Linewrap PR description to 72 chars automatically (#923)

## Summary * Linewrap the PR description to 72 chars automatically when merging a PR * Codeblocks (```` ``` ````, `~~~` seperated or indented with 4 or more spaces, see https://github.github.com/gfm/#indented-code-block ) are not linewrapped, and codespans `like this` are treated as single words ## Details * Remove the manual 72 char requirement and merge the top comment into the comment below the section break * Restrict the removal of html comment introduced in a862810 to non-codeblocks --------- Co-authored-by: Saem Ghani <saemghani+github@gmail.com>
nim-works · Sep 26, 2023 · 4cd20fa · 4cd20fa
1 parent 11751e9
commit 4cd20fa
Show file tree

Hide file tree

Showing 4 changed files with 232 additions and 6 deletions.
diff --git a/.github/pull_request_template.md b/.github/pull_request_template.md
@@ -1,7 +1,3 @@
-<!--- The Pull Request (=PR) message is what will get automatically used as
-the commit message when the PR is merged. Make sure that no line is longer
-than 72 characters -->
-
 ## Summary
 * what changed and how?
 * why are we changing it?
@@ -14,7 +10,8 @@ than 72 characters -->
 Fixes full_issue_url
 
 ---
-<!-- Note: section break (`---`) onwards is not in CI merge commit -->
+<!--- Anything before this section break (`---`) will be used as
+the commit message when the PR is merged. -->
 
 ## Notes for Reviewers
 * leave additional context for reviewers

diff --git a/.github/workflows/slash-command-merge.yml b/.github/workflows/slash-command-merge.yml
@@ -15,6 +15,10 @@ jobs:
     name: Trim description and merge
     runs-on: ubuntu-latest
     steps:
+      - uses: actions/checkout@v4
+        with:
+          sparse-checkout: |
+            tools/ci_format_pr_description.js
       - id: pr-data
         name: Get PR data and trim PR body
         uses: actions/github-script@v6
@@ -51,8 +55,10 @@ jobs:
               return;
             }
 
+            const formatPR = require('../../tools/ci_format_pr_description.js');
+
             let [ body, body_tail ] = pr_info.body.split(/^---\s*$/m, 2);
-            body = body.replace(/<\!--.*?-->/s, "");
+            body = formatPR.formatPRdescription(body, 72);
             body = body.trimEnd();
             body_tail = body_tail?.trimEnd() || '';
             // Only runs when the PR is still open and is not staged for merging.

diff --git a/tests/tools/tci_format_pr_description.nim b/tests/tools/tci_format_pr_description.nim
@@ -0,0 +1,170 @@
+discard """
+description: "Test the PR description formatter"
+joinable: false
+target: js
+"""
+
+import std/jsffi
+
+let formatPR = require "../../tools/ci_format_pr_description.js"
+
+proc formatPRdescription(formatPR: JSObject, str: cstring, maxLineLength: int): cstring {.importjs.}
+
+let testingText = cstring"""Lorem ipsum dolor sit amet, consectetur i adipiscing elit. Nullam blandit mauris id venenatis tincidunt. Vestibulum at gravida sapien. Mauris tellus augue, aliquet sed laoreet blandit, pulvinar sed felis. Phasellus nec est vitae enim blandit facilisis.
+Vestibulum fermentum ligula sit amet volutpat fermentum. Sed in faucibus orci. Pellentesque a dui ex. Curabitur sollicitudin, nulla id dignissim lacinia, odio mauris blandit nisi, eget auctor arcu odio nec est.
+
+=========================================
+
+## Summary
+* The quick brown fox now jumps over the lazy dog
+
+
+## Details
+* THE QUICK BROWN FOX JUMPED OVER THE LAZY DOG'S BACK 1234567890
+* Hamburgevons
+
+<!--This HTML comment should be removed-->
+
+```
+this is a triple quote delimited codeblock:
+  it should not be linewrapped as should no other style of codeblock be
+    there are also some birds sitting on a html comment:
+                                       ___     ___
+                                      (o o)   (o o)
+                                     (  V  ) (  V  )
+                             <!--------m-m-----m-m--------->
+```
+Here's some text.
+ And some indented text, that will be linewrapped,
+  but must not be dedented
+
+<!--- This HTML comment
+ spanning multiple lines should be removed
+ too! -->
+
+    and this is an indented codeblock:
+      with a cat:
+        !"#%&'()*+,-./012  _._     _,-'""`-._      3456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefgh
+        "#%&'()*+,-./0123 (,-.'._,'(       |\`-/|  456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghi
+        #%&'()*+,-./01234     '-.-' \ )-`( , o o)  56789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghij
+        %&'()*+,-./012345           '-    \`_`"'-  6789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghijk
+                       <!--and-a-html-comment-too--->
+
+
+
+https://www.some.url/that/is/quite/long/abcdefghijklmnopqrstuvw123467890xyz
+
+(test) =========================================
+(test) ======================================= i
+
+
+   ~~~tokipona
+<!-This is important!-->
+
+	jan ali li kama lon nasin ni: ona li ken tawa li ken pali.
+	jan ali li kama lon sama. jan ali li jo e ken pi pilin suli.
+	jan ali li ken pali e wile pona ona.
+	jan ali li jo e ken pi sona pona e ken pi pali pona.
+	jan ali li wile pali nasin ni: ona li jan pona pi jan ante.
+
+~~~
+
+=================================== split this
+================================= `keep this`
+
+there are trailing spaces after this sentence      
+and they must not leak into this line
+
+the next line is indented
+  and it should stay this way
+"""
+
+let testingTextWrapped = cstring"""Lorem ipsum dolor sit amet, consectetur i
+adipiscing elit. Nullam blandit mauris id
+venenatis tincidunt. Vestibulum at
+gravida sapien. Mauris tellus augue,
+aliquet sed laoreet blandit, pulvinar sed
+felis. Phasellus nec est vitae enim
+blandit facilisis.
+Vestibulum fermentum ligula sit amet
+volutpat fermentum. Sed in faucibus orci.
+Pellentesque a dui ex. Curabitur
+sollicitudin, nulla id dignissim lacinia,
+odio mauris blandit nisi, eget auctor
+arcu odio nec est.
+
+=========================================
+
+## Summary
+* The quick brown fox now jumps over the
+lazy dog
+
+
+## Details
+* THE QUICK BROWN FOX JUMPED OVER THE
+LAZY DOG'S BACK 1234567890
+* Hamburgevons
+
+
+
+```
+this is a triple quote delimited codeblock:
+  it should not be linewrapped as should no other style of codeblock be
+    there are also some birds sitting on a html comment:
+                                       ___     ___
+                                      (o o)   (o o)
+                                     (  V  ) (  V  )
+                             <!--------m-m-----m-m--------->
+```
+Here's some text.
+ And some indented text, that will be
+linewrapped,
+  but must not be dedented
+
+
+
+    and this is an indented codeblock:
+      with a cat:
+        !"#%&'()*+,-./012  _._     _,-'""`-._      3456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefgh
+        "#%&'()*+,-./0123 (,-.'._,'(       |\`-/|  456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghi
+        #%&'()*+,-./01234     '-.-' \ )-`( , o o)  56789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghij
+        %&'()*+,-./012345           '-    \`_`"'-  6789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_abcdefghijk
+                       <!--and-a-html-comment-too--->
+
+
+
+https://www.some.url/that/is/quite/long/abcdefghijklmnopqrstuvw123467890xyz
+
+(test)
+=========================================
+(test)
+======================================= i
+
+
+   ~~~tokipona
+<!-This is important!-->
+
+	jan ali li kama lon nasin ni: ona li ken tawa li ken pali.
+	jan ali li kama lon sama. jan ali li jo e ken pi pilin suli.
+	jan ali li ken pali e wile pona ona.
+	jan ali li jo e ken pi sona pona e ken pi pali pona.
+	jan ali li wile pali nasin ni: ona li jan pona pi jan ante.
+
+~~~
+
+=================================== split
+this
+================================= 
+`keep this`
+
+there are trailing spaces after this
+sentence      
+and they must not leak into this line
+
+the next line is indented
+  and it should stay this way
+"""
+
+echo formatPR.formatPRdescription(testingText, 41)
+assert formatPR.formatPRdescription(testingText, 41) == testingTextWrapped
+
diff --git a/tools/ci_format_pr_description.js b/tools/ci_format_pr_description.js
@@ -0,0 +1,53 @@
+// This file is imported by .github/workflows/slash-command-merge.yml
+// and tested by tests/tools/tci_forrmat_pr_description.nim
+
+function getSegments(regex, str) {
+  // Returns the segments that `regex`
+  // divides `str` into, along with the
+  // info of wether each segment was matching
+  // the regex or not
+  let segments = [];
+  let lastIndex = 0;
+  let match;
+  regex.lastIndex = 0; // In case there's a dangling previous search
+  while (match = regex.exec(str)) {
+    if (match.index > lastIndex) {
+      segments.push([false, str.substring(lastIndex, match.index)]);
+    }
+    segments.push([true, match[0]]);
+    lastIndex = match.index + match[0].length;
+  }
+  if (lastIndex < str.length) {
+    segments.push([false, str.substring(lastIndex)]);
+  }
+  return segments;
+}
+
+exports.formatPRdescription = (text, maxLineLength) =>
+  // Returns `text` stripped of HTML comments and linewrapped,
+  // respecting codeblocks and codespans
+  getSegments(/(?:^ {0,3}(```|~~~)[^]*?^ {0,3}\1 *$)|^(?:\n {4,}.*)+/gm, text)
+  .map( ([isCodeBlock, segment]) =>
+    isCodeBlock
+    ? segment // Don't touch code blocks
+    : segment
+      // Remove HTML comments first:
+      .replace(/<\!--.*?-->/s, "")
+      .split('\n').map(line =>
+        line.length <= maxLineLength
+        ? line // Line isn't too long, nothing to do
+        : getSegments(/(`+).*?\1/g, line)
+          // Treat codespans as single words:
+          .flatMap( ([isCodeSpan, segment]) => isCodeSpan ? segment : segment.split(' ') )
+          .reduce( ([wrappedLine, lineLength, isStart], word) =>
+            // Don't touch indentation nor start with a line break:
+            isStart
+            ? [wrappedLine + word, lineLength + word.length, false]
+            : // Don't touch trailing spaces:
+              word.length > 0 && lineLength + 1 + word.length > maxLineLength
+              ? [wrappedLine + '\n' + word, word.length, false]
+              : [wrappedLine + ' ' + word, lineLength + 1 + word.length, false]
+            , ['', 0, true]
+          )[0]
+      ).join('\n')
+  ).join('')