Switch to ansi_up for ansi rendering in actions (#25401)

Fixes: https://github.com/go-gitea/gitea/issues/24777

4 files changed, 67 insertions, 76 deletions

diff --git a/web_src/js/components/RepoActionView.test.js b/web_src/js/components/RepoActionView.test.js
deleted file mode 100644
index 011c20d8d1..0000000000
--- a/web_src/js/components/RepoActionView.test.js
+++ /dev/null
@@ -1,30 +0,0 @@
-import {expect, test} from 'vitest';
-
-import {ansiLogToHTML} from './RepoActionView.vue';
-import AnsiToHTML from 'ansi-to-html';
-
-test('processConsoleLine', () => {
-  expect(ansiLogToHTML('abc')).toEqual('abc');
-  expect(ansiLogToHTML('abc\n')).toEqual('abc');
-  expect(ansiLogToHTML('abc\r\n')).toEqual('abc');
-  expect(ansiLogToHTML('\r')).toEqual('');
-  expect(ansiLogToHTML('\rx\rabc')).toEqual('x\nabc');
-  expect(ansiLogToHTML('\rabc\rx\r')).toEqual('abc\nx');
-
-  expect(ansiLogToHTML('\x1b[30mblack\x1b[37mwhite')).toEqual('<span style="color:#000">black<span style="color:#AAA">white</span></span>');
-  expect(ansiLogToHTML('<script>')).toEqual('&lt;script&gt;');
-
-
-  // upstream AnsiToHTML has bugs when processing "\033[1A" and "\033[1B", we fixed these control sequences in our code
-  // if upstream could fix these bugs, we can remove these tests and remove our patch code
-  const ath = new AnsiToHTML({escapeXML: true});
-  expect(ath.toHtml('\x1b[1A\x1b[2Ktest\x1b[1B\x1b[1A\x1b[2K')).toEqual('AtestBA'); // AnsiToHTML bug
-  expect(ath.toHtml('\x1b[1A\x1b[2K\rtest\r\x1b[1B\x1b[1A\x1b[2K')).toEqual('A\rtest\rBA'); // AnsiToHTML bug
-
-  // test our patched behavior
-  expect(ansiLogToHTML('\x1b[1A\x1b[2Ktest\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
-  expect(ansiLogToHTML('\x1b[1A\x1b[2K\rtest\r\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
-
-  // treat "\033[0K" and "\033[0J" (Erase display/line) as "\r", then it will be covered to "\n" finally.
-  expect(ansiLogToHTML('a\x1b[Kb\x1b[2Jc')).toEqual('a\nb\nc');
-});
diff --git a/web_src/js/components/RepoActionView.vue b/web_src/js/components/RepoActionView.vue
index ebc78b049b..fd40bd7ee5 100644
--- a/web_src/js/components/RepoActionView.vue
+++ b/web_src/js/components/RepoActionView.vue
@@ -119,14 +119,12 @@
 import {SvgIcon} from '../svg.js';
 import ActionRunStatus from './ActionRunStatus.vue';
 import {createApp} from 'vue';
-import AnsiToHTML from 'ansi-to-html';
 import {toggleElem} from '../utils/dom.js';
 import {getCurrentLocale} from '../utils.js';
+import {renderAnsi} from '../render/ansi.js';
 
 const {csrfToken} = window.config;
 
-const ansiLogRender = new AnsiToHTML({escapeXML: true});
-
 const sfc = {
   name: 'RepoActionView',
   components: {
@@ -304,7 +302,7 @@ const sfc = {
 
       const logMessage = document.createElement('span');
       logMessage.className = 'log-msg';
-      logMessage.innerHTML = ansiLogToHTML(line.message);
+      logMessage.innerHTML = renderAnsi(line.message);
       div.append(logTimeStamp);
       div.append(logMessage);
       div.append(logTimeSeconds);
@@ -470,48 +468,6 @@ export function initRepositoryActionView() {
   view.mount(el);
 }
 
-// some unhandled control sequences by AnsiToHTML
-// https://man7.org/linux/man-pages/man4/console_codes.4.html
-const ansiRegexpRemove = /\x1b\[\d+[A-H]/g; // Move cursor, treat them as no-op.
-const ansiRegexpNewLine = /\x1b\[\d?[JK]/g; // Erase display/line, treat them as a Carriage Return
-
-function ansiCleanControlSequences(line) {
-  if (line.includes('\x1b')) {
-    line = line.replace(ansiRegexpRemove, '');
-    line = line.replace(ansiRegexpNewLine, '\r');
-  }
-  return line;
-}
-
-export function ansiLogToHTML(line) {
-  if (line.endsWith('\r\n')) {
-    line = line.substring(0, line.length - 2);
-  } else if (line.endsWith('\n')) {
-    line = line.substring(0, line.length - 1);
-  }
-
-  // usually we do not need to process control chars like "\033[", let AnsiToHTML do it
-  // but AnsiToHTML has bugs, so we need to clean some control sequences first
-  line = ansiCleanControlSequences(line);
-
-  if (!line.includes('\r')) {
-    return ansiLogRender.toHtml(line);
-  }
-
-  // handle "\rReading...1%\rReading...5%\rReading...100%",
-  // convert it into a multiple-line string: "Reading...1%\nReading...5%\nReading...100%"
-  const lines = [];
-  for (const part of line.split('\r')) {
-    if (part === '') continue;
-    const partHtml = ansiLogRender.toHtml(part);
-    if (partHtml !== '') {
-      lines.push(partHtml);
-    }
-  }
-  // the log message element is with "white-space: break-spaces;", so use "\n" to break lines
-  return lines.join('\n');
-}
-
 </script>
 
 <style scoped>
diff --git a/web_src/js/render/ansi.js b/web_src/js/render/ansi.js
new file mode 100644
index 0000000000..68e8f2c544
--- /dev/null
+++ b/web_src/js/render/ansi.js
@@ -0,0 +1,44 @@
+import AnsiUp from 'ansi_up';
+
+const replacements = [
+  [/\x1b\[\d+[A-H]/g, ''], // Move cursor, treat them as no-op
+  [/\x1b\[\d?[JK]/g, '\r'], // Erase display/line, treat them as a Carriage Return
+];
+
+// render ANSI to HTML
+export function renderAnsi(line) {
+  // create a fresh ansi_up instance because otherwise previous renders can influence
+  // the output of future renders, because ansi_up is stateful and remembers things like
+  // unclosed opening tags for colors.
+  const ansi_up = new (AnsiUp.default || AnsiUp)();
+
+  if (line.endsWith('\r\n')) {
+    line = line.substring(0, line.length - 2);
+  } else if (line.endsWith('\n')) {
+    line = line.substring(0, line.length - 1);
+  }
+
+  if (line.includes('\x1b')) {
+    for (const [regex, replacement] of replacements) {
+      line = line.replace(regex, replacement);
+    }
+  }
+
+  if (!line.includes('\r')) {
+    return ansi_up.ansi_to_html(line);
+  }
+
+  // handle "\rReading...1%\rReading...5%\rReading...100%",
+  // convert it into a multiple-line string: "Reading...1%\nReading...5%\nReading...100%"
+  const lines = [];
+  for (const part of line.split('\r')) {
+    if (part === '') continue;
+    const partHtml = ansi_up.ansi_to_html(part);
+    if (partHtml !== '') {
+      lines.push(partHtml);
+    }
+  }
+
+  // the log message element is with "white-space: break-spaces;", so use "\n" to break lines
+  return lines.join('\n');
+}
diff --git a/web_src/js/render/ansi.test.js b/web_src/js/render/ansi.test.js
new file mode 100644
index 0000000000..7542ba0ede
--- /dev/null
+++ b/web_src/js/render/ansi.test.js
@@ -0,0 +1,21 @@
+import {expect, test} from 'vitest';
+import {renderAnsi} from './ansi.js';
+
+test('renderAnsi', () => {
+  expect(renderAnsi('abc')).toEqual('abc');
+  expect(renderAnsi('abc\n')).toEqual('abc');
+  expect(renderAnsi('abc\r\n')).toEqual('abc');
+  expect(renderAnsi('\r')).toEqual('');
+  expect(renderAnsi('\rx\rabc')).toEqual('x\nabc');
+  expect(renderAnsi('\rabc\rx\r')).toEqual('abc\nx');
+  expect(renderAnsi('\x1b[30mblack\x1b[37mwhite')).toEqual('<span style="color:rgb(0,0,0)">black</span><span style="color:rgb(255,255,255)">white</span>'); // unclosed
+  expect(renderAnsi('<script>')).toEqual('&lt;script&gt;');
+  expect(renderAnsi('\x1b[1A\x1b[2Ktest\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
+  expect(renderAnsi('\x1b[1A\x1b[2K\rtest\r\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
+  expect(renderAnsi('\x1b[1A\x1b[2Ktest\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
+  expect(renderAnsi('\x1b[1A\x1b[2K\rtest\r\x1b[1B\x1b[1A\x1b[2K')).toEqual('test');
+
+  // treat "\033[0K" and "\033[0J" (Erase display/line) as "\r", then it will be covered to "\n" finally.
+  expect(renderAnsi('a\x1b[Kb\x1b[2Jc')).toEqual('a\nb\nc');
+  expect(renderAnsi('\x1b[48;5;88ma\x1b[38;208;48;5;159mb\x1b[m')).toEqual(`<span style="background-color:rgb(135,0,0)">a</span><span style="background-color:rgb(175,255,255)">b</span>`);
+});