From e97634b7bf4d20d77dde39f0ed2c949d739ca0eb Mon Sep 17 00:00:00 2001
From: "E. S." <the@rootless.one>
Date: Sun, 28 Jul 2024 21:47:18 +0300
Subject: [PATCH] files/collections: more robust sphinxsearch reindexing; add
 reindex command to tools/cli

---
 lib/files.php      | 14 +++++++++-----
 skin/files.phps    |  2 +-
 tools/cli_util.php | 11 +++++++++++
 3 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/lib/files.php b/lib/files.php
index e1d33e4..185d13f 100644
--- a/lib/files.php
+++ b/lib/files.php
@@ -320,13 +320,14 @@ function wff_reindex(): void {
     $q = $db->query("SELECT * FROM wff_collection");
     while ($row = $db->fetch($q)) {
         $item = new WFFCollectionItem($row);
+        $text = '';
         if ($item->isFile()) {
-            $txt = file_get_contents('/home/user/nsa/txt/'.str_replace('.pdf', '.txt', basename($item->path)));
-        } else {
-            $txt = '';
+            $text_q = $db->query("SELECT text FROM wff_texts WHERE wff_id=?", $item->id);
+            if ($db->numRows($text_q))
+                $text = $db->result($text_q);
         }
         sphinx_execute("INSERT INTO ".WFF_ARCHIVE_SPHINX_RTINDEX." (id, document_id, title, text, is_folder, parent_id) VALUES (?, ?, ?, ?, ?, ?)",
-            $item->id, $item->getDocumentId(), $item->title, $txt, (int)$item->isFolder(), $item->parentId);
+            $item->id, $item->getDocumentId(), $item->title, $text, (int)$item->isFolder(), $item->parentId);
     }
 }
 
@@ -585,7 +586,10 @@ function baconiana_reindex(): void {
     $db = DB();
     $baconiana = baconiana_get(null);
     foreach ($baconiana as $item) {
-        $text = $db->result($db->query("SELECT text FROM baconiana_texts WHERE bcn_id=?", $item->id));
+        $text_q = $db->query("SELECT text FROM baconiana_texts WHERE bcn_id=?", $item->id);
+        if (!$db->numRows($text_q))
+            continue;
+        $text = $db->result($text_q);
         sphinx_execute("INSERT INTO ".BACONIANA_ARCHIVE_SPHINX_RTINDEX." (id, title, year, text) VALUES (?, ?, ?, ?)",
             $item->id, "$item->year ($item->issues)", $item->year, $text);
     }
diff --git a/skin/files.phps b/skin/files.phps
index c9305df..ecaa952 100644
--- a/skin/files.phps
+++ b/skin/files.phps
@@ -186,7 +186,7 @@ function file(SkinContext $ctx,
               FilesItemInterface $file,
               ?SkinString $unsafe_query = null,
               bool $disabled = false,
-              ?array $text_excerpts = null,) {
+              ?array $text_excerpts = null) {
 $icons = svg();
 if ($file instanceof BookItem && $file->fileType == BookFileType::BOOK)
     $icon = $icons->book_20();
diff --git a/tools/cli_util.php b/tools/cli_util.php
index 39eceb2..b3405bd 100755
--- a/tools/cli_util.php
+++ b/tools/cli_util.php
@@ -79,6 +79,17 @@ require_once 'lib/admin.php';
     echo "upload id: $id\n";
 })
 
+->on('collection-reindex', function() {
+    require_once 'lib/files.php';
+    $collections = array_map(fn($c) => $c->value, FilesCollection::cases());
+    $s = cli::input('Enter collection to reindex (variants: '.implode(', ', $collections).': ');
+    $c = FilesCollection::from($s);
+    $f = "{$s}_reindex";
+    echo "calling $f()... ";
+    call_user_func($f);
+    echo "done\n";
+})
+
 ->run();
 
 function _get_admin_login_password_input(): array {