fix(html): Fix bugs of parser and codegen (#4461)

2024-12-22 13:11:31 +03:00 · 2022-04-28 20:57:05 +03:00 · 2022-04-28 20:57:05 +03:00 · 8bdfcd996a
commit 8bdfcd996a
parent 44567bb065
13 changed files with 585 additions and 32 deletions
--- a/crates/swc_html_codegen/src/ctx.rs
+++ b/crates/swc_html_codegen/src/ctx.rs
@ -0,0 +1,50 @@
+use std::ops::{Deref, DerefMut};
+
+use crate::{writer::HtmlWriter, CodeGenerator};
+
+impl<W> CodeGenerator<W>
+where
+    W: HtmlWriter,
+{
+    /// Original context is restored when returned guard is dropped.
+    #[inline]
+    pub(super) fn with_ctx(&mut self, ctx: Ctx) -> WithCtx<W> {
+        let orig_ctx = self.ctx;
+
+        self.ctx = ctx;
+
+        WithCtx {
+            orig_ctx,
+            inner: self,
+        }
+    }
+}
+
+#[derive(Debug, Default, Clone, Copy)]
+pub(crate) struct Ctx {
+    pub skip_escape_text: bool,
+}
+
+pub(super) struct WithCtx<'w, I: 'w + HtmlWriter> {
+    inner: &'w mut CodeGenerator<I>,
+    orig_ctx: Ctx,
+}
+
+impl<'w, I: HtmlWriter> Deref for WithCtx<'w, I> {
+    type Target = CodeGenerator<I>;
+
+    fn deref(&self) -> &CodeGenerator<I> {
+        self.inner
+    }
+}
+impl<'w, I: HtmlWriter> DerefMut for WithCtx<'w, I> {
+    fn deref_mut(&mut self) -> &mut CodeGenerator<I> {
+        self.inner
+    }
+}
+
+impl<'w, I: HtmlWriter> Drop for WithCtx<'w, I> {
+    fn drop(&mut self) {
+        self.inner.ctx = self.orig_ctx;
+    }
+}
--- a/crates/swc_html_codegen/src/lib.rs
+++ b/crates/swc_html_codegen/src/lib.rs
@ -9,10 +9,11 @@ use swc_html_codegen_macros::emitter;
 use writer::HtmlWriter;

 pub use self::emit::*;
-use self::list::ListFormat;
+use self::{ctx::Ctx, list::ListFormat};

 #[macro_use]
 mod macros;
+mod ctx;
 mod emit;
 mod list;
 pub mod writer;
@ -28,6 +29,7 @@ where
 {
    wr: W,
    config: CodegenConfig,
+    ctx: Ctx,
 }

 impl<W> CodeGenerator<W>
@ -35,7 +37,11 @@ where
    W: HtmlWriter,
 {
    pub fn new(wr: W, config: CodegenConfig) -> Self {
-        CodeGenerator { wr, config }
+        CodeGenerator {
+            wr,
+            config,
+            ctx: Default::default(),
+        }
    }

    #[emitter]
@ -152,7 +158,26 @@ where
        }

        if !n.children.is_empty() {
-            self.emit_list(&n.children, ListFormat::NotDelimited)?;
+            let skip_escape_text = matches!(
+                &*n.tag_name,
+                "style"
+                | "script"
+                | "xmp"
+                | "iframe"
+                | "noembed"
+                | "noframes"
+                | "plaintext"
+                // TODO we need option here
+                | "noscript"
+            );
+
+            let ctx = Ctx {
+                skip_escape_text,
+                ..self.ctx
+            };
+
+            self.with_ctx(ctx)
+                .emit_list(&n.children, ListFormat::NotDelimited)?;
        }

        write_raw!(self, "<");
@ -193,25 +218,29 @@ where

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

-        for c in n.value.chars() {
-            match c {
-                '&' => {
-                    text.push_str(&String::from("&amp;"));
+            for c in n.value.chars() {
+                match c {
+                    '&' => {
+                        text.push_str(&String::from("&amp;"));
+                    }
+                    '<' => {
+                        text.push_str(&String::from("&lt;"));
+                    }
+                    '>' => {
+                        text.push_str(&String::from("&gt;"));
+                    }
+                    '\u{00A0}' => text.push_str(&String::from("&nbsp;")),
+                    _ => text.push(c),
                }
-                '<' => {
-                    text.push_str(&String::from("&lt;"));
-                }
-                '>' => {
-                    text.push_str(&String::from("&gt;"));
-                }
-                '\u{00A0}' => text.push_str(&String::from("&nbsp;")),
-                _ => text.push(c),
            }
-        }

-        write_str!(self, n.span, &text);
+            write_str!(self, n.span, &text);
+        }
    }

    #[emitter]
--- a/crates/swc_html_minifier/tests/fixture/attribute/script-type/input.html
+++ b/crates/swc_html_minifier/tests/fixture/attribute/script-type/input.html
@ -7,4 +7,10 @@
 <script type="module">
    console.log();
 </script>
-<script>window.jQuery || document.write('<script src="jquery.js"><\/script>')</script>
+<script>window.jQuery || document.write('<script src="jquery.js"><\/script>')</script>
+<script type="text/html">
+    <div>
+        test
+    </div>
+    <!-- aa -->\n
+</script>
--- a/crates/swc_html_minifier/tests/fixture/attribute/script-type/output.min.html
+++ b/crates/swc_html_minifier/tests/fixture/attribute/script-type/output.min.html
@ -7,4 +7,10 @@
 <script type=module>
    console.log();
 </script>
-<script>window.jQuery || document.write('&lt;script src="jquery.js"&gt;&lt;\/script&gt;')</script></head><body></body></html>
+<script>window.jQuery || document.write('<script src="jquery.js"><\/script>')</script>
+<script type=text/html>
+    <div>
+        test
+    </div>
+    <!-- aa -->\n
+</script></head><body></body></html>
--- a/crates/swc_html_minifier/tests/fixture/attribute/textarea/output.min.html
+++ b/crates/swc_html_minifier/tests/fixture/attribute/textarea/output.min.html
@ -3,5 +3,5 @@
 </head>
 <body>
 <textarea name=test id=test></textarea>
-&lt;/&gt;
-&lt;/&gt;</body></html>
+&lt;/body&gt;
+&lt;/html&gt;</body></html>
--- a/crates/swc_html_minifier/tests/fixture/comment/output.min.html
+++ b/crates/swc_html_minifier/tests/fixture/comment/output.min.html
@ -23,16 +23,16 @@

    
    <div>baz</div>
-    <script>&lt;!-- alert(1) --&gt;</script>
-    <script>&lt;!--
+    <script><!-- alert(1) --></script>
+    <script><!--
    alert(1);
-    --&gt;</script>
-    <style>&lt;!-- p { color: red } --&gt;</style>
-    <script>/*&lt;![CDATA[*/alert(8)/*]]&gt;*/</script>
+    --></script>
+    <style><!-- p { color: red } --></style>
+    <script>/*<![CDATA[*/alert(8)/*]]>*/</script>
    <script> /*
-      &lt;![CDATA[  */
+      <![CDATA[  */
    alert(10)
-    /*  ]]&gt;  */
+    /*  ]]>  */
    </script>

 </body></html>
--- a/crates/swc_html_parser/src/lexer/mod.rs
+++ b/crates/swc_html_parser/src/lexer/mod.rs
@ -353,6 +353,17 @@ where
        }
    }

+    fn emit_temporary_buffer(&mut self) {
+        if let Some(temporary_buffer) = self.temporary_buffer.take() {
+            for c in temporary_buffer.chars() {
+                self.emit_token(Token::Character {
+                    value: c,
+                    raw: None,
+                });
+            }
+        }
+    }
+
    fn read_token_and_span(&mut self) -> LexResult<TokenAndSpan> {
        if self.finished {
            return Err(ErrorKind::Eof);
@ -885,6 +896,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rcdata;
                            self.reconsume();
                        }
@ -905,6 +917,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rcdata;
                            self.reconsume();
                        }
@ -926,6 +939,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rcdata;
                            self.reconsume();
                        }
@ -1011,6 +1025,7 @@ where
                            value: '/',
                            raw: None,
                        });
+                        self.emit_temporary_buffer();
                        self.state = State::Rcdata;
                        self.reconsume();
                    }
@ -1100,6 +1115,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rawtext;
                            self.reconsume()
                        }
@ -1120,6 +1136,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rawtext;
                            self.reconsume()
                        }
@ -1141,6 +1158,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::Rawtext;
                            self.reconsume()
                        }
@ -1226,6 +1244,7 @@ where
                            value: '/',
                            raw: None,
                        });
+                        self.emit_temporary_buffer();
                        self.state = State::Rawtext;
                        self.reconsume()
                    }
@ -1329,6 +1348,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptData;
                            self.reconsume()
                        }
@ -1349,6 +1369,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptData;
                            self.reconsume()
                        }
@ -1370,6 +1391,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptData;
                            self.reconsume()
                        }
@ -1455,6 +1477,7 @@ where
                            value: '/',
                            raw: None,
                        });
+                        self.emit_temporary_buffer();
                        self.state = State::ScriptData;
                        self.reconsume()
                    }
@ -1759,6 +1782,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptDataEscaped;
                            self.reconsume()
                        }
@ -1779,6 +1803,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptDataEscaped;
                            self.reconsume()
                        }
@ -1800,6 +1825,7 @@ where
                                value: '/',
                                raw: None,
                            });
+                            self.emit_temporary_buffer();
                            self.state = State::ScriptDataEscaped;
                            self.reconsume()
                        }
@ -1885,6 +1911,7 @@ where
                            value: '/',
                            raw: None,
                        });
+                        self.emit_temporary_buffer();
                        self.state = State::ScriptDataEscaped;
                        self.reconsume()
                    }
--- a/crates/swc_html_parser/src/parser/mod.rs
+++ b/crates/swc_html_parser/src/parser/mod.rs
@ -543,7 +543,7 @@ where
            //      namespace
            //
            //      Acknowledge the token's self-closing flag, and then act as
-            //      described in the steps for      a "script" end tag below.
+            //      described in the steps for a "script" end tag below.
            //
            // Otherwise
            //      Pop the current node off the stack of open elements and acknowledge the token's
--- a/crates/swc_html_parser/tests/fixture/element/embed/output.json
+++ b/crates/swc_html_parser/tests/fixture/element/embed/output.json
@ -215,7 +215,7 @@
                    "end": 225,
                    "ctxt": 0
                  },
-                  "value": "\n        <h1>Alternative content</1>\n    "
+                  "value": "\n        <h1>Alternative content</h1>\n    "
                }
              ],
              "content": null
--- a/crates/swc_html_parser/tests/fixture/tag/script/input.html
+++ b/crates/swc_html_parser/tests/fixture/tag/script/input.html
@ -0,0 +1,14 @@
+<!doctype html>
+<html lang="en">
+<head>
+    <title>Document</title>
+</head>
+<body>
+    <script type="text/html">
+        <div>
+            test
+        </div>
+        <!-- aa -->\n
+    </script>
+</body>
+</html>
--- a/crates/swc_html_parser/tests/fixture/tag/script/output.json
+++ b/crates/swc_html_parser/tests/fixture/tag/script/output.json
@ -0,0 +1,181 @@
+{
+  "type": "Document",
+  "span": {
+    "start": 0,
+    "end": 210,
+    "ctxt": 0
+  },
+  "mode": "no-quirks",
+  "children": [
+    {
+      "type": "DocumentType",
+      "span": {
+        "start": 0,
+        "end": 15,
+        "ctxt": 0
+      },
+      "name": "html",
+      "publicId": null,
+      "systemId": null
+    },
+    {
+      "type": "Element",
+      "span": {
+        "start": 16,
+        "end": 203,
+        "ctxt": 0
+      },
+      "tagName": "html",
+      "namespace": "http://www.w3.org/1999/xhtml",
+      "attributes": [
+        {
+          "type": "Attribute",
+          "span": {
+            "start": 0,
+            "end": 0,
+            "ctxt": 0
+          },
+          "namespace": null,
+          "prefix": null,
+          "name": "lang",
+          "value": "en"
+        }
+      ],
+      "children": [
+        {
+          "type": "Element",
+          "span": {
+            "start": 33,
+            "end": 68,
+            "ctxt": 0
+          },
+          "tagName": "head",
+          "namespace": "http://www.w3.org/1999/xhtml",
+          "attributes": [],
+          "children": [
+            {
+              "type": "Text",
+              "span": {
+                "start": 39,
+                "end": 44,
+                "ctxt": 0
+              },
+              "value": "\n    "
+            },
+            {
+              "type": "Element",
+              "span": {
+                "start": 44,
+                "end": 59,
+                "ctxt": 0
+              },
+              "tagName": "title",
+              "namespace": "http://www.w3.org/1999/xhtml",
+              "attributes": [],
+              "children": [
+                {
+                  "type": "Text",
+                  "span": {
+                    "start": 51,
+                    "end": 59,
+                    "ctxt": 0
+                  },
+                  "value": "Document"
+                }
+              ],
+              "content": null
+            },
+            {
+              "type": "Text",
+              "span": {
+                "start": 67,
+                "end": 68,
+                "ctxt": 0
+              },
+              "value": "\n"
+            }
+          ],
+          "content": null
+        },
+        {
+          "type": "Text",
+          "span": {
+            "start": 75,
+            "end": 76,
+            "ctxt": 0
+          },
+          "value": "\n"
+        },
+        {
+          "type": "Element",
+          "span": {
+            "start": 76,
+            "end": 203,
+            "ctxt": 0
+          },
+          "tagName": "body",
+          "namespace": "http://www.w3.org/1999/xhtml",
+          "attributes": [],
+          "children": [
+            {
+              "type": "Text",
+              "span": {
+                "start": 82,
+                "end": 87,
+                "ctxt": 0
+              },
+              "value": "\n    "
+            },
+            {
+              "type": "Element",
+              "span": {
+                "start": 87,
+                "end": 185,
+                "ctxt": 0
+              },
+              "tagName": "script",
+              "namespace": "http://www.w3.org/1999/xhtml",
+              "attributes": [
+                {
+                  "type": "Attribute",
+                  "span": {
+                    "start": 0,
+                    "end": 0,
+                    "ctxt": 0
+                  },
+                  "namespace": null,
+                  "prefix": null,
+                  "name": "type",
+                  "value": "text/html"
+                }
+              ],
+              "children": [
+                {
+                  "type": "Text",
+                  "span": {
+                    "start": 112,
+                    "end": 185,
+                    "ctxt": 0
+                  },
+                  "value": "\n        <div>\n            test\n        </div>\n        <!-- aa -->\\n\n    "
+                }
+              ],
+              "content": null
+            },
+            {
+              "type": "Text",
+              "span": {
+                "start": 194,
+                "end": 203,
+                "ctxt": 0
+              },
+              "value": "\n\n"
+            }
+          ],
+          "content": null
+        }
+      ],
+      "content": null
+    }
+  ]
+}
--- a/crates/swc_html_parser/tests/fixture/tag/script/span.rust-debug
+++ b/crates/swc_html_parser/tests/fixture/tag/script/span.rust-debug
@ -0,0 +1,240 @@
+
+  x Document
+    ,-[$DIR/tests/fixture/tag/script/input.html:1:1]
+  1 | ,-> <!doctype html>
+  2 | |   <html lang="en">
+  3 | |   <head>
+  4 | |       <title>Document</title>
+  5 | |   </head>
+  6 | |   <body>
+  7 | |       <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | |       </script>
+ 13 | |   </body>
+ 14 | `-> </html>
+    `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:1:1]
+ 1 | <!doctype html>
+   : ^^^^^^^^^^^^^^^
+   `----
+
+  x DocumentType
+   ,-[$DIR/tests/fixture/tag/script/input.html:1:1]
+ 1 | <!doctype html>
+   : ^^^^^^^^^^^^^^^
+   `----
+
+  x Child
+    ,-[$DIR/tests/fixture/tag/script/input.html:2:1]
+  2 | ,-> <html lang="en">
+  3 | |   <head>
+  4 | |       <title>Document</title>
+  5 | |   </head>
+  6 | |   <body>
+  7 | |       <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | |       </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
+
+  x Element
+    ,-[$DIR/tests/fixture/tag/script/input.html:2:1]
+  2 | ,-> <html lang="en">
+  3 | |   <head>
+  4 | |       <title>Document</title>
+  5 | |   </head>
+  6 | |   <body>
+  7 | |       <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | |       </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
+
+  x Attribute
+   ,-[$DIR/tests/fixture/tag/script/input.html:1:1]
+ 1 | <!doctype html>
+   : ^
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:3:1]
+ 3 | ,-> <head>
+ 4 | `->     <title>Document</title>
+ 5 |     </head>
+   `----
+
+  x Element
+   ,-[$DIR/tests/fixture/tag/script/input.html:3:1]
+ 3 | ,-> <head>
+ 4 | `->     <title>Document</title>
+ 5 |     </head>
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:3:1]
+ 3 | ,-> <head>
+ 4 | `->     <title>Document</title>
+   `----
+
+  x Text
+   ,-[$DIR/tests/fixture/tag/script/input.html:3:1]
+ 3 | ,-> <head>
+ 4 | `->     <title>Document</title>
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   : ^^^^^^^^^^^^^^^
+   `----
+
+  x Element
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   : ^^^^^^^^^^^^^^^
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   :        ^^^^^^^^
+   `----
+
+  x Text
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   :        ^^^^^^^^
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   :                        ^
+ 5 | </head>
+   `----
+
+  x Text
+   ,-[$DIR/tests/fixture/tag/script/input.html:4:5]
+ 4 | <title>Document</title>
+   :                        ^
+ 5 | </head>
+   `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:5:1]
+ 5 | </head>
+   :        ^
+ 6 | <body>
+   `----
+
+  x Text
+   ,-[$DIR/tests/fixture/tag/script/input.html:5:1]
+ 5 | </head>
+   :        ^
+ 6 | <body>
+   `----
+
+  x Child
+    ,-[$DIR/tests/fixture/tag/script/input.html:6:1]
+  6 | ,-> <body>
+  7 | |       <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | |       </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
+
+  x Element
+    ,-[$DIR/tests/fixture/tag/script/input.html:6:1]
+  6 | ,-> <body>
+  7 | |       <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | |       </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
+
+  x Child
+   ,-[$DIR/tests/fixture/tag/script/input.html:6:1]
+ 6 | ,-> <body>
+ 7 | `->     <script type="text/html">
+   `----
+
+  x Text
+   ,-[$DIR/tests/fixture/tag/script/input.html:6:1]
+ 6 | ,-> <body>
+ 7 | `->     <script type="text/html">
+   `----
+
+  x Child
+    ,-[$DIR/tests/fixture/tag/script/input.html:7:5]
+  7 | ,-> <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | `->     </script>
+    `----
+
+  x Element
+    ,-[$DIR/tests/fixture/tag/script/input.html:7:5]
+  7 | ,-> <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | `->     </script>
+    `----
+
+  x Child
+    ,-[$DIR/tests/fixture/tag/script/input.html:7:5]
+  7 | ,-> <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | `->     </script>
+    `----
+
+  x Text
+    ,-[$DIR/tests/fixture/tag/script/input.html:7:5]
+  7 | ,-> <script type="text/html">
+  8 | |           <div>
+  9 | |               test
+ 10 | |           </div>
+ 11 | |           <!-- aa -->\n
+ 12 | `->     </script>
+    `----
+
+  x Child
+    ,-[$DIR/tests/fixture/tag/script/input.html:12:5]
+ 12 | ,-> </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
+
+  x Text
+    ,-[$DIR/tests/fixture/tag/script/input.html:12:5]
+ 12 | ,-> </script>
+ 13 | `-> </body>
+ 14 |     </html>
+    `----
--- a/crates/swc_html_parser/tests/recovery/element/embed-closed/output.json
+++ b/crates/swc_html_parser/tests/recovery/element/embed-closed/output.json
@ -215,7 +215,7 @@
                    "end": 225,
                    "ctxt": 0
                  },
-                  "value": "\n        <h1>Alternative content</1>\n    "
+                  "value": "\n        <h1>Alternative content</h1>\n    "
                }
              ],
              "content": null