Interfaces for IO that make use of promises.

Q-IO now subsumes all of Q-HTTP and Q-FS.

The Q-IO package does not export a main module. You must reach in directly for q-io/fs, q-io/http, and q-io/http-apps.

var FS = require("q-io/fs");

File system API for Q promises with method signatures patterned after CommonJS/Fileystem/A but returning promises and promise streams.

Open returns a promise for either a buffer or string Reader or a Writer depending on the flags.

The options can be omitted, abbreviated to a flags string, or expanded to an options object.

• flags: r, w, a, b, default of r, not bytewise
• charset: default of utf-8
• bufferSize: in bytes
• mode: UNIX permissions
• begin first byte to read (defaults to zero)
• end one past the last byte to read. end - begin == length

read is a shortcut for opening a file and reading the entire contents into memory. It returns a promise for the whole file contents. By default, read provides a string decoded from UTF-8. With the bytewise mode flag, provides a Buffer.

The options argument is identical to that of open.

return FS.read(__filename, "b")
.then(function (content) {
    // ...
})

return FS.read(__filename, {
    flags: "b"
})

write is a shortcut for opening a file and writing its entire content from a single string or buffer.

The options are identical to that of open, but the "w" flag is implied, and the "b" flag is implied if the content is a buffer.

return FS.write("hello.txt", "Hello, World!\n")
.then(function () {
    return FS.read("hello.txt")
})
.then(function (hello) {
    expect(hello).toBe("Hello, World!\n")
})

append is a shorthand for opening a file for writing from the end of the existing content from a single string or buffer.

The options are identical to that of open, but the "w+" flags are implied, and the "b" flag is implied if the content is a buffer.

Copies a single file from one path to another. The target must be the full path, including the file name. Unlike at the shell, the file name is not inferred from the source path if the target turns out to be a directory.

Returns a promise for the completion of the operation.

Copies a file or tree of files from one path to another. Symbolic links are copied but not followed.

Returns a promise for the completion of the operation.

Returns a promise for a list of file names in a directory. The file names are relative to the given path.

Returns a promise for a list of files in a directory and all the directories it contains. Does not follow symbolic links.

The second argument is an optional guard function that determines what files to include and whether to traverse into another directory. It receives the path of the file, relative to the starting path, and also the stats object for that file. The guard must return a value like:

• true indicates that the entry should be included
• false indicates that the file should be excluded, but should still be traversed if it is a directory.
• null indiciates that a directory should not be traversed.

Returns a promise for a deep list of directories.

Makes a directory at a given path. Fails if the parent directory does not exist. Returns a promise for the completion of the operation.

The mode is an optional Unix mode as an integer or string of octal digits.

Finishes a path of directories. For any branch of the path that does not exist, creates a directory. Fails if any branch of the path already exists but is not a directory.

Makes any directories with the given Unix mode.

Removes a file at the given path. Fails if a directory exists at the given path or if no file exists at the path.

Removes a file or directory at a given path, recursively removing any contained files and directories, without following symbolic links.

Moves a file or directory from one path to another using the underlying rename(2) implementation, thus it cannot move a file across devices.

Moves a file or directory from one path to another. If the source and target are on different devices, falls back to copying and removing, using copyTree(source, target) and, if completely successful, removeTree(source).

Creates a hard link from the source

Creates a relative symbolic link from the target to the source with an effect that resembles copying a file.

The type is important for Windows. It is "file" by default, but may be "directory" or "junction".

Creates a symbolic link at the target path. The link may be absolute or relative. The type must be "file", "directory", or "junction" and is mandatory to encourage Windows portability.

Changes the owner for a path using Unix user-id and group-id numbers.

Changes the Unix mode for a path. Returns a promise.

Follows all symbolic links along a path and returns a promise for the metadata about a path as a Stats object. The Stats object implements:

• size the size of the file in bytes
• isDirectory(): returns whether the path refers to a directory with entries for other paths.
• isFile(): returns whether the path refers to a file physically stored by the file system.
• isBlockDevice(): returns whether the path refers to a Unix device driver, in which case there is no actual data in storage but the operating system may allow you to communicate with the driver as a blocks of memory.
• isCharacterDevice(): returns whether the path refers to a Unix device driver, in which case there is no actual data in storage but the operating system may allow you to communicate with the driver as a stream.
• isSymbolicLink(): returns whether the path refers to a symbolic link or junction. Stats for symbolic links are only discoverable through statLink since stat follows symbolic links.
• isFIFO(): returns whether the path refers to a Unix named pipe.
• isSocket(): returns whether the path refers to a Unix domain socket.
• lastModified(): returns the last time the path was opened for writing as a Date
• lastAccessed(): returns the last time the path was opened for reading or writing as a Date

Returns a promise for the Stats for a path without following symbolic links.

Returns a promise for the Stats for a Unix file descriptor number.

Follows symbolic links and returns a promise for whether an entry exists at a given path.

Follows symbolic links and returns a promise for whether a file exists at a given path and does not cause an exception if nothing exists at that path.

Follows symbolic links and returns a promise for whether a directory exists at a given path and does not cause an exception if nothing exists at that path.

Returns a promise for whether a symbolic link exists at a given path and does not cause an exception if nothing exists at that path.

Follows symbolic links and returns a promise for the Date when the entry at the given path was last opened for writing, but causes an exception if no file exists at that path.

Follows symbolic links and returns a promise for the Date when the entry at the given path was last opened for reading or writing, but causes an exception if no file exists at that path.

Splits a path into the names of entries along the path. If the path is absolute, the first component is either a drive (with a colon) on Windows or an empty string for the root of a Unix file system.

Joins a sequence of paths into a single normalized path. All but the last path are assumed to refer to directories.

Like join but treats each path like a relative URL, so a terminating slash indicates that a path is to a directory and the next path begins at that directory.

Takes a single path or sequence of paths and joins them into a single path, eliminating self . and parent .. entries when possible.

Joins and normalizes a path from the current working directory, returning a string.

Returns a promise for the absolute, canonical location of a given path, following symbolic links and normalizing path components. An entry does not need to exist at the end of the path.

Returns a promise for the path string of a symbolic link at a given path.

For any two absolute or relative paths, computes whether the parent path is an ancestor of the child path.

Returns a promise for the relative path from one path to another using .. parent links where necessary. This operation is asynchronous because it is necessary to determine whether the source path refers to a directory or a file.

Assuming that the source path refers to a file, returns a string for the relative path from the source to the target path.

Assuming that the source path refers to a directory, returns a string for the relative path from the source to the target path.

Returns whether a path begins at the root of a Unix file system or a Windows drive.

Returns whether a path does not begin at the root of a Unix file system or Windows drive.

Returns whether a path is to the root of a Unix file system or a Windows drive.

Returns the Windows drive that contains a given path, or the root of a Unix file system.

Returns the path to the directory containing the given path.

Returns the last entry of a path. If an extension is provided and matches the extension of the path, removes that extension.

Returns the extension for a path (everything following the last dot . in a path, unless that dot is at the beginning of the entry).

Returns an attenuated file system that uses the given path as its root. The resulting file system object is identical to the parent except that the child cannot open any file that is not within the root. Hard links are effectively inside the root regardless, but symbolic links cannot be followed outside of the jail.

Reads every file in the file system under a given path and returns a promise for an object that contains the absolute path and a Buffer for each of those files.

Not yet implemented

Not yet implemented

Q-IO provides a mock filesystem interface. The mock filesystem has the same interface as the real one and has most of the same features, but operates on a purely in-memory, in-process, in-javascript filesystem.

A mock filesystem can be created from a data structure. Objects are directories. Keys are paths. A buffer is a file’s contents. Anything else is coerced to a string, then to a buffer in the UTF-8 encoding.

var MockFs = require("q-io/fs-mock");
var mockFs = MockFs({
    "a": {
        "b": {
            "c.txt": "Content of a/b/c.txt"
        }
    },
    "a/b/d.txt": new Buffer("Content of a/b/d.txt", "utf-8")
})

You can also instantiate a mock file system with the content of a subtree of a real file system. You receive a promise for the mock filesystem.

var FS = require("q-io/fs");
FS.mock(__dirname)
.then(function (fs) {
    //
})
.done();

The HTTP module resembles CommonJS/JSGI.

var HTTP = require("q-io/http");

The http module exports a Server constructor.

• accepts an application, returns a server.
• calls the application function when requests are received.
  • if the application returns a response object, sends that response.
• listen(port)
  • accepts a port number.
  • returns a promise for undefined when the server has begun listening.
• stop()
  • returns a promise for undefined when the server has stopped.

The http module exports a request function that returns a promise for a response.

• accepts a request object or a URL string.
• returns a promise for a response object.

The http module exports a read function, analogous to Fs.read(path), but returning a promise for the content of an OK HTTP response.

• accepts a request object or a URL string.
• returns a promise for the response body as a string provided that the request is successful with a 200 status.
  • rejects the promise with the response as the reason for failure if the request fails.

• coerces URLs into request objects.
• completes an incomplete request object based on its url.

• coerces strings, arrays, and other objects supporting forEach into proper response objects.
• if it receives undefined, it returns undefined. This is used as a singal to the requester that the responder has taken control of the response stream.

A complete request object has the following properties.

• url the full URL of the request as a string
• path the full path as a string
• scriptName the routed portion of the path, like "" for http://example.com/ if no routing has occurred.
• pathInfo the part of the path that remains to be routed, like / for http://example.com or http://example.com/ if no routing has occurred.
• version the requested HTTP version as an array of strings.
• method like "GET"
• scheme like "http"
• host like "example.com" in case of default ports (80 or 443), otherwise like example.com:8080
• hostname like "example.com"
• port the port number, like 80
• remoteHost
• remotePort
• headers corresponding values, possibly an array for multiple headers of the same name.
• agent a custom node HTTP or HTTPS agent. HTTP and HTTPS agents can implement custom socket pools, allow use of SSL client certificates and self-signed certificates.
• body an array of string or node buffers
• timeout an optional socket timeout in miliseconds to thread through to the HTTP agent.
• node the wrapped Node request object
• cancelled a promise that indicates by rejection that the HTTP request should be aborted and the response promise should be rejected with the given error. Such a promise must not be fulfilled.

A complete response object has the following properties.

• status the HTTP status code as a number, like 200.
• headers
• body any forEachable, such as an array of strings
• onclose is an optional function that this library will call when a response concludes.
• node the wrapped Node response object.

Headers are an object mapping lower-case header-names to corresponding values, possibly an array for multiple headers of the same name, for both requests and responses.

body is a representation of a readable stream, either for the content of a request or a response. It is implemented as a Q-IO reader.

• forEach(callback)
  • accepts a callback(chunk) function
    • accepts a chunk as either a string or a Buffer
    • returns undefined or a promise for undefined when the chunk has been flushed.
  • returns undefined or a promise for undefined when the stream is finished writing.
  • the forEach function for arrays of strings or buffers is sufficient for user-provided bodies
• the forEach function is the only necessary function for bodies provided to this library.
• in addition to forEach, bodies provided by this library support the entire readable stream interface provided by q-io.
• read()
  • returns a promise for the entire body as a string or a buffer.

An HTTP application is a function that accepts a request and returns a response. The request function itself is an application. Applications can be chained and combined to make advanced servers and clients.

• accepts a request
• returns a response, a promise for a response, or nothing if no response should be sent.

Reader instances have the following methods:

• read()
• forEach(callback)
• close()
• node the underlying node reader

Additionally, the Reader constructor has the following methods:

• read(stream, charset) accepts any foreachable and returns either a buffer or a string if given a charset.
• join(buffers) consolidates an array of buffers into a single buffer. The buffers array is collapsed in place and the new first and only buffer is returned.

The reader module exports a function that accepts a Node reader and returns a Q reader.

Writer instances have the following methods:

• write(content) writes a chunk of content, either from a string or a buffer.
• flush() returns a promise to drain the outbound content all the way to its destination.
• close()
• destroy()
• node the underlying node writer

The writer module exports a function that accepts a Node writer and returns a Q writer.

var BufferStream = require("q-io/buffer-stream");
var stream = BufferStream(new Buffer("Hello, World!\n", "utf-8"), "utf-8")

The HTTP applications module provides a comprehensive set of JSGI-alike applications and application factories, suitable for use with the http server and client.

var Apps = require("q-io/http-apps");

Creates an HTTP 200 Ok response with the given content, content type, and status.

The content may be a string, buffer, array of strings, array of buffers, a readable stream of strings or buffers, or (generally) anything that implements forEach.

The default content type is text/plain.

The default status is 200.

An application that returns an HTTP 400 Bad request response for any request.

An application that returns an HTTP 404 Not found response for any request.

An application that returns an HTTP 405 Method not allowed response for any request. This is suitable for any endpoint where there is no viable handler for the request method.

An application that returns an HTTP 406 Not acceptable response for any request. This is suitable for any situation where content negotiation has failed, for example, if you cannot response with any of the accepted encoding, charset, or language.

Not to be confused with an HTTP application, this is a utility that generates redirect responses.

The returns response issues a redirect to the given location. The utility fully qualifies the location.

This particular method should be used directly to generate an HTTP 301 Temporary redirect response, but passing 307 in the status argument turns it into an HTTP 307 Permanent redirect response.

This particular method should be used to send all requests to a specific location, but setting the tree argument to true causes the redirect to follow the remaining unrouted path from the redirect location, so if you move an entire directory tree from one location to another, this redirect can forward to all of them.

Produces an HTTP 301 Temporary redirect from one directory tree to another, using redirect.

Produces an HTTP 307 Permanent redirect to a given location, using redirect.

Produces an HTTP 307 Permanent redirect from one directory tree to another, using redirect.

Produces an HTTP response with the file at a given path. By default, it infers the content type from the extension of the path.

The file utility produces an e-tag header suitable for cache control, and may produce an HTTP 304 Not modified if the requested resource has the same entity tag.

The file utility may produce an HTTP 206 Partial content response with a content-range header if the request has a range header. If the partial range request cannot be satisified, it may respond HTTP 416 Not satisfiable.

In all cases, the response body is streamed from the file system.

Computes an entity tag for a file system Stats object, using the node.ino, size, and last modification time.

This is not yet implemented.

Returns an HTTP 200 Ok response from some JSON, using the same argumensts as JSON.stringify.

A factory that produces an HTTP application that will always respond with the given content, content type, and status. The default content type is text/plain and the default status is 200.

The body may be a string, array of strings or buffers, or a readable stream of strings or buffers.

A factory that produces an HTTP application that will always respond with the file at the given path. The content type is inferred from the path extension by default, but can be overridden with contentType.

A factory that produces an HTTP application that responds to all requests with files within a branch of the file system starting at the given path and using any unprocessed portion of the request location.

Options include:

• notFound(request, response): alternate 404 responder, defaults to HttpApps.notFound
• file(request, path, contentType, fs): alternate file responder, defaults to HttpApps.file
• contentType: forces the content type of file requests, forwarded to the file handler.
• directory(request, path, contentType, fs): alternate directory responder, defaults to HttpApps.directory.
• redirectSymbolicLinks: directs the client to use a redirect response for symbolic links instead of following links internally.
• permanent: symbolic links that are turned into HTTP redirects will be permanent.
• followInsecureSymbolicLinks: directs FileTree to serve files that are outside the root path of the file tree if a symbolic link traverses there.
• fs: alternate file system, defaults to the fs module.

A factory that produces an HTTP application that temporarily redirects to the given path.

A factory that produces an HTTP application that redirects all requests under the requested path to parallel locations at the given path.

A factory that produces an HTTP application that redirects all requests to an exact location and instructs the requester's cache never to ask again.

A factory that produces an HTTP application that redirects all requests under the request path to a parallel location under the given path and instructs the requester's cache never to ask again.

A factory that produces an HTTP application that will cause an HTTP 404 Not found response if the request has not reached the end of its route (meaning pathInfo is not "" or "/"), or will forward to the given application.

Several routing application factories have the same form. They all take an object as their first argument and an optional fallback application as their second. The object maps each of the supported options for keys to an HTTP application for handling that option.

• Branch(paths, notFound) routes the next unprocessed path component
• Method(methods, notFound) routes the HTTP method. Methods are upper-case.
• ContentType(contentTypes, notAcceptable) routes based on the "accept" request header and produces a "content-type" response header.
• Langauge(languages, notAcceptable) routes based on the "accept-language" header and produces a "language" response header.
• Charaset(charsets, notAcceptable) routes based on the "accept-charset" header and produces a "charset" response header.
• Encoding(encodings, notAcceptable) routes based on the "accept-encoding" request header and produces an "encoding" response header.
• Host(hosts, notFound) routes based on the host name of the request "host" header, which defaults to "*". This is equivalent to virtual host mapping.

Produces an HTTP application that uses a function to determine the next application to route. The selector is a function that accepts the request and returns an HTTP application.

Returns an HTTP application that attempts to respond with each of a series of applications and returns the first response that does not have a 404 Not found status, or whatever response comes last.

Wraps an application such that any exceptions get converted into HTTP 500 Server error responses. If debug is enabled, produces the exception and stack traces in the body of the response.

Wraps an application such that request and response times are logged. The log function reports to console.log by default. The stamp(message) function prefixes an ISO time stamp by default.

Adds an x-response-time header to the response, with the time from receiving starting the request to starting the response in miliseconds.

Adds a date header to the response with the current date for cache control purposes.

Wraps an application such that the tap function receives the request first. If the tap returns nothing, the request goes to the app. If the tap returns a response, the app never gets called.

Wraps an application such that the trap function receives the response. If it returns nothing, the response if forwarded. If the trap returns a response, the original response is discarded.

Wraps an application such that the query string is parsed and placed in request.parse.

Use npm run cover to generate and view a coverage report of Q-IO.

Copyright 2009–2013 Kristopher Michael Kowal MIT License (enclosed)