fix(2.9.1): extractFromDocx — drop PHPWord dependency, parse word/document.xml directly

Root cause of the Shira hallucination incident in production case 46 Friedman:
DocumentAnalyzer::extractFromDocx relied on PhpOffice\PhpWord, but on the
EspoCRM prod image the PHPWord files are present in vendor/ yet are NOT
registered in the composer PSR-4 autoload map — class_exists() silently
returned false, the method returned null, and Shira's read_document fell
back to OCR. OCR then only saw the signature image, the AI got "[signature]"
as document content and fabricated the entire CTS appeal as a knee injury.

This change rewrites extractFromDocx to use ZipArchive + a small regex
parser of word/document.xml. Independent of PHPWord, more robust on
tables/footnotes/hyperlinks (which PHPWord's element walker missed at depth
>1), and verified on prod against the same Friedman appeal: 80 paragraphs /
16633 clean chars extracted (vs 0 before).

The paired Python fix in shira-hermes (commit 7b517e1) makes the OCR
fallback also read document.xml, so even if this PHP fix regresses again,
the AI will not be fed "[signature]" as document content.

Refs Task Master #5

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

files/custom/Espo/Modules/SmartAssistant/Services/DocumentAnalyzer.php

@@ -533,20 +533,37 @@ class DocumentAnalyzer
 
     private function extractFromDocx(string $tmpFile): ?string
     {
-        if (!class_exists(\PhpOffice\PhpWord\IOFactory::class)) return null;
+        // PHPWord (vendor/phpoffice/phpword) ships with the EspoCRM image but
-        $phpWord = \PhpOffice\PhpWord\IOFactory::load($tmpFile);
+        // is NOT in the composer PSR-4 autoload map — class_exists silently
-        $text = '';
+        // returns false. Rather than fix the autoloader (rebuild of the image)
-        foreach ($phpWord->getSections() as $section) {
+        // we parse word/document.xml directly. This is also more robust:
-            foreach ($section->getElements() as $element) {
+        // PHPWord's element walker only goes one level deep, missing text in
-                if (method_exists($element, 'getText')) $text .= $element->getText() . "\n";
+        // tables, hyperlinks, and footnotes.
-                elseif (method_exists($element, 'getElements')) {
+        $zip = new \ZipArchive();
-                    foreach ($element->getElements() as $child) {
+        if ($zip->open($tmpFile) !== true) {
-                        if (method_exists($child, 'getText')) $text .= $child->getText() . "\n";
+            $this->log->warning("SmartAssistant: extractFromDocx: not a valid zip: $tmpFile");
-                    }
+            return null;
+        }
+        $xml = $zip->getFromName('word/document.xml');
+        $zip->close();
+        if ($xml === false || $xml === '') {
+            $this->log->warning("SmartAssistant: extractFromDocx: word/document.xml missing in $tmpFile");
+            return null;
+        }
+
+        // Each <w:p> is a paragraph; concatenate <w:t> runs inside it.
+        // `(?:\s[^>]*)?` keeps us from also matching <w:tab> / <w:tbl>.
+        $paragraphs = [];
+        if (preg_match_all('#<w:p(?:\s[^>]*)?>(.*?)</w:p>#s', $xml, $pMatches)) {
+            foreach ($pMatches[1] as $block) {
+                if (preg_match_all('#<w:t(?:\s[^>]*)?>(.*?)</w:t>#s', $block, $tMatches)) {
+                    $line = trim(implode('', $tMatches[1]));
+                    if ($line !== '') $paragraphs[] = html_entity_decode($line, ENT_XML1 | ENT_QUOTES, 'UTF-8');
                 }
             }
         }
-        return $text ?: null;
+        $text = trim(implode("\n", $paragraphs));
+        return $text !== '' ? $text : null;
     }
 
     private function updateDocumentEntity(string $oldPath, string $newPath): void

manifest.json

@@ -3,11 +3,11 @@
     "module": "SmartAssistant",
     "description": "Unified AI Assistant for Legal CRM — floating chat with case memory, office alerts, and AI Gateway integration",
     "author": "klear",
-    "version": "2.9.0",
+    "version": "2.9.1",
     "acceptableVersions": [
         ">=8.0.0"
     ],
-    "releaseDate": "2026-05-14",
+    "releaseDate": "2026-05-26",
     "php": [
         ">=8.1"
     ]