From 5cdcba36d5678d67eaa7b84e70ca51353c521bbd Mon Sep 17 00:00:00 2001
From: Khaled Hosny <khaled@aliftype.com>
Date: Sun, 19 Jul 2026 19:46:37 +0300
Subject: [PATCH] Reject input that spans multiple paragraphs

raqm expects the text to be a single paragraph. Inputting multiple
paragraphs at once is undefined behaviour.

Enforce this by making raqm_layout() fail when the text contains a
paragraph separator, and use stronger words about this requirement on
raqm_set_text() docs.

Also drop the disabled cursor-position-GB3 test, at it contains CR+LF.
---
 src/raqm.c                     | 20 ++++++++++++---
 tests/cursor-position-GB3.test | 45 ----------------------------------
 tests/meson.build              |  1 -
 3 files changed, 17 insertions(+), 49 deletions(-)
 delete mode 100644 tests/cursor-position-GB3.test

diff --git a/src/raqm.c b/src/raqm.c
index 05685afd..07eb2da9 100644
--- a/src/raqm.c
+++ b/src/raqm.c
@@ -512,8 +512,8 @@ raqm_clear_contents (raqm_t *rq)
  * @len: the length of @text.
  *
  * Adds @text to @rq to be used for layout. It must be a valid UTF-32 text, any
- * invalid characters will be replaced with U+FFFD. The text should typically
- * represent a full paragraph, since doing the layout of chunks of text
+ * invalid characters will be replaced with U+FFFD. The text must
+ * represent a single full paragraph, since doing the layout of chunks of text
  * separately can give improper output.
  *
  * Return value:
@@ -1300,6 +1300,15 @@ _raqm_itemize (raqm_t *rq);
 static bool
 _raqm_shape (raqm_t *rq);
 
+static bool
+_raqm_is_paragraph_separator (uint32_t ch)
+{
+  /* Unicode characters with Bidi class B (Paragraph_Separator). */
+  return ch == 0x000A || ch == 0x000D ||
+         ch == 0x001C || ch == 0x001D || ch == 0x001E ||
+         ch == 0x0085 || ch == 0x2029;
+}
+
 /**
  * raqm_layout:
  * @rq: a #raqm_t.
@@ -1309,7 +1318,8 @@ _raqm_shape (raqm_t *rq);
  * text shaping, and any other part of the layout process.
  *
  * Return value:
- * `true` if the layout process was successful, `false` otherwise.
+ * `true` if the layout process was successful, `false` otherwise. It also fails
+ * if the text is not a single paragraph (see raqm_set_text()).
  *
  * Since: 0.1
  */
@@ -1329,6 +1339,10 @@ raqm_layout (raqm_t *rq)
   {
       if (!rq->text_info[i].ftface)
           return false;
+
+      /* The text must be a single paragraph. */
+      if (_raqm_is_paragraph_separator (rq->text[i]))
+          return false;
   }
 
   if (!_raqm_itemize (rq))
diff --git a/tests/cursor-position-GB3.test b/tests/cursor-position-GB3.test
deleted file mode 100644
index f35d867b..00000000
--- a/tests/cursor-position-GB3.test
+++ /dev/null
@@ -1,45 +0,0 @@
-fonts/sha1sum/bcb3b98eb67ece19b8b709f77143d91bcb3d95eb.ttf
-om\r\nan
---cluster 2 --position 3000
-Direction is: DEFAULT
-
-Before script detection:
-script for ch[0]	Latn
-script for ch[1]	Latn
-script for ch[2]	Zyyy
-script for ch[3]	Zyyy
-script for ch[4]	Latn
-script for ch[5]	Latn
-
-After script detection:
-script for ch[0]	Latn
-script for ch[1]	Latn
-script for ch[2]	Latn
-script for ch[3]	Latn
-script for ch[4]	Latn
-script for ch[5]	Latn
-
-Number of runs before script itemization: 1
-
-BiDi Runs:
-run[0]:	 start: 0	length: 6	level: 0
-
-Number of runs after script itemization: 1
-
-Final Runs:
-run[0]:	 start: 0	length: 6	direction: ltr	script: Latn	font: Amiri
-
-Glyph information:
-glyph [82]	x_offset: 0	y_offset: 0	x_advance: 1018	font: Amiri
-glyph [80]	x_offset: 0	y_offset: 0	x_advance: 1580	font: Amiri
-glyph [0]	x_offset: 0	y_offset: 0	x_advance: 748	font: Amiri
-glyph [0]	x_offset: 0	y_offset: 0	x_advance: 748	font: Amiri
-glyph [68]	x_offset: 0	y_offset: 0	x_advance: 862	font: Amiri
-glyph [81]	x_offset: 0	y_offset: 0	x_advance: 1064	font: Amiri
-
-UTF-32 clusters: 00 01 02 03 04 05
-UTF-8 clusters:  00 01 02 03 04 05
-
-The position is 4094 at index 3
-
-The start-index is 3  at position 3000 
diff --git a/tests/meson.build b/tests/meson.build
index 4316f2e8..c6b35b88 100644
--- a/tests/meson.build
+++ b/tests/meson.build
@@ -16,7 +16,6 @@ tests = [
     'cursor-position-2.test',
     'cursor-position-3.test',
     'cursor-position-4.test',
-    #'cursor-position-GB3.test',
     'cursor-position-GB4.test',
     'cursor-position-GB5.test',
     'cursor-position-GB8a.test',