fix: handle fence indents in md processor

.size-limit.json

@@ -33,7 +33,7 @@
       "build/globals.js",
       "build/deno.js"
     ],
-    "limit": "849.55 kB",
+    "limit": "849.90 kB",
     "brotli": false,
     "gzip": false
   },
@@ -66,7 +66,7 @@
       "README.md",
       "LICENSE"
     ],
-    "limit": "911.30 kB",
+    "limit": "911.60 kB",
     "brotli": false,
     "gzip": false
   }

build/cli.cjs

@@ -62,75 +62,78 @@ var import_util2 = require("./util.cjs");
 var import_util = require("./util.cjs");
 function transformMarkdown(buf) {
   var _a2;
-  const output = [];
+  const out = [];
   const tabRe = /^(  +|\t)/;
-  const codeBlockRe = new RegExp("^(?<fence>(`{3,20}|~{3,20}))(?:(?<js>(js|javascript|ts|typescript))|(?<bash>(sh|shell|bash))|.*)$");
+  const fenceRe = new RegExp("^(?<indent> {0,3})(?<fence>(`{3,20}|~{3,20}))(?:(?<js>js|javascript|ts|typescript)|(?<bash>sh|shell|bash)|.*)$");
   let state = "root";
-  let codeBlockEnd = "";
-  let prevLineIsEmpty = true;
+  let prevEmpty = true;
+  let fenceChar = "";
+  let stripRe = /^/;
+  let endRe = /^$/;
+  let linePrefix = "";
+  let closeOut = "";
+  let closeBlank = false;
+  const isEnd = (s) => fenceChar && endRe.test(s);
   for (const line of (0, import_util.bufToString)(buf).split(/\r?\n/)) {
     switch (state) {
-      case "root":
-        if (tabRe.test(line) && prevLineIsEmpty) {
-          output.push(line);
-          state = "tab";
-          continue;
+      case "root": {
+        const g = (_a2 = line.match(fenceRe)) == null ? void 0 : _a2.groups;
+        if (g == null ? void 0 : g.fence) {
+          fenceChar = g.fence[0];
+          stripRe = g.indent ? new RegExp(`^ {0,${g.indent.length}}`) : /^/;
+          endRe = new RegExp(`^ {0,3}${fenceChar}{${g.fence.length},}[ \\t]*$`);
+          if (g.js) {
+            out.push("");
+            linePrefix = "";
+            closeOut = "";
+            closeBlank = true;
+          } else if (g.bash) {
+            out.push("await $`");
+            linePrefix = "";
+            closeOut = "`";
+            closeBlank = false;
+          } else {
+            out.push("");
+            linePrefix = "// ";
+            closeOut = "";
+            closeBlank = true;
+          }
+          state = "fence";
+          prevEmpty = false;
+          break;
         }
-        const { fence, js, bash } = ((_a2 = line.match(codeBlockRe)) == null ? void 0 : _a2.groups) || {};
-        if (!fence) {
-          prevLineIsEmpty = line === "";
-          output.push("// " + line);
+        if (prevEmpty && tabRe.test(line)) {
+          out.push(line);
+          state = "tab";
           continue;
         }
-        codeBlockEnd = fence;
-        if (js) {
-          state = "js";
-          output.push("");
-        } else if (bash) {
-          state = "bash";
-          output.push("await $`");
-        } else {
-          state = "other";
-          output.push("");
-        }
-        break;
+        prevEmpty = line === "";
+        out.push("// " + line);
+        continue;
+      }
       case "tab":
-        if (line === "") {
-          output.push("");
-        } else if (tabRe.test(line)) {
-          output.push(line);
-        } else {
-          output.push("// " + line);
+        if (line === "") out.push("");
+        else if (tabRe.test(line)) out.push(line);
+        else {
+          out.push("// " + line);
           state = "root";
         }
+        prevEmpty = line === "";
         break;
-      case "js":
-        if (line === codeBlockEnd) {
-          output.push("");
-          state = "root";
-        } else {
-          output.push(line);
-        }
-        break;
-      case "bash":
-        if (line === codeBlockEnd) {
-          output.push("`");
-          state = "root";
-        } else {
-          output.push(line);
-        }
-        break;
-      case "other":
-        if (line === codeBlockEnd) {
-          output.push("");
+      case "fence":
+        if (isEnd(line)) {
+          if (closeOut) out.push(closeOut);
+          else if (closeBlank) out.push("");
           state = "root";
+          prevEmpty = true;
         } else {
-          output.push("// " + line);
+          out.push(linePrefix + line.replace(stripRe, ""));
+          prevEmpty = false;
         }
         break;
     }
   }
-  return output.join("\n");
+  return out.join("\n");
 }
 
 // src/cli.ts

src/md.ts

@@ -16,74 +16,88 @@
 import { bufToString } from './util.ts'
 
 export function transformMarkdown(buf: Buffer | string): string {
-  const output = []
+  const out: string[] = []
   const tabRe = /^(  +|\t)/
-  const codeBlockRe =
-    /^(?<fence>(`{3,20}|~{3,20}))(?:(?<js>(js|javascript|ts|typescript))|(?<bash>(sh|shell|bash))|.*)$/
+  const fenceRe =
+    /^(?<indent> {0,3})(?<fence>(`{3,20}|~{3,20}))(?:(?<js>js|javascript|ts|typescript)|(?<bash>sh|shell|bash)|.*)$/
+
   let state = 'root'
-  let codeBlockEnd = ''
-  let prevLineIsEmpty = true
+  let prevEmpty = true
+
+  let fenceChar = ''
+  let stripRe = /^/
+  let endRe = /^$/
+  let linePrefix = ''
+  let closeOut = ''
+  let closeBlank = false
+
+  const isEnd = (s: string) => fenceChar && endRe.test(s)
+
   for (const line of bufToString(buf).split(/\r?\n/)) {
     switch (state) {
-      case 'root':
-        if (tabRe.test(line) && prevLineIsEmpty) {
-          output.push(line)
-          state = 'tab'
-          continue
+      case 'root': {
+        const g = line.match(fenceRe)?.groups
+        if (g?.fence) {
+          fenceChar = g.fence[0]
+          stripRe = g.indent ? new RegExp(`^ {0,${g.indent.length}}`) : /^/
+          endRe = new RegExp(`^ {0,3}${fenceChar}{${g.fence.length},}[ \\t]*$`)
+
+          if (g.js) {
+            out.push('')
+            linePrefix = ''
+            closeOut = ''
+            closeBlank = true
+          } else if (g.bash) {
+            out.push('await $`')
+            linePrefix = ''
+            closeOut = '`'
+            closeBlank = false
+          } else {
+            out.push('')
+            linePrefix = '// '
+            closeOut = ''
+            closeBlank = true
+          }
+
+          state = 'fence'
+          prevEmpty = false
+          break
         }
-        const { fence, js, bash } = line.match(codeBlockRe)?.groups || {}
-        if (!fence) {
-          prevLineIsEmpty = line === ''
-          output.push('// ' + line)
+
+        if (prevEmpty && tabRe.test(line)) {
+          out.push(line)
+          state = 'tab'
           continue
         }
-        codeBlockEnd = fence
-        if (js) {
-          state = 'js'
-          output.push('')
-        } else if (bash) {
-          state = 'bash'
-          output.push('await $`')
-        } else {
-          state = 'other'
-          output.push('')
-        }
-        break
+
+        prevEmpty = line === ''
+        out.push('// ' + line)
+        continue
+      }
+
       case 'tab':
-        if (line === '') {
-          output.push('')
-        } else if (tabRe.test(line)) {
-          output.push(line)
-        } else {
-          output.push('// ' + line)
+        if (line === '') out.push('')
+        else if (tabRe.test(line)) out.push(line)
+        else {
+          out.push('// ' + line)
           state = 'root'
         }
+        prevEmpty = line === ''
         break
-      case 'js':
-        if (line === codeBlockEnd) {
-          output.push('')
-          state = 'root'
-        } else {
-          output.push(line)
-        }
-        break
-      case 'bash':
-        if (line === codeBlockEnd) {
-          output.push('`')
-          state = 'root'
-        } else {
-          output.push(line)
-        }
-        break
-      case 'other':
-        if (line === codeBlockEnd) {
-          output.push('')
+
+      case 'fence':
+        if (isEnd(line)) {
+          if (closeOut) out.push(closeOut)
+          else if (closeBlank) out.push('')
           state = 'root'
+          prevEmpty = true
         } else {
-          output.push('// ' + line)
+          out.push(linePrefix + line.replace(stripRe, ''))
+          prevEmpty = false
         }
         break
     }
   }
-  return output.join('\n')
+
+  return out.join('\n')
 }

test/md.test.ts

@@ -12,22 +12,34 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-import { test, describe } from 'node:test'
+import { describe, test } from 'node:test'
 import assert from 'node:assert'
 import { transformMarkdown } from '../src/md.ts'
 
-describe('md', () => {
-  test('transformMarkdown()', () => {
-    assert.equal(transformMarkdown('\n'), '// \n// ')
-    assert.equal(transformMarkdown('  \n    '), '  \n    ')
-    assert.equal(
-      transformMarkdown(`
+describe('transformMarkdown()', () => {
+  describe('root handling', () => {
+    test('comments out plain lines (including empty line)', () => {
+      assert.equal(transformMarkdown('\n'), '// \n// ')
+    })
+
+    test('preserves tab-indented blocks after a blank line (legacy behavior)', () => {
+      assert.equal(transformMarkdown('  \n    '), '  \n    ')
+    })
+
+    test('does not treat a mid-paragraph fence as a fenced block (legacy behavior)', () => {
+      assert.equal(
+        transformMarkdown(`
 \t~~~js
 console.log('js')`),
-      `// \n\t~~~js\n// console.log('js')`
-    )
-    // prettier-ignore
-    assert.equal(transformMarkdown(`
+        `// \n\t~~~js\n// console.log('js')`
+      )
+    })
+  })
+
+  describe('fenced code blocks', () => {
+    test('converts js/ts to raw code, bash to await $`...` and comments unknown fences', () => {
+      // prettier-ignore
+      assert.equal(transformMarkdown(`
 # Title
 
 ~~~js
@@ -68,5 +80,40 @@ echo foo
 \`
 // 
 // `)
+    })
+
+    test('accepts fences indented up to 3 spaces (CommonMark) and converts them', () => {
+      const input = `# h1
+
+paragraph
+
+## h2
+
+### h3
+
+\`\`\`bash
+echo "1"
+\`\`\`
+
+### h3
+
+- item 1
+
+   \`\`\`bash
+   echo "2"
+   \`\`\`
+
+### h3
+
+\`\`\`bash
+echo "4"
+\`\`\`
+`
+      const result = transformMarkdown(input)
+
+      assert.ok(!/```|~~~/.test(result), 'no raw markdown fences should remain')
+      assert.equal((result.match(/await \$`/g) ?? []).length, 3)
+      assert.equal((result.match(/^`$/gm) ?? []).length, 3)
+    })
   })
 })