From e876dba8bad1a2af9b96027a24ca713ccbde6ad1 Mon Sep 17 00:00:00 2001
From: nuno maduro <enunomaduro@gmail.com>
Date: Thu, 23 Apr 2026 09:29:56 -0700
Subject: [PATCH] wip

---
 src/Concerns/Testable.php        |   2 +-
 src/Plugins/Tia/ChangedFiles.php |  84 +++++++++++++++++++++-
 src/Plugins/Tia/ContentHash.php  | 118 +++++++++++++++++++++++++++++++
 src/Plugins/Tia/Fingerprint.php  |   7 +-
 4 files changed, 206 insertions(+), 5 deletions(-)
 create mode 100644 src/Plugins/Tia/ContentHash.php

diff --git a/src/Concerns/Testable.php b/src/Concerns/Testable.php
index 1e35cc2a..3f8e2dc9 100644
--- a/src/Concerns/Testable.php
+++ b/src/Concerns/Testable.php
@@ -395,7 +395,7 @@ trait Testable
             $tia = Container::getInstance()->get(Tia::class);
             $assertions = $tia->getCachedAssertions($this::class.'::'.$this->name());
 
-            $this->addToAssertionCount($assertions > 0 ? $assertions : 1);
+            $this->addToAssertionCount($assertions);
 
             return null;
         }
diff --git a/src/Plugins/Tia/ChangedFiles.php b/src/Plugins/Tia/ChangedFiles.php
index 171b007a..2b511a39 100644
--- a/src/Plugins/Tia/ChangedFiles.php
+++ b/src/Plugins/Tia/ChangedFiles.php
@@ -85,7 +85,7 @@ final readonly class ChangedFiles
                 continue;
             }
 
-            $hash = @hash_file('xxh128', $absolute);
+            $hash = ContentHash::of($absolute);
 
             if ($hash === false || $hash !== $snapshot) {
                 $remaining[] = $file;
@@ -119,7 +119,7 @@ final readonly class ChangedFiles
                 continue;
             }
 
-            $hash = @hash_file('xxh128', $absolute);
+            $hash = ContentHash::of($absolute);
 
             if ($hash !== false) {
                 $out[$file] = $hash;
@@ -167,7 +167,85 @@ final readonly class ChangedFiles
             $unique[$file] = true;
         }
 
-        return array_keys($unique);
+        $candidates = array_keys($unique);
+
+        // Behavioural de-noising: for every file git calls "changed", hash
+        // the current content and the content at `$sha` through
+        // `ContentHash::of()`. A change that only touched comments /
+        // whitespace / blade `{{-- --}}` blocks produces the same hash on
+        // both sides and gets dropped before it can invalidate any test.
+        // Without this, a single-comment edit on a migration re-runs the
+        // entire DB-touching suite.
+        if ($sha !== null && $sha !== '') {
+            return $this->filterBehaviourallyUnchanged($candidates, $sha);
+        }
+
+        return $candidates;
+    }
+
+    /**
+     * @param  array<int, string>  $files
+     * @return array<int, string>
+     */
+    private function filterBehaviourallyUnchanged(array $files, string $sha): array
+    {
+        $remaining = [];
+
+        foreach ($files as $file) {
+            $absolute = $this->projectRoot.DIRECTORY_SEPARATOR.$file;
+
+            if (! is_file($absolute)) {
+                // Deleted on disk — a genuine change, keep it.
+                $remaining[] = $file;
+
+                continue;
+            }
+
+            $currentHash = ContentHash::of($absolute);
+
+            if ($currentHash === false) {
+                $remaining[] = $file;
+
+                continue;
+            }
+
+            $baselineContent = $this->contentAtSha($sha, $file);
+
+            if ($baselineContent === null) {
+                // Couldn't read the baseline (new file, binary, `git show`
+                // failed). Err on the side of re-running.
+                $remaining[] = $file;
+
+                continue;
+            }
+
+            $baselineHash = ContentHash::ofContent($file, $baselineContent);
+
+            if ($currentHash !== $baselineHash) {
+                $remaining[] = $file;
+            }
+        }
+
+        return $remaining;
+    }
+
+    /**
+     * Reads `$path` at `$sha` via `git show`. Returns null when the file
+     * didn't exist at that SHA, when git errors, or when the content
+     * isn't valid UTF-8-safe bytes (rare — binary files that happen to
+     * be tracked).
+     */
+    private function contentAtSha(string $sha, string $path): ?string
+    {
+        $process = new Process(['git', 'show', $sha.':'.$path], $this->projectRoot);
+        $process->setTimeout(5.0);
+        $process->run();
+
+        if (! $process->isSuccessful()) {
+            return null;
+        }
+
+        return $process->getOutput();
     }
 
     private function shouldIgnore(string $path): bool
diff --git a/src/Plugins/Tia/ContentHash.php b/src/Plugins/Tia/ContentHash.php
new file mode 100644
index 00000000..3c56b6cd
--- /dev/null
+++ b/src/Plugins/Tia/ContentHash.php
@@ -0,0 +1,118 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Pest\Plugins\Tia;
+
+/**
+ * Per-file hashing that ignores changes which can't alter behaviour —
+ * comments and whitespace for PHP, `{{-- … --}}` comments and whitespace
+ * runs for Blade templates. Every other file type falls back to a plain
+ * xxh128 of the raw bytes.
+ *
+ * Why it matters: TIA's file diff signals drive which tests re-run. A
+ * one-line comment tweak on a migration is a behavioural no-op, but the
+ * raw-bytes hash still differs, so every test that talks to the DB would
+ * currently re-execute. Normalising to the parsed-token / compiled-shape
+ * keeps the drift signal honest: edits that can't change runtime
+ * behaviour don't invalidate the replay cache.
+ *
+ * Important: this hash is stored in the graph's last-run tree, so any
+ * format change here must be paired with a `Fingerprint::SCHEMA_VERSION`
+ * bump — otherwise stale hashes from older graphs would be compared
+ * against normalised hashes from the new code and everything would
+ * appear changed.
+ *
+ * @internal
+ */
+final class ContentHash
+{
+    /**
+     * xxh128 hex of the file's "behavioural" shape, or `false` when the
+     * file can't be read. Callers should treat `false` the same way they
+     * treated a failed `hash_file()` previously.
+     */
+    public static function of(string $absolute): string|false
+    {
+        $raw = @file_get_contents($absolute);
+
+        if ($raw === false) {
+            return false;
+        }
+
+        return self::ofContent($absolute, $raw);
+    }
+
+    /**
+     * Same as `of()` but accepts the file contents in memory. Used when
+     * we already have the bytes (e.g. from `git show <sha>:<path>`) and
+     * want to avoid a disk round-trip.
+     */
+    public static function ofContent(string $path, string $raw): string
+    {
+        $lower = strtolower($path);
+
+        if (str_ends_with($lower, '.blade.php')) {
+            return self::hashBladeContent($raw);
+        }
+
+        if (str_ends_with($lower, '.php')) {
+            return self::hashPhpContent($raw);
+        }
+
+        return hash('xxh128', $raw);
+    }
+
+    /**
+     * Tokenise the content and hash the concatenated values of every
+     * token except whitespace / comment / docblock. `token_get_all()`
+     * is built-in, fast, and enough to collapse any formatting-only
+     * edit. If tokenisation fails (rare syntax error), fall back to
+     * the raw hash so the caller still gets a deterministic signal.
+     */
+    private static function hashPhpContent(string $raw): string
+    {
+        $tokens = @token_get_all($raw);
+
+        if ($tokens === []) {
+            return hash('xxh128', $raw);
+        }
+
+        $normalised = '';
+
+        foreach ($tokens as $token) {
+            if (is_array($token)) {
+                if ($token[0] === T_WHITESPACE) {
+                    continue;
+                }
+                if ($token[0] === T_COMMENT) {
+                    continue;
+                }
+                if ($token[0] === T_DOC_COMMENT) {
+                    continue;
+                }
+                $normalised .= $token[1];
+            } else {
+                $normalised .= $token;
+            }
+        }
+
+        return hash('xxh128', $normalised);
+    }
+
+    /**
+     * Blade templates aren't PHP syntactically, so `token_get_all()`
+     * doesn't help. Strip `{{-- … --}}` comments (the only Blade-native
+     * comment form) and collapse whitespace runs. Output differences
+     * that would survive the Blade compiler (markup reordering, new
+     * directives, changed interpolation) still flip the hash; pure
+     * reformatting does not.
+     */
+    private static function hashBladeContent(string $raw): string
+    {
+        $stripped = preg_replace('/\{\{--.*?--\}\}/s', '', $raw) ?? $raw;
+        $stripped = preg_replace('/\s+/', ' ', $stripped) ?? $stripped;
+
+        return hash('xxh128', trim($stripped));
+    }
+}
diff --git a/src/Plugins/Tia/Fingerprint.php b/src/Plugins/Tia/Fingerprint.php
index 076273a3..fd6dbec2 100644
--- a/src/Plugins/Tia/Fingerprint.php
+++ b/src/Plugins/Tia/Fingerprint.php
@@ -29,7 +29,12 @@ final readonly class Fingerprint
 {
     // Bump this whenever the set of inputs or the hash algorithm changes,
     // so older graphs are invalidated automatically.
-    private const int SCHEMA_VERSION = 4;
+    //
+    //   v5: ChangedFiles now hashes via `ContentHash` (normalises PHP
+    //       tokens + Blade whitespace/comments) instead of raw bytes.
+    //       Old graphs' run-tree hashes are incompatible and must be
+    //       rebuilt.
+    private const int SCHEMA_VERSION = 5;
 
     /**
      * @return array{