializeOptions {
+  // a primeira linha da saída é o cabeçalho (CSV, ASCII, LaTeX, HTML); formatos com chaves sempre usam os nomes
+  header?: boolean
+  // números, true/false e null viram tipos de verdade (JSON, YAML, SQL)
+  types?: boolean
+  emptyAsNull?: boolean
+  decimal?: Decimal
+  outputDelimiter?: Delimiter
+  jsonShape?: JsonShape
+  indent?: Indent
+  tableName?: string
+  dialect?: SqlDialect
+  multiRow?: boolean
+  createTable?: boolean
+  align?: Align
+  pretty?: boolean
+  // nome da lista/elemento raiz (XML, TOML) ou do tipo (Go, TypeScript)
+  xmlRoot?: string
+  xmlRow?: string
+  asciiStyle?: AsciiStyle
+  // nome do tipo gerado (Go, TypeScript)
+  typeName?: string
+}
+
+export const EXTENSIONS: Record<OutputFormat, { ext: string, mime: string }> = {
+  csv: { ext: "csv", mime: "text/csv" },
+  json: { ext: "json", mime: "application/json" },
+  jsonl: { ext: "jsonl", mime: "application/jsonl" },
+  markdown: { ext: "md", mime: "text/markdown" },
+  html: { ext: "html", mime: "text/html" },
+  sql: { ext: "sql", mime: "application/sql" },
+  yaml: { ext: "yaml", mime: "application/yaml" },
+  xml: { ext: "xml", mime: "application/xml" },
+  ascii: { ext: "txt", mime: "text/plain" },
+  latex: { ext: "tex", mime: "application/x-tex" },
+  xlsx: { ext: "xlsx", mime: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet" },
+  toml: { ext: "toml", mime: "application/toml" },
+  ini: { ext: "ini", mime: "text/plain" },
+  jira: { ext: "txt", mime: "text/plain" },
+  mediawiki: { ext: "wiki", mime: "text/plain" },
+  asciidoc: { ext: "adoc", mime: "text/asciidoc" },
+  rst: { ext: "rst", mime: "text/x-rst" },
+  python: { ext: "py", mime: "text/x-python" },
+  pandas: { ext: "py", mime: "text/x-python" },
+  typescript: { ext: "ts", mime: "text/typescript" },
+  go: { ext: "go", mime: "text/x-go" },
+  php: { ext: "php", mime: "application/x-php" },
+}
+
+// Largura visível (emojis e acentos combinados contam 1; CJK e emoji ocupam 2 colunas no terminal)
+export function displayWidth(text: string): number {
+  let width = 0
+  for (const ch of text.replace(/\p{M}/gu, "")) {
+    const code = ch.codePointAt(0) ?? 0
+    width += (code >= 0x1100 && code <= 0x115f) || (code >= 0x2e80 && code <= 0xa4cf) || (code >= 0xac00 && code <= 0xd7a3) || (code >= 0xf900 && code <= 0xfaff) || (code >= 0xff00 && code <= 0xff60) || (code >= 0x1f300 && code <= 0x1faff) ? 2 : 1
+  }
+  return width
+}
+
+const pad = (text: string, width: number, align: Align = "left") => {
+  const gap = Math.max(0, width - displayWidth(text))
+  if (align === "right") return " ".repeat(gap) + text
+  if (align === "center") return " ".repeat(Math.floor(gap / 2)) + text + " ".repeat(Math.ceil(gap / 2))
+  return text + " ".repeat(gap)
+}
+
+export type ColumnKind = "number" | "integer" | "boolean" | "text"
+
+// Tipos decididos por coluna, não por célula: se um valor da coluna não é número (ex.: "8500,50" com
+// decimal "dot"), a coluna inteira fica como texto, para o JSON/SQL não misturar 7200 com "8500,50".
+// Vazios e "null" não decidem o tipo.
+export function columnKinds(t: Table, o: SerializeOptions = {}): ColumnKind[] {
+  return t.columns.map((_, i) => {
+    if (!(o.types ?? true)) return "text"
+    const values = t.rows.map((r) => r[i] ?? "").filter((c) => c !== "" && c.toLowerCase() !== "null").map((c) => inferValue(c, false, o.decimal))
+    if (values.length && values.every((v) => typeof v === "number")) return values.every((v) => Number.isInteger(v)) ? "integer" : "number"
+    if (values.length && values.every((v) => typeof v === "boolean")) return "boolean"
+    return "text"
+  })
+}
+
+export function typedRows(t: Table, o: SerializeOptions = {}, kinds = columnKinds(t, o)): Scalar[][] {
+  return t.rows.map((r) =>
+    t.columns.map((_, i) => {
+      const cell = r[i] ?? ""
+      if (cell === "") return o.emptyAsNull ? null : kinds[i] === "text" ? "" : null
+      if (kinds[i] === "text") return cell
+      return cell.toLowerCase() === "null" ? null : inferValue(cell, false, o.decimal)
+    }),
+  )
+}
+
+const indentOf = (indent: Indent = "2") => (indent === "tab" ? "\t" : indent === "min" ? undefined : Number(indent))
+const records = (t: Table, o: SerializeOptions) => {
+  const keys = uniqueKeys(t.columns)
+  return typedRows(t, o).map((r) => Object.fromEntries(keys.map((k, i) => [k, r[i] ?? null])))
+}
+
+// ---------- texto delimitado ----------
+
+export function toCsv(t: Table, o: SerializeOptions = {}): string {
+  const delimiter = DELIMITERS[o.outputDelimiter ?? "comma"]
+  const quote = (cell: string) => (cell.includes(delimiter) || /["\r\n]/.test(cell) || /^\s|\s$/.test(cell) ? `"${cell.replace(/"/g, '""')}"` : cell)
+  const lines = (o.header ?? true ? [t.columns, ...t.rows] : t.rows).map((r) => r.map(quote).join(delimiter))
+  return lines.join("\n")
+}
+
+// ---------- JSON ----------
+
+export function toJson(t: Table, o: SerializeOptions = {}): string {
+  const space = indentOf(o.indent)
+  const keys = uniqueKeys(t.columns)
+  let data: unknown
+  switch (o.jsonShape ?? "objects") {
+    case "objects":
+      data = records(t, o)
+      break
+    case "arrays":
+      data = [t.columns, ...typedRows(t, o)]
+      break
+    case "columns": {
+      const rows = typedRows(t, o)
+      data = Object.fromEntries(keys.map((k, i) => [k, rows.map((r) => r[i] ?? null)]))
+      break
+    }
+    case "keyed": {
+      // primeira coluna vira a chave do objeto
+      const rows = typedRows(t, o)
+      data = Object.fromEntries(t.rows.map((r, n) => [r[0] ?? "", Object.fromEntries(keys.slice(1).map((k, i) => [k, rows[n]?.[i + 1] ?? null]))]))
+      break
+    }
+  }
+  return JSON.stringify(data, null, space)
+}
+
+export const toJsonLines = (t: Table, o: SerializeOptions = {}) => records(t, o).map((r) => JSON.stringify(r)).join("\n")
+
+// ---------- Markdown ----------
+
+export function toMarkdown(t: Table, o: SerializeOptions = {}): string {
+  const escape = (cell: string) => cell.replace(/\\/g, "\\\\").replace(/\|/g, "\\|").replace(/\r?\n/g, "<br>")
+  const head = t.columns.map(escape)
+  const body = t.rows.map((r) => r.map(escape))
+  const align = o.align ?? "none"
+  const pretty = o.pretty ?? true
+  const widths = head.map((h, i) => Math.max(3, displayWidth(h), ...body.map((r) => displayWidth(r[i] ?? ""))))
+  const rule = (w: number) => {
+    const dashes = pretty ? w : 3
+    if (align === "left") return `:${"-".repeat(dashes - 1)}`
+    if (align === "right") return `${"-".repeat(dashes - 1)}:`
+    if (align === "center") return `:${"-".repeat(Math.max(1, dashes - 2))}:`
+    return "-".repeat(dashes)
+  }
+  const line = (cells: string[]) => `| ${cells.map((c, i) => (pretty ? pad(c, widths[i] ?? 0, align === "none" ? "left" : align) : c)).join(" | ")} |`
+  return [line(head), `| ${widths.map(rule).join(" | ")} |`, ...body.map(line)].join("\n")
+}
+
+// ---------- HTML ----------
+
+export const escapeHtml = (text: string) => text.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/"/g, "&quot;")
+
+export function toHtml(t: Table, o: SerializeOptions = {}): string {
+  const pretty = o.pretty ?? true
+  const [nl, i1, i2, i3] = pretty ? ["\n", "  ", "    ", "      "] : ["", "", "", ""]
+  const cell = (tag: string, text: string) => `${i3}<${tag}>${escapeHtml(text).replace(/\r?\n/g, "<br>")}</${tag}>`
+  const row = (tag: string, cells: string[]) => `${i2}<tr>${nl}${cells.map((c) => cell(tag, c)).join(nl)}${nl}${i2}</tr>`
+  const parts = ["<table>"]
+  if (o.header ?? true) parts.push(`${i1}<thead>`, row("th", t.columns), `${i1}</thead>`)
+  parts.push(`${i1}<tbody>`, ...t.rows.map((r) => row("td", r)), `${i1}</tbody>`, "</table>")
+  return parts.join(nl)
+}
+
+// ---------- SQL ----------
+
+const sqlIdent = (name: string, dialect: SqlDialect) => {
+  if (dialect === "mysql") return `\`${name.replace(/`/g, "``")}\``
+  if (dialect === "mssql") return `[${name.replace(/]/g, "]]")}]`
+  return `"${name.replace(/"/g, '""')}"`
+}
+
+const sqlLiteral = (v: Scalar, dialect: SqlDialect) => {
+  if (v === null) return "NULL"
+  if (typeof v === "number") return String(v)
+  if (typeof v === "boolean") return dialect === "mssql" ? (v ? "1" : "0") : v ? "TRUE" : "FALSE"
+  return `'${v.replace(/'/g, "''")}'`
+}
+
+// Tipo de coluna inferido dos valores: inteiro, decimal, booleano ou texto (VARCHAR com folga)
+export function sqlColumnType(cells: string[], dialect: SqlDialect, decimal: Decimal = "dot", infer = true): string {
+  const values = infer ? cells.filter((c) => c !== "" && c.toLowerCase() !== "null").map((c) => inferValue(c, false, decimal)) : []
+  const longest = Math.max(1, ...cells.map((c) => c.length))
+  const text = dialect === "sqlite" ? "TEXT" : dialect === "mssql" ? `NVARCHAR(${longest > 4000 ? "MAX" : Math.min(4000, Math.ceil(longest * 1.5))})` : longest > 255 ? "TEXT" : `VARCHAR(${Math.min(255, Math.max(16, Math.ceil(longest * 1.5)))})`
+  if (values.length === 0) return text
+  if (values.every((v) => typeof v === "number" && Number.isInteger(v))) return dialect === "sqlite" ? "INTEGER" : values.some((v) => Math.abs(v as number) > 2_147_483_647) ? "BIGINT" : "INT"
+  if (values.every((v) => typeof v === "number")) return dialect === "sqlite" ? "REAL" : "DECIMAL(18,6)"
+  if (values.every((v) => typeof v === "boolean")) return dialect === "mssql" ? "BIT" : dialect === "sqlite" ? "INTEGER" : "BOOLEAN"
+  return text
+}
+
+export function toSql(t: Table, o: SerializeOptions = {}): string {
+  const dialect = o.dialect ?? "mysql"
+  const table = sqlIdent(o.tableName?.trim() || "my_table", dialect)
+  const keys = uniqueKeys(t.columns)
+  const columns = keys.map((k) => sqlIdent(k, dialect)).join(", ")
+  const typed = typedRows(t, o)
+  const tuple = (r: Scalar[]) => `(${keys.map((_, i) => sqlLiteral(r[i] ?? null, dialect)).join(", ")})`
+  const parts: string[] = []
+  if (o.createTable) {
+    const defs = keys.map((k, i) => `  ${sqlIdent(k, dialect)} ${sqlColumnType(t.rows.map((r) => r[i] ?? ""), dialect, o.decimal, o.types ?? true)}`)
+    parts.push(`CREATE TABLE ${table} (\n${defs.join(",\n")}\n);`)
+  }
+  if (t.rows.length) {
+    // MS SQL aceita no máximo 1000 linhas por VALUES
+    const batch = o.multiRow ?? true ? (dialect === "mssql" ? 1000 : 500) : 1
+    for (let i = 0; i < typed.length; i += batch) {
+      const chunk = typed.slice(i, i + batch)
+      parts.push(chunk.length === 1 ? `INSERT INTO ${table} (${columns}) VALUES ${tuple(chunk[0] as Scalar[])};` : `INSERT INTO ${table} (${columns}) VALUES\n${chunk.map((r) => `  ${tuple(r)}`).join(",\n")};`)
+    }
+  }
+  return parts.join("\n\n")
+}
+
+// ---------- YAML ----------
+
+const YAML_PLAIN_UNSAFE = /^(?:[-?:,[\]{}#&*!|>'"%@`]|\s)|[:\s]$|: | #|[\r\n\t]/
+const YAML_RESERVED = /^(?:true|false|yes|no|on|off|y|n|null|~|[-+]?(?:\.inf|\.nan)|[-+]?(?:\d[\d_]*)?(?:\.\d*)?(?:e[-+]?\d+)?|0x[\da-f]+|0o[0-7]+)$/i
+
+export function yamlScalar(v: Scalar): string {
+  if (v === null) return "null"
+  if (typeof v !== "string") return String(v)
+  if (v === "" || YAML_PLAIN_UNSAFE.test(v) || YAML_RESERVED.test(v)) return JSON.stringify(v)
+  return v
+}
+
+export function toYaml(t: Table, o: SerializeOptions = {}): string {
+  if (t.rows.length === 0) return "[]"
+  const keys = uniqueKeys(t.columns).map(yamlScalar)
+  return typedRows(t, o).map((r) => keys.map((k, i) => `${i === 0 ? "- " : "  "}${k}: ${yamlScalar(r[i] ?? null)}`).join("\n")).join("\n")
+}
+
+// ---------- XML ----------
+
+export function xmlName(name: string): string {
+  const cleaned = name.trim().replace(/[^\p{L}\p{N}_.-]+/gu, "_")
+  return !cleaned || /^[^\p{L}_]/u.test(cleaned) || /^xml/i.test(cleaned) ? `_${cleaned}` : cleaned
+}
+
+const escapeXml = (text: string) => escapeHtml(text).replace(/'/g, "&apos;")
+
+export function toXml(t: Table, o: SerializeOptions = {}): string {
+  const root = xmlName(o.xmlRoot || "rows")
+  const rowName = xmlName(o.xmlRow || "row")
+  const keys = uniqueKeys(t.columns).map(xmlName)
+  const rows = t.rows.map((r) => `  <${rowName}>\n${keys.map((k, i) => (r[i] ? `    <${k}>${escapeXml(r[i] ?? "")}</${k}>` : `    <${k}/>`)).join("\n")}\n  </${rowName}>`)
+  return [`<?xml version="1.0" encoding="UTF-8"?>`, rows.length ? `<${root}>\n${rows.join("\n")}\n</${root}>` : `<${root}/>`].join("\n")
+}
+
+// ---------- ASCII ----------
+
+export function toAscii(t: Table, o: SerializeOptions = {}): string {
+  const box = (o.asciiStyle ?? "unicode") === "unicode" ? { h: "─", v: "│", tl: "┌", tm: "┬", tr: "┐", ml: "├", mm: "┼", mr: "┤", bl: "└", bm: "┴", br: "┘" } : { h: "-", v: "|", tl: "+", tm: "+", tr: "+", ml: "+", mm: "+", mr: "+", bl: "+", bm: "+", br: "+" }
+  const flat = (c: string) => c.replace(/\r?\n/g, " ")
+  const grid = (o.header ?? true ? [t.columns, ...t.rows] : t.rows).map((r) => r.map(flat))
+  const width = t.columns.length
+  if (width === 0) return ""
+  const widths = Array.from({ length: width }, (_, i) => Math.max(1, ...grid.map((r) => displayWidth(r[i] ?? ""))))
+  const right = widths.map((_, i) => t.rows.length > 0 && t.rows.every((r) => /^-?[\d.,]+%?$/.test((r[i] ?? "").trim()) || !r[i]))
+  const rule = (l: string, m: string, r: string) => l + widths.map((w) => box.h.repeat(w + 2)).join(m) + r
+  const line = (cells: string[], head: boolean) => box.v + widths.map((w, i) => ` ${pad(cells[i] ?? "", w, right[i] && !head ? "right" : "left")} `).join(box.v) + box.v
+  const out = [rule(box.tl, box.tm, box.tr)]
+  grid.forEach((r, i) => {
+    out.push(line(r, i === 0 && (o.header ?? true)))
+    if (i === 0 && (o.header ?? true) && grid.length > 1) out.push(rule(box.ml, box.mm, box.mr))
+  })
+  out.push(rule(box.bl, box.bm, box.br))
+  return out.join("\n")
+}
+
+// ---------- LaTeX ----------
+
+const LATEX: Record<string, string> = { "\\": "\\textbackslash{}", "&": "\\&", "%": "\\%", $: "\\$", "#": "\\#", _: "\\_", "{": "\\{", "}": "\\}", "~": "\\textasciitilde{}", "^": "\\textasciicircum{}" }
+export const escapeLatex = (text: string) => text.replace(/[\\&%$#_{}~^]/g, (ch) => LATEX[ch] ?? ch).replace(/\r?\n/g, " \\newline ")
+
+export function toLatex(t: Table, o: SerializeOptions = {}): string {
+  const align = { none: "l", left: "l", center: "c", right: "r" }[o.align ?? "none"]
+  const row = (cells: string[]) => `  ${cells.map(escapeLatex).join(" & ")} \\\\`
+  const lines = [`\\begin{tabular}{|${t.columns.map(() => align).join("|")}|}`, "  \\hline"]
+  if (o.header ?? true) lines.push(row(t.columns), "  \\hline")
+  lines.push(...t.rows.map(row), "  \\hline", "\\end{tabular}")
+  return lines.join("\n")
+}
+
+// ---------- TOML ----------
+
+const tomlKey = (key: string) => (/^[A-Za-z0-9_-]+$/.test(key) ? key : JSON.stringify(key))
+
+// Lista de tabelas ([[rows]]). TOML não tem null: a chave é omitida.
+export function toToml(t: Table, o: SerializeOptions = {}): string {
+  const name = tomlKey(o.xmlRoot || "rows")
+  const keys = uniqueKeys(t.columns).map(tomlKey)
+  return typedRows(t, o)
+    .map((r) => [`[[${name}]]`, ...keys.flatMap((k, i) => (r[i] === null || r[i] === undefined ? [] : [`${k} = ${typeof r[i] === "string" ? JSON.stringify(r[i]) : String(r[i])}`]))].join("\n"))
+    .join("\n\n")
+}
+
+// ---------- INI ----------
+
+// Uma seção por linha, com o nome tirado da primeira coluna
+export function toIni(t: Table): string {
+  const keys = uniqueKeys(t.columns)
+  const quote = (v: string) => (/[;#="]|^\s|\s$|\n/.test(v) ? `"${v.replace(/\\/g, "\\\\").replace(/"/g, '\\"').replace(/\r?\n/g, "\\n")}"` : v)
+  return t.rows
+    .map((r, n) => {
+      const section = (r[0] ?? "").replace(/[[\]\r\n]/g, "").trim() || `row ${n + 1}`
+      return [`[${section}]`, ...keys.slice(1).map((k, i) => `${k.replace(/[=\s]+/g, "_")} = ${quote(r[i + 1] ?? "")}`)].join("\n")
+    })
+    .join("\n\n")
+}
+
+// ---------- wikis e documentação ----------
+
+export function toJira(t: Table, o: SerializeOptions = {}): string {
+  const cell = (c: string) => c.replace(/\|/g, "\\|").replace(/\r?\n/g, " \\\\ ") || " "
+  const lines = t.rows.map((r) => `|${r.map(cell).join("|")}|`)
+  if (o.header ?? true) lines.unshift(`||${t.columns.map(cell).join("||")}||`)
+  return lines.join("\n")
+}
+
+export function toMediaWiki(t: Table, o: SerializeOptions = {}): string {
+  const cell = (c: string) => c.replace(/\|/g, "&#124;").replace(/\r?\n/g, "<br>")
+  const lines = ['{| class="wikitable"']
+  if (o.header ?? true) lines.push("|-", `! ${t.columns.map(cell).join(" !! ")}`)
+  for (const r of t.rows) lines.push("|-", `| ${r.map(cell).join(" || ")}`)
+  lines.push("|}")
+  return lines.join("\n")
+}
+
+export function toAsciiDoc(t: Table, o: SerializeOptions = {}): string {
+  const cell = (c: string) => c.replace(/\|/g, "\\|").replace(/\r?\n/g, " +\n")
+  const row = (r: string[]) => r.map((c) => `| ${cell(c)}`).join(" ")
+  const header = o.header ?? true
+  const lines = [`[cols="${t.columns.map(() => "1").join(",")}"${header ? ', options="header"' : ""}]`, "|==="]
+  if (header) lines.push(row(t.columns), "")
+  lines.push(...t.rows.map(row), "|===")
+  return lines.join("\n")
+}
+
+// reStructuredText: grid table, com "=" separando o cabeçalho
+export function toRst(t: Table, o: SerializeOptions = {}): string {
+  if (t.columns.length === 0) return ""
+  const header = o.header ?? true
+  const flat = (c: string) => c.replace(/\r?\n/g, " ")
+  const grid = (header ? [t.columns, ...t.rows] : t.rows).map((r) => r.map(flat))
+  const widths = t.columns.map((_, i) => Math.max(1, ...grid.map((r) => displayWidth(r[i] ?? ""))))
+  const rule = (ch: string) => `+${widths.map((w) => ch.repeat(w + 2)).join("+")}+`
+  const out = [rule("-")]
+  grid.forEach((r, i) => {
+    out.push(`|${widths.map((w, c) => ` ${pad(r[c] ?? "", w)} `).join("|")}|`)
+    out.push(rule(i === 0 && header ? "=" : "-"))
+  })
+  return out.join("\n")
+}
+
+export function serialize(t: Table, format: TextFormat, o: SerializeOptions = {}): string {
+  switch (format) {
+    case "csv":
+      return toCsv(t, o)
+    case "json":
+      return toJson(t, o)
+    case "jsonl":
+      return toJsonLines(t, o)
+    case "markdown":
+      return toMarkdown(t, o)
+    case "html":
+      return toHtml(t, o)
+    case "sql":
+      return toSql(t, o)
+    case "yaml":
+      return toYaml(t, o)
+    case "xml":
+      return toXml(t, o)
+    case "ascii":
+      return toAscii(t, o)
+    case "latex":
+      return toLatex(t, o)
+    case "toml":
+      return toToml(t, o)
+    case "ini":
+      return toIni(t)
+    case "jira":
+      return toJira(t, o)
+    case "mediawiki":
+      return toMediaWiki(t, o)
+    case "asciidoc":
+      return toAsciiDoc(t, o)
+    case "rst":
+      return toRst(t, o)
+    case "python":
+      return toPython(t, o)
+    case "pandas":
+      return toPandas(t, o)
+    case "typescript":
+      return toTypeScript(t, o)
+    case "go":
+      return toGo(t, o)
+    case "php":
+      return toPhp(t, o)
+  }
+}

package/src/table-convert/transform.ts

@@ -0,0 +1,133 @@
+import { parseNumber, type Decimal, type Table } from "./model"
+
+export const CASES = ["none", "upper", "lower", "title"] as const
+export type CaseMode = (typeof CASES)[number]
+
+// Transformações não destrutivas: a entrada colada não muda, então desfazer é só desligar a opção
+// (e o link compartilhado reproduz o mesmo resultado).
+export interface TransformOptions {
+  trim?: boolean
+  dropEmpty?: boolean
+  find?: string
+  replace?: string
+  regex?: boolean
+  ignoreCase?: boolean
+  caseMode?: CaseMode
+  // índice da coluna (na tabela lida) para ordenar; -1 = sem ordenação
+  sortBy?: number
+  descending?: boolean
+  // colunas exibidas, na ordem, por índice na tabela lida; vazio = todas
+  columns?: number[]
+  dedupe?: boolean
+  transpose?: boolean
+  // como reconhecer números ao ordenar
+  decimal?: Decimal
+}
+
+export interface Transformed {
+  table: Table
+  // para cada coluna do resultado, o índice na tabela lida (undefined depois de transpor)
+  source: (number | undefined)[]
+  // colunas que podem aparecer (as vazias removidas por dropEmpty ficam de fora), para restaurar ocultas
+  available: number[]
+  removedDuplicates: number
+  removedEmpty: number
+}
+
+export class FindPatternError extends Error {}
+
+const titleCase = (text: string) => text.toLocaleLowerCase().replace(/(^|[\s\-_/(])(\p{L})/gu, (_, sep: string, ch: string) => sep + ch.toLocaleUpperCase())
+
+function cellMapper(options: TransformOptions): ((cell: string) => string) | undefined {
+  const steps: ((cell: string) => string)[] = []
+  if (options.trim) steps.push((c) => c.trim().replace(/[ \t]+/g, " "))
+  if (options.find) {
+    const flags = `g${options.ignoreCase ? "i" : ""}u`
+    let pattern: RegExp
+    try {
+      pattern = new RegExp(options.regex ? options.find : options.find.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"), flags)
+    } catch (error) {
+      throw new FindPatternError(error instanceof Error ? error.message : String(error))
+    }
+    const replacement = options.replace ?? ""
+    // sem regex, "$1" na substituição é texto literal
+    steps.push(options.regex ? (c) => c.replace(pattern, replacement) : (c) => c.replace(pattern, () => replacement))
+  }
+  if (options.caseMode === "upper") steps.push((c) => c.toLocaleUpperCase())
+  if (options.caseMode === "lower") steps.push((c) => c.toLocaleLowerCase())
+  if (options.caseMode === "title") steps.push(titleCase)
+  if (steps.length === 0) return undefined
+  return (cell) => steps.reduce((value, step) => step(value), cell)
+}
+
+const collator = new Intl.Collator(undefined, { numeric: true, sensitivity: "base" })
+
+export function compareCells(a: string, b: string, decimal: Decimal = "dot"): number {
+  // números comparam como números; o resto em ordem "natural" (item2 antes de item10)
+  const x = parseNumber(a, decimal)
+  const y = parseNumber(b, decimal)
+  if (x !== undefined && y !== undefined) return x - y
+  return collator.compare(a, b)
+}
+
+export function transposeTable(table: Table): Table {
+  const grid = [table.columns, ...table.rows]
+  const [first = [], ...rest] = grid[0]?.map((_, c) => grid.map((row) => row[c] ?? "")) ?? []
+  return { columns: first, rows: rest }
+}
+
+// Ordem: limpeza → localizar/substituir → caixa → ordenar → escolher colunas → remover duplicadas → transpor
+export function transformTable(input: Table, options: TransformOptions = {}): Transformed {
+  const map = cellMapper(options)
+  const columns = map ? input.columns.map(map) : [...input.columns]
+  let rows = map ? input.rows.map((r) => r.map(map)) : input.rows.map((r) => [...r])
+
+  let removedEmpty = 0
+  if (options.dropEmpty) {
+    const before = rows.length
+    rows = rows.filter((r) => r.some((c) => c.trim() !== ""))
+    removedEmpty = before - rows.length
+    const keep = columns.map((name, i) => name.trim() !== "" && !/^column\d+$/.test(name) ? true : rows.some((r) => (r[i] ?? "").trim() !== ""))
+    if (keep.includes(false) && keep.includes(true)) {
+      // colunas vazias saem; a numeração de origem é preservada pelo índice
+      const indices = keep.flatMap((k, i) => (k ? [i] : []))
+      removedEmpty += keep.length - indices.length
+      return finish({ columns, rows }, options, indices, removedEmpty)
+    }
+  }
+  return finish({ columns, rows }, options, columns.map((_, i) => i), removedEmpty)
+}
+
+function finish(table: Table, options: TransformOptions, available: number[], removedEmpty: number): Transformed {
+  let rows = table.rows
+  const sortBy = options.sortBy ?? -1
+  if (sortBy >= 0 && sortBy < table.columns.length) {
+    const direction = options.descending ? -1 : 1
+    rows = [...rows].sort((a, b) => {
+      const x = a[sortBy] ?? ""
+      const y = b[sortBy] ?? ""
+      if (x === "" || y === "") return x === y ? 0 : x === "" ? 1 : -1
+      return compareCells(x, y, options.decimal) * direction
+    })
+  }
+
+  const chosen = (options.columns ?? []).filter((i) => available.includes(i))
+  const source = chosen.length ? chosen : available
+  let result: Table = { columns: source.map((i) => table.columns[i] ?? ""), rows: rows.map((r) => source.map((i) => r[i] ?? "")) }
+
+  let removedDuplicates = 0
+  if (options.dedupe) {
+    const seen = new Set<string>()
+    const unique = result.rows.filter((r) => {
+      const key = JSON.stringify(r)
+      if (seen.has(key)) return false
+      seen.add(key)
+      return true
+    })
+    removedDuplicates = result.rows.length - unique.length
+    result = { ...result, rows: unique }
+  }
+
+  if (options.transpose) return { table: transposeTable(result), source: [], available, removedDuplicates, removedEmpty }
+  return { table: result, source, available, removedDuplicates, removedEmpty }
+}