From 14b3999705f4c8eab15db3697f5da01bdff4c7fb Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Tue, 10 Aug 2021 17:45:37 +1000
Subject: [PATCH] Add Cfg option to disable DOCTYPE minification

---
 cli/src/main.rs                               | 15 ++++++++++++
 .../in/wilsonl/minifyhtml/Configuration.java  | 10 ++++++++
 java/src/main/rust/lib.rs                     |  5 ++++
 nodejs/binding.c                              |  2 ++
 nodejs/index.d.ts                             |  2 ++
 nodejs/native/src/lib.rs                      |  2 ++
 python/src/lib.template.rs                    | 23 +++++++++++--------
 ruby/src/lib.rs                               |  1 +
 rust/main/src/cfg/mod.rs                      |  3 +++
 rust/main/src/minify/doctype.rs               |  8 +++++--
 10 files changed, 59 insertions(+), 12 deletions(-)
diff --git a/cli/src/main.rs b/cli/src/main.rs
index 3e61e77..ec0aa8f 100644
--- a/cli/src/main.rs
+++ b/cli/src/main.rs
@@ -15,33 +15,47 @@ struct Cli {
     /// File to minify; omit for stdin.
     #[structopt(parse(from_os_str))]
     input: Option<std::path::PathBuf>,
+
     /// Output destination; omit for stdout.
     #[structopt(short, long, parse(from_os_str))]
     output: Option<std::path::PathBuf>,
+
     /// Minify JS in `<script>` tags that have a valid or no `type` attribute value.
     #[structopt(long)]
     minify_js: bool,
+
     /// Minify CSS in `<style>` tags and `style` attributes.
     #[structopt(long)]
     minify_css: bool,
+
+    #[structopt(long)]
+    /// Do not minify DOCTYPEs. Minified DOCTYPEs may not be spec compliant.
+    do_not_minify_doctype: bool,
+
     /// Ensure all unquoted attribute values in the output do not contain any characters prohibited by the WHATWG specification.
     #[structopt(long)]
     ensure_spec_compliant_unquoted_attribute_values: bool,
+
     /// Do not omit closing tags when possible.
     #[structopt(long)]
     keep_closing_tags: bool,
+
     /// Do not omit `<html>` and `<head>` opening tags when they don't have attributes.
     #[structopt(long)]
     keep_html_and_head_opening_tags: bool,
+
     /// Keep spaces between attributes when possible to conform to HTML standards.
     #[structopt(long)]
     keep_spaces_between_attributes: bool,
+
     /// Keep all comments.
     #[structopt(long)]
     keep_comments: bool,
+
     /// Remove all bangs.
     #[structopt(long)]
     remove_bangs: bool,
+
     /// Remove all processing_instructions.
     #[structopt(long)]
     remove_processing_instructions: bool,
@@ -74,6 +88,7 @@ fn main() {
     let out_code = minify(
         &src_code,
         &Cfg {
+            do_not_minify_doctype: args.do_not_minify_doctype,
             ensure_spec_compliant_unquoted_attribute_values: args
                 .ensure_spec_compliant_unquoted_attribute_values,
             keep_closing_tags: args.keep_closing_tags,
diff --git a/java/src/main/java/in/wilsonl/minifyhtml/Configuration.java b/java/src/main/java/in/wilsonl/minifyhtml/Configuration.java
index 6ce07a3..c772147 100644
--- a/java/src/main/java/in/wilsonl/minifyhtml/Configuration.java
+++ b/java/src/main/java/in/wilsonl/minifyhtml/Configuration.java
@@ -4,6 +4,7 @@ package in.wilsonl.minifyhtml;
  * Class representing minification configuration.
  */
 public class Configuration {
+  public final boolean do_not_minify_doctype;
   public final boolean ensure_spec_compliant_unquoted_attribute_values;
   public final boolean keep_closing_tags;
   public final boolean keep_comments;
@@ -15,6 +16,7 @@ public class Configuration {
   public final boolean remove_processing_instructions;
 
   public Configuration(
+    boolean do_not_minify_doctype,
     boolean ensure_spec_compliant_unquoted_attribute_values,
     boolean keep_closing_tags,
     boolean keep_comments,
@@ -25,6 +27,7 @@ public class Configuration {
     boolean remove_bangs,
     boolean remove_processing_instructions
   ) {
+    this.do_not_minify_doctype = do_not_minify_doctype;
     this.ensure_spec_compliant_unquoted_attribute_values = ensure_spec_compliant_unquoted_attribute_values;
     this.keep_closing_tags = keep_closing_tags;
     this.keep_comments = keep_comments;
@@ -40,6 +43,7 @@ public class Configuration {
    * Builder to help create configuration.
    */
   public static class Builder {
+    private boolean do_not_minify_doctype = false;
     private boolean ensure_spec_compliant_unquoted_attribute_values = false;
     private boolean keep_closing_tags = false;
     private boolean keep_comments = false;
@@ -50,6 +54,11 @@ public class Configuration {
     private boolean remove_bangs = false;
     private boolean remove_processing_instructions = false;
 
+    public Builder setDoNotMinifyDoctype(boolean val) {
+      this.do_not_minify_doctype = val;
+      return this;
+    }
+
     public Builder setEnsureSpecCompliantUnquotedAttributeValues(boolean val) {
       this.ensure_spec_compliant_unquoted_attribute_values = val;
       return this;
@@ -98,6 +107,7 @@ public class Configuration {
 
     public Configuration build() {
       return new Configuration(
+        this.do_not_minify_doctype,
         this.ensure_spec_compliant_unquoted_attribute_values,
         this.keep_closing_tags,
         this.keep_comments,
diff --git a/java/src/main/rust/lib.rs b/java/src/main/rust/lib.rs
index 85d19b0..d9f8207 100644
--- a/java/src/main/rust/lib.rs
+++ b/java/src/main/rust/lib.rs
@@ -6,6 +6,11 @@ use std::str::from_utf8;
 
 fn build_cfg(env: &JNIEnv, obj: &JObject) -> Cfg {
     Cfg {
+        do_not_minify_doctype: env
+            .get_field(*obj, "do_not_minify_doctype", "Z")
+            .unwrap()
+            .z()
+            .unwrap(),
         ensure_spec_compliant_unquoted_attribute_values: env
             .get_field(*obj, "ensure_spec_compliant_unquoted_attribute_values", "Z")
             .unwrap()
diff --git a/nodejs/binding.c b/nodejs/binding.c
index f9c6e53..e6aebc4 100644
--- a/nodejs/binding.c
+++ b/nodejs/binding.c
@@ -71,6 +71,7 @@ napi_value node_method_create_configuration(napi_env env, napi_callback_info inf
     /* It's OK if this fails. */ napi_get_value_bool(env, prop##_value, &prop); \
   }
 
+  GET_CFG_PROP(do_not_minify_doctype);
   GET_CFG_PROP(ensure_spec_compliant_unquoted_attribute_values);
   GET_CFG_PROP(keep_closing_tags);
   GET_CFG_PROP(keep_comments);
@@ -82,6 +83,7 @@ napi_value node_method_create_configuration(napi_env env, napi_callback_info inf
   GET_CFG_PROP(remove_processing_instructions);
 
   Cfg const* cfg = ffi_create_cfg(
+    do_not_minify_doctype,
     ensure_spec_compliant_unquoted_attribute_values,
     keep_closing_tags,
     keep_comments,
diff --git a/nodejs/index.d.ts b/nodejs/index.d.ts
index 404fa16..30ebb84 100644
--- a/nodejs/index.d.ts
+++ b/nodejs/index.d.ts
@@ -8,6 +8,8 @@ export type Cfg = { __doNotUseCfgDirectly: string & { __itIsANapiExternalValue:
  * @returns An opaque value that can be passed to minify functions
  */
 export function createConfiguration (options: {
+  /** Do not minify DOCTYPEs. Minified DOCTYPEs may not be spec compliant. */
+  do_not_minify_doctype?: boolean,
   /** Ensure all unquoted attribute values in the output do not contain any characters prohibited by the WHATWG specification. */
   ensure_spec_compliant_unquoted_attribute_values?: boolean;
   /** Do not omit closing tags when possible. */
diff --git a/nodejs/native/src/lib.rs b/nodejs/native/src/lib.rs
index 877e3d0..89c556b 100644
--- a/nodejs/native/src/lib.rs
+++ b/nodejs/native/src/lib.rs
@@ -3,6 +3,7 @@ use std::{mem, slice};
 
 #[no_mangle]
 pub extern "C" fn ffi_create_cfg(
+    do_not_minify_doctype: bool,
     ensure_spec_compliant_unquoted_attribute_values: bool,
     keep_closing_tags: bool,
     keep_comments: bool,
@@ -14,6 +15,7 @@ pub extern "C" fn ffi_create_cfg(
     remove_processing_instructions: bool,
 ) -> *const Cfg {
     Box::into_raw(Box::new(Cfg {
+        do_not_minify_doctype,
         ensure_spec_compliant_unquoted_attribute_values,
         keep_closing_tags,
         keep_comments,
diff --git a/python/src/lib.template.rs b/python/src/lib.template.rs
index 4e6a950..9cd43d0 100644
--- a/python/src/lib.template.rs
+++ b/python/src/lib.template.rs
@@ -4,19 +4,21 @@ use pyo3::wrap_pyfunction;
 use std::string::String;
 
 #[pyfunction(
-    py_args="*",
-    ensure_spec_compliant_unquoted_attribute_values="false",
-    keep_closing_tags="false",
-    keep_comments="false",
-    keep_html_and_head_opening_tags="false",
-    keep_spaces_between_attributes="false",
-    minify_css="false",
-    minify_js="false",
-    remove_bangs="false",
-    remove_processing_instructions="false",
+    py_args = "*",
+    do_not_minify_doctype = "false",
+    ensure_spec_compliant_unquoted_attribute_values = "false",
+    keep_closing_tags = "false",
+    keep_comments = "false",
+    keep_html_and_head_opening_tags = "false",
+    keep_spaces_between_attributes = "false",
+    minify_css = "false",
+    minify_js = "false",
+    remove_bangs = "false",
+    remove_processing_instructions = "false",
 )]
 fn minify(
     code: String,
+    do_not_minify_doctype: bool,
     ensure_spec_compliant_unquoted_attribute_values: bool,
     keep_closing_tags: bool,
     keep_comments: bool,
@@ -29,6 +31,7 @@ fn minify(
 ) -> PyResult<String> {
     let code = code.into_bytes();
     let out_code = minify_html_native(&code, &Cfg {
+        do_not_minify_doctype,
         ensure_spec_compliant_unquoted_attribute_values,
         keep_closing_tags,
         keep_comments,
diff --git a/ruby/src/lib.rs b/ruby/src/lib.rs
index 640386d..30f1c83 100644
--- a/ruby/src/lib.rs
+++ b/ruby/src/lib.rs
@@ -29,6 +29,7 @@ methods! {
             .unwrap();
 
         let cfg = &Cfg {
+            do_not_minify_doctype: get_cfg_hash_prop!(cfg_hash, "do_not_minify_doctype"),
             ensure_spec_compliant_unquoted_attribute_values: get_cfg_hash_prop!(cfg_hash, "ensure_spec_compliant_unquoted_attribute_values"),
             keep_closing_tags: get_cfg_hash_prop!(cfg_hash, "keep_closing_tags"),
             keep_comments: get_cfg_hash_prop!(cfg_hash, "keep_comments"),
diff --git a/rust/main/src/cfg/mod.rs b/rust/main/src/cfg/mod.rs
index 74a8193..72d1b27 100644
--- a/rust/main/src/cfg/mod.rs
+++ b/rust/main/src/cfg/mod.rs
@@ -2,6 +2,8 @@
 /// minification approach.
 #[derive(Default)]
 pub struct Cfg {
+    /// Do not minify DOCTYPEs. Minified DOCTYPEs may not be spec compliant.
+    pub do_not_minify_doctype: bool,
     /// Ensure all unquoted attribute values in the output do not contain any characters prohibited by the [WHATWG specification](https://html.spec.whatwg.org/multipage/syntax.html#attributes-2).
     pub ensure_spec_compliant_unquoted_attribute_values: bool,
     /// Do not omit closing tags when possible.
@@ -37,6 +39,7 @@ impl Cfg {
 
     pub fn spec_compliant() -> Cfg {
         Cfg {
+            do_not_minify_doctype: true,
             ensure_spec_compliant_unquoted_attribute_values: true,
             keep_spaces_between_attributes: true,
             ..Cfg::default()
diff --git a/rust/main/src/minify/doctype.rs b/rust/main/src/minify/doctype.rs
index 4aa4798..4b37aa0 100644
--- a/rust/main/src/minify/doctype.rs
+++ b/rust/main/src/minify/doctype.rs
@@ -1,7 +1,11 @@
 use crate::cfg::Cfg;
 
-pub fn minify_doctype(_cfg: &Cfg, out: &mut Vec<u8>, legacy: &[u8], ended: bool) {
-    out.extend_from_slice(b"<!doctypehtml");
+pub fn minify_doctype(cfg: &Cfg, out: &mut Vec<u8>, legacy: &[u8], ended: bool) {
+    out.extend_from_slice(b"<!doctype");
+    if cfg.do_not_minify_doctype {
+        out.push(b' ');
+    }
+    out.extend_from_slice(b"html");
     if !legacy.is_empty() {
         out.push(b' ');
         out.extend_from_slice(legacy);
