From f50d2acf8ae31ef37606360f265e3f28fee7b5ca Mon Sep 17 00:00:00 2001
From: RickyRister <ricky.rister.wang@gmail.com>
Date: Tue, 11 Mar 2025 23:44:26 -0700
Subject: [PATCH] get thing to work

---
 oracle/src/parsehelpers.cpp | 49 ++++++++++++++++++++++++++-----------
 1 file changed, 35 insertions(+), 14 deletions(-)

diff --git a/oracle/src/parsehelpers.cpp b/oracle/src/parsehelpers.cpp
index 779ba956f..893378eec 100644
--- a/oracle/src/parsehelpers.cpp
+++ b/oracle/src/parsehelpers.cpp
@@ -3,30 +3,51 @@
 #include <QRegularExpression>
 
 /**
- * Parses the card text to determine if the card enters the battlefield tapped.
+ * Parses the card text to determine if the card should have the cipt tag
  *
- * The parsing logic will be able to handle the following cases:
- * - "<name> enters tapped..."
- * - shortname, if it appears at the start of the name.
- * - "This <type> enters tapped..."
+ * The parsing logic is able to handle the following cases:
+ * - "<name> enters tapped"
+ * - "<shortname> enters tapped", if the card name starts with the shortname
+ * - "This <type> enters tapped"
+ * - "..., it enters tapped"
  * - Any naming scheme that appends a non-alphanumeric character plus extra text to the end of the name.
- * (e.g. name is "Card Name_SET" or "Card Name (Set)" and text contains "Card Name enters tapped...")
+ * (e.g. name is "Card Name_SET" or "Card Name (Set)" and text contains "Card Name enters tapped")
  *
- * However, it will still miss certain cases:
- * - shortnames that aren't the first word in the card name
+ * However, it will still miss on certain cases:
+ * - shortnames that aren't the at the beginning of the card name
+ *
+ * Note that "...enters tapped unless..." returns false.
  *
  * @param name The name of the card
  * @param text The oracle text of the card
  */
 bool parseCipt(const QString &name, const QString &text)
 {
-    static auto WHITESPACE_REGEX = QRegularExpression("\\s+");
-    QString shortname = name.split(WHITESPACE_REGEX, Qt::SkipEmptyParts).first();
+    // try to split shortname on all non-alphanumeric characters (including _)
+    static auto SHORTNAME_DIVIDERS = QRegularExpression("[^\\w]|_");
 
-    QString namePattern = QString("( it|this [^ ]* |%1|%2.*)")
-                              .arg(QRegularExpression::escape(name), QRegularExpression::escape(shortname));
+    // Try all possible shortnames.
+    // This also handles the case of extra text appended at end.
+    QStringList possibleNames;
+    qsizetype i = 0;
+    while ((i = name.indexOf(SHORTNAME_DIVIDERS, i)) != -1) {
+        possibleNames.append(QRegularExpression::escape(name.first(i)));
+        ++i;
+    }
 
-    QRegularExpression ciptRegex(namePattern + " enters( the battlefield)? tapped(?! unless)");
+    // and the full name
+    possibleNames.append(QRegularExpression::escape(name));
 
-    return ciptRegex.match(text).hasMatch();
+    QString subject = "(it|"            // "..., it enters tapped"
+                      "(T|t)his [^ ]+|" // "This <type> enters tapped"
+                      + possibleNames.join("|") + ")";
+
+    auto ciptPattern = QRegularExpression(
+        // cipt phrase is either first sentence of line, or is after a punctuation mark
+        "(^|(, |\\. ))" + subject +
+            // support old wording, and exclude the "unless" case
+            " enters( the battlefield)? tapped(?! unless)",
+        QRegularExpression::MultilineOption);
+
+    return ciptPattern.match(text).hasMatch();
 }