std/path/windows/parse.ts

// Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.
// This module is browser compatible.

import { CHAR_COLON, CHAR_DOT } from "../_common/constants.ts";
import type { ParsedPath } from "../types.ts";
import { assertPath } from "../_common/assert_path.ts";
import { isPathSeparator, isWindowsDeviceRoot } from "./_util.ts";

export type { ParsedPath } from "../types.ts";

/**
 * Return a `ParsedPath` object of the `path`.
 *
 * @example Usage
 * ```ts
 * import { parse } from "@std/path/windows/parse";
 * import { assertEquals } from "@std/assert";
 *
 * const parsed = parse("C:\\foo\\bar\\baz.ext");
 * assertEquals(parsed, {
 *   root: "C:\\",
 *   dir: "C:\\foo\\bar",
 *   base: "baz.ext",
 *   ext: ".ext",
 *   name: "baz",
 * });
 * ```
 *
 * @param path The path to parse.
 * @returns The `ParsedPath` object.
 */
export function parse(path: string): ParsedPath {
  assertPath(path);

  const ret: ParsedPath = { root: "", dir: "", base: "", ext: "", name: "" };

  const len = path.length;
  if (len === 0) return ret;

  let rootEnd = 0;
  let code = path.charCodeAt(0);

  // Try to match a root
  if (len > 1) {
    if (isPathSeparator(code)) {
      // Possible UNC root

      rootEnd = 1;
      if (isPathSeparator(path.charCodeAt(1))) {
        // Matched double path separator at beginning
        let j = 2;
        let last = j;
        // Match 1 or more non-path separators
        for (; j < len; ++j) {
          if (isPathSeparator(path.charCodeAt(j))) break;
        }
        if (j < len && j !== last) {
          // Matched!
          last = j;
          // Match 1 or more path separators
          for (; j < len; ++j) {
            if (!isPathSeparator(path.charCodeAt(j))) break;
          }
          if (j < len && j !== last) {
            // Matched!
            last = j;
            // Match 1 or more non-path separators
            for (; j < len; ++j) {
              if (isPathSeparator(path.charCodeAt(j))) break;
            }
            if (j === len) {
              // We matched a UNC root only

              rootEnd = j;
            } else if (j !== last) {
              // We matched a UNC root with leftovers

              rootEnd = j + 1;
            }
          }
        }
      }
    } else if (isWindowsDeviceRoot(code)) {
      // Possible device root

      if (path.charCodeAt(1) === CHAR_COLON) {
        rootEnd = 2;
        if (len > 2) {
          if (isPathSeparator(path.charCodeAt(2))) {
            if (len === 3) {
              // `path` contains just a drive root, exit early to avoid
              // unnecessary work
              ret.root = ret.dir = path;
              ret.base = "\\";
              return ret;
            }
            rootEnd = 3;
          }
        } else {
          // `path` contains just a relative drive root, exit early to avoid
          // unnecessary work
          ret.root = ret.dir = path;
          return ret;
        }
      }
    }
  } else if (isPathSeparator(code)) {
    // `path` contains just a path separator, exit early to avoid
    // unnecessary work
    ret.root = ret.dir = path;
    ret.base = "\\";
    return ret;
  }

  if (rootEnd > 0) ret.root = path.slice(0, rootEnd);

  let startDot = -1;
  let startPart = rootEnd;
  let end = -1;
  let matchedSlash = true;
  let i = path.length - 1;

  // Track the state of characters (if any) we see before our first dot and
  // after any path separator we find
  let preDotState = 0;

  // Get non-dir info
  for (; i >= rootEnd; --i) {
    code = path.charCodeAt(i);
    if (isPathSeparator(code)) {
      // If we reached a path separator that was not part of a set of path
      // separators at the end of the string, stop now
      if (!matchedSlash) {
        startPart = i + 1;
        break;
      }
      continue;
    }
    if (end === -1) {
      // We saw the first non-path separator, mark this as the end of our
      // extension
      matchedSlash = false;
      end = i + 1;
    }
    if (code === CHAR_DOT) {
      // If this is our first dot, mark it as the start of our extension
      if (startDot === -1) startDot = i;
      else if (preDotState !== 1) preDotState = 1;
    } else if (startDot !== -1) {
      // We saw a non-dot and non-path separator before our dot, so we should
      // have a good chance at having a non-empty extension
      preDotState = -1;
    }
  }

  if (
    startDot === -1 ||
    end === -1 ||
    // We saw a non-dot character immediately before the dot
    preDotState === 0 ||
    // The (right-most) trimmed path component is exactly '..'
    (preDotState === 1 && startDot === end - 1 && startDot === startPart + 1)
  ) {
    if (end !== -1) {
      ret.base = ret.name = path.slice(startPart, end);
    }
  } else {
    ret.name = path.slice(startPart, startDot);
    ret.base = path.slice(startPart, end);
    ret.ext = path.slice(startDot, end);
  }

  // Fallback to '\' in case there is no basename
  ret.base = ret.base || "\\";

  // If the directory is the root, use the entire root as the `dir` including
  // the trailing slash if any (`C:\abc` -> `C:\`). Otherwise, strip out the
  // trailing slash (`C:\abc\def` -> `C:\abc`).
  if (startPart > 0 && startPart !== rootEnd) {
    ret.dir = path.slice(0, startPart - 1);
  } else ret.dir = ret.root;

  return ret;
}