From eb33e6cf2dd17e6aef0e257bd7d15dacfd88a93c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= <mike_faehrmann@web.de>
Date: Fri, 4 Nov 2022 21:37:36 +0100
Subject: [PATCH] add 'text.extr()'

a stripped-down version of text.extract() that
- always returns a string (like 'extract_from')
- only returns a string
- does not deal with 'pos' arguments
- is ~20% faster
---
 gallery_dl/text.py |  9 +++++++++
 test/test_text.py  | 13 +++++++++++++
 2 files changed, 22 insertions(+)

diff --git a/gallery_dl/text.py b/gallery_dl/text.py
index 79cf016e..cbba063d 100644
--- a/gallery_dl/text.py
+++ b/gallery_dl/text.py
@@ -120,6 +120,15 @@ def extract(txt, begin, end, pos=0):
         return None, pos
 
 
+def extr(txt, begin, end):
+    """Stripped-down version of 'extract()'"""
+    try:
+        first = txt.index(begin) + len(begin)
+        return txt[first:txt.index(end, first)]
+    except (ValueError, TypeError, AttributeError):
+        return ""
+
+
 def rextract(txt, begin, end, pos=-1):
     try:
         lbeg = len(begin)
diff --git a/test/test_text.py b/test/test_text.py
index 0ac77671..a1b30887 100644
--- a/test/test_text.py
+++ b/test/test_text.py
@@ -203,6 +203,19 @@ class TestText(unittest.TestCase):
             self.assertEqual(f(txt  , value, ">")  , (None, 0))
             self.assertEqual(f(txt  , "<"  , value), (None, 0))
 
+    def test_extr(self, f=text.extr):
+        txt = "<a><b>"
+        self.assertEqual(f(txt, "X", ">"), "")
+        self.assertEqual(f(txt, "<", "X"), "")
+        self.assertEqual(f(txt, "<", ">"), "a")
+        self.assertEqual(f(txt, "><", ">"), "b")
+
+        # invalid arguments
+        for value in INVALID:
+            self.assertEqual(f(value, "<"  , ">")  , "")
+            self.assertEqual(f(txt  , value, ">")  , "")
+            self.assertEqual(f(txt  , "<"  , value), "")
+
     def test_rextract(self, f=text.rextract):
         txt = "<a><b>"
         self.assertEqual(f(txt, "<", ">"), ("b" , 3))