Release v0.32.1

- Latest version is supported
- Report vulnerabilities via e-mail

.github/SECURITY.md

@@ -0,0 +1,18 @@
+# Security Policy
+
+## Supported Versions
+
+The latest version of `sharp` as published to npm
+and reported by `npm view sharp dist-tags.latest`
+is supported with security updates.
+
+## Reporting a Vulnerability
+
+Please use
+[e-mail](https://github.com/lovell/sharp/blob/main/package.json#L5)
+to report a vulnerability.
+
+You can expect a response within 48 hours
+if you are a human reporting a genuine issue.
+
+Thank you in advance.

README.md

@@ -98,8 +98,6 @@ readableStream
 A [guide for contributors](https://github.com/lovell/sharp/blob/main/.github/CONTRIBUTING.md)
 covers reporting bugs, requesting features and submitting code changes.
 
-[![Node-API v5](https://img.shields.io/badge/Node--API-v5-green.svg)](https://nodejs.org/dist/latest/docs/api/n-api.html#n_api_n_api_version_matrix)
-
 ## Licensing
 
 Copyright 2013 Lovell Fuller and others.

docs/api-colour.md

@@ -55,7 +55,8 @@ Alternative spelling of `greyscale`.
 Set the pipeline colourspace.
 
 The input image will be converted to the provided colourspace at the start of the pipeline.
-All operations will use this colourspace before converting to the output colourspace, as defined by [toColourspace](#toColourspace).
+All operations will use this colourspace before converting to the output colourspace,
+as defined by [toColourspace](#tocolourspace).
 
 This feature is experimental and has not yet been fully-tested with all operations.
 

docs/api-constructor.md

@@ -51,9 +51,9 @@ Implements the [stream.Duplex](http://nodejs.org/api/stream.html#stream_class_st
 | [options.text.text] | <code>string</code> |  | text to render as a UTF-8 string. It can contain Pango markup, for example `<i>Le</i>Monde`. |
 | [options.text.font] | <code>string</code> |  | font name to render with. |
 | [options.text.fontfile] | <code>string</code> |  | absolute filesystem path to a font file that can be used by `font`. |
-| [options.text.width] | <code>number</code> | <code>0</code> | integral number of pixels to word-wrap at. Lines of text wider than this will be broken at word boundaries. |
-| [options.text.height] | <code>number</code> | <code>0</code> | integral number of pixels high. When defined, `dpi` will be ignored and the text will automatically fit the pixel resolution defined by `width` and `height`. Will be ignored if `width` is not specified or set to 0. |
-| [options.text.align] | <code>string</code> | <code>&quot;&#x27;left&#x27;&quot;</code> | text alignment (`'left'`, `'centre'`, `'center'`, `'right'`). |
+| [options.text.width] | <code>number</code> | <code>0</code> | Integral number of pixels to word-wrap at. Lines of text wider than this will be broken at word boundaries. |
+| [options.text.height] | <code>number</code> | <code>0</code> | Maximum integral number of pixels high. When defined, `dpi` will be ignored and the text will automatically fit the pixel resolution defined by `width` and `height`. Will be ignored if `width` is not specified or set to 0. |
+| [options.text.align] | <code>string</code> | <code>&quot;&#x27;left&#x27;&quot;</code> | Alignment style for multi-line text (`'left'`, `'centre'`, `'center'`, `'right'`). |
 | [options.text.justify] | <code>boolean</code> | <code>false</code> | set this to true to apply justification to the text. |
 | [options.text.dpi] | <code>number</code> | <code>72</code> | the resolution (size) at which to render the text. Does not take effect if `height` is specified. |
 | [options.text.rgba] | <code>boolean</code> | <code>false</code> | set this to true to enable RGBA output. This is useful for colour emoji rendering, or support for pango markup features like `<span foreground="red">Red!</span>`. |

docs/api-operation.md

@@ -93,7 +93,7 @@ Perform an affine transform on an image. This operation will always occur after
 
 You must provide an array of length 4 or a 2x2 affine transformation matrix.
 By default, new pixels are filled with a black background. You can provide a background color with the `background` option.
-A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolator` Object e.g. `sharp.interpolator.nohalo`.
+A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolators` Object e.g. `sharp.interpolators.nohalo`.
 
 In the case of a 2x2 matrix, the transform is:
 - X = `matrix[0, 0]` \* (x + `idx`) + `matrix[0, 1]` \* (y + `idy`) + `odx`
@@ -127,7 +127,7 @@ where:
 const pipeline = sharp()
   .affine([[1, 0.3], [0.1, 0.7]], {
      background: 'white',
-     interpolate: sharp.interpolators.nohalo
+     interpolator: sharp.interpolators.nohalo
   })
   .toBuffer((err, outputBuffer, info) => {
      // outputBuffer contains the transformed image
@@ -265,6 +265,31 @@ await sharp(rgbaInput)
 ```
 
 
+## unflatten
+Ensure the image has an alpha channel
+with all white pixel values made fully transparent.
+
+Existing alpha channel values for non-white pixels remain unchanged.
+
+This feature is experimental and the API may change.
+
+
+**Since**: 0.32.1  
+**Example**  
+```js
+await sharp(rgbInput)
+  .unflatten()
+  .toBuffer();
+```
+**Example**  
+```js
+await sharp(rgbInput)
+  .threshold(128, { grayscale: false }) // converter bright pixels to white
+  .unflatten()
+  .toBuffer();
+```
+
+
 ## gamma
 Apply a gamma correction by reducing the encoding (darken) pre-resize at a factor of `1/gamma`
 then increasing the encoding (brighten) post-resize at a factor of `gamma`.

docs/api-output.md

@@ -6,7 +6,7 @@ with JPEG, PNG, WebP, AVIF, TIFF, GIF, DZI, and libvips' V format supported.
 Note that raw pixel data is only supported for buffer output.
 
 By default all metadata will be removed, which includes EXIF-based orientation.
-See [withMetadata](#withMetadata) for control over this.
+See [withMetadata](#withmetadata) for control over this.
 
 The caller is responsible for ensuring directory structures and permissions exist.
 
@@ -42,12 +42,12 @@ sharp(input)
 Write output to a Buffer.
 JPEG, PNG, WebP, AVIF, TIFF, GIF and raw pixel data output are supported.
 
-Use [toFormat](#toFormat) or one of the format-specific functions such as [jpeg](#jpeg), [png](#png) etc. to set the output format.
+Use [toFormat](#toformat) or one of the format-specific functions such as [jpeg](#jpeg), [png](#png) etc. to set the output format.
 
 If no explicit format is set, the output format will match the input image, except SVG input which becomes PNG output.
 
 By default all metadata will be removed, which includes EXIF-based orientation.
-See [withMetadata](#withMetadata) for control over this.
+See [withMetadata](#withmetadata) for control over this.
 
 `callback`, if present, gets three arguments `(err, data, info)` where:
 - `err` is an error, if any.
@@ -140,12 +140,18 @@ sharp('input.jpg')
 ```
 **Example**  
 ```js
-// Set "IFD0-Copyright" in output EXIF metadata
+// Set output EXIF metadata
 const data = await sharp(input)
   .withMetadata({
     exif: {
       IFD0: {
-        Copyright: 'Wernham Hogg'
+        Copyright: 'The National Gallery'
+      },
+      IFD3: {
+        GPSLatitudeRef: 'N',
+        GPSLatitude: '51/1 30/1 3230/100',
+        GPSLongitudeRef: 'W',
+        GPSLongitude: '0/1 7/1 4366/100'
       }
     }
   })
@@ -368,12 +374,55 @@ await sharp('in.gif', { animated: true })
   .gif({ interFrameMaxError: 8 })
   .toFile('optim.gif');
 ```
+<a name="jp2"></a>
+
+## jp
+Use these JP2 options for output image.
+
+Requires libvips compiled with support for OpenJPEG.
+The prebuilt binaries do not include this - see
+[installing a custom libvips](https://sharp.pixelplumbing.com/install#custom-libvips).
+
+
+**Throws**:
+
+- <code>Error</code> Invalid options
+
+**Since**: 0.29.1  
+
+| Param | Type | Default | Description |
+| --- | --- | --- | --- |
+| [options] | <code>Object</code> |  | output options |
+| [options.quality] | <code>number</code> | <code>80</code> | quality, integer 1-100 |
+| [options.lossless] | <code>boolean</code> | <code>false</code> | use lossless compression mode |
+| [options.tileWidth] | <code>number</code> | <code>512</code> | horizontal tile size |
+| [options.tileHeight] | <code>number</code> | <code>512</code> | vertical tile size |
+| [options.chromaSubsampling] | <code>string</code> | <code>&quot;&#x27;4:4:4&#x27;&quot;</code> | set to '4:2:0' to use chroma subsampling |
+
+**Example**  
+```js
+// Convert any input to lossless JP2 output
+const data = await sharp(input)
+  .jp2({ lossless: true })
+  .toBuffer();
+```
+**Example**  
+```js
+// Convert any input to very high quality JP2 output
+const data = await sharp(input)
+  .jp2({
+    quality: 100,
+    chromaSubsampling: '4:4:4'
+  })
+  .toBuffer();
+```
 
 
 ## tiff
 Use these TIFF options for output image.
 
-The `density` can be set in pixels/inch via [withMetadata](#withMetadata) instead of providing `xres` and `yres` in pixels/mm.
+The `density` can be set in pixels/inch via [withMetadata](#withmetadata)
+instead of providing `xres` and `yres` in pixels/mm.
 
 
 **Throws**:

docs/changelog.md

@@ -4,6 +4,34 @@
 
 Requires libvips v8.14.2
 
+### v0.32.1 - 27th April 2023
+
+* Add experimental `unflatten` operation.
+  [#3461](https://github.com/lovell/sharp/pull/3461)
+  [@antonmarsden](https://github.com/antonmarsden)
+
+* Ensure use of `flip` operation forces random access read (regression in 0.32.0).
+  [#3600](https://github.com/lovell/sharp/issues/3600)
+
+* Ensure `linear` operation works with 16-bit input (regression in 0.31.3).
+  [#3605](https://github.com/lovell/sharp/issues/3605)
+
+* Install: ensure proxy URLs are logged correctly.
+  [#3615](https://github.com/lovell/sharp/pull/3615)
+  [@TomWis97](https://github.com/TomWis97)
+
+* Ensure profile-less CMYK to CMYK roundtrip skips colourspace conversion.
+  [#3620](https://github.com/lovell/sharp/issues/3620)
+
+* Add support for `modulate` operation when using non-sRGB pipeline colourspace.
+  [#3620](https://github.com/lovell/sharp/issues/3620)
+
+* Ensure `trim` operation works with CMYK images (regression in 0.31.0).
+  [#3636](https://github.com/lovell/sharp/issues/3636)
+
+* Install: coerce libc version to semver.
+  [#3641](https://github.com/lovell/sharp/issues/3641)
+
 ### v0.32.0 - 24th March 2023
 
 * Default to using sequential rather than random access read where possible.

docs/performance.md

@@ -98,7 +98,7 @@ Note: jimp does not support premultiply/unpremultiply.
 | jimp               | buffer | buffer |    7.62 |     19.1 |
 | imagemagick        | file   | file   |    7.96 |     19.9 |
 | sharp              | file   | file   |   12.97 |     32.4 |
-| sharp              | buffer | buffer |   13.12 |     45.0 |
+| sharp              | buffer | buffer |   13.12 |     32.8 |
 
 ## Running the benchmark test
 

install/libvips.js

@@ -11,6 +11,7 @@ const zlib = require('zlib');
 const { createHash } = require('crypto');
 
 const detectLibc = require('detect-libc');
+const semverCoerce = require('semver/functions/coerce');
 const semverLessThan = require('semver/functions/lt');
 const semverSatisfies = require('semver/functions/satisfies');
 const simpleGet = require('simple-get');
@@ -77,7 +78,11 @@ const verifyIntegrity = function (platformAndArch) {
     flush: function (done) {
       const digest = `sha512-${hash.digest('base64')}`;
       if (expected !== digest) {
-        libvips.removeVendoredLibvips();
+        try {
+          libvips.removeVendoredLibvips();
+        } catch (err) {
+          libvips.log(err.message);
+        }
         libvips.log(`Integrity expected: ${expected}`);
         libvips.log(`Integrity received: ${digest}`);
         done(new Error(`Integrity check failed for ${platformAndArch}`));
@@ -135,17 +140,19 @@ try {
       throw new Error(`BSD/SunOS systems require manual installation of libvips >= ${minimumLibvipsVersion}`);
     }
     // Linux libc version check
-    const libcFamily = detectLibc.familySync();
-    const libcVersion = detectLibc.versionSync();
-    if (libcFamily === detectLibc.GLIBC && libcVersion && minimumGlibcVersionByArch[arch]) {
-      const libcVersionWithoutPatch = libcVersion.split('.').slice(0, 2).join('.');
-      if (semverLessThan(`${libcVersionWithoutPatch}.0`, `${minimumGlibcVersionByArch[arch]}.0`)) {
-        handleError(new Error(`Use with glibc ${libcVersion} requires manual installation of libvips >= ${minimumLibvipsVersion}`));
+    const libcVersionRaw = detectLibc.versionSync();
+    if (libcVersionRaw) {
+      const libcFamily = detectLibc.familySync();
+      const libcVersion = semverCoerce(libcVersionRaw).version;
+      if (libcFamily === detectLibc.GLIBC && minimumGlibcVersionByArch[arch]) {
+        if (semverLessThan(libcVersion, semverCoerce(minimumGlibcVersionByArch[arch]).version)) {
+          handleError(new Error(`Use with glibc ${libcVersionRaw} requires manual installation of libvips >= ${minimumLibvipsVersion}`));
+        }
       }
-    }
-    if (libcFamily === detectLibc.MUSL && libcVersion) {
-      if (semverLessThan(libcVersion, '1.1.24')) {
-        handleError(new Error(`Use with musl ${libcVersion} requires manual installation of libvips >= ${minimumLibvipsVersion}`));
+      if (libcFamily === detectLibc.MUSL) {
+        if (semverLessThan(libcVersion, '1.1.24')) {
+          handleError(new Error(`Use with musl ${libcVersionRaw} requires manual installation of libvips >= ${minimumLibvipsVersion}`));
+        }
       }
     }
     // Node.js minimum version check

lib/agent.js

@@ -30,7 +30,7 @@ module.exports = function (log) {
     const proxyAuth = proxy.username && proxy.password
       ? `${decodeURIComponent(proxy.username)}:${decodeURIComponent(proxy.password)}`
       : null;
-    log(`Via proxy ${proxy.protocol}://${proxy.hostname}:${proxy.port} ${proxyAuth ? 'with' : 'no'} credentials`);
+    log(`Via proxy ${proxy.protocol}//${proxy.hostname}:${proxy.port} ${proxyAuth ? 'with' : 'no'} credentials`);
     return tunnel({
       proxy: {
         port: Number(proxy.port),

lib/colour.js

@@ -70,7 +70,8 @@ function grayscale (grayscale) {
  * Set the pipeline colourspace.
  *
  * The input image will be converted to the provided colourspace at the start of the pipeline.
- * All operations will use this colourspace before converting to the output colourspace, as defined by {@link toColourspace}.
+ * All operations will use this colourspace before converting to the output colourspace,
+ * as defined by {@link #tocolourspace|toColourspace}.
  *
  * This feature is experimental and has not yet been fully-tested with all operations.
  *

lib/constructor.js

@@ -154,9 +154,9 @@ const debuglog = util.debuglog('sharp');
  * @param {string} [options.text.text] - text to render as a UTF-8 string. It can contain Pango markup, for example `<i>Le</i>Monde`.
  * @param {string} [options.text.font] - font name to render with.
  * @param {string} [options.text.fontfile] - absolute filesystem path to a font file that can be used by `font`.
- * @param {number} [options.text.width=0] - integral number of pixels to word-wrap at. Lines of text wider than this will be broken at word boundaries.
- * @param {number} [options.text.height=0] - integral number of pixels high. When defined, `dpi` will be ignored and the text will automatically fit the pixel resolution defined by `width` and `height`. Will be ignored if `width` is not specified or set to 0.
- * @param {string} [options.text.align='left'] - text alignment (`'left'`, `'centre'`, `'center'`, `'right'`).
+ * @param {number} [options.text.width=0] - Integral number of pixels to word-wrap at. Lines of text wider than this will be broken at word boundaries.
+ * @param {number} [options.text.height=0] - Maximum integral number of pixels high. When defined, `dpi` will be ignored and the text will automatically fit the pixel resolution defined by `width` and `height`. Will be ignored if `width` is not specified or set to 0.
+ * @param {string} [options.text.align='left'] - Alignment style for multi-line text (`'left'`, `'centre'`, `'center'`, `'right'`).
  * @param {boolean} [options.text.justify=false] - set this to true to apply justification to the text.
  * @param {number} [options.text.dpi=72] - the resolution (size) at which to render the text. Does not take effect if `height` is specified.
  * @param {boolean} [options.text.rgba=false] - set this to true to enable RGBA output. This is useful for colour emoji rendering, or support for pango markup features like `<span foreground="red">Red!</span>`.
@@ -217,6 +217,7 @@ const Sharp = function (input, options) {
     tintB: 128,
     flatten: false,
     flattenBackground: [0, 0, 0],
+    unflatten: false,
     negate: false,
     negateAlpha: true,
     medianSize: 0,

lib/index.d.ts

@@ -356,7 +356,7 @@ declare namespace sharp {
          * Perform an affine transform on an image. This operation will always occur after resizing, extraction and rotation, if any.
          * You must provide an array of length 4 or a 2x2 affine transformation matrix.
          * By default, new pixels are filled with a black background. You can provide a background color with the `background` option.
-         * A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolator` Object e.g. `sharp.interpolator.nohalo`.
+         * A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolators` Object e.g. `sharp.interpolators.nohalo`.
          *
          * In the case of a 2x2 matrix, the transform is:
          * X = matrix[0, 0] * (x + idx) + matrix[0, 1] * (y + idy) + odx
@@ -427,6 +427,13 @@ declare namespace sharp {
          */
         flatten(flatten?: boolean | FlattenOptions): Sharp;
 
+        /**
+         * Ensure the image has an alpha channel with all white pixel values made fully transparent.
+         * Existing alpha channel values for non-white pixels remain unchanged.
+         * @returns A sharp instance that can be used to chain operations
+         */
+        unflatten(): Sharp;
+
         /**
          * Apply a gamma correction by reducing the encoding (darken) pre-resize at a factor of 1/gamma then increasing the encoding (brighten) post-resize at a factor of gamma.
          * This can improve the perceived brightness of a resized image in non-linear colour spaces.

lib/libvips.js

@@ -88,8 +88,7 @@ const hasVendoredLibvips = function () {
 
 /* istanbul ignore next */
 const removeVendoredLibvips = function () {
-  const rm = fs.rmSync ? fs.rmSync : fs.rmdirSync;
-  rm(vendorPath, { recursive: true, maxRetries: 3, force: true });
+  fs.rmSync(vendorPath, { recursive: true, maxRetries: 3, force: true });
 };
 
 /* istanbul ignore next */

lib/operation.js

@@ -114,7 +114,7 @@ function flop (flop) {
  *
  * You must provide an array of length 4 or a 2x2 affine transformation matrix.
  * By default, new pixels are filled with a black background. You can provide a background color with the `background` option.
- * A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolator` Object e.g. `sharp.interpolator.nohalo`.
+ * A particular interpolator may also be specified. Set the `interpolator` option to an attribute of the `sharp.interpolators` Object e.g. `sharp.interpolators.nohalo`.
  *
  * In the case of a 2x2 matrix, the transform is:
  * - X = `matrix[0, 0]` \* (x + `idx`) + `matrix[0, 1]` \* (y + `idy`) + `odx`
@@ -131,7 +131,7 @@ function flop (flop) {
  * const pipeline = sharp()
  *   .affine([[1, 0.3], [0.1, 0.7]], {
  *      background: 'white',
- *      interpolate: sharp.interpolators.nohalo
+ *      interpolator: sharp.interpolators.nohalo
  *   })
  *   .toBuffer((err, outputBuffer, info) => {
  *      // outputBuffer contains the transformed image
@@ -405,6 +405,32 @@ function flatten (options) {
   return this;
 }
 
+/**
+ * Ensure the image has an alpha channel
+ * with all white pixel values made fully transparent.
+ *
+ * Existing alpha channel values for non-white pixels remain unchanged.
+ *
+ * This feature is experimental and the API may change.
+ *
+ * @since 0.32.1
+ *
+ * @example
+ * await sharp(rgbInput)
+ *   .unflatten()
+ *   .toBuffer();
+ *
+ * @example
+ * await sharp(rgbInput)
+ *   .threshold(128, { grayscale: false }) // converter bright pixels to white
+ *   .unflatten()
+ *   .toBuffer();
+ */
+function unflatten () {
+  this.options.unflatten = true;
+  return this;
+}
+
 /**
  * Apply a gamma correction by reducing the encoding (darken) pre-resize at a factor of `1/gamma`
  * then increasing the encoding (brighten) post-resize at a factor of `gamma`.
@@ -875,6 +901,7 @@ module.exports = function (Sharp) {
     median,
     blur,
     flatten,
+    unflatten,
     gamma,
     negate,
     normalise,

lib/output.js

@@ -43,7 +43,7 @@ const bitdepthFromColourCount = (colours) => 1 << 31 - Math.clz32(Math.ceil(Math
  * Note that raw pixel data is only supported for buffer output.
  *
  * By default all metadata will be removed, which includes EXIF-based orientation.
- * See {@link withMetadata} for control over this.
+ * See {@link #withmetadata|withMetadata} for control over this.
  *
  * The caller is responsible for ensuring directory structures and permissions exist.
  *
@@ -95,12 +95,12 @@ function toFile (fileOut, callback) {
  * Write output to a Buffer.
  * JPEG, PNG, WebP, AVIF, TIFF, GIF and raw pixel data output are supported.
  *
- * Use {@link toFormat} or one of the format-specific functions such as {@link jpeg}, {@link png} etc. to set the output format.
+ * Use {@link #toformat|toFormat} or one of the format-specific functions such as {@link jpeg}, {@link png} etc. to set the output format.
  *
  * If no explicit format is set, the output format will match the input image, except SVG input which becomes PNG output.
  *
  * By default all metadata will be removed, which includes EXIF-based orientation.
- * See {@link withMetadata} for control over this.
+ * See {@link #withmetadata|withMetadata} for control over this.
  *
  * `callback`, if present, gets three arguments `(err, data, info)` where:
  * - `err` is an error, if any.
@@ -177,12 +177,18 @@ function toBuffer (options, callback) {
  *   .then(info => { ... });
  *
  * @example
- * // Set "IFD0-Copyright" in output EXIF metadata
+ * // Set output EXIF metadata
  * const data = await sharp(input)
  *   .withMetadata({
  *     exif: {
  *       IFD0: {
- *         Copyright: 'Wernham Hogg'
+ *         Copyright: 'The National Gallery'
+ *       },
+ *       IFD3: {
+ *         GPSLatitudeRef: 'N',
+ *         GPSLatitude: '51/1 30/1 3230/100',
+ *         GPSLongitudeRef: 'W',
+ *         GPSLongitude: '0/1 7/1 4366/100'
  *       }
  *     }
  *   })
@@ -626,6 +632,7 @@ function gif (options) {
   return this._updateFormatOut('gif', options);
 }
 
+/* istanbul ignore next */
 /**
  * Use these JP2 options for output image.
  *
@@ -659,7 +666,6 @@ function gif (options) {
  * @returns {Sharp}
  * @throws {Error} Invalid options
  */
-/* istanbul ignore next */
 function jp2 (options) {
   if (!this.constructor.format.jp2k.output.buffer) {
     throw errJp2Save();
@@ -740,7 +746,8 @@ function trySetAnimationOptions (source, target) {
 /**
  * Use these TIFF options for output image.
  *
- * The `density` can be set in pixels/inch via {@link withMetadata} instead of providing `xres` and `yres` in pixels/mm.
+ * The `density` can be set in pixels/inch via {@link #withmetadata|withMetadata}
+ * instead of providing `xres` and `yres` in pixels/mm.
  *
  * @example
  * // Convert SVG input to LZW-compressed, 1 bit per pixel TIFF output

package.json

@@ -1,7 +1,7 @@
 {
   "name": "sharp",
   "description": "High performance Node.js image processing, the fastest module to resize JPEG, PNG, WebP, GIF, AVIF and TIFF images",
-  "version": "0.32.0",
+  "version": "0.32.1",
   "author": "Lovell Fuller <npm@lovell.info>",
   "homepage": "https://github.com/lovell/sharp",
   "contributors": [
@@ -134,9 +134,9 @@
   "dependencies": {
     "color": "^4.2.3",
     "detect-libc": "^2.0.1",
-    "node-addon-api": "^6.0.0",
+    "node-addon-api": "^6.1.0",
     "prebuild-install": "^7.1.1",
-    "semver": "^7.3.8",
+    "semver": "^7.5.0",
     "simple-get": "^4.0.1",
     "tar-fs": "^2.1.1",
     "tunnel-agent": "^0.6.0"
@@ -147,7 +147,7 @@
     "cc": "^3.0.1",
     "exif-reader": "^1.2.0",
     "extract-zip": "^2.0.1",
-    "icc": "^2.0.0",
+    "icc": "^3.0.0",
     "jsdoc-to-markdown": "^8.0.0",
     "license-checker": "^25.0.1",
     "mocha": "^10.2.0",
@@ -155,7 +155,7 @@
     "nyc": "^15.1.0",
     "prebuild": "^11.0.4",
     "semistandard": "^16.0.1",
-    "tsd": "^0.28.0"
+    "tsd": "^0.28.1"
   },
   "license": "Apache-2.0",
   "config": {

src/common.cc

@@ -65,16 +65,6 @@ namespace sharp {
     }
     return vector;
   }
-  Napi::Buffer<char> NewOrCopyBuffer(Napi::Env env, char* data, size_t len) {
-    try {
-      return Napi::Buffer<char>::New(env, data, len, FreeCallback);
-    } catch (Napi::Error const &err) {
-      static_cast<void>(err);
-    }
-    Napi::Buffer<char> buf = Napi::Buffer<char>::Copy(env, data, len);
-    FreeCallback(nullptr, data);
-    return buf;
-  }
 
   // Create an InputDescriptor instance from a Napi::Object describing an input image
   InputDescriptor* CreateInputDescriptor(Napi::Object input) {

src/common.h

@@ -126,7 +126,6 @@ namespace sharp {
     return static_cast<T>(
       vips_enum_from_nick(nullptr, type, AttrAsStr(obj, attr).data()));
   }
-  Napi::Buffer<char> NewOrCopyBuffer(Napi::Env env, char* data, size_t len);
 
   // Create an InputDescriptor instance from a Napi::Object describing an input image
   InputDescriptor* CreateInputDescriptor(Napi::Object input);

src/metadata.cc

@@ -230,20 +230,21 @@ class MetadataWorker : public Napi::AsyncWorker {
         info.Set("orientation", baton->orientation);
       }
       if (baton->exifLength > 0) {
-        info.Set("exif", sharp::NewOrCopyBuffer(env, baton->exif, baton->exifLength));
+        info.Set("exif", Napi::Buffer<char>::NewOrCopy(env, baton->exif, baton->exifLength, sharp::FreeCallback));
       }
       if (baton->iccLength > 0) {
-        info.Set("icc", sharp::NewOrCopyBuffer(env, baton->icc, baton->iccLength));
+        info.Set("icc", Napi::Buffer<char>::NewOrCopy(env, baton->icc, baton->iccLength, sharp::FreeCallback));
       }
       if (baton->iptcLength > 0) {
-        info.Set("iptc", sharp::NewOrCopyBuffer(env, baton->iptc, baton->iptcLength));
+        info.Set("iptc", Napi::Buffer<char>::NewOrCopy(env, baton->iptc, baton->iptcLength, sharp::FreeCallback));
       }
       if (baton->xmpLength > 0) {
-        info.Set("xmp", sharp::NewOrCopyBuffer(env, baton->xmp, baton->xmpLength));
+        info.Set("xmp", Napi::Buffer<char>::NewOrCopy(env, baton->xmp, baton->xmpLength, sharp::FreeCallback));
       }
       if (baton->tifftagPhotoshopLength > 0) {
         info.Set("tifftagPhotoshop",
-          sharp::NewOrCopyBuffer(env, baton->tifftagPhotoshop, baton->tifftagPhotoshopLength));
+          Napi::Buffer<char>::NewOrCopy(env, baton->tifftagPhotoshop,
+            baton->tifftagPhotoshopLength, sharp::FreeCallback));
       }
       Callback().MakeCallback(Receiver().Value(), { env.Null(), info });
     } else {

src/operations.cc

@@ -186,6 +186,7 @@ namespace sharp {
 
   VImage Modulate(VImage image, double const brightness, double const saturation,
                   int const hue, double const lightness) {
+    VipsInterpretation colourspaceBeforeModulate = image.interpretation();
     if (HasAlpha(image)) {
       // Separate alpha channel
       VImage alpha = image[image.bands() - 1];
@@ -195,7 +196,7 @@ namespace sharp {
           { brightness, saturation, 1},
           { lightness, 0.0, static_cast<double>(hue) }
         )
-        .colourspace(VIPS_INTERPRETATION_sRGB)
+        .colourspace(colourspaceBeforeModulate)
         .bandjoin(alpha);
     } else {
       return image
@@ -204,7 +205,7 @@ namespace sharp {
           { brightness, saturation, 1 },
           { lightness, 0.0, static_cast<double>(hue) }
         )
-        .colourspace(VIPS_INTERPRETATION_sRGB);
+        .colourspace(colourspaceBeforeModulate);
     }
   }
 
@@ -268,30 +269,20 @@ namespace sharp {
     if (image.width() < 3 && image.height() < 3) {
       throw VError("Image to trim must be at least 3x3 pixels");
     }
-
-    // Scale up 8-bit values to match 16-bit input image
-    double multiplier = sharp::Is16Bit(image.interpretation()) ? 256.0 : 1.0;
-    threshold *= multiplier;
-
-    std::vector<double> backgroundAlpha(1);
     if (background.size() == 0) {
       // Top-left pixel provides the default background colour if none is given
       background = image.extract_area(0, 0, 1, 1)(0, 0);
-      multiplier = 1.0;
+    } else if (sharp::Is16Bit(image.interpretation())) {
+      for (size_t i = 0; i < background.size(); i++) {
+        background[i] *= 256.0;
+      }
+      threshold *= 256.0;
     }
-    if (HasAlpha(image) && background.size() == 4) {
-      // Just discard the alpha because flattening the background colour with
-      // itself (effectively what find_trim() does) gives the same result
-      backgroundAlpha[0] = background[3] * multiplier;
-    }
-    if (image.bands() > 2) {
-      background = {
-        background[0] * multiplier,
-        background[1] * multiplier,
-        background[2] * multiplier
-      };
+    std::vector<double> backgroundAlpha({ background.back() });
+    if (HasAlpha(image)) {
+      background.pop_back();
     } else {
-      background[0] = background[0] * multiplier;
+      background.resize(image.bands());
     }
     int left, top, width, height;
     left = image.find_trim(&top, &width, &height, VImage::option()
@@ -332,12 +323,26 @@ namespace sharp {
     if (a.size() > bands) {
       throw VError("Band expansion using linear is unsupported");
     }
+    bool const uchar = !Is16Bit(image.interpretation());
     if (HasAlpha(image) && a.size() != bands && (a.size() == 1 || a.size() == bands - 1 || bands - 1 == 1)) {
       // Separate alpha channel
       VImage alpha = image[bands - 1];
-      return RemoveAlpha(image).linear(a, b, VImage::option()->set("uchar", TRUE)).bandjoin(alpha);
+      return RemoveAlpha(image).linear(a, b, VImage::option()->set("uchar", uchar)).bandjoin(alpha);
     } else {
-      return image.linear(a, b, VImage::option()->set("uchar", TRUE));
+      return image.linear(a, b, VImage::option()->set("uchar", uchar));
+    }
+  }
+
+  /*
+   * Unflatten
+   */
+  VImage Unflatten(VImage image) {
+    if (HasAlpha(image)) {
+      VImage alpha = image[image.bands() - 1];
+      VImage noAlpha = RemoveAlpha(image);
+      return noAlpha.bandjoin(alpha & (noAlpha.colourspace(VIPS_INTERPRETATION_B_W) < 255));
+    } else {
+      return image.bandjoin(image.colourspace(VIPS_INTERPRETATION_B_W) < 255);
     }
   }
 

src/operations.h

@@ -86,6 +86,11 @@ namespace sharp {
    */
   VImage Linear(VImage image, std::vector<double> const a,  std::vector<double> const b);
 
+  /*
+   * Unflatten
+   */
+  VImage Unflatten(VImage image);
+
   /*
    * Recomb with a Matrix of the given bands/channel size.
    * Eg. RGB will be a 3x3 matrix.

src/pipeline.cc

@@ -322,7 +322,10 @@ class PipelineWorker : public Napi::AsyncWorker {
         } catch(...) {
           sharp::VipsWarningCallback(nullptr, G_LOG_LEVEL_WARNING, "Invalid embedded profile", nullptr);
         }
-      } else if (image.interpretation() == VIPS_INTERPRETATION_CMYK) {
+      } else if (
+        image.interpretation() == VIPS_INTERPRETATION_CMYK &&
+        baton->colourspaceInput != VIPS_INTERPRETATION_CMYK
+      ) {
         image = image.icc_transform(processingProfile, VImage::option()
           ->set("input_profile", "cmyk")
           ->set("intent", VIPS_INTENT_PERCEPTUAL));
@@ -470,6 +473,7 @@ class PipelineWorker : public Napi::AsyncWorker {
 
             image = image.smartcrop(baton->width, baton->height, VImage::option()
               ->set("interesting", baton->position == 16 ? VIPS_INTERESTING_ENTROPY : VIPS_INTERESTING_ATTENTION)
+              ->set("premultiplied", shouldPremultiplyAlpha)
               ->set("attention_x", &attention_x)
               ->set("attention_y", &attention_y));
             baton->hasCropOffset = true;
@@ -550,7 +554,9 @@ class PipelineWorker : public Napi::AsyncWorker {
       if (baton->medianSize > 0) {
         image = image.median(baton->medianSize);
       }
+
       // Threshold - must happen before blurring, due to the utility of blurring after thresholding
+      // Threshold - must happen before unflatten to enable non-white unflattening
       if (baton->threshold != 0) {
         image = sharp::Threshold(image, baton->threshold, baton->thresholdGrayscale);
       }
@@ -560,6 +566,11 @@ class PipelineWorker : public Napi::AsyncWorker {
         image = sharp::Blur(image, baton->blurSigma);
       }
 
+      // Unflatten the image
+      if (baton->unflatten) {
+        image = sharp::Unflatten(image);
+      }
+
       // Convolve
       if (shouldConv) {
         image = sharp::Convolve(image,
@@ -1222,8 +1233,8 @@ class PipelineWorker : public Napi::AsyncWorker {
         // Add buffer size to info
         info.Set("size", static_cast<uint32_t>(baton->bufferOutLength));
         // Pass ownership of output data to Buffer instance
-        Napi::Buffer<char> data = sharp::NewOrCopyBuffer(env, static_cast<char*>(baton->bufferOut),
-          baton->bufferOutLength);
+        Napi::Buffer<char> data = Napi::Buffer<char>::NewOrCopy(env, static_cast<char*>(baton->bufferOut),
+          baton->bufferOutLength, sharp::FreeCallback);
         Callback().MakeCallback(Receiver().Value(), { env.Null(), data, info });
       } else {
         // Add file size to info
@@ -1460,6 +1471,7 @@ Napi::Value pipeline(const Napi::CallbackInfo& info) {
   // Operators
   baton->flatten = sharp::AttrAsBool(options, "flatten");
   baton->flattenBackground = sharp::AttrAsVectorOfDouble(options, "flattenBackground");
+  baton->unflatten = sharp::AttrAsBool(options, "unflatten");
   baton->negate = sharp::AttrAsBool(options, "negate");
   baton->negateAlpha = sharp::AttrAsBool(options, "negateAlpha");
   baton->blurSigma = sharp::AttrAsDouble(options, "blurSigma");
@@ -1662,6 +1674,7 @@ Napi::Value pipeline(const Napi::CallbackInfo& info) {
       baton->rotationAngle != 0.0 ||
       baton->tileAngle != 0 ||
       baton->useExifOrientation ||
+      baton->flip ||
       baton->claheWidth != 0 ||
       !baton->affineMatrix.empty()
     ) {

src/pipeline.h

@@ -73,6 +73,7 @@ struct PipelineBaton {
   double tintB;
   bool flatten;
   std::vector<double> flattenBackground;
+  bool unflatten;
   bool negate;
   bool negateAlpha;
   double blurSigma;
@@ -239,6 +240,7 @@ struct PipelineBaton {
     tintB(128.0),
     flatten(false),
     flattenBackground{ 0.0, 0.0, 0.0 },
+    unflatten(false),
     negate(false),
     negateAlpha(true),
     blurSigma(0.0),

test/unit/agent.js

@@ -21,7 +21,7 @@ describe('HTTP agent', function () {
     assert.strictEqual(123, proxy.options.proxy.port);
     assert.strictEqual('user:pass', proxy.options.proxy.proxyAuth);
     assert.strictEqual(443, proxy.defaultPort);
-    assert.strictEqual(logMsg, 'Via proxy https:://secure:123 with credentials');
+    assert.strictEqual(logMsg, 'Via proxy https://secure:123 with credentials');
   });
 
   it('HTTPS proxy with auth from HTTPS_PROXY using credentials containing special characters', function () {
@@ -34,7 +34,7 @@ describe('HTTP agent', function () {
     assert.strictEqual(789, proxy.options.proxy.port);
     assert.strictEqual('user,:pass=', proxy.options.proxy.proxyAuth);
     assert.strictEqual(443, proxy.defaultPort);
-    assert.strictEqual(logMsg, 'Via proxy https:://secure:789 with credentials');
+    assert.strictEqual(logMsg, 'Via proxy https://secure:789 with credentials');
   });
 
   it('HTTP proxy without auth from npm_config_proxy', function () {
@@ -47,6 +47,6 @@ describe('HTTP agent', function () {
     assert.strictEqual(456, proxy.options.proxy.port);
     assert.strictEqual(null, proxy.options.proxy.proxyAuth);
     assert.strictEqual(443, proxy.defaultPort);
-    assert.strictEqual(logMsg, 'Via proxy http:://plaintext:456 no credentials');
+    assert.strictEqual(logMsg, 'Via proxy http://plaintext:456 no credentials');
   });
 });

test/unit/colourspace.js

@@ -93,6 +93,19 @@ describe('Colour space conversion', function () {
       });
   });
 
+  it('Profile-less CMYK roundtrip', async () => {
+    const [c, m, y, k] = await sharp(fixtures.inputJpgWithCmykNoProfile)
+      .pipelineColourspace('cmyk')
+      .toColourspace('cmyk')
+      .raw()
+      .toBuffer();
+
+    assert.deepStrictEqual(
+      { c, m, y, k },
+      { c: 55, m: 27, y: 0, k: 0 }
+    );
+  });
+
   it('From sRGB with RGB16 pipeline, resize with gamma, to sRGB', function (done) {
     sharp(fixtures.inputPngGradients)
       .pipelineColourspace('rgb16')

test/unit/linear.js

@@ -51,6 +51,16 @@ describe('Linear adjustment', function () {
       });
   });
 
+  it('applies linear levels adjustment to 16-bit w alpha ch', function (done) {
+    sharp(fixtures.inputPngWithTransparency16bit)
+      .linear(a, b)
+      .png({ compressionLevel: 0 })
+      .toBuffer(function (err, data) {
+        if (err) throw err;
+        fixtures.assertSimilar(fixtures.expected('linear-16bit.png'), data, done);
+      });
+  });
+
   it('applies slope level adjustment w alpha ch', function (done) {
     sharp(fixtures.inputPngOverlayLayer1)
       .resize(240)

test/unit/text.js

@@ -8,7 +8,9 @@ const assert = require('assert');
 const sharp = require('../../');
 const fixtures = require('../fixtures');
 
-describe('Text to image', () => {
+describe('Text to image', function () {
+  this.retries(3);
+
   it('text with default values', async () => {
     const output = fixtures.path('output.text-default.png');
     const text = sharp({

test/unit/trim.js

@@ -153,6 +153,32 @@ describe('Trim borders', function () {
     assert.strictEqual(trimOffsetLeft, -12);
   });
 
+  it('Ensure CMYK image can be trimmed', async () => {
+    const cmyk = await sharp({
+      create: {
+        width: 16,
+        height: 8,
+        channels: 3,
+        background: 'red'
+      }
+    })
+      .extend({ left: 12, right: 24, background: 'blue' })
+      .toColourspace('cmyk')
+      .jpeg()
+      .toBuffer();
+
+    const { info } = await sharp(cmyk)
+      .trim()
+      .raw()
+      .toBuffer({ resolveWithObject: true });
+
+    const { width, height, trimOffsetTop, trimOffsetLeft } = info;
+    assert.strictEqual(width, 16);
+    assert.strictEqual(height, 8);
+    assert.strictEqual(trimOffsetTop, 0);
+    assert.strictEqual(trimOffsetLeft, -12);
+  });
+
   it('Ensure trim of image with all pixels same is no-op', async () => {
     const { info } = await sharp({
       create: {

test/unit/unflatten.js

@@ -0,0 +1,31 @@
+// Copyright 2013 Lovell Fuller and others.
+// SPDX-License-Identifier: Apache-2.0
+
+'use strict';
+
+const sharp = require('../../');
+const fixtures = require('../fixtures');
+
+describe('Unflatten', function () {
+  it('unflatten white background', function (done) {
+    sharp(fixtures.inputPng).unflatten()
+      .toBuffer(function (err, data) {
+        if (err) throw err;
+        fixtures.assertSimilar(fixtures.expected('unflatten-white-transparent.png'), data, { threshold: 0 }, done);
+      });
+  });
+  it('unflatten transparent image', function (done) {
+    sharp(fixtures.inputPngTrimSpecificColourIncludeAlpha).unflatten()
+      .toBuffer(function (err, data) {
+        if (err) throw err;
+        fixtures.assertSimilar(fixtures.expected('unflatten-flag-white-transparent.png'), data, { threshold: 0 }, done);
+      });
+  });
+  it('unflatten using threshold', function (done) {
+    sharp(fixtures.inputPngPalette).unflatten().threshold(128, { grayscale: false })
+      .toBuffer(function (err, data) {
+        if (err) throw err;
+        fixtures.assertSimilar(fixtures.expected('unflatten-swiss.png'), data, { threshold: 1 }, done);
+      });
+  });
+});