Reading a diff of any size means moving in bigger steps than a line at a
time, which the staging view offers and the main view didn't. So the same
hunk keys work here, and n / N step from file to file — worth having only
here, since a diff spanning several files is something the staging view
never showed.
Where "the next file" begins isn't in the text: a diff renderer may print
whatever it likes above a file's content. So navigation lands on the first
row that states which file it belongs to, which is the file's header when
the source says so and its first content line otherwise. The anchor's own
file is found by scanning down rather than to the nearest row either way —
having just landed on a header, the nearest row above belongs to the file
we came from, and taking it would send the next press back where we
started.
A file you go to is brought to the top of the view, since the file is what
you went there for and the more of it is on screen the better. That only
applies where the view has to scroll at all: a file already on screen
leaves it where it is. In hunk mode what ends up selected is the file's
first change rather than the row the file begins at, and a large context
size can put that change further down than a screenful; the selection is
scrolled into view afterwards as any other jump's is, and the alignment
gives way where the two can't both hold.
The diff loads lazily, so a target below the loaded portion isn't there to
be found; rather than doing nothing, we read the rest in and look again.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Selecting a range with the mouse is the obvious gesture once clicking
selects a line, and it's the only way to get a range without knowing the
keybindings.
The range has to be anchored where the mouse went down, which the view
can't tell us: a click in hunk mode selects a whole block, leaving the
view's own range anchor at the block's far end, so a drag from there would
grow the selection from the wrong end. So the clicked line is remembered
when the click happens, and the drag anchors there.
A drag that reaches the edge of the view keeps scrolling, using the same
autoscroller the staging view does — mouse capture means the pointer can be
dragged past the edge, and there is more diff down there than fits on
screen. Unlike the staging view, whose content is a string that is always
there in full, this diff loads lazily, so scrolling down has to keep
reading it in or the autoscroll would stop at the loaded edge.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Users who set hunk mode as their default get it when entering the staging
view, and the focused main view is on its way to replacing that view, so it
has to behave the same: focusing selects a whole change block, and clicking
a change line selects that line's block, ready to act on. A click on a
context line still selects only that line, since the click points at it
precisely — you may well want to edit it. A click inside the block that is
already selected does the same, giving hunk mode up and going back to line
by line. That keeps a single line reachable with the mouse in a block too
long to see the end of, and matches how a click inside a range selection
already collapses it.
The block offered up is the first one that begins on screen, so that its
whole extent can be seen before acting on it. Only when none does — a change
too long to fit on the screen — is the block that reaches into the view from
above taken instead, and the selection then extends to its first line off
screen: focusing a diff must not move it, so nothing here scrolls.
The staging view makes one exception, and so must we, or a file that is one
solid block of changes (a file you just created, or deleted) would come up
with all of it selected. The staging view asks the parsed patch; we ask the
rendered diff the same question, which needs no second git invocation and
works over a whole commit's diff, where the answer differs per file.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Focusing the main view is about to establish a hunk selection, and that
happens in a free function shared with the controller that focuses the
main view from a side panel, which has no main view controller to hand.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The mode of a diff view's selection lives on the context of the pane showing
it, and until now only the controllers that drive the selection needed to
name that pane, all of which have it as the concrete context it is.
A side panel is about to be handed the pane a command was invoked in, so
that it can act on what is selected there — and it can't be handed a
context, the interface it is handed one through being declared where a
context's concrete type isn't available. So the state moves to where such an
interface can speak of it, and the panes gain one: a context that has a diff
selection.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The focused main view could be scrolled but not pointed at: there was no
way to say "this line" or "these lines", which is what every line-level
interaction needs — editing a line, copying part of a diff, and, later,
staging. So a diff main view now always carries a selection while focused,
starting at the first change already on screen so that focusing doesn't
move the view.
The mode of the selection lives on the main context (a single line, a
range from a fixed anchor, or the change block around the cursor); the
selected line and the range anchor stay in the view itself, whose native
range select draws them, so there is no new highlight machinery. The modes
and their keys are the ones the staging view has, including that the arrow
keys step from block to block in hunk mode.
A pane holds something to select only when what it is showing is a diff
with changes in it, which the render works out once its content is final: a
placeholder message is not a diff, and neither is a diff with nothing in it
at all, such as a binary file's or an empty commit's. Whether a selection
is then drawn there follows from the context stack, as it does for every
other view.
A click in the focused main view now has a meaning again: it selects the
line it points at.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
ReadToEnd holds a gocui task while it reads, so that lazygit doesn't count as
idle while something waits on the result. That has nothing to do with reading
all the way to the end, and the next commit needs it for a bounded read too, so
the two are pulled apart.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The count comes from the view's buffer, which a rendering task appends to on
its own goroutine, so reading it without the write mutex is a data race. Nobody
called it until now, which is why nothing has tripped over it; the next commit
does, from the UI thread while a render is still loading.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Both callers pass a main context, and focusing one is about to need more
of it than the Context interface offers. Saying so in the signature also
retires the type assertion that was there only to reach ClearSearchString.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
With both implementations gone, nothing is left that lets a side panel
handle a click in the focused main view, so the mechanism for attaching
one to a context can go as well.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Clicking a line of the focused main view entered the staging or patch
building panel at that line. The focused main view is about to gain a
selection of its own, which is what a click there should set — and with
the explorers on their way out, the dive has nowhere to go.
Nothing replaces the gesture yet, so for the next couple of commits a
click in the focused main view does nothing.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The focused main view is about to show a selection, but only where there
are diff lines to act on: a branch's commit log or the status dashboard
has nothing to select. Rather than have the main view guess from the
rendered content, let the side panels say so, since each of them knows
what it renders.
The classification is finer than a yes/no because acting on a selection
means different things per panel — staging into the working tree for the
files panel, taking lines into a custom patch for the commit panels — and
those actions want the same one answer as this. Only whether a panel
shows a diff at all is read for now.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The focused main view is about to get a real diff selection, which makes
its up/down/page/top/bottom keys mode-aware: what they do depends on the
select state the main view controller owns. Keeping them in a separate
controller would mean either duplicating that state or reaching across
controllers for it, so move them to where the state will live. The two
controllers were attached to the same pair of contexts and nothing else,
so nothing else can notice.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
A renderer that speaks the protocol emits nothing unless it is asked to,
so that its output stays plain wherever it is used outside lazygit. The
variable names the versions we understand.
git is one of the renderers we ask. It has no pager to spawn, and so no
terminal to spawn one in, but it still renders the diff itself — for the
word-diff formats, whose markup nothing else could resolve — so the
request has to be made before we decide a pty isn't needed.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Where a renderer states which line of which file it is rendering, take
that as the line's identity. The renderer knows, and for a rendering
that no longer looks like a diff nothing else does. Reading the rendered
text stays the way for the renderings that keep a diff's structure, and
for the renderers that say nothing at all.
Which of the two is used is settled for the rendering as a whole, not
row by row. A rendering with records is not parsed at all, not even for
the rows the renderer says nothing about. Its text is no unified diff,
and a row of it can read like one. Under delta, a commit that adds a
test whose input is a diff shows the test's "diff --git a/img.png
b/img.png" line as a row of its own; parsed, that row opens a file
section that runs to the end of the rendering, and every row delta puts
between hunks and files comes out as a header of a file called img.png.
So the untagged rows of such a rendering stay unresolved, as the
protocol has it (spec section 6.4).
Header records are accepted too. This lets a renderer point at a file
with no content lines, such as a pure rename, a mode change or a binary
file: for those files the header is the only row in the diff.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A diff line's identity is about to become recoverable from a second
source, a diff renderer's own records, and a renderer states the path
however it likes — absolute paths included. The field can't promise
repo-relative any more.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Wherever two diff lines end up on one rendered line, a renderer emits
their records back to back: the deletion and the addition of a
modification collapsed into a single column, or a banner announcing a
file and its first hunk at once. A changed line that is empty is
rendered as its record alone. Attaching a record only to the cells it
precedes loses all of these — the last record of a run wins, and an
empty changed line becomes a line we can say nothing about at all.
Give such a record a cell of its own instead. It renders nothing, so the
diff looks the same, but the line keeps every record it was given.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Before the diff, a conforming renderer emits one OSC 1717 record that
carries only the version — its way of announcing that it speaks the
protocol at all, without a host having to inspect what it renders. It
describes no line, so keeping it would give the first line of the diff a
record that says nothing about it.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
A diff renderer that restructures the diff — into columns, or with the
+/- markers replaced by colour — leaves us no way to tell which line of
which file a rendered row came from, which is what acting on the row
requires. The OSC 1717 protocol has the renderer say so directly: it
prefixes each line it renders with a record naming the file and the
line's position in the old and new versions of it.
Attach each record to the cells it precedes, so that a row's records
survive wrapping and the columns of a side-by-side rendering, and hand
them to readers together with the row's text: the two have to describe
the same buffer, and a re-render can rebuild it between two reads.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The OSC parser dispatched on a single character, so only the
single-digit OSC 8 could ever be recognized; the diff-line metadata
protocol we are about to read uses OSC 1717.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
git doesn't just print the path: it terminates it with a tab when the
path contains a space, and C-quotes the whole field when the path
contains anything it won't print raw — with core.quotePath, which is on
by default, that means any non-ASCII byte, so a file called café shows
up as "b/caf\303\251".
Taking the field verbatim therefore gives a path that doesn't exist, for
a whole class of perfectly ordinary file names. The quoting is Go's own
string syntax, octal escapes and all, so decoding it is a call to
strconv.Unquote.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
To act on the line the user is pointing at in a diff view — to stage it,
to open it in an editor, to keep the cursor on it while the diff is
regenerated — we need to know which file it belongs to and where it sits
in that file. Only the diff the view was rendered from knows that, and
by the time it's on screen all we have is text.
So parse it back: the view's contents are (usually) a unified diff, and
running them through the patch parser recovers each row's file, kind and
line numbers. "Usually" is why this goes behind a seam, and why it
answers "I don't know" rather than guessing: a diff renderer is free to
restructure what it prints, and a wrong answer here means acting on the
wrong line of the wrong file.
Parsing a whole buffer is a separate entry point from parsing a single
line, because a caller resolving every row of a large diff must not
re-parse a file's section once per line of it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A reader that parses a view's content rather than showing it wants the
text the writer wrote, and the cells don't always spell it. A tab is
expanded into the spaces it fills, so a line read back from the cells
ends in one to four spaces where the writer put a tab. A carriage
return moves the write cursor back to the start of the line, so the
text written after it overwrites what came before.
The parser of the main view's diff, which the next commit adds, meets
the first case in every header of a file whose path contains a space.
git terminates the path field of a "---" or "+++" line with a tab
then, and a parser reading the cells takes the spaces the tab became
for part of the path. That path names a file that doesn't exist, so
the file's lines can't be acted on.
Keep the text as written per line, from the first character on that
the cells spell differently, so that a line without a tab or a
carriage return costs nothing. LinesAsWritten hands it out the way
BufferLines hands out the cells' text.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
OverwriteLines is asked for a line and writes the one below it when the
write before it ended in a newline. The view holds such a newline back
until more content arrives, so that it doesn't end in an empty line,
and OverwriteLines moved the write cursor without letting go of it, so
the write that followed advanced to the next line first.
Move the cursor through SetWritePos, which drops the pending newline
along with it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A view holds back the newline that ends a write until more content
arrives, so that it doesn't end in an empty line. OverwriteLines moves
the write cursor to the line it is given without letting go of that
pending newline, so the write that follows advances first and lands on
the line below. Nothing in lazygit overwrites lines right after such a
write today.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Everything about a diff's content — which file and line a row belongs
to, whether it's a change — is a property of the unwrapped buffer line,
while the cursor, clicks and the range selection all speak in view
lines, which count wrapped segments. Reading the content under the
cursor therefore needs the mapping in both directions, and doing it
outside gocui isn't possible: the wrapping is internal, and the caller
couldn't take the view's lock across the lookup and the read.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Parse is lenient: it takes any text and reads a diff out of it, which is
what we want when we hand it a diff, but it has no way to say "this
isn't one". We're about to parse the *rendered* contents of a diff view,
which a diff renderer is free to restructure — putting the line numbers
in a gutter, say, shifts the +/- marker off the start of each body line,
so every line reads as context and the parse silently lies about which
lines are changes.
Comparing each hunk's body against the lengths its header declares
catches exactly that, without teaching us anything about any particular
renderer's layout: a faithful unified diff agrees with its headers, a
restructured one doesn't.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Identifying a change line of a diff by its file line number needs both
sides: two consecutive deletions sit at the same new-file position, so
only their old-file line numbers tell them apart. LineNumberOfLine only
answers for the new file, which leaves deletions ambiguous.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Wrapping the content at another width moves every line of it to a
different view line. The positions into the view are all view lines: the
scroll offset, the cursor, a range's anchor. Each of them is then left
pointing at a line it was never on. Committing the last of the staged
changes widens the main view by half a screen, and that moves a selected
hunk somewhere else entirely.
Carry the positions through the lines of content they were on. A position
always meant a line of content rather than a view line. The line the
cursor is on keeps the row it was drawn on, so it stays in front of the
user rather than the view scrolling under it; a view with no cursor on
screen keeps its own place instead. A range's ends go on the outermost
segments of their lines, since a range covers lines of content and not
the segments those lines are drawn as.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The scroll offset, the cursor and a range's anchor are all view lines, which
count the segments each line of the content is wrapped into. A change of
width wraps the content differently, so every one of them ends up on a
different line than the one it was put on.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
A test asks which lines of a view are selected; a wrapping view's cursor and
range anchor answer in view lines, which count the segments each line is
drawn as. Going through the segment-to-line mapping keeps the answer in the
terms the question was asked in, and a line the selection covers several
segments of is reported once.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
The lines a selection covers are asked for by view line, which counts the
segments a wrapping view breaks a line into, and then used to index the
content, whose lines are unwrapped. The two agree only for a view whose
content doesn't wrap.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Tags can only be pushed one at a time. Let the push command of the tags
panel work on a range selection of tags too. It asks for one remote and
pushes all selected tags to it with a single git command.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This lets pushing a range selection of tags push them all with a single
git command.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Tags can only be deleted one at a time. If many of them have piled up,
for example backup tags made before rewriting a branch, deleting them
takes a lot of keypresses.
Let the delete menu work on a range selection of tags. It deletes the
selected tags with one git command for the local tags and one for the
remote tags, and asks for a single remote for all of them. After
deleting local tags, collapse the range selection to its first line;
otherwise it would select the tags that moved up into the place of the
deleted ones.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Deleting a remote tag and deleting a local and remote tag each have
their own copy of the code that asks for the remote and for a
confirmation. Deleting a range selection of tags needs to change it in
the same way for both.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This lets deleting a range selection of tags delete them all with a
single git command, the same way it works for branches.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Deleting and pushing a range selection of tags will need to show their
operation on all the selected tags. Move the code from BranchesHelper
into a generic function that takes the items and the key of the context
that shows them.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
When somebody else rebases a stack of branches and force-pushes it,
pulling the topmost branch leaves the branches below it diverged. Since
#6067 they can be updated manually with `f`, but pushing the stack
offers to push the branches
below along with the current one, so pulling should be symmetric, and
this PR adds that.
When somebody else rebases a stack and force-pushes it, pulling the
topmost branch leaves the branches below it diverged. They can be
updated with `f`, but pushing the stack offers to push the branches
below along with the current one, and pulling should be symmetric.
When pulling, find the branches stacked below the current one that can
be updated without losing anything. These are those that are behind
their upstream branches, and those that diverged from them only because
they were rewritten. If there are any, show a menu that lists them and
offers to update them in addition to pulling the current branch, or to
pull only the current branch. The current branch is pulled as before.
The decision is based on the last fetch, not on a new one, so that the
menu appears right away. If the remote has changed since then, a branch
might not be offered although it could be updated; it then shows as
diverged after the pull, and pulling again offers it. The update itself
fetches the upstream branches and checks them again, so it never loses
commits.
Branches that are checked out in another worktree are not offered,
since updating them would change the files there.
The branches below are updated before the current branch is pulled. If
one of them pointed into the commits that a rebasing pull rebases, the
pull would move it when rebase.updateRefs is set, and updating it
afterwards would fail. If updating them fails, the current branch isn't
pulled.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Pushing a stack and fast-forwarding a range of branches each show their
operation on all the branches involved, with their own copy of the same
code. Pulling a stack will need it too, so move it to BranchesHelper.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
FastForwardBranches looks up the worktrees of the branches in the model
on the UI thread, and then does the rest on a worker, with the branches
shown as being fast-forwarded. The next commit needs to update branches
as part of pulling the current one, in the pull's own task.
Move everything except the inline status into PrepareFastForward, which
returns the part to run on a worker.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Lazygit can auto-forward main branches after fetching, which is useful
to always keep your `main` branch up to date. However, checking out
`main` in another worktree stopped this from happening, on the
assumption that we don't want to touch branches that are checked out
somewhere. For a main branch this is not an issue (as long as the
worktree doesn't have modified files); forwarding it in that case too is
useful for the setup where you always keep main checked out in the main
worktree, and work on feature branches in linked worktrees.
Pressing `f` on such a branch already supported fast-forwarding it even
when checked out in a worktree, so now auto-forwarding after fetching
does the same.
If a main branch is checked out in another worktree, auto-forwarding
after a fetch skips it. People who keep the main branch checked out in
their main worktree, and work on feature branches in linked worktrees,
therefore never get it forwarded.
Fast-forward such a branch in the worktree it is checked out in, as `f`
does. Skip the worktree if it has changes to tracked files, since
nobody asked for its files to change. Also skip it if it is mid-rebase
or mid-bisect, because its HEAD is then detached from the branch. A
branch in the current worktree is never forwarded. This covers the case
of the current worktree being mid-rebase on a main branch, whose ref
would otherwise be forwarded under the rebase.
Auto-forwarding now writes the same reflog message as `f`
("lazygit: update to upstream branch").
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
If `f` is used on several branches that are checked out in other
worktrees, the first worktree in which the fast-forward fails stops the
whole operation. For example, an untracked file that the fast-forward
would overwrite leaves the branches after it untouched, even though
nothing is wrong with them.
Try all the branches, and report the errors of the failed ones
together. The checks that refuse to fast-forward any of the branches
still run before anything is moved.
The next commit uses forwardBranches for auto-forwarding branches in
other worktrees, where stopping at the first failure is worse. A
worktree that fails every time would prevent the ones after it from
ever being forwarded. For this, forwardBranches also takes the git
commands to use, since a background worker doesn't block switching
repos.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
AutoForwardBranches runs in the Then callback of the refresh after a
fetch. This is on the UI thread, so the git command for updating the
branches blocks the UI while it runs. This is fine for a single
update-ref, but the next commit adds a git status and a git merge for
each branch that is checked out in another worktree.
Keep reading the branches from the model on the UI thread, and move the
git work to a worker.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Lazygit shows local branches that have diverged from their upstream with
a yellow `↓3↑5`. There can be two reasons for diverging, and right now
it's impossible to tell which of these is the case: (1) you rebased the
branch locally (or rewrote its history), in which case you want to
force-push it, or (2) someone else rebased the branch, in which case you
want to pull it. (We'll ignore the case where both of these happened, in
which case you need to manually reconcile the work that each side did,
e.g. through cherry-picking; you want to avoid this situation.)
With this PR, lazygit distinguishes the two scenarios, and for (2) it
shows the `↓3↑5` in a dim yellow. What's more, you don't even have to
check out the branch to pull it; you can press `f` to reset it to its
upstream without checking it out, like you would fast-forward a branch
that is behind its upstream. In fact the command is still called
"Fast-forward", which is technically not quite correct, but it feels the
same to me. Fast-forwarding a diverged branch in this way is very useful
if an entire stack of branches was rebased remotely; you can simply
range-select all those branches and hit `f` to bring them up to date
with their upstreams.
The worktree loader associates a worktree that is mid-rebase or
mid-bisect with the branch involved, even though its HEAD is detached.
Fast-forwarding such a branch therefore ran the fast-forward in that
worktree. This moved the detached HEAD and put the upstream commits
into the rebase in progress, while the branch stayed where it was.
Remember in the worktree model when its branch comes from a rebase or
bisect, and refuse to fast-forward the branch in that case. Updating
only the ref is no option for a rebase; the rebase writes the branch
when it finishes.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
If a branch is being rebased in a worktree, pressing `f` on it moves
the detached HEAD of that worktree instead of the branch. The upstream
commits end up in the rebase in progress, and the branch itself stays
where it was. The same happens for a branch being bisected.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
