← All changes | includes/class-mxchat-word-handler.php +38 -23 3.2.17 → 3.2.23 View file →

		@@ -42,12 +42,12 @@
42	42	$original_filename = sanitize_text_field($file['name']);
43	43
44	44	// Update session owner if it changed (e.g. IP changed due to network switch)
45	45	$current_user_identifier = MxChat_User::mxchat_get_user_identifier();
46		- $session_owner = get_option("mxchat_session_owner_{$session_id}");
	46	+ $session_owner = MxChat_Session_Store::get($session_id, 'owner');
47	47
48	48	if (!$session_owner || $session_owner !== $current_user_identifier) {
49		- update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
	49	+ MxChat_Session_Store::set($session_id, 'owner', $current_user_identifier);
50	50	}
51	51
52	52	// Check file type
53	53	$allowed_types = array(
		@@ -98,8 +98,33 @@
98	98	]);
99	99	}
100	100
101	101	/**
	102	+ * Extract plain text from a .docx file. THE one .docx parser — the
	103	+ * visitor-facing toolbar upload (via mxchat_process_word_document) and the
	104	+ * admin knowledge importer (plan 0485e5) both call this; do not fork a
	105	+ * second copy. Returns the cleaned text, or false on unreadable/empty
	106	+ * files (not a Zip, no word/document.xml, no text content).
	107	+ */
	108	+ public static function extract_docx_text($file_path) {
	109	+ try {
	110	+ $zip = new ZipArchive();
	111	+ if ($zip->open($file_path) !== true) {
	112	+ return false;
	113	+ }
	114	+ $content = $zip->getFromName('word/document.xml');
	115	+ $zip->close();
	116	+ if ($content === false) {
	117	+ return false;
	118	+ }
	119	+ $text = self::clean_word_xml($content);
	120	+ return $text === '' ? false : $text;
	121	+ } catch (Exception $e) {
	122	+ return false;
	123	+ }
	124	+ }
	125	+
	126	+ /**
102	127	* Process Word document and generate embeddings
103	128	*/
104	129	private function mxchat_process_word_document($file_path) {
105	130	// Get the maximum number of pages allowed from admin settings
		@@ -105,24 +130,13 @@
105	130	// Get the maximum number of pages allowed from admin settings
106	131	$max_pages = isset($this->options['pdf_max_pages']) ? intval($this->options['pdf_max_pages']) : 69; // Use same setting as PDF
107	132
108	133	try {
109		- $zip = new ZipArchive();
110		- if ($zip->open($file_path) !== true) {
	134	+ $text = self::extract_docx_text($file_path);
	135	+ if ($text === false) {
111	136	return false;
112	137	}
113	138
114		- // Extract main document content
115		- $content = $zip->getFromName('word/document.xml');
116		- $zip->close();
117		-
118		- if ($content === false) {
119		- return false;
120		- }
121		-
122		- // Clean up the content
123		- $text = $this->mxchat_clean_word_content($content);
124		-
125	139	// Count pages (roughly estimate based on paragraphs)
126	140	$paragraphs = explode("\n\n", $text);
127	141	$estimated_pages = ceil(count($paragraphs) / 3); // Assume ~3 paragraphs per page
128	142
		@@ -159,25 +173,26 @@
159	173	return false;
160	174	}
161	175	}
162	176	/**
163		- * Clean Word XML content
	177	+ * Clean Word XML content (word/document.xml) to plain text.
	178	+ * Static so both callers of extract_docx_text share one implementation.
164	179	*/
165		- private function mxchat_clean_word_content($content) {
	180	+ public static function clean_word_xml($content) {
166	181	// Remove XML namespaces
167	182	$content = preg_replace('/xmlns[^=]*="[^"]*"/i', '', $content);
168		-
	183	+
169	184	// Convert Word XML elements to text
170	185	$content = str_replace('</w:p>', "\n", $content);
171	186	$content = str_replace('</w:tr>', "\n", $content);
172		-
	187	+
173	188	// Strip remaining XML tags
174	189	$content = strip_tags($content);
175		-
	190	+
176	191	// Clean up whitespace
177	192	$content = preg_replace('/\s+/', ' ', $content);
178	193	$content = preg_replace('/\n\s*\n/', "\n\n", $content);
179		-
	194	+
180	195	return trim($content);
181	196	}
182	197
183	198	/**
		@@ -227,11 +242,11 @@
227	242	// history endpoint (plan-mxchat-20260731-d42bec). Possession of the session
228	243	// id is the credential, so this records/refreshes the owner rather than
229	244	// refusing a mismatch; see the fuller note in handle_pdf_remove().
230	245	$current_user_identifier = MxChat_User::mxchat_get_user_identifier();
231		- $session_owner = get_option("mxchat_session_owner_{$session_id}");
	246	+ $session_owner = MxChat_Session_Store::get($session_id, 'owner');
232	247	if (!$session_owner || $session_owner !== $current_user_identifier) {
233		- update_option("mxchat_session_owner_{$session_id}", $current_user_identifier, 'no');
	248	+ MxChat_Session_Store::set($session_id, 'owner', $current_user_identifier);
234	249	}
235	250
236	251	$word_path = get_transient('mxchat_word_url_' . $session_id);
237	252