From bf9460fddeec7366df117ddae13b7d31d3354313 Mon Sep 17 00:00:00 2001 From: Titus Wormer Date: Thu, 30 Jun 2022 12:49:55 +0200 Subject: Add support for stripping tags in image `alt` --- src/compiler.rs | 104 ++++++++++++++++++++++++++---------------------- tests/heading_setext.rs | 8 ++-- tests/image.rs | 32 +++++++-------- tests/link_resource.rs | 11 +++-- 4 files changed, 81 insertions(+), 74 deletions(-) diff --git a/src/compiler.rs b/src/compiler.rs index fbc4792..acba76a 100644 --- a/src/compiler.rs +++ b/src/compiler.rs @@ -241,6 +241,7 @@ struct CompileContext<'a> { pub definitions: HashMap, /// Fields used to influance the current compilation. pub slurp_one_line_ending: bool, + pub tags: bool, pub ignore_encode: bool, pub last_was_tag: bool, /// Configuration @@ -272,6 +273,7 @@ impl<'a> CompileContext<'a> { media_stack: vec![], definitions: HashMap::new(), slurp_one_line_ending: false, + tags: true, ignore_encode: false, last_was_tag: false, protocol_href: if options.allow_dangerous_protocol { @@ -309,6 +311,17 @@ impl<'a> CompileContext<'a> { .last_mut() .expect("Cannot push w/o buffer") .push(value); + self.last_was_tag = false; + } + + pub fn tag(&mut self, value: String) { + if self.tags { + self.buffers + .last_mut() + .expect("Cannot push w/o buffer") + .push(value); + self.last_was_tag = true; + } } /// Get the last chunk of current buffer. @@ -323,13 +336,6 @@ impl<'a> CompileContext<'a> { .expect("at least one buffer should exist") } - /// Get the mutable last chunk of current buffer. - pub fn buf_tail_mut(&mut self) -> &mut Vec { - self.buffers - .last_mut() - .expect("at least one buffer should exist") - } - /// Optionally encode. pub fn encode_opt(&self, value: &str) -> String { if self.ignore_encode { @@ -600,7 +606,7 @@ fn on_enter_buffer(context: &mut CompileContext) { fn on_enter_code_indented(context: &mut CompileContext) { context.code_flow_seen_data = Some(false); context.line_ending_if_needed(); - context.push("
".to_string());
+    context.tag("
".to_string());
 }
 
 /// Handle [`Enter`][EventType::Enter]:[`CodeFenced`][TokenType::CodeFenced].
@@ -608,13 +614,13 @@ fn on_enter_code_fenced(context: &mut CompileContext) {
     context.code_flow_seen_data = Some(false);
     context.line_ending_if_needed();
     // Note that no `>` is used, which is added later.
-    context.push("
".to_string());
+    context.tag("".to_string());
     context.buffer();
 }
 
@@ -662,7 +668,7 @@ fn on_enter_image(context: &mut CompileContext) {
         destination: None,
         title: None,
     });
-    // tags = undefined // Disallow tags.
+    context.tags = false; // Disallow tags.
 }
 
 /// Handle [`Enter`][EventType::Enter]:[`Link`][TokenType::Link].
@@ -679,7 +685,7 @@ fn on_enter_link(context: &mut CompileContext) {
 
 /// Handle [`Enter`][EventType::Enter]:[`Paragraph`][TokenType::Paragraph].
 fn on_enter_paragraph(context: &mut CompileContext) {
-    context.buf_tail_mut().push("
".to_string());
+    context.tag("
".to_string());
 }
 
 /// Handle [`Enter`][EventType::Enter]:[`Resource`][TokenType::Resource].
@@ -704,11 +710,16 @@ fn on_exit_autolink_email(context: &mut CompileContext) {
         &from_exit_event(context.events, context.index),
         false,
     );
-    context.push(format!(
-        "{}",
-        sanitize_uri(slice.as_str(), &context.protocol_href),
-        context.encode_opt(&slice)
+    // To do:
+    context.tag(format!(
+        "",
+        sanitize_uri(
+            format!("mailto:{}", slice.as_str()).as_str(),
+            &context.protocol_href
+        )
     ));
+    context.push(context.encode_opt(&slice));
+    context.tag("".to_string());
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`AutolinkProtocol`][TokenType::AutolinkProtocol].
@@ -718,16 +729,17 @@ fn on_exit_autolink_protocol(context: &mut CompileContext) {
         &from_exit_event(context.events, context.index),
         false,
     );
-    context.push(format!(
-        "{}",
-        sanitize_uri(slice.as_str(), &context.protocol_href),
-        context.encode_opt(&slice)
+    context.tag(format!(
+        "",
+        sanitize_uri(slice.as_str(), &context.protocol_href)
     ));
+    context.push(context.encode_opt(&slice));
+    context.tag("".to_string());
 }
 
 /// Handle [`Exit`][EventType::Exit]:{[`HardBreakEscape`][TokenType::HardBreakEscape],[`HardBreakTrailing`][TokenType::HardBreakTrailing]}.
 fn on_exit_break(context: &mut CompileContext) {
-    context.push("
".to_string());
+    context.tag("
".to_string());
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`CharacterReferenceMarker`][TokenType::CharacterReferenceMarker].
@@ -785,8 +797,7 @@ fn on_exit_code_fenced_fence(context: &mut CompileContext) {
     };
 
     if count == 0 {
-        context.push(">".to_string());
-        // tag = true;
+        context.tag(">".to_string());
         context.slurp_one_line_ending = true;
     }
 
@@ -796,8 +807,7 @@ fn on_exit_code_fenced_fence(context: &mut CompileContext) {
 /// Handle [`Exit`][EventType::Exit]:[`CodeFencedFenceInfo`][TokenType::CodeFencedFenceInfo].
 fn on_exit_code_fenced_fence_info(context: &mut CompileContext) {
     let value = context.resume();
-    context.push(format!(" class=\"language-{}\"", value));
-    // tag = true;
+    context.tag(format!(" class=\"language-{}\"", value));
 }
 
 /// Handle [`Exit`][EventType::Exit]:{[`CodeFenced`][TokenType::CodeFenced],[`CodeIndented`][TokenType::CodeIndented]}.
@@ -823,7 +833,7 @@ fn on_exit_code_flow(context: &mut CompileContext) {
         context.line_ending_if_needed();
     }
 
-    context.push("
".to_string());
+    context.tag("
".to_string());
 
     if let Some(count) = context.code_fenced_fences_count.take() {
         if count < 2 {
@@ -855,7 +865,7 @@ fn on_exit_code_text(context: &mut CompileContext) {
     } else {
         result
     });
-    context.push("".to_string());
+    context.tag("".to_string());
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`CodeTextLineEnding`][TokenType::CodeTextLineEnding].
@@ -873,7 +883,6 @@ fn on_exit_drop(context: &mut CompileContext) {
 /// Handle [`Exit`][EventType::Exit]:{[`CodeTextData`][TokenType::CodeTextData],[`Data`][TokenType::Data],[`CharacterEscapeValue`][TokenType::CharacterEscapeValue]}.
 fn on_exit_data(context: &mut CompileContext) {
     // Just output it.
-    // last_was_tag = false;
     context.push(context.encode_opt(&serialize(
         context.codes,
         &from_exit_event(context.events, context.index),
@@ -930,7 +939,7 @@ fn on_exit_heading_atx(context: &mut CompileContext) {
         .take()
         .expect("`atx_opening_sequence_size` must be set in headings");
 
-    context.push(format!("", rank));
+    context.tag(format!("", rank));
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`HeadingAtxSequence`][TokenType::HeadingAtxSequence].
@@ -944,7 +953,7 @@ fn on_exit_heading_atx_sequence(context: &mut CompileContext) {
         )
         .len();
         context.atx_opening_sequence_size = Some(rank);
-        context.push(format!("", rank));
+        context.tag(format!("", rank));
     }
 }
 
@@ -973,7 +982,9 @@ fn on_exit_heading_setext_underline(context: &mut CompileContext) {
     )[0];
     let level: usize = if head == Code::Char('-') { 2 } else { 1 };
 
-    context.push(format!("{}", level, text, level));
+    context.tag(format!("", level));
+    context.push(text);
+    context.tag(format!("", level));
 }
 
 /// Handle [`Exit`][EventType::Exit]:{[`HtmlFlow`][TokenType::HtmlFlow],[`HtmlText`][TokenType::HtmlText]}.
@@ -988,7 +999,6 @@ fn on_exit_html_data(context: &mut CompileContext) {
         &from_exit_event(context.events, context.index),
         false,
     );
-    // last_was_tag = false;
     context.push(context.encode_opt(&slice));
 }
 
@@ -1029,6 +1039,7 @@ fn on_exit_line_ending(context: &mut CompileContext) {
 fn on_exit_media(context: &mut CompileContext) {
     let mut is_in_image = false;
     let mut index = 0;
+
     // Skip current.
     while index < (context.media_stack.len() - 1) {
         if context.media_stack[index].image {
@@ -1038,7 +1049,7 @@ fn on_exit_media(context: &mut CompileContext) {
         index += 1;
     }
 
-    // context.tags = is_in_image;
+    context.tags = !is_in_image;
 
     let media = context.media_stack.pop().unwrap();
     let id = media
@@ -1070,28 +1081,27 @@ fn on_exit_media(context: &mut CompileContext) {
         "".to_string()
     };
 
-    let result = if media.image {
-        format!(
-            "\"{}\"{}",
+    if media.image {
+        context.tag(format!(
+            "\"",", title));
     } else {
-        format!(
-            "{}",
+        context.tag(format!(
+            "",
             sanitize_uri(&destination, &context.protocol_href),
             title,
-            label
-        )
+        ));
+        context.push(label);
+        context.tag("".to_string());
     };
-
-    context.push(result);
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`Paragraph`][TokenType::Paragraph].
 fn on_exit_paragraph(context: &mut CompileContext) {
-    context.push("
".to_string());
+    context.tag("
".to_string());
 }
 
 /// Handle [`Exit`][EventType::Exit]:[`ReferenceString`][TokenType::ReferenceString].
@@ -1123,5 +1133,5 @@ fn on_exit_resource_title_string(context: &mut CompileContext) {
 
 /// Handle [`Exit`][EventType::Exit]:[`ThematicBreak`][TokenType::ThematicBreak].
 fn on_exit_thematic_break(context: &mut CompileContext) {
-    context.push("
".to_string());
+    context.tag("
".to_string());
 }
diff --git a/tests/heading_setext.rs b/tests/heading_setext.rs
index 92a5b43..ecf22a8 100644
--- a/tests/heading_setext.rs
+++ b/tests/heading_setext.rs
@@ -3,28 +3,28 @@ use micromark::micromark;
 
 #[test]
 fn heading_setext() {
-    // To do: emphasis.
+    // To do: attention.
     // assert_eq!(
     //     micromark("Foo *bar*\n========="),
     //     "
Foo bar
",
     //     "should support a heading w/ an equals to (rank of 1)"
     // );
 
-    // To do: emphasis.
+    // To do: attention.
     // assert_eq!(
     //     micromark("Foo *bar*\n---------"),
     //     "
Foo bar
",
     //     "should support a heading w/ a dash (rank of 2)"
     // );
 
-    // To do: emphasis.
+    // To do: attention.
     // assert_eq!(
     //     micromark("Foo *bar\nbaz*\n===="),
     //     "
Foo bar\nbaz
",
     //     "should support line endings in setext headings"
     // );
 
-    // To do: emphasis, trim.
+    // To do: attention, trim.
     // assert_eq!(
     //     micromark("  Foo *bar\nbaz*\t\n===="),
     //     "
Foo bar\nbaz
",
diff --git a/tests/image.rs b/tests/image.rs
index 9be056f..68b9717 100644
--- a/tests/image.rs
+++ b/tests/image.rs
@@ -14,35 +14,33 @@ fn image() {
         "should support image w/ resource"
     );
 
-    // To do: attention, tags in images.
+    // To do: attention.
     // assert_eq!(
     //     micromark("[foo *bar*]: train.jpg \"train & tracks\"\n\n![foo *bar*]"),
     //     "
\"foo
",
     //     "should support image as shortcut reference"
     // );
 
-    // To do: tags in images.
-    // assert_eq!(
-    //     micromark("![foo ![bar](/url)](/url2)"),
-    //     "
\"foo
",
-    //     "should “support” images in images"
-    // );
+    assert_eq!(
+        micromark("![foo ![bar](/url)](/url2)"),
+        "
\"foo
",
+        "should “support” images in images"
+    );
 
-    // To do: tags in images.
-    // assert_eq!(
-    //     micromark("![foo [bar](/url)](/url2)"),
-    //     "
\"foo
",
-    //     "should “support” links in images"
-    // );
+    assert_eq!(
+        micromark("![foo [bar](/url)](/url2)"),
+        "
\"foo
",
+        "should “support” links in images"
+    );
 
-    // To do: attention, tags in images.
+    // To do: attention.
     // assert_eq!(
     //     micromark("[foo *bar*]: train.jpg \"train & tracks\"\n\n![foo *bar*][]"),
     //     "
\"foo
",
     //     "should support “content” in images"
     // );
 
-    // To do: attention, tags in images.
+    // To do: attention.
     // assert_eq!(
     //     micromark("[FOOBAR]: train.jpg \"train & tracks\"\n\n![foo *bar*][foobar]"),
     //     "
\"foo
",
@@ -91,7 +89,7 @@ fn image() {
         "should support collapsed references (1)"
     );
 
-    // To do: attention, tags in images.
+    // To do: attention.
     // assert_eq!(
     //     micromark("[*foo* bar]: /url \"title\"\n\n![*foo* bar][]"),
     //     "
\"foo
",
@@ -117,7 +115,7 @@ fn image() {
         "should support shortcut references (1)"
     );
 
-    // To do: attention, tags in images.
+    // To do: attention.
     // assert_eq!(
     //     micromark("[*foo* bar]: /url \"title\"\n\n![*foo* bar]"),
     //     "
\"foo
",
diff --git a/tests/link_resource.rs b/tests/link_resource.rs
index 992c7d2..d75736e 100644
--- a/tests/link_resource.rs
+++ b/tests/link_resource.rs
@@ -257,12 +257,11 @@ fn link_resource() {
     //     "should not support links in links (2)"
     // );
 
-    // To do: tags in images.
-    // assert_eq!(
-    //     micromark("![[[foo](uri1)](uri2)](uri3)"),
-    //     "
\"[foo](uri2)\"
",
-    //     "should not support links in links (3)"
-    // );
+    assert_eq!(
+        micromark("![[[foo](uri1)](uri2)](uri3)"),
+        "
\"[foo](uri2)\"
",
+        "should not support links in links (3)"
+    );
 
     assert_eq!(
         micromark("*[foo*](/uri)"),
-- 
cgit