WARNING: this PR won't compile unless [PR 132 from pattern language
repository](https://github.com/WerWolv/PatternLanguage/pull/132) is
merged first. Some changes here are shared by at least another PR to
this repository but there should not be any conflicts as the shared
changes are identical.
### Problem description
fix: Editing patterns with breakpoints sets behaves unexpectedly. As a
simple example, set a breakpoint and insert a blank line somewhere
before the breakpoint location. The breakpoint will appear to move but
in reality it hasn't. To see this set another breakpoint elsewhere in
the file and the old one will be displayed where it is really located
at.
The reason for this and many other problems with breakpoints is that
currently ImHex keeps two set of breakpoints in text editor and in
evaluator that are independent of each other, ie, changes to one don't
affect the other. This PR aims at synchronizing the two sets through the
per provider breakpoints that exist in view pattern editor.
### Implementation description
It accomplishes this by making the text editor version of breakpoints
the principal source of vectors and the ones in evaluator the effective
version. The first allows one to modify the text around and at the
breakpoint and notify others that the changes have induced changes in
the breakpoint locations. The effective breakpoints allow the insertion
and deletion of breakpoints.
View pattern editor is where breakpoints are updated. It receives
notifications from text editor about changes and then makes sure the
version in evaluator is updated with those changes. View pattern editor
also manages breakpoint addition and deletion so before making changes
it gets a copy of the current ones from text editor, sets the ones in
evaluator, uses the evaluator functions to add or delete breakpoints and
finally sets the text editor version with the new version.
This code fixes issue #1950 This PR depends on PR [#27
](https://github.com/WerWolv/libwolv/pull/27) on libwolv repo
### Problem description
See description of the problem on the issue linked above
### Implementation description
The code in the PR uses function added to the libwolv repository to
remove tabs if they exist on the provider. Additionally, any pasted tab
on the pattern editor will be translated to spaces automatically. This
code was tested successfully using the pattern posted in the issue.
std::forward ensures that the arguments are perfectly forwarded to
loadScript
By reserving space in the features vector based on the size of scripts,
we can avoid multiple memory allocations during the loop. If an
exception occurs, returning an empty vector immediately clarifies that
no scripts were loaded. Without reservation, each call to emplace_back
could potentially trigger a reallocation if the current capacity is
exceeded, which is costly in terms of performance. This leads to more
efficient memory management and can significantly speed up the execution
time when dealing with a large number of scripts.
### Problem description
fix: pressing F5 (start pattern execution ) while being in a break point
hangs ImHex.
The reason for hanging ImHex was that the shortcut procedure was not
checking if the pattern was already running, and it attempted to start
another one. This makes ImHex wait indefinitely for a lock to be
released
### Implementation description
To fix the hanging of ImHex we check the runtime for current evaluation
and if detected we stop it. In all cases the evaluation is started
again.
### Problem description
Fixes provided for the following unreported bugs.
- Environment variables are set to be per provider but used as if they
are not. When a project is loaded all the environment variables for each
provider are assigned to the first provider making it impossible to add
new ones to the other providers.
- When switching providers, the text editor selection, the text editor
breakpoints, the console text, the console selection and the console
cursor position of the old provider are being assigned to the new
provider
### Implementation description
This PR aims at fixing both errors by:
- using variable defined to be per provider so that they affect their
provider only when necessary.
- creating new per provider variables and using them so that each
provider has their own console, selections and breakpoints.
In order to support the newly added per provided features new functions
were added to the text processor for selections and breakpoints. All the
new per provider variables are defined and used in view pattern editor.
Fixes https://github.com/WerWolv/ImHex/issues/1807
### Problem description
Build fails for MacOS x86-64 due to using a wrong architecture header.
### Implementation description
Used the headers for mach that auto select the correct architecture.
### Additional things
This is being used in https://github.com/NixOS/nixpkgs/pull/330303 as a
patch
Co-authored-by: Nik <werwolv98@gmail.com>
### Problem description
Moving error markers to be underwaved created s couple of errors:
- Markers that were 1 char long and located at the end of the line
didn't show on text editor.
- Markers that had zero length or no stack trace didn't show on text
editor.
### Implementation description
- The first error was caused by an off by one error in the column index.
- There was no implementation for errors that had no location or zero
length. Now errors with no location will be shown at the beginning of
the line where they occur. Errors with zero length will be marked from
their location to the end of the line, therefore, errors with no stack
trace will be marked from the beginning to the end of the line.
### Problem description
The only time textures would update was through the file picker. Once UV
coordinates were used, disabling them didn't have any effect because the
vertex array didn't invalidate the corresponding buffer array. When
`shouldUpdate` is true, the light source needs to know as well,
otherwise it will set the location at 0,0.
### Implementation description
Textures were fixed by creating a member variable that holds the file
name of the texture used in the last model rendering.
Instead of invalidating buffer arrays it is much simpler to define a
default UV coordinate set for the case when no UV is specified. if a
texture is not present then the values of UV will not be used. If there
is a texture it must be a minimum of 1 pixel in size. So we choose the
UV coordinates so that every vertex gets assigned the color at the 0,0
coordinate of the texture.
When `shouldUpdate` is on, we also turn `shouldUpdateLightSource` on.
Also included are some formatting changes that are purely aesthetic.
---------
Co-authored-by: Nik <werwolv98@gmail.com>
### Problem description
The process memory provider currently doesn't function correctly on
Linux due to incorrect handling of the special procfs file
`/proc/<pid>/maps`. I don't know if some of this behavior could vary by
distro and/or kernel version, but I've observed the following issues in
my Ubuntu 24.04 environment.
- The current code in master calls `file.readString()` which attempts to
determine the size of the file by [seeking to the
end](https://github.com/WerWolv/libwolv/blob/master/libs/io/source/io/file_unix.cpp#L148).
However, procfs files don't have a defined size, so this fails with a
return of -1. libwolv [interprets this as the file size and attempts to
allocate an enormous
buffer](https://github.com/WerWolv/libwolv/blob/master/libs/io/source/io/file.cpp#L30),
which results in an exception, so ultimately the process memory provider
is unusable on the current code.
- The previous version of the code that went out in 1.35.4 was calling
`readString` with a fixed maximum size of `0xF'FFFF`. This avoids the
seek issue, but when working with special files, a single `read` call
isn't guaranteed to read the requested number of bytes even if that many
bytes are available. In practice, on my machine, this call only ever
reads the first few dozen lines of the file. So the feature works in
this version, but it's unable to see the vast majority of the process'
address space.
- On a more minor note, on rows in the `maps` file that have a filename,
the filenames are visually aligned by padding spaces between the inode
column and filename column. ImHex includes these spaces as part of the
filename, resulting in most of the path being pushed out of the visible
area of the window.
### Implementation description
- To ensure the entire `maps` file is read, I've changed the code to
read from the file in a loop until we stop getting data. I've also set a
fixed limit on the maximum number of bytes to read in one go to avoid
issues with trying to determine the file size.
- I've added a `trim` call to remove any padding around the filename.
### Screenshots
Exception in `file.readString()` in current code (for some reason this
also causes the window to become transparent):
Abridged memory region list in 1.35.4:
Complete memory region list after this PR:
### Additional things
I was focused on fixing this ImHex feature here, but I wonder if some of
this should be addressed in libwolv. Maybe `readBuffer` in file_unix.cpp
should read in a loop until it has the requested number of bytes or
encounters EOF/error?
---------
Co-authored-by: Justus Garbe <55301990+jumanji144@users.noreply.github.com>
### Problem description
Currently, the pattern editor does not remember where the cursor is
located in each provider. For example, suppose you have 2 providers in
your project, and you scrolled down to line 200 in the first pattern to
make some changes and remembered that the code you want to insert is in
the second provider. Then you switch to the second provider, look for
the code and find it in line 235. Switch back to the first one, and you
are at the beginning of the file. So you again look for the line to edit
paste it to realize that it needs code a few lines before the place you
found it. You switch to the second provider, and you are at the top
again. This gets annoying very fast.
### Implementation description
This PR ensures that, when you return to the pattern in the editor for
any of the opened providers, the cursor will still be at the same place
it was when you switched to a different one. Each provider pattern saves
its cursor position and returns to it when you switch to that provider.
It does that by creating a PerProvider variable and using it when
providers are first opened to set it to the origin and when switching
providers it first saves the position of the old provider and then loads
and sets the saved position of the new provider.
---------
Co-authored-by: Nik <werwolv98@gmail.com>
changes needed were:
1) add an extra dummy empty widget so that width of child and parent are
not always the same. 2) Width of reset button needs to be adjusted when
the width of the child is decreased so that the parent is decreased too.
<!--
Please provide as much information as possible about what your PR aims
to do.
PRs with no description will most likely be closed until more
information is provided.
If you're planing on changing fundamental behaviour or add big new
features, please open a GitHub Issue first before starting to work on
it.
If it's not something big and you still want to contact us about it,
feel free to do so !
-->
### Problem description
<!-- Describe the bug that you fixed/feature request that you
implemented, or link to an existing issue describing it -->
### Implementation description
<!-- Explain what you did to correct the problem -->
### Screenshots
<!-- If your change is visual, take a screenshot showing it. Ideally,
make before/after sceenshots -->
### Additional things
<!-- Anything else you would like to say -->
<!--
Please provide as much information as possible about what your PR aims
to do.
PRs with no description will most likely be closed until more
information is provided.
If you're planing on changing fundamental behaviour or add big new
features, please open a GitHub Issue first before starting to work on
it.
If it's not something big and you still want to contact us about it,
feel free to do so !
-->
### Problem description
I was unable to use the LZMA de-compressor even though 7zip was able to
identify them as LZMA compressed and decompress them without a problem.
Under the debugger the `lzma_code()` call was returning
`LZMA_MEMLIMIT_ERROR`.
### Implementation description
I found online that the error can be fixed using `lzma_memlimit_set()`
using a value obtained with `lzma_memusage()`. I tried to avoid having
to call `lzma_code()` twice but if the functions are called before the
first `lzma_code()`the values of memory obtained fall short and error
occurs again.
I suspect that there are better ways to deal with this other than the
code proposed in this PR, but I haven't been able to find any.
---------
Co-authored-by: Nik <werwolv98@gmail.com>
### Problem description
There are some obvious constrains in the various array data sizes that
we can use to determine if their sizes and in some cases their values
are correct.
### Implementation description
To test their validity, the most important of the various arrays are the
indices, so their definition and use has been simplified a great deal.
In order to treat all variables in a similar manner the changes to the
uv variable posted in an earlier PR were also added here.
With this one, all the recently opened PR's combined should contain
exactly the same changes and fixes as the older 3-d visualizer PR #1850
which I will close shortly after this one is opened.
---------
Co-authored-by: Nik <werwolv98@gmail.com>
### Problem description
Currently when errors are found the entire line where the error occurred
is highlighted and one has to look at the error message in order to find
where the error is located on the line. With this PR the line will no
longer be highlighted and the location of the error will be marked with
an red waved line under the error location. Hovering over the text where
the error occurred produces an error overlay so if several errors occur
on the same line they can all be seen separately.
### Implementation description
The definition of error marker was switched to include column and size
as well as line and message like before.
This change required changing the way view pattern editor draws the
error markers because the errors themselves don't have size information.
Also, a new errorHoverBoxes type was defined to help in the detection of
the floating error messages when error is hovered.
Note that the underwave code depends on having a monospaced. If font is
not monospaced the underwaved text can be short/long or displaced.
### Screenshots
### Problem description
There are some recent issues about Mac keys not configured properly for
the pattern editor. This PR moves all the shortcuts to the shortcut
manager, so they can be edited at will. Even if the key is not
identified correctly it should be possible to use preferred keys for any
action.
---------
Co-authored-by: Nik <werwolv98@gmail.com>
<!--
Please provide as much information as possible about what your PR aims
to do.
PRs with no description will most likely be closed until more
information is provided.
If you're planing on changing fundamental behaviour or add big new
features, please open a GitHub Issue first before starting to work on
it.
If it's not something big and you still want to contact us about it,
feel free to do so !
-->
### Problem description
https://github.com/WerWolv/ImHex/issues/1895
### Implementation description
Added code that handles streamed zstd data. It is based around the
[official
documentation](http://facebook.github.io/zstd/zstd_manual.html) and the
[example](https://github.com/facebook/zstd/blob/dev/examples/simple_decompression.c)
provided at the main zstd repo.
The loop around the non-streamed version was also removed because I
don't think it was doing anything (no `continue`s, `sourceSize` was
always being set to 0).
### Additional things
To test, I generated streamed zstd data with this python script:
```py
import io; import pyzstd;
with open("data.zstd", "wb") as f:
pyzstd.compress_stream(io.BytesIO(b'ab' * 100), f)
```
And then I ran this pattern script:
```
import std.mem;
import hex.dec;
u8 data[while(!std::mem::eof())] @ 0x00;
std::mem::Section data_sec = std::mem::create_section("data_sec");
hex::dec::zstd_decompress(data, data_sec);
```
Inspecting the section shows the correct data:
Co-authored-by: Nik <werwolv98@gmail.com>
### Problem description
Older image format use to store color values in a small lookup table so
that the image could simply store an index to the table. The pattern
language is not designed to handle this sort of operation which makes
this extension necessary to be able to visualize images of this type.
### Implementation description
The changes add an optional parameter to the bitmap visualizer which
holds the color lookup table. If the image contains values that are
outside the range of possible colors then the first color in the map is
chosen. The dimensions of the image can be equal to or smaller than
rows*columns depending on how the indices to the color map are stored.
For example, you can use 4 bit indices which would make the image half
the size (in bytes) but that limits the number of colors to 16. To store
colors in sizes larger than one byte use an array of the appropriate
sized integers.
### Screenshots
### Problem description
1. Old Chinese text json, the order is chaotic. It doesn't compare well
with English.
2. Missing a batch of new English strings.
### Implementation description
1. Synchronize the key sequence of English files.
2. Added a missing batch of text, and translate to CN.
### Screenshots
None
### Additional things
None
### Problem description
Attempting to do an MD5 hash of a large region (e.g. 2 MiB, ``u8
data[0x200000]``) crashes with a segfault.
### Implementation description
In ``hex::plugin::hashes::hashProviderRegionWithHashLib()``,
``hashFunction->TransformBytes()`` is called with an offset of 0,
because it iterates over ``data`` and not the entire region.
