Merge remote-tracking branch 'origin/master' into move_to_libzim7

Changelog

@@ -1,3 +1,10 @@
+1.11.12:
+* FIX: ZIM illustration now always in PNG format
+* FIX: Somehow cropped ZIM illustration
+* NEW: Add support for mw:File typeof
+* NEW: Use keep-alive TCP flag
+* UPDATE: Redis 8 & Node 18 in Docker image
+
 1.11.11:
 * FIX: Workaround wrong Wep mime-type returns sporadically by Mediawiki
 * UPDATE: Better speed computation

package.json

@@ -1,6 +1,6 @@
 {
   "name": "mwoffliner",
-  "version": "1.11.11",
+  "version": "1.11.12",
   "description": "Mediawiki ZIM scraper",
   "main": "lib/mwoffliner.lib.js",
   "license": "GPL-3.0",

res/content.parsoid.css

@@ -51,18 +51,21 @@ sub {
  * Block media items
  */
 
+figure[typeof*='mw:File'],
 figure[typeof*='mw:Image'],
 figure[typeof*='mw:Video'],
 figure[typeof*='mw:Audio'] {
     margin: 0;
 }
 
+figure[typeof*='mw:File'] a,
 figure[typeof*='mw:Image'] a,
 figure[typeof*='mw:Video'] a,
 figure[typeof*='mw:Audio'] a {
     border: 0;
 }
 
+figure[typeof*='mw:File'].mw-halign-right,
 figure[typeof*='mw:Image'].mw-halign-right,
 figure[typeof*='mw:Video'].mw-halign-right,
 figure[typeof*='mw:Audio'].mw-halign-right {
@@ -74,6 +77,7 @@ figure[typeof*='mw:Audio'].mw-halign-right {
     float: right;
 }
 
+figure[typeof*='mw:File'].mw-halign-left,
 figure[typeof*='mw:Image'].mw-halign-left,
 figure[typeof*='mw:Video'].mw-halign-left,
 figure[typeof*='mw:Audio'].mw-halign-left {
@@ -85,6 +89,7 @@ figure[typeof*='mw:Audio'].mw-halign-left {
     float: left;
 }
 
+figure[typeof*='mw:File'].mw-halign-none,
 figure[typeof*='mw:Image'].mw-halign-none,
 figure[typeof*='mw:Video'].mw-halign-none,
 figure[typeof*='mw:Audio'].mw-halign-none {
@@ -93,6 +98,7 @@ figure[typeof*='mw:Audio'].mw-halign-none {
     float: none;
 }
 
+figure[typeof*='mw:File'].mw-halign-center,
 figure[typeof*='mw:Image'].mw-halign-center,
 figure[typeof*='mw:Video'].mw-halign-center,
 figure[typeof*='mw:Audio'].mw-halign-center {
@@ -102,6 +108,7 @@ figure[typeof*='mw:Audio'].mw-halign-center {
     float: none;
 }
 
+figure[typeof*='mw:File']>figcaption,
 figure[typeof*='mw:Image']>figcaption,
 figure[typeof*='mw:Video']>figcaption,
 figure[typeof*='mw:Audio']>figcaption {
@@ -120,16 +127,19 @@ figure[typeof*='mw:Audio']>figcaption {
     background-color: #f8f9fa;
 }
 
+figure[typeof*='mw:File']>figcaption table,
 figure[typeof*='mw:Image']>figcaption table,
 figure[typeof*='mw:Video']>figcaption table,
 figure[typeof*='mw:Audio']>figcaption table {
     /* reset caption side for tables inside figcaptions */
     caption-side: top;
 }
 
+figure[typeof~='mw:File/Thumb'],
 figure[typeof~='mw:Image/Thumb'],
 figure[typeof~='mw:Video/Thumb'],
 figure[typeof~='mw:Audio/Thumb'],
+figure[typeof~='mw:File/Frame'],
 figure[typeof~='mw:Image/Frame'],
 figure[typeof~='mw:Video/Frame'],
 figure[typeof~='mw:Audio/Frame'] {
@@ -145,12 +155,20 @@ figure[typeof~='mw:Audio/Frame'] {
     float: right;
 }
 
+figure[typeof~='mw:File/Thumb']>*:first-child>img,
+figure[typeof~='mw:File/Thumb']>*:first-child>video,
+figure[typeof~='mw:File/Thumb']>*:first-child>audio,
 figure[typeof~='mw:Image/Thumb']>*:first-child>img,
 figure[typeof~='mw:Video/Thumb']>*:first-child>video,
 figure[typeof~='mw:Audio/Thumb']>*:first-child>video,
+figure[typeof~='mw:Audio/Thumb']>*:first-child>audio,
+figure[typeof~='mw:File/Frame']>*:first-child>img,
+figure[typeof~='mw:File/Frame']>*:first-child>video,
+figure[typeof~='mw:File/Frame']>*:first-child>audio,
 figure[typeof~='mw:Image/Frame']>*:first-child>img,
 figure[typeof~='mw:Video/Frame']>*:first-child>video,
-figure[typeof~='mw:Audio/Frame']>*:first-child>video {
+figure[typeof~='mw:Audio/Frame']>*:first-child>video,
+figure[typeof~='mw:Audio/Frame']>*:first-child>audio {
     border: 1px solid #c8ccd1;
     margin: 3px;
     background: #fff;
@@ -159,17 +177,20 @@ figure[typeof~='mw:Audio/Frame']>*:first-child>video {
 /* Same as img.thumbborder in content.css */
 
 .mw-image-border>*:first-child>img,
-.mw-image-border>*:first-child>video {
+.mw-image-border>*:first-child>video,
+.mw-image-border>*:first-child>audio {
     border: 1px solid #eaecf0;
 }
 
 /* Hide the caption for frameless and plain floated images */
 
+figure[typeof~='mw:File/Frameless']>figcaption,
 figure[typeof~='mw:Image/Frameless']>figcaption,
 figure[typeof~='mw:Video/Frameless']>figcaption,
 figure[typeof~='mw:Audio/Frameless']>figcaption,
+figure[typeof~='mw:File']>figcaption,
 figure[typeof~='mw:Image']>figcaption,
 figure[typeof~='mw:Video']>figcaption,
 figure[typeof~='mw:Audio']>figcaption {
     display: none;
-}
+}

src/mwoffliner.lib.ts

@@ -113,8 +113,7 @@ async function execute(argv: any) {
     optimisationCacheUrl,
     noLocalParserFallback,
     forceLocalParser,
-    customFlavour,
-    zstd,
+    customFlavour
   } = argv;
 
   (process as any).verbose = !!verbose;

src/parameterList.ts

@@ -36,5 +36,4 @@ export const parameterDescriptions = {
   osTmpDir: 'Override default operating system temporary directory path environment variable',
   customFlavour: 'A custom processor that can filter and process articles (see extensions/*.js)',
   optimisationCacheUrl: 'S3 url, including credentials and bucket name',
-  zstd: 'Use Zstandard as ZIM compression (Zstd by default, left for backwards compatability)',
 };

src/util/saveArticles.ts

@@ -694,7 +694,7 @@ export async function treatMedias(parsoidDoc: DominoElement, mw: MediaWiki, dump
 
     /* Improve image frames */
     const figures = parsoidDoc.getElementsByTagName('figure');
-    const spans = parsoidDoc.querySelectorAll('span[typeof=mw:Image/Frameless]');
+    const spans = parsoidDoc.querySelectorAll('span[typeof~=mw:Image/Frameless],span[typeof~=mw:File/Frameless]');
     const imageNodes = Array.prototype.slice.call(figures).concat(Array.prototype.slice.call(spans));
     for (const imageNode of imageNodes) {
         treatImageFrames(mw, dump, parsoidDoc, imageNode);