fix: improve changelog generation for non-tagged commits and edge cases

CHANGELOG.md

@@ -52,6 +52,10 @@ To update your existing Danger workflows:
 
 - Updater - Prevent script injection vulnerabilities through workflow inputs ([#98](https://github.com/getsentry/github-workflows/pull/98))
 
+### Fixes
+
+- Improve changelog generation for non-tagged commits and edge cases ([#115](https://github.com/getsentry/github-workflows/pull/115))
+
 ## 2.14.1
 
 ### Fixes

updater/scripts/get-changelog.ps1

@@ -5,122 +5,154 @@ param(
 )
 
 Set-StrictMode -Version latest
+$PSNativeCommandErrorActionPreference = $true
+$ErrorActionPreference = 'Stop'
 
 $prefix = 'https?://(www\.)?github.com/'
-if (-not ($RepoUrl -match "^$prefix"))
-{
-    Write-Warning "Only github.com repositories are currently supported. Given RepoUrl doesn't look like one: $RepoUrl"
+if (-not ($RepoUrl -match "^$prefix([^/]+)/([^/]+?)(?:\.git)?/?$")) {
+    Write-Warning "Only https://github.com repositories are currently supported. Could not parse repository from URL: $RepoUrl"
     return
 }
 
+$repoOwner = $matches[2]
+$repoName = $matches[3]
+$apiRepo = "$repoOwner/$repoName"
+
+# Create temporary directory for changelog files
 $tmpDir = Join-Path ([System.IO.Path]::GetTempPath()) ([System.Guid]::NewGuid())
 New-Item -ItemType Directory $tmpDir | Out-Null
 
-try
-{
-    git clone --depth 1 $RepoUrl $tmpDir
+# Function to try different changelog filenames
+function Get-ChangelogContent {
+    param($ref, $filePath)
+
+    $changelogNames = @('CHANGELOG.md', 'changelog.md', 'CHANGELOG.txt', 'changelog.txt', 'CHANGELOG')
+
+    foreach ($name in $changelogNames) {
+        try {
+            # Try fetching directly from raw.githubusercontent.com
+            $rawUrl = "https://raw.githubusercontent.com/$apiRepo/$ref/$name"
+            $content = Invoke-RestMethod -Uri $rawUrl -Method Get -ErrorAction SilentlyContinue
+            if ($content) {
+                Set-Content -Path $filePath -Value $content -Encoding UTF8
+                Write-Host "Found $name for ref $ref"
+                return $true
+            }
+        } catch {
+            # Continue to next filename
+        }
+    }
+    return $false
+}
+
+try {
+    Write-Host 'Fetching CHANGELOG files for comparison...'
 
-    $file = $(Get-ChildItem -Path $tmpDir | Where-Object { $_.Name -match '^changelog(\.md|\.txt|)$' } )
-    if ("$file" -eq '')
-    {
-        Write-Warning "Couldn't find a changelog"
+    # Fetch old changelog
+    $oldChangelogPath = Join-Path $tmpDir 'old-changelog.md'
+    if (-not (Get-ChangelogContent $OldTag $oldChangelogPath)) {
+        Write-Warning "Could not find changelog at $OldTag"
         return
     }
-    elseif ($file -is [Array])
-    {
-        Write-Warning "Multiple changelogs found: $file"
+
+    # Fetch new changelog
+    $newChangelogPath = Join-Path $tmpDir 'new-changelog.md'
+    if (-not (Get-ChangelogContent $NewTag $newChangelogPath)) {
+        Write-Warning "Could not find changelog at $NewTag"
         return
     }
-    Write-Host "Found changelog: $file"
-    [string[]]$lines = Get-Content $file
-}
-finally
-{
-    Write-Host "Removing $tmpDir"
-    Remove-Item -Recurse -Force -ErrorAction Continue -Path $tmpDir
-}
 
-$startIndex = -1
-$endIndex = -1
-$changelog = ''
-for ($i = 0; $i -lt $lines.Count; $i++)
-{
-    $line = $lines[$i]
-
-    if ($startIndex -lt 0)
-    {
-        if ($line -match "^#+ +v?$NewTag\b")
-        {
-            $startIndex = $i
+    Write-Host "Generating changelog diff between $OldTag and $NewTag..."
+
+    # Generate diff using git diff --no-index
+    # git diff returns exit code 1 when differences are found, which is expected behavior
+    # We need to handle this properly when PSNativeCommandErrorActionPreference is enabled
+    $fullDiff = & {
+        $oldErrorActionPreference = $ErrorActionPreference
+        $ErrorActionPreference = 'Continue'
+        try {
+            git diff --no-index $oldChangelogPath $newChangelogPath
+        } finally {
+            $ErrorActionPreference = $oldErrorActionPreference
         }
     }
-    elseif ($line -match "^#+ +v?$OldTag\b")
-    {
-        $endIndex = $i - 1
-        break
+
+    # The first lines are diff metadata, skip them
+    $fullDiff = $fullDiff -split "`n" | Select-Object -Skip 4
+    if ([string]::IsNullOrEmpty("$fullDiff")) {
+        Write-Host "No differences found between $OldTag and $NewTag"
+        return
+    } else {
+        Write-Host "Successfully created a changelog diff - $($fullDiff.Count) lines"
     }
-}
 
-# If the changelog doesn't have a section for the oldTag, stop at the first SemVer that's lower than oldTag.
-if ($endIndex -lt 0)
-{
-    $endIndex = $lines.Count - 1 # fallback, may be overwritten below
-    try
-    {
-        $semverOldTag = [System.Management.Automation.SemanticVersion]::Parse($OldTag)
-        for ($i = $startIndex; $i -lt $lines.Count; $i++)
-        {
-            $line = $lines[$i]
-            if ($line -match '^#+ +v?([0-9]+.*)$')
-            {
-                try
-                {
-                    if ($semverOldTag -ge [System.Management.Automation.SemanticVersion]::Parse($matches[1]))
-                    {
-                        $endIndex = $i - 1
+    # Extract only the added lines (lines starting with + but not ++)
+    $addedLines = $fullDiff | Where-Object { $_ -match '^[+][^+]*' } | ForEach-Object { $_.Substring(1) }
+
+    if ($addedLines.Count -gt 0) {
+        # Create clean changelog from added lines
+        $changelog = ($addedLines -join "`n").Trim()
+
+        # Apply formatting to clean changelog
+        if ($changelog.Length -gt 0) {
+            # Add header
+            if (-not ($changelog -match '^(##|#) Changelog')) {
+                $changelog = "## Changelog`n`n$changelog"
+            }
+
+            # Increase header level by one for content (not the main header)
+            $changelog = $changelog -replace '(^|\n)(#+) ', '$1$2# ' -replace '^### Changelog', '## Changelog'
+
+            # Only add details section if there are deletions or modifications (not just additions)
+            $hasModifications = $fullDiff | Where-Object { $_ -match '^[-]' -and $_ -notmatch '^[-]{3}' }
+            if ($hasModifications) {
+                $changelog += "`n`n<details>`n<summary>Full CHANGELOG.md diff</summary>`n`n"
+                $changelog += '```diff' + "`n"
+                $changelog += $fullDiff -join "`n"
+                $changelog += "`n" + '```' + "`n`n</details>"
+            }
+
+            # Apply standard formatting
+            # Remove at-mentions.
+            $changelog = $changelog -replace '@', ''
+            # Make PR/issue references into links to the original repository (unless they already are links).
+            $changelog = $changelog -replace '(?<!\[)#([0-9]+)(?![\]0-9])', ('[#$1](' + $RepoUrl + '/issues/$1)')
+            # Replace any links pointing to github.com so that the target PRs/Issues don't get na notification.
+            $changelog = $changelog -replace ('\(' + $prefix), '(https://github-redirect.dependabot.com/'
+
+            # Limit the changelog length to ~60k to allow for other text in the PR body (total PR limit is 65536 characters).
+            $limit = 60000
+            if ($changelog.Length -gt $limit) {
+                $oldLength = $changelog.Length
+                Write-Warning "Truncating changelog because it's $($changelog.Length - $limit) characters longer than the limit $limit."
+                while ($changelog.Length -gt $limit) {
+                    $lastNewlineIndex = $changelog.LastIndexOf("`n")
+                    if ($lastNewlineIndex -eq -1) {
+                        # No newlines found, just truncate to limit
+                        $changelog = $changelog.Substring(0, $limit)
                         break
                     }
+                    $changelog = $changelog.Substring(0, $lastNewlineIndex)
                 }
-                catch {}
+                $changelog += "`n`n> :warning: **Changelog content truncated by $($oldLength - $changelog.Length) characters because it was over the limit ($limit) and wouldn't fit into PR description.**"
             }
+
+            Write-Host "Final changelog length: $($changelog.Length) characters"
+            Write-Output $changelog
         }
     }
-    catch {}
-}
-
-# Slice changelog lines from startIndex to endIndex.
-if ($startIndex -ge 0)
-{
-    $changelog = ($lines[$startIndex..$endIndex] -join "`n").Trim()
-}
-else
-{
-    $changelog = ''
-}
-if ($changelog.Length -gt 1)
-{
-    $changelog = "# Changelog`n$changelog"
-    # Increase header level by one.
-    $changelog = $changelog -replace '(^|\n)(#+) ', '$1$2# '
-    # Remove at-mentions.
-    $changelog = $changelog -replace '@', ''
-    # Make PR/issue references into links to the original repository (unless they already are links).
-    $changelog = $changelog -replace '(?<!\[)#([0-9]+)(?![\]0-9])', ('[#$1](' + $RepoUrl + '/issues/$1)')
-    # Replace any links pointing to github.com so that the target PRs/Issues don't get na notification.
-    $changelog = $changelog -replace ('\(' + $prefix), '(https://github-redirect.dependabot.com/'
-}
 
-# Limit the changelog length to ~60k to allow for other text in the PR body (total PR limit is 65536 characters).
-$limit = 60000
-if ($changelog.Length -gt $limit)
-{
-    $oldLength = $changelog.Length
-    Write-Warning "Truncating changelog because it's $($changelog.Length - $limit) characters longer than the limit $limit."
-    while ($changelog.Length -gt $limit)
-    {
-        $changelog = $changelog.Substring(0, $changelog.LastIndexOf("`n"))
+    Write-Host "No changelog additions found between $OldTag and $NewTag"
+} catch {
+    Write-Warning "Failed to get changelog: $($_.Exception.Message)"
+} finally {
+    if (Test-Path $tmpDir) {
+        Write-Host 'Cleaning up temporary files...'
+        Remove-Item -Recurse -Force -ErrorAction Continue $tmpDir
+        Write-Host 'Cleanup complete.'
     }
-    $changelog += "`n`n> :warning: **Changelog content truncated by $($oldLength - $changelog.Length) characters because it was over the limit ($limit) and wouldn't fit into PR description.**"
 }
 
-$changelog
+# This resets the $LASTEXITCODE set by git diff above.
+# Note that this only runs in the successful path.
+exit 0