Merge pull request #1070 from lightpanda-io/dump_strip_mode

Replace --noscript with more advanced --strip_mode

Author: karlseguin

src/browser/dump.zig

@@ -26,7 +26,13 @@ pub const Opts = struct {
     // set to include element shadowroots in the dump
     page: ?*const Page = null,
 
-    exclude_scripts: bool = false,
+    strip_mode: StripMode = .{},
+
+    pub const StripMode = struct {
+        js: bool = false,
+        ui: bool = false,
+        css: bool = false,
+    };
 };
 
 // writer must be a std.io.Writer
@@ -67,7 +73,7 @@ pub fn writeNode(node: *parser.Node, opts: Opts, writer: *std.Io.Writer) anyerro
         .element => {
             // open the tag
             const tag_type = try parser.nodeHTMLGetTagType(node) orelse .undef;
-            if (opts.exclude_scripts and try isScriptOrRelated(tag_type, node)) {
+            if (try isStripped(tag_type, node, opts.strip_mode)) {
                 return;
             }
 
@@ -159,9 +165,22 @@ pub fn writeChildren(root: *parser.Node, opts: Opts, writer: *std.Io.Writer) !vo
     }
 }
 
-// When `exclude_scripts` is passed to dump, we don't include <script> tags.
-// We also want to omit <link rel=preload as=ascript>
-fn isScriptOrRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+fn isStripped(tag_type: parser.Tag, node: *parser.Node, strip_mode: Opts.StripMode) !bool {
+    if (strip_mode.js and try isJsRelated(tag_type, node)) {
+        return true;
+    }
+
+    if (strip_mode.css and try isCssRelated(tag_type, node)) {
+        return true;
+    }
+
+    if (strip_mode.ui and try isUIRelated(tag_type, node)) {
+        return true;
+    }
+    return false;
+}
+
+fn isJsRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
     if (tag_type == .script) {
         return true;
     }
@@ -178,6 +197,34 @@ fn isScriptOrRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
     return false;
 }
 
+fn isCssRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+    if (tag_type == .style) {
+        return true;
+    }
+    if (tag_type == .link) {
+        const el = parser.nodeToElement(node);
+        const rel = try parser.elementGetAttribute(el, "rel") orelse return false;
+        return std.ascii.eqlIgnoreCase(rel, "stylesheet");
+    }
+    return false;
+}
+
+fn isUIRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+    if (try isCssRelated(tag_type, node)) {
+        return true;
+    }
+    if (tag_type == .img or tag_type == .picture or tag_type == .video) {
+        return true;
+    }
+    if (tag_type == .undef) {
+        const name = try parser.nodeLocalName(node);
+        if (std.mem.eql(u8, name, "svg")) {
+            return true;
+        }
+    }
+    return false;
+}
+
 // area, base, br, col, embed, hr, img, input, link, meta, source, track, wbr
 // https://html.spec.whatwg.org/#void-elements
 fn isVoid(elem: *parser.Element) !bool {

src/browser/page.zig

@@ -181,7 +181,7 @@ pub const Page = struct {
         // set to include element shadowroots in the dump
         page: ?*const Page = null,
         with_base: bool = false,
-        exclude_scripts: bool = false,
+        strip_mode: Dump.Opts.StripMode = .{},
     };
 
     // dump writes the page content into the given file.
@@ -228,7 +228,7 @@ pub const Page = struct {
 
         try Dump.writeHTML(doc, .{
             .page = opts.page,
-            .exclude_scripts = opts.exclude_scripts,
+            .strip_mode = opts.strip_mode,
         }, out);
     }
 

src/main.zig

@@ -24,6 +24,7 @@ const log = @import("log.zig");
 const App = @import("app.zig").App;
 const Server = @import("server.zig").Server;
 const Browser = @import("browser/browser.zig").Browser;
+const DumpStripMode = @import("browser/dump.zig").Opts.StripMode;
 
 const build_config = @import("build_config");
 
@@ -184,7 +185,7 @@ fn run(alloc: Allocator) !void {
                 try page.dump(.{
                     .page = page,
                     .with_base = opts.withbase,
-                    .exclude_scripts = opts.noscript,
+                    .strip_mode = opts.strip_mode,
                 }, &writer.interface);
                 try writer.interface.flush();
             }
@@ -292,8 +293,8 @@ const Command = struct {
         url: []const u8,
         dump: bool = false,
         common: Common,
-        noscript: bool = false,
         withbase: bool = false,
+        strip_mode: DumpStripMode = .{},
     };
 
     const Common = struct {
@@ -372,7 +373,14 @@ const Command = struct {
             \\Options:
             \\--dump          Dumps document to stdout.
             \\                Defaults to false.
-            \\--noscript      Exclude <script> tags in dump. Defaults to false.
+            \\
+            \\--strip_mode    Comma separated list of tag groups to remove from dump
+            \\                the dump. e.g. --strip_mode js,css
+            \\                  - "js" script and link[as=script, rel=preload]
+            \\                  - "ui" includes img, picture, video, css and svg
+            \\                  - "css" includes style and link[rel=stylesheet]
+            \\                  - "full" includes js, ui and css
+            \\
             \\--with_base     Add a <base> tag in dump. Defaults to false.
             \\
         ++ common_options ++
@@ -460,6 +468,10 @@ fn inferMode(opt: []const u8) ?App.RunMode {
         return .fetch;
     }
 
+    if (std.mem.eql(u8, opt, "--strip_mode")) {
+        return .fetch;
+    }
+
     if (std.mem.eql(u8, opt, "--with_base")) {
         return .fetch;
     }
@@ -545,10 +557,10 @@ fn parseFetchArgs(
     args: *std.process.ArgIterator,
 ) !Command.Fetch {
     var dump: bool = false;
-    var noscript: bool = false;
     var withbase: bool = false;
     var url: ?[]const u8 = null;
     var common: Command.Common = .{};
+    var strip_mode: DumpStripMode = .{};
 
     while (args.next()) |opt| {
         if (std.mem.eql(u8, "--dump", opt)) {
@@ -557,7 +569,11 @@ fn parseFetchArgs(
         }
 
         if (std.mem.eql(u8, "--noscript", opt)) {
-            noscript = true;
+            log.warn(.app, "deprecation warning", .{
+                .feature = "--noscript argument",
+                .hint = "use '--strip_mode js' instead",
+            });
+            strip_mode.js = true;
             continue;
         }
 
@@ -566,6 +582,32 @@ fn parseFetchArgs(
             continue;
         }
 
+        if (std.mem.eql(u8, "--strip_mode", opt)) {
+            const str = args.next() orelse {
+                log.fatal(.app, "missing argument value", .{ .arg = "--strip_mode" });
+                return error.InvalidArgument;
+            };
+
+            var it = std.mem.splitScalar(u8, str, ',');
+            while (it.next()) |part| {
+                const trimmed = std.mem.trim(u8, part, &std.ascii.whitespace);
+                if (std.mem.eql(u8, trimmed, "js")) {
+                    strip_mode.js = true;
+                } else if (std.mem.eql(u8, trimmed, "ui")) {
+                    strip_mode.ui = true;
+                } else if (std.mem.eql(u8, trimmed, "css")) {
+                    strip_mode.css = true;
+                } else if (std.mem.eql(u8, trimmed, "full")) {
+                    strip_mode.js = true;
+                    strip_mode.ui = true;
+                    strip_mode.css = true;
+                } else {
+                    log.fatal(.app, "invalid option choice", .{ .arg = "--strip_mode", .value = trimmed });
+                }
+            }
+            continue;
+        }
+
         if (try parseCommonArg(allocator, opt, args, &common)) {
             continue;
         }
@@ -591,8 +633,8 @@ fn parseFetchArgs(
         .url = url.?,
         .dump = dump,
         .common = common,
-        .noscript = noscript,
         .withbase = withbase,
+        .strip_mode = strip_mode,
     };
 }