Merge pull request #115 from jugglerchris/update_tracing

Update tracing and fix ignoring of !important CSS properties.

.circleci/config.yml

@@ -20,14 +20,14 @@ jobs:
       - run: cargo --version
       - run: cargo build --features=css
       - run: cargo test --features=css
-  build-1-60:
+  build-1-63:
     docker:
-      - image: cimg/rust:1.60
+      - image: cimg/rust:1.63
     steps:
       - checkout
       - run: cargo --version
       - run: cargo build
-      - run: cargo test
+        #      - run: cargo test
   build-windows:
     executor:
       name: win/default
@@ -63,5 +63,5 @@ workflows:
     jobs:
       - "build-stable"
       - "build-css"
-      - "build-1-60"
+      - "build-1-63"
       - "build-windows"

CHANGELOG.md

@@ -9,6 +9,12 @@ Possible log types:
 - `[fixed]` for any bug fixes.
 - `[security]` to invite users to upgrade in case of vulnerabilities.
 
+### 0.11.0
+
+- [fixed] CSS: rules marked !important were ignored.
+- [changed] html\_trace feature now uses the `log` crate.
+- [changed] Bumped MSRV to 1.63 (matching Debian stable) due to some dependencies.
+
 ### 0.10.3
 
 - [fixed] A panic on some unlucky text wrapping coincidences.

Cargo.toml

@@ -1,13 +1,13 @@
 [package]
 name = "html2text"
-version = "0.10.3"
+version = "0.11.0"
 authors = ["Chris Emerson <[email protected]>"]
 description = "Render HTML as plain text."
 repository = "https://github.com/jugglerchris/rust-html2text/"
 readme = "README.md"
 documentation = "https://docs.rs/html2text/"
 edition = "2021"
-rust-version = "1.60"
+rust-version = "1.63"
 categories = ["text-processing"]
 
 keywords = ["html", "text"]
@@ -24,10 +24,11 @@ thiserror = "1.0.50"
 lightningcss = { version = "1.0.0-alpha.51", optional=true }
 # Keep dashmap back; 5.5 has a higher MSRV.
 dashmap = "~5.4"
+log = { version = "0.4.20", optional = true }
 
 [features]
-html_trace = []
-html_trace_bt = ["dep:backtrace"]
+html_trace = ["dep:log"]
+html_trace_bt = ["html_trace", "dep:backtrace"]
 default = []
 css = ["dep:lightningcss"]
 
@@ -40,7 +41,9 @@ name = "html2text"
 path = "examples/html2text.rs"
 
 [dev-dependencies]
+env_logger = "0.10.1"
 argparse = "0.2.2"
+log = "0.4.20"
 
 [target.'cfg(unix)'.dev-dependencies]
 termion = "2.0"

examples/html2text.rs

@@ -5,6 +5,7 @@ use html2text::config::{Config, self};
 use html2text::render::text_renderer::{TextDecorator, TrivialDecorator};
 use std::io;
 use std::io::Write;
+use log::trace;
 
 #[cfg(unix)]
 use html2text::render::text_renderer::RichAnnotation;
@@ -19,6 +20,7 @@ fn default_colour_map(annotations: &[RichAnnotation], s: &str) -> String {
     let mut have_explicit_colour = false;
     let mut start = Vec::new();
     let mut finish = Vec::new();
+    trace!("default_colour_map: str={s}, annotations={annotations:?}");
     for annotation in annotations.iter() {
         match annotation {
             Default => {}
@@ -79,6 +81,7 @@ fn default_colour_map(annotations: &[RichAnnotation], s: &str) -> String {
     for s in finish {
         result.push_str(&s);
     }
+    trace!("default_colour_map: output={result}");
     result
 }
 
@@ -129,6 +132,9 @@ struct Flags {
 }
 
 fn main() {
+    #[cfg(feature = "html_trace")]
+    env_logger::init();
+
     let mut infile: Option<String> = None;
     let mut outfile: Option<String> = None;
     let mut flags = Flags {

src/css.rs

@@ -178,10 +178,13 @@ pub struct StyleData {
 }
 
 pub(crate) fn parse_style_attribute(text: &str) -> Result<Vec<Style>> {
+    html_trace_quiet!("Parsing inline style: {text}");
     let sattr = StyleAttribute::parse(text, ParserOptions::default())
         .map_err(|_| crate::Error::CssParseError)?;
 
-    Ok(styles_from_properties(&sattr.declarations))
+    let styles = styles_from_properties(&sattr.declarations);
+    html_trace_quiet!("Parsed inline style: {:?}", styles);
+    Ok(styles)
 }
 
 fn is_transparent(color: &CssColor) -> bool {
@@ -198,7 +201,9 @@ fn is_transparent(color: &CssColor) -> bool {
 
 fn styles_from_properties(decls: &DeclarationBlock<'_>) -> Vec<Style> {
     let mut styles = Vec::new();
-    for decl in &decls.declarations {
+    html_trace_quiet!("styles:from_properties: {decls:?}");
+    for decl in decls.declarations.iter().chain(decls.important_declarations.iter()) {
+        html_trace_quiet!("styles:from_properties: {decl:?}");
         match decl {
             Property::Color(color) => {
                 if is_transparent(&color) {
@@ -224,7 +229,9 @@ fn styles_from_properties(decls: &DeclarationBlock<'_>) -> Vec<Style> {
                     styles.push(Style::DisplayNone);
                 }
             }
-            _ => {}
+            _ => {
+                html_trace_quiet!("CSS: Unhandled property {:?}", decl);
+            }
         }
     }
     styles
@@ -245,12 +252,15 @@ impl StyleData {
                         for selector in &style.selectors.0 {
                             match Selector::try_from(selector) {
                                 Ok(selector) => {
-                                    self.rules.push(Ruleset {
+                                    let ruleset = Ruleset {
                                         selector,
                                         styles: styles.clone()
-                                    });
+                                    };
+                                    html_trace_quiet!("Adding ruleset {ruleset:?}");
+                                    self.rules.push(ruleset);
                                 }
                                 Err(_) => {
+                                    html_trace!("Ignoring selector {:?}", selector);
                                     continue;
                                 }
                             }
@@ -322,7 +332,7 @@ fn combine_vecs(vecs: Vec<Vec<String>>) -> Vec<String> {
 
 fn extract_style_nodes<'a, 'b, T: Write>(
     handle: Handle,
-    err_out: &'b mut T,
+    _err_out: &'b mut T,
 ) -> TreeMapResult<'a, (), Handle, Vec<String>> {
     use TreeMapResult::*;
 
@@ -345,9 +355,7 @@ fn extract_style_nodes<'a, 'b, T: Write>(
                     Finished(vec![result])
                 }
                 _ => {
-                    html_trace!("Unhandled element: {:?}\n", name.local);
                     pending(handle, |_, cs| Ok(Some(combine_vecs(cs))))
-                    //None
                 }
             }
         }
@@ -356,7 +364,6 @@ fn extract_style_nodes<'a, 'b, T: Write>(
         }
         _ => {
             // NodeData doesn't have a Debug impl.
-            write!(err_out, "Unhandled node type.\n").unwrap();
             Nothing
         }
     }

src/macros.rs

@@ -17,22 +17,22 @@ macro_rules! html_trace {
          #[cfg(feature = "html_trace_bt")]
          {
              let bt = ::backtrace::Backtrace::new();
-             eprintln!( concat!($fmt, " at {:?}"), bt );
+             log::info!( concat!($fmt, " at {:?}"), bt );
          }
          #[cfg(not(feature = "html_trace_bt"))]
          {
-             eprintln!($fmt);
+             log::info!($fmt);
          }
     };
     ($fmt:expr, $( $args:expr ),*) => {
          #[cfg(feature = "html_trace_bt")]
          {
              let bt = ::backtrace::Backtrace::new();
-             eprintln!( concat!($fmt, " at {:?}"), $( $args ),* , bt );
+             log::info!( concat!($fmt, " at {:?}"), $( $args ),* , bt );
          }
          #[cfg(not(feature = "html_trace_bt"))]
          {
-             eprintln!($fmt, $( $args ),*);
+             log::info!($fmt, $( $args ),*);
          }
     };
 }
@@ -53,10 +53,10 @@ macro_rules! html_trace {
 #[doc(hidden)]
 macro_rules! html_trace_quiet {
     ($fmt:expr) => {
-         eprintln!( $fmt );
+         log::trace!( $fmt );
     };
     ($fmt:expr, $( $args:expr ),*) => {
-         eprintln!( $fmt, $( $args ),* );
+         log::trace!( $fmt, $( $args ),* );
     };
 }
 

src/tests.rs

@@ -2017,6 +2017,32 @@ text
 "#);
     }
 
+    #[test]
+    fn test_coloured_important()
+    {
+        use super::test_colour_map;
+        let config = crate::config::rich()
+                    .use_doc_css();
+        let dom = config.parse_html(&br##"
+          <style>
+              .red {
+                  color:#FF0000 !important;
+              }
+          </style>
+        <p>Test paragraph with <span class="red">red</span> text</p>
+        "##[..]).unwrap();
+        let rt = config.dom_to_render_tree(&dom).unwrap();
+        assert_eq!(config.render_coloured(rt.clone(), 10, test_colour_map).unwrap(),
+        r#"Test
+paragraph
+with <R>red</R>
+text
+"#);
+        assert_eq!(config.render_coloured(rt.clone(), 100, test_colour_map).unwrap(),
+        r#"Test paragraph with <R>red</R> text
+"#);
+    }
+
     #[test]
     fn test_wrap_word_boundaries() {
         let html = br#"<head><style>em { color: white; }</style></head>