From 658db66c2bb18c83d78873f82470adee6575f3b2 Mon Sep 17 00:00:00 2001
From: Alex Cabal <alex@alexcabal.com>
Date: Tue, 8 Dec 2020 14:27:01 -0600
Subject: [PATCH] Include ToC entries in search results for short and poetry
 compilations

---
 lib/Ebook.php            | 23 +++++++++++++++++++++++
 templates/SearchForm.php |  2 +-
 www/ebooks/index.php     |  2 +-
 3 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/lib/Ebook.php b/lib/Ebook.php
index 2bd683af..a4137232 100644
--- a/lib/Ebook.php
+++ b/lib/Ebook.php
@@ -57,6 +57,7 @@ class Ebook{
 	public $ModifiedTimestamp;
 	public $TextUrl;
 	public $TextSinglePageUrl;
+	public $TocEntries = null; // A list of non-Roman ToC entries ONLY IF the work has the 'shorts' or 'poetry' tag, null otherwise
 
 	public function __construct(string $wwwFilesystemPath){
 		// First, construct a source repo path from our WWW filesystem path.
@@ -194,8 +195,23 @@ class Ebook{
 		}
 
 		// Get SE tags
+		$includeToc = false;
 		foreach($xml->xpath('/package/metadata/meta[@property="se:subject"]') ?: [] as $tag){
 			$this->Tags[] = new Tag($tag);
+
+			if($tag == 'Shorts' || $tag == 'Poetry'){
+				$includeToc = true;
+			}
+		}
+
+		// Fill the ToC if necessary
+		if($includeToc){
+			$this->TocEntries = [];
+			$tocDom = new SimpleXMLElement(str_replace('xmlns=', 'ns=', file_get_contents($wwwFilesystemPath . '/toc.xhtml') ?: ''));
+			$tocDom->registerXPathNamespace('epub', 'http://www.idpf.org/2007/ops');
+			foreach($tocDom->xpath('/html/body//nav[@epub:type="toc"]//a[not(contains(@epub:type, "z3998:roman")) and not(text() = "Titlepage" or text() = "Imprint" or text() = "Colophon" or text() = "Endnotes" or text() = "Uncopyright") and not(contains(@href, "halftitle"))]') as $item){
+				$this->TocEntries[] = (string)$item;
+			}
 		}
 
 		// Get SE collections
@@ -423,6 +439,7 @@ class Ebook{
 
 	public function Contains(string $query): bool{
 		// When searching an ebook, we search the title, alternate title, author(s), SE tags, series data, and LoC tags.
+		// Also, if the ebook is shorts or poetry, search the ToC as well.
 
 		$searchString = $this->FullTitle ?? $this->Title;
 
@@ -444,6 +461,12 @@ class Ebook{
 			$searchString .= ' ' . $tag;
 		}
 
+		if($this->TocEntries !== null){
+			foreach($this->TocEntries as $item){
+				$searchString .= ' ' . $item;
+			}
+		}
+
 		// Remove diacritics and non-alphanumeric characters
 		$searchString = trim(preg_replace('|[^a-zA-Z0-9 ]|ius', ' ', Formatter::RemoveDiacritics($searchString)) ?? '');
 		$query = trim(preg_replace('|[^a-zA-Z0-9 ]|ius', ' ', Formatter::RemoveDiacritics($query)) ?? '');
diff --git a/templates/SearchForm.php b/templates/SearchForm.php
index c1f11324..35ab3bc0 100644
--- a/templates/SearchForm.php
+++ b/templates/SearchForm.php
@@ -7,7 +7,7 @@ $allSelected = sizeof($tags) == 0 || in_array('all', $tags);
 			<option value="all">All</option>
 		<? foreach(Library::GetTags() as $tag){
 			$lcTag = mb_strtolower($tag); ?>
-			<option value="<?= $lcTag ?>"<? if(!$allSelected && in_array($lcTag, $tags)){ ?> selected="selected"<? } ?>><?= $tag ?></option>
+			<option value="<?= Formatter::ToPlainText($lcTag) ?>"<? if(!$allSelected && in_array($lcTag, $tags)){ ?> selected="selected"<? } ?>><?= Formatter::ToPlainText($tag) ?></option>
 		<? } ?>
 		</select>
 	</label>
diff --git a/www/ebooks/index.php b/www/ebooks/index.php
index 43e62bd8..dc08de10 100644
--- a/www/ebooks/index.php
+++ b/www/ebooks/index.php
@@ -121,7 +121,7 @@ try{
 		}
 	}
 
-	$queryString = preg_replace('/^&amp;/ius', '', $queryString);
+	$queryString = Formatter::ToPlainText(preg_replace('/^&amp;/ius', '', $queryString));
 }
 catch(\Exception $ex){
 	http_response_code(404);