On Sat, May 18, 2024, at 10:15 AM, Alexander Dunlap wrote:
> On Sat, May 18, 2024, at 9:52 AM, Alexander Dunlap wrote:
>> On Sat, May 18, 2024, at 2:30 AM, Niranjana K M wrote:
>>> Dear LyX developers,
>>> Thank you and congratulations for heading towards much awaited LyX-2.4.
>>> I have been using LyX-2.4-RC1 from last 6 months and yesterday accidentally
>>> i caught this strange behavior. And I even installed recently released
>>> LyX-2.4 and the behavior persisted in this version also. But i am not
>>> facing this problem in LyX-2.3.7.
>>> The problem is, when I try to copy, by selecting and pressing Ctrl+C
>>> (or Copy from edit menu), a math block which is boxed and also having deep
>>> nested subscripts to at least 4 levels then my computer memory rapidly got
>>> consumed to the full (starting from 1GB to ~10GB) and in the end LyX closed
>>> with following message.
>>>
>>> $ /usr/local/bin/lyx
>>> Error: Software exception Detected
>>> ----------------------------------------
>>> LyX has caught an exception, it will now attempt to save all unsaved
>>> documents and exit.
>>> Exception: std::bad_alloc
>>> Aborted
>>>
>>> Following is the math block which caused it, (LyX code)
>>> \begin_inset Formula
>>> \[
>>> \boxed{v=\left[v_{p}\right]_{a_{0}}}
>>> \]
>>> \end_inset
>>>
>>> The file is also attached in the end and here is the screenshot of the math
>>> block.
>>> Screenshot from 2024-05-18 11-43-00.png
>>>
>>> It is a legitimate construction and it gives no problem in LyX-2.3.7
>>> (confirmed).
>>> If either the box is removed or the subscript 0 is removed from *a*, then
>>> it has no issues in copying.
>>>
>>> All these versions I mentioned were compiled from source on Gentoo Linux.
>>> This LyX-2.4 was compiled with the following configuration,
>>> $ ./configure --without-included-boost
>>> Configuration
>>> Host type: x86_64-pc-linux-gnu
>>> Special build flags: build=release callback-printing use-hunspell
>>> use-aspell use-enchant
>>> Bundled libraries: nod
>>> C++ Compiler: g++ (13.2.1)
>>> C++ Compiler flags: -fPIC -O2 -std=c++17
>>> C++ Compiler user flags: -std=c++17 -march=native -O2 -pipe
>>> Linker flags: -rdynamic
>>> Linker user flags:
>>> Qt Frontend:
>>> Qt version: 5.15.11
>>> Packaging: posix
>>> LyX binary dir: /usr/local/bin
>>> LyX files dir: /usr/local/share/lyx
>>>
>>> I loved the new features in LyX-2.4 and don't want to go back to LyX-2.3.7
>>> due to this issue. Please help in this regard.
>>
>> I can reproduce this. I investigated a little and it seems that the problem
>> is that the "while (true)" loop in the splitAndWrapInMText function in
>> InsetMathBox.cpp is never terminating. I believe that the reason for this is
>> that the logic in this function is not correctly identifying the matching
>> XHTML end tag for a given start tag, since it doesn't take into account
>> nested tags properly.
>
> Here is a patch that fixes the problem for me. It needs extensive
> checking/testing though because I do not know too much about XHTML or what
> all sorts of input might be seen in this function.
This patch is indeed not good enough, because it fails if there is text inside
the tag that contains the tag name. This is another case of the problem in the
2.4 RC's: if you type \left(msub\right)_{0} then it also hangs when you try to
copy it. The attached patches attempt to fix this issue but it seems maybe a
proper parsing system would be better.....
>
>>
>>
>>>
>>> *Regards*
>>> *Niranjana*
>>> --
>>> lyx-devel mailing list
>>> lyx-devel@lists.lyx.org
>>> http://lists.lyx.org/mailman/listinfo/lyx-devel
>>>
>>>
>>> *Attachments:*
>>> • test.24.lyx
>>
>
>
> *Attachments:*
> • 0001-Improving-logic-in-splitAndWrapInMText.patch
From a9d3e6f960265b9097a321ab8fdff41658e31b96 Mon Sep 17 00:00:00 2001
From: Alexander Dunlap <alexander.dun...@gmail.com>
Date: Sat, 18 May 2024 10:12:04 -0400
Subject: [PATCH 1/2] Improving logic in splitAndWrapInMText
---
src/mathed/InsetMathBox.cpp | 29 ++++++++++++++++++++++++++---
1 file changed, 26 insertions(+), 3 deletions(-)
diff --git a/src/mathed/InsetMathBox.cpp b/src/mathed/InsetMathBox.cpp
index 6feddf9cdd..4295ac0780 100644
--- a/src/mathed/InsetMathBox.cpp
+++ b/src/mathed/InsetMathBox.cpp
@@ -101,9 +101,32 @@ void splitAndWrapInMText(MathMLStream & ms, MathData const & cell,
const std::size_t tag_name_length = tag_name_end - 1;
const docstring tag_name = inset_contents.substr(1, tag_name_length);
- const std::size_t end_tag_start =
- inset_contents.find(tag_name, tag_name_end + 1);
- const std::size_t end_tag = inset_contents.find('>', end_tag_start);
+ // Here we find the matching end tag. We need to take into
+ // account that there may be new tags with the same name opening // up.
+ int depth = 0;
+ std::size_t end_tag = tag_name_end;
+ std::size_t end_tag_start;
+ while (true) {
+ // Find the next tag with the name we are looking for
+ end_tag_start = inset_contents.find(tag_name, end_tag + 1);
+ end_tag = inset_contents.find('>', end_tag_start);
+ if (inset_contents[end_tag_start-1] == '/') {
+ // if it's a closing tag, then reduce the depth, or if
+ // we are already at depth 0, then we have found the
+ // matching tag
+ if (depth == 0) {
+ break;
+ } else {
+ depth--;
+ }
+ } else {
+ // if it's not a closing tag, then it's a nested opening
+ // tag, so we increase the depth: now we need an
+ // additional closing tag before we have found the
+ // matching one
+ depth++;
+ }
+ }
parts.emplace_back(inset_contents.substr(0, end_tag + 1));
inset_contents = inset_contents.substr(end_tag + 1);
--
2.45.1
From 1446a258f20be1da3665fe5541eae400c48fe679 Mon Sep 17 00:00:00 2001
From: Alexander Dunlap <alexander.dun...@gmail.com>
Date: Sat, 18 May 2024 12:35:05 -0400
Subject: [PATCH 2/2] Making sure it is an actual HTML tag
---
src/mathed/InsetMathBox.cpp | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/src/mathed/InsetMathBox.cpp b/src/mathed/InsetMathBox.cpp
index 4295ac0780..c21a6df045 100644
--- a/src/mathed/InsetMathBox.cpp
+++ b/src/mathed/InsetMathBox.cpp
@@ -108,9 +108,10 @@ void splitAndWrapInMText(MathMLStream & ms, MathData const & cell,
std::size_t end_tag_start;
while (true) {
// Find the next tag with the name we are looking for
- end_tag_start = inset_contents.find(tag_name, end_tag + 1);
+ end_tag_start = std::min(inset_contents.find("<"+tag_name, end_tag + 1),inset_contents.find("</"+tag_name, end_tag + 1));
+;
end_tag = inset_contents.find('>', end_tag_start);
- if (inset_contents[end_tag_start-1] == '/') {
+ if (inset_contents[end_tag_start+1] == '/') {
// if it's a closing tag, then reduce the depth, or if
// we are already at depth 0, then we have found the
// matching tag
--
2.45.1
--
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel
