feat(html/codegen): Support context element (#4887)

2024-10-04 20:28:43 +03:00 · 2022-06-05 23:43:56 +03:00 · 2022-06-05 23:43:56 +03:00 · ae1ff1e55b
commit ae1ff1e55b
parent 3812fb2eb0
7 changed files with 138 additions and 131 deletions
--- a/crates/swc_html_codegen/src/ctx.rs
+++ b/crates/swc_html_codegen/src/ctx.rs
@ -22,7 +22,7 @@ where

 #[derive(Debug, Default, Clone, Copy)]
 pub(crate) struct Ctx {
-    pub skip_escape_text: bool,
+    pub need_escape_text: bool,
    pub need_extra_newline_in_text: bool,
 }

--- a/crates/swc_html_codegen/src/lib.rs
+++ b/crates/swc_html_codegen/src/lib.rs
@ -18,10 +18,18 @@ mod emit;
 mod list;
 pub mod writer;

-#[derive(Debug, Clone, Copy, Default)]
+#[derive(Debug, Clone, Default)]
 pub struct CodegenConfig {
    pub minify: bool,
    pub scripting_enabled: bool,
+    /// Should be used only for `DocumentFragment` code generation
+    pub context_element: Option<Element>,
+}
+
+enum TagOmissionParent<'a> {
+    Document(&'a Document),
+    DocumentFragment(&'a DocumentFragment),
+    Element(&'a Element),
 }

 #[derive(Debug)]
@ -52,23 +60,7 @@ where
    #[emitter]
    fn emit_document(&mut self, n: &Document) -> Result {
        if self.config.minify {
-            for (idx, node) in n.children.iter().enumerate() {
-                match node {
-                    Child::Element(element) => {
-                        let prev = if idx > 0 {
-                            n.children.get(idx - 1)
-                        } else {
-                            None
-                        };
-                        let next = n.children.get(idx + 1);
-
-                        self.basic_emit_element(element, None, prev, next)?;
-                    }
-                    _ => {
-                        emit!(self, node)
-                    }
-                }
-            }
+            self.emit_list_for_tag_omission(TagOmissionParent::Document(n))?;
        } else {
            self.emit_list(&n.children, ListFormat::NotDelimited)?;
        }
@ -76,26 +68,18 @@ where

    #[emitter]
    fn emit_document_fragment(&mut self, n: &DocumentFragment) -> Result {
-        if self.config.minify {
-            for (idx, node) in n.children.iter().enumerate() {
-                match node {
-                    Child::Element(element) => {
-                        let prev = if idx > 0 {
-                            n.children.get(idx - 1)
-                        } else {
-                            None
-                        };
-                        let next = n.children.get(idx + 1);
-
-                        self.basic_emit_element(element, None, prev, next)?;
-                    }
-                    _ => {
-                        emit!(self, node)
-                    }
-                }
-            }
+        let ctx = if let Some(context_element) = &self.config.context_element {
+            self.create_context_for_element(context_element)
        } else {
-            self.emit_list(&n.children, ListFormat::NotDelimited)?;
+            Default::default()
+        };
+
+        if self.config.minify {
+            self.with_ctx(ctx)
+                .emit_list_for_tag_omission(TagOmissionParent::DocumentFragment(n))?;
+        } else {
+            self.with_ctx(ctx)
+                .emit_list(&n.children, ListFormat::NotDelimited)?;
        }
    }

@ -183,6 +167,10 @@ where
        prev: Option<&Child>,
        next: Option<&Child>,
    ) -> Result {
+        if self.is_plaintext {
+            return Ok(());
+        }
+
        let has_attributes = !n.attributes.is_empty();
        let can_omit_start_tag = self.config.minify
            && !has_attributes
@ -342,29 +330,18 @@ where
            return Ok(());
        }

-        self.is_plaintext = matches!(&*n.tag_name, "plaintext");
+        if !self.is_plaintext {
+            self.is_plaintext = matches!(&*n.tag_name, "plaintext");
+        }

        if let Some(content) = &n.content {
            emit!(self, content);
        } else if !n.children.is_empty() {
-            let skip_escape_text = match &*n.tag_name {
-                "style" | "script" | "xmp" | "iframe" | "noembed" | "noframes" => true,
-                "noscript" => self.config.scripting_enabled,
-                _ if self.is_plaintext => true,
-                _ => false,
-            };
-            let need_extra_newline_in_text =
-                n.namespace == Namespace::HTML && matches!(&*n.tag_name, "textarea" | "pre");
-
-            let ctx = Ctx {
-                skip_escape_text,
-                need_extra_newline_in_text,
-                ..self.ctx
-            };
+            let ctx = self.create_context_for_element(n);

            if self.config.minify {
                self.with_ctx(ctx)
-                    .emit_list_for_tag_omission(n, &n.children)?;
+                    .emit_list_for_tag_omission(TagOmissionParent::Element(n))?;
            } else {
                self.with_ctx(ctx)
                    .emit_list(&n.children, ListFormat::NotDelimited)?;
@ -713,9 +690,7 @@ where

    #[emitter]
    fn emit_text(&mut self, n: &Text) -> Result {
-        if self.ctx.skip_escape_text {
-            write_str!(self, n.span, &n.value);
-        } else {
+        if self.ctx.need_escape_text {
            let mut data = String::with_capacity(n.value.len());

            if self.ctx.need_extra_newline_in_text && n.value.contains('\n') {
@ -729,6 +704,8 @@ where
            }

            write_str!(self, n.span, &data);
+        } else {
+            write_str!(self, n.span, &n.value);
        }
    }

@ -743,6 +720,23 @@ where
        write_str!(self, n.span, &comment);
    }

+    fn create_context_for_element(&self, n: &Element) -> Ctx {
+        let need_escape_text = match &*n.tag_name {
+            "style" | "script" | "xmp" | "iframe" | "noembed" | "noframes" | "plaintext" => false,
+            "noscript" => !self.config.scripting_enabled,
+            _ if self.is_plaintext => false,
+            _ => true,
+        };
+        let need_extra_newline_in_text =
+            n.namespace == Namespace::HTML && matches!(&*n.tag_name, "textarea" | "pre");
+
+        Ctx {
+            need_escape_text,
+            need_extra_newline_in_text,
+            ..self.ctx
+        }
+    }
+
    #[emitter]
    fn emit_token_and_span(&mut self, n: &TokenAndSpan) -> Result {
        let span = n.span;
@ -1002,6 +996,34 @@ where
        }
    }

+    fn emit_list_for_tag_omission(&mut self, parent: TagOmissionParent) -> Result {
+        let nodes = match &parent {
+            TagOmissionParent::Document(document) => &document.children,
+            TagOmissionParent::DocumentFragment(document_fragment) => &document_fragment.children,
+            TagOmissionParent::Element(element) => &element.children,
+        };
+        let parent = match parent {
+            TagOmissionParent::Element(element) => Some(element),
+            _ => None,
+        };
+
+        for (idx, node) in nodes.iter().enumerate() {
+            match node {
+                Child::Element(element) => {
+                    let prev = if idx > 0 { nodes.get(idx - 1) } else { None };
+                    let next = nodes.get(idx + 1);
+
+                    self.basic_emit_element(element, parent, prev, next)?;
+                }
+                _ => {
+                    emit!(self, node)
+                }
+            }
+        }
+
+        Ok(())
+    }
+
    fn emit_list<N>(&mut self, nodes: &[N], format: ListFormat) -> Result
    where
        Self: Emit<N>,
@ -1022,24 +1044,6 @@ where
        Ok(())
    }

-    fn emit_list_for_tag_omission(&mut self, parent: &Element, nodes: &[Child]) -> Result {
-        for (idx, node) in nodes.iter().enumerate() {
-            match node {
-                Child::Element(element) => {
-                    let prev = if idx > 0 { nodes.get(idx - 1) } else { None };
-                    let next = nodes.get(idx + 1);
-
-                    self.basic_emit_element(element, Some(parent), prev, next)?;
-                }
-                _ => {
-                    emit!(self, node)
-                }
-            }
-        }
-
-        Ok(())
-    }
-
    fn write_delim(&mut self, f: ListFormat) -> Result {
        match f & ListFormat::DelimitersMask {
            ListFormat::None => {}
--- a/crates/swc_html_codegen/tests/fixture.rs
+++ b/crates/swc_html_codegen/tests/fixture.rs
@ -254,11 +254,13 @@ fn verify_document_fragment(
        Some(writer_config) => writer_config,
        _ => BasicHtmlWriterConfig::default(),
    };
-    let codegen_config = match codegen_config {
+    let mut codegen_config = match codegen_config {
        Some(codegen_config) => codegen_config,
        _ => CodegenConfig::default(),
    };

+    codegen_config.context_element = Some(context_element.clone());
+
    testing::run_test2(false, |cm, handler| {
        let fm = cm.load_file(input).unwrap();
        let mut errors = vec![];
@ -330,6 +332,7 @@ fn test_document(input: PathBuf) {
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: false,
+            ..Default::default()
        }),
    );
    print_document(
@ -339,44 +342,42 @@ fn test_document(input: PathBuf) {
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: true,
+            ..Default::default()
        }),
    );
 }

 #[testing::fixture("tests/document_fragment/**/input.html")]
 fn test_document_fragment(input: PathBuf) {
+    let context_element = Element {
+        span: Default::default(),
+        tag_name: "template".into(),
+        namespace: Namespace::HTML,
+        attributes: vec![],
+        children: vec![],
+        content: None,
+    };
+
    print_document_fragment(
        &input,
-        Element {
-            span: Default::default(),
-            tag_name: "template".into(),
-            namespace: Namespace::HTML,
-            attributes: vec![],
-            children: vec![],
-            content: None,
-        },
+        context_element.clone(),
        None,
        None,
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: false,
+            ..Default::default()
        }),
    );
    print_document_fragment(
        &input,
-        Element {
-            span: Default::default(),
-            tag_name: "template".into(),
-            namespace: Namespace::HTML,
-            attributes: vec![],
-            children: vec![],
-            content: None,
-        },
+        context_element,
        None,
        None,
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: true,
+            ..Default::default()
        }),
    );
 }
@ -405,6 +406,7 @@ fn parser_verify(input: PathBuf) {
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: true,
+            ..Default::default()
        }),
        false,
    );
@ -419,7 +421,6 @@ fn parser_verify(input: PathBuf) {
        "element/foreign-context/input.html",
        "element/a-4/input.html",
        "element/b-3/input.html",
-        "element/template-1/input.html",
    )
 )]
 fn parser_recovery_verify(input: PathBuf) {
@ -430,41 +431,37 @@ fn parser_recovery_verify(input: PathBuf) {
        Some(CodegenConfig {
            scripting_enabled: false,
            minify: true,
+            ..Default::default()
        }),
        true,
    );
 }

-// Non conforming document (i.e. broken HTML), so restore them how it was
-// originally is impossible ,
-//
-// Example - `<!DOCTYPE html><html><body><!-- Test`
-//
-// Here we have unclosed comment, so serialization will be not the same
-//
-// TODO There is only bugs for `fragment` because we should allow to
-// pass context element for codegen too
+// Tag omission only works for valid HTML documents (i.e. without errors)
+static IGNORE_TAG_OMISSION: &[&str] = &[
+    "adoption01_dat/5/input.html",
+    "adoption01_dat/6/input.html",
+    "adoption01_dat/7/input.html",
+    "adoption01_dat/8/input.html",
+    "adoption02_dat/0/input.html",
+    "tests1_dat/68/input.html",
+    "tests1_dat/69/input.html",
+    "tests1_dat/70/input.html",
+    "tests1_dat/71/input.html",
+    "tests15_dat/0/input.html",
+    "tests15_dat/1/input.html",
+    "template_dat/68/input.html",
+    "tricky01_dat/6/input.html",
+];
+
 #[testing::fixture(
    "../swc_html_parser/tests/html5lib-tests-fixture/**/*.html",
    exclude(
-        "adoption01_dat/5/input.html",
-        "adoption01_dat/6/input.html",
-        "adoption01_dat/7/input.html",
-        "adoption01_dat/8/input.html",
-        "adoption02_dat/0/input.html",
        "tests1_dat/30/input.html",
-        "tests1_dat/68/input.html",
-        "tests1_dat/69/input.html",
-        "tests1_dat/70/input.html",
-        "tests1_dat/71/input.html",
        "tests1_dat/77/input.html",
        "tests1_dat/90/input.html",
        "tests1_dat/103/input.html",
        "tests2_dat/12/input.html",
-        "tests4_dat/3.fragment_style/input.html",
-        "tests4_dat/4.fragment_plaintext/input.html",
-        "tests15_dat/0/input.html",
-        "tests15_dat/1/input.html",
        "tests16_dat/31/input.html",
        "tests16_dat/32/input.html",
        "tests16_dat/33/input.html",
@ -494,14 +491,11 @@ fn parser_recovery_verify(input: PathBuf) {
        "tests18_dat/8/input.html",
        "tests18_dat/9/input.html",
        "tests18_dat/12/input.html",
-        "tests18_dat/15/input.html",
        "tests18_dat/21/input.html",
        "tests19_dat/103/input.html",
        "tests20_dat/41/input.html",
        "tests26_dat/2/input.html",
-        "tricky01_dat/6/input.html",
        "plain-text-unsafe_dat/0/input.html",
-        "template_dat/68/input.html",
        "template_dat/107/input.html",
    )
 )]
@ -515,10 +509,12 @@ fn html5lib_tests_verify(input: PathBuf) {
    let codegen_config = CodegenConfig {
        minify: false,
        scripting_enabled,
+        ..Default::default()
    };
    let minified_codegen_config = CodegenConfig {
        minify: true,
        scripting_enabled,
+        ..Default::default()
    };

    if parent.contains("fragment") {
@ -583,12 +579,20 @@ fn html5lib_tests_verify(input: PathBuf) {
            Some(codegen_config),
            true,
        );
-        verify_document(
-            &input,
-            Some(parser_config),
-            None,
-            Some(minified_codegen_config),
-            true,
-        );
+
+        let relative_path = input.to_string_lossy().replace('-', "_").replace('\\', "/");
+
+        if !IGNORE_TAG_OMISSION
+            .iter()
+            .any(|ignored| relative_path.contains(&**ignored))
+        {
+            verify_document(
+                &input,
+                Some(parser_config),
+                None,
+                Some(minified_codegen_config),
+                true,
+            );
+        }
    }
 }
--- a/crates/swc_html_codegen/tests/fixture/attribute/input.html
+++ b/crates/swc_html_codegen/tests/fixture/attribute/input.html
@ -56,8 +56,7 @@
 <p x='x"'>x</p>
 <a href="#"><p>Click me</p></a>
 <span><button>Hit me</button></span>
-<!--TODO FIX ME -->
-<!--<object type="image/svg+xml" data="image.svg"><div>[fallback image]</div></object>-->
+<object type="image/svg+xml" data="image.svg"><div>[fallback image]</div></object>

 <img class="user-image" src="http:&#47;&#47;adasdasdasd.cloudfront.net&#47;users&#47;2011&#47;05&#47;24&#47;4asdasd&#47;asdasd.jpg" />

--- a/crates/swc_html_codegen/tests/fixture/attribute/output.html
+++ b/crates/swc_html_codegen/tests/fixture/attribute/output.html
@ -52,8 +52,7 @@
 <p x="x&quot;">x</p>
 <a href="#"><p>Click me</p></a>
 <span><button>Hit me</button></span>
-<!--TODO FIX ME -->
-<!--<object type="image/svg+xml" data="image.svg"><div>[fallback image]</div></object>-->
+<object type="image/svg+xml" data="image.svg"><div>[fallback image]</div></object>

 <img class="user-image" src="http://adasdasdasd.cloudfront.net/users/2011/05/24/4asdasd/asdasd.jpg">

--- a/crates/swc_html_codegen/tests/fixture/attribute/output.min.html
+++ b/crates/swc_html_codegen/tests/fixture/attribute/output.min.html
@ -50,8 +50,7 @@
 <p x='x"'>x</p>
 <a href=#><p>Click me</p></a>
 <span><button>Hit me</button></span>
-<!--TODO FIX ME -->
-<!--<object type="image/svg+xml" data="image.svg"><div>[fallback image]</div></object>-->
+<object type=image/svg+xml data=image.svg><div>[fallback image]</div></object>

 <img class=user-image src=http://adasdasdasd.cloudfront.net/users/2011/05/24/4asdasd/asdasd.jpg>

--- a/crates/swc_html_minifier/tests/fixture.rs
+++ b/crates/swc_html_minifier/tests/fixture.rs
@ -47,6 +47,7 @@ fn minify_fixtures(input: PathBuf) {
                CodegenConfig {
                    scripting_enabled: false,
                    minify: true,
+                    ..Default::default()
                },
            );

@ -99,6 +100,7 @@ fn minify_recovery(input: PathBuf) {
                CodegenConfig {
                    scripting_enabled: false,
                    minify: true,
+                    ..Default::default()
                },
            );