vim-patch:9.1.1243: diff mode is lacking for changes within lines

Problem:  Diff mode's inline highlighting is lackluster. It only
          performs a line-by-line comparison, and calculates a single
          shortest range within a line that could encompass all the
          changes. In lines with multiple changes, or those that span
          multiple lines, this approach tends to end up highlighting
          much more than necessary.

Solution: Implement new inline highlighting modes by doing per-character
          or per-word diff within the diff block, and highlight only the
          relevant parts, add "inline:simple" to the defaults (which is
          the old behaviour)

This change introduces a new diffopt option "inline:". Setting to
"none" will disable all inline highlighting, "simple" (the default) will
use the old behavior, "char" / "word" will perform a character/word-wise
diff of the texts within each diff block and only highlight the
differences.

The new char/word inline diff only use the internal xdiff, and will
respect diff options such as algorithm choice, icase, and misc iwhite
options. indent-heuristics is always on to perform better sliding.

For character highlight, a post-process of the diff results is first
applied before we show the highlight. This is because a naive diff will
create a result with a lot of small diff chunks and gaps, due to the
repetitive nature of individual characters. The post-process is a
heuristic-based refinement that attempts to merge adjacent diff blocks
if they are separated by a short gap (1-3 characters), and can be
further tuned in the future for better results. This process results in
more characters than necessary being highlighted but overall less visual
noise.

For word highlight, always use first buffer's iskeyword definition.
Otherwise if each buffer has different iskeyword settings we would not
be able to group words properly.

The char/word diffing is always per-diff block, not per line, meaning
that changes that span multiple lines will show up correctly.
Added/removed newlines are not shown by default, but if the user has
'list' set (with "eol" listchar defined), the eol character will be be
highlighted correctly for the specific newline characters.

Also, add a new "DiffTextAdd" highlight group linked to "DiffText" by
default. It allows color schemes to use different colors for texts that
have been added within a line versus modified.

This doesn't interact with linematch perfectly currently. The linematch
feature splits up diff blocks into multiple smaller blocks for better
visual matching, which makes inline highlight less useful especially for
multi-line change (e.g. a line is broken into two lines). This could be
addressed in the future.

As a side change, this also removes the bounds checking introduced to
diff_read() as they were added to mask existing logic bugs that were
properly fixed in vim/vim#16768.

closes: vim/vim#16881

vim/vim@9943d47

Co-authored-by: Yee Cheng Chin 

Author: zeertzjq

runtime/colors/vim.lua

@@ -47,6 +47,7 @@ hi('WildMenu',     { fg = 'Black', bg = 'Yellow',       ctermfg = 'Black', cterm
 hi('VertSplit',      { link = 'Normal' })
 hi('WinSeparator',   { link = 'VertSplit' })
 hi('WinBarNC',       { link = 'WinBar' })
+hi('DiffTextAdd',    { link = 'DiffText' })
 hi('EndOfBuffer',    { link = 'NonText' })
 hi('LineNrAbove',    { link = 'LineNr' })
 hi('LineNrBelow',    { link = 'LineNr' })

runtime/doc/diff.txt

@@ -214,14 +214,29 @@ The diffs are highlighted with these groups:
 |hl-DiffAdd|	DiffAdd		Added (inserted) lines.  These lines exist in
 				this buffer but not in another.
 |hl-DiffChange|	DiffChange	Changed lines.
-|hl-DiffText|	DiffText	Changed text inside a Changed line.  Vim
-				finds the first character that is different,
-				and the last character that is different
-				(searching from the end of the line).  The
-				text in between is highlighted.  This means
-				that parts in the middle that are still the
-				same are highlighted anyway.  The 'diffopt'
-				flags "iwhite" and "icase" are used here.
+|hl-DiffText|	DiffText	Changed text inside a Changed line.  Exact
+				behavior depends on the `inline:` setting in
+				'diffopt'.
+				With `inline:` set to "simple", Vim finds the
+				first character that is different, and the
+				last character that is different (searching
+				from the end of the line).  The text in
+				between is highlighted.  This means that parts
+				in the middle that are still the same are
+				highlighted anyway.  The 'diffopt' flags
+				"iwhite" and "icase" are used here.
+				With `inline:` set to "char" or "word", Vim
+				uses the internal diff library to perform a
+				detailed diff between the changed blocks and
+				highlight the exact difference between the
+				two.  Will respect any 'diffopt' flag that
+				affects internal diff.
+				Not used when `inline:` set to "none".
+|hl-DiffTextAdd|  DiffTextAdd	Added text inside a Changed line. Similar to
+				DiffText, but used when there is no
+				corresponding text in other buffers.  Will not
+				be used when `inline:` is set to "simple" or
+				"none".
 |hl-DiffDelete|	DiffDelete	Deleted lines.  Also called filler lines,
 				because they don't really exist in this
 				buffer.

runtime/doc/news.txt

@@ -66,7 +66,7 @@ EVENTS
 
 HIGHLIGHTS
 
-• todo
+• |hl-DiffTextAdd| highlights added text within a changed line.
 
 LSP
 
@@ -78,7 +78,7 @@ LUA
 
 OPTIONS
 
-• todo
+• 'diffopt' "inline" configures diff highlighting for changes within a line.
 
 PLUGINS
 

runtime/doc/options.txt

@@ -2056,7 +2056,7 @@ A jump table for the options with a short description can be found at |Q_op|.
 	security reasons.
 
 						*'diffopt'* *'dip'*
-'diffopt' 'dip'		string	(default "internal,filler,closeoff,linematch:40")
+'diffopt' 'dip'		string	(default "internal,filler,closeoff,inline:simple,linematch:40")
 			global
 	Option settings for diff mode.  It can consist of the following items.
 	All are optional.  Items must be separated by a comma.
@@ -2119,6 +2119,21 @@ A jump table for the options with a short description can be found at |Q_op|.
 				Use the indent heuristic for the internal
 				diff library.
 
+		inline:{text}	Highlight inline differences within a change.
+				See |view-diffs|.  Supported values are:
+
+				none    Do not perform inline highlighting.
+				simple  Highlight from first different
+					character to the last one in each
+					line. This is the default if nothing
+					is set.
+				char    Use internal diff to perform a
+					character-wise diff and highlight the
+					difference.
+				word    Use internal diff to perform a
+					|word|-wise diff and highlight the
+					difference.
+
 		internal	Use the internal diff library.  This is
 				ignored when 'diffexpr' is set.  *E960*
 				When running out of memory when writing a

runtime/doc/syntax.txt

@@ -5177,8 +5177,11 @@ DiffChange	Diff mode: Changed line. |diff.txt|
 DiffDelete	Diff mode: Deleted line. |diff.txt|
 							*hl-DiffText*
 DiffText	Diff mode: Changed text within a changed line. |diff.txt|
+							*hl-DiffTextAdd*
+DiffTextAdd	Diff mode: Added text within a changed line.  Linked to
+		|hl-DiffText| by default. |diff.txt|
 							*hl-EndOfBuffer*
-EndOfBuffer	Filler lines (~) after the end of the buffer.
+EndOfBuffer	Filler lines (~) after the last line in the buffer.
 		By default, this is highlighted like |hl-NonText|.
 							*hl-TermCursor*
 TermCursor	Cursor in a focused terminal.

runtime/doc/vim_diff.txt

@@ -48,7 +48,7 @@ Defaults                                            *defaults* *nvim-defaults*
 - 'complete' excludes "i"
 - 'completeopt' defaults to "menu,popup"
 - 'define' defaults to "". The C ftplugin sets it to "^\\s*#\\s*define"
-- 'diffopt' defaults to "internal,filler,closeoff,linematch:40"
+- 'diffopt' defaults to "internal,filler,closeoff,inline:simple,linematch:40"
 - 'directory' defaults to ~/.local/state/nvim/swap// (|xdg|), auto-created
 - 'display' defaults to "lastline"
 - 'encoding' is UTF-8 (cf. 'fileencoding' for file-content encoding)

runtime/lua/vim/_meta/options.lua

@@ -60,7 +60,7 @@ syn case ignore
 syn keyword vimGroup contained	Comment Constant String Character Number Boolean Float Identifier Function Statement Conditional Repeat Label Operator Keyword Exception PreProc Include Define Macro PreCondit Type StorageClass Structure Typedef Special SpecialChar Tag Delimiter SpecialComment Debug Underlined Ignore Error Todo
 
 " Default highlighting groups {{{2
-syn keyword vimHLGroup contained ErrorMsg IncSearch ModeMsg NonText StatusLine StatusLineNC EndOfBuffer VertSplit DiffText PmenuSbar TabLineSel TabLineFill Cursor lCursor QuickFixLine CursorLineSign CursorLineFold CurSearch PmenuKind PmenuKindSel PmenuMatch PmenuMatchSel PmenuExtra PmenuExtraSel ComplMatchIns Normal Directory LineNr CursorLineNr MoreMsg Question Search SpellBad SpellCap SpellRare SpellLocal PmenuThumb Pmenu PmenuSel SpecialKey Title WarningMsg WildMenu Folded FoldColumn SignColumn Visual DiffAdd DiffChange DiffDelete TabLine CursorColumn CursorLine ColorColumn MatchParen StatusLineTerm StatusLineTermNC CursorIM LineNrAbove LineNrBelow
+syn keyword vimHLGroup contained ErrorMsg IncSearch ModeMsg NonText StatusLine StatusLineNC EndOfBuffer VertSplit DiffText DiffTextAdd PmenuSbar TabLineSel TabLineFill Cursor lCursor QuickFixLine CursorLineSign CursorLineFold CurSearch PmenuKind PmenuKindSel PmenuMatch PmenuMatchSel PmenuExtra PmenuExtraSel ComplMatchIns Normal Directory LineNr CursorLineNr MoreMsg Question Search SpellBad SpellCap SpellRare SpellLocal PmenuThumb Pmenu PmenuSel SpecialKey Title WarningMsg WildMenu Folded FoldColumn SignColumn Visual DiffAdd DiffChange DiffDelete TabLine CursorColumn CursorLine ColorColumn MatchParen StatusLineTerm StatusLineTermNC CursorIM LineNrAbove LineNrBelow
 syn match vimHLGroup contained "\<Conceal\>"
 syn keyword vimOnlyHLGroup contained	Menu Scrollbar ToolbarButton ToolbarLine Tooltip VisualNOS
 syn keyword nvimHLGroup contained	FloatBorder FloatFooter FloatTitle MsgSeparator NormalFloat NormalNC Substitute TermCursor VisualNC Whitespace WinBar WinBarNC WinSeparator

runtime/syntax/vim.vim

@@ -743,25 +743,53 @@ struct file_buffer {
 // Stuff for diff mode.
 #define DB_COUNT 8     // up to four buffers can be diff'ed
 
-// Each diffblock defines where a block of lines starts in each of the buffers
-// and how many lines it occupies in that buffer.  When the lines are missing
-// in the buffer the df_count[] is zero.  This is all counted in
-// buffer lines.
-// There is always at least one unchanged line in between the diffs.
-// Otherwise it would have been included in the diff above or below it.
-// df_lnum[] + df_count[] is the lnum below the change.  When in one buffer
-// lines have been inserted, in the other buffer df_lnum[] is the line below
-// the insertion and df_count[] is zero.  When appending lines at the end of
-// the buffer, df_lnum[] is one beyond the end!
-// This is using a linked list, because the number of differences is expected
-// to be reasonable small.  The list is sorted on lnum.
+/// Each diffblock defines where a block of lines starts in each of the buffers
+/// and how many lines it occupies in that buffer.  When the lines are missing
+/// in the buffer the df_count[] is zero.  This is all counted in
+/// buffer lines.
+/// There is always at least one unchanged line in between the diffs (unless
+/// linematch is used).  Otherwise it would have been included in the diff above
+/// or below it.
+/// df_lnum[] + df_count[] is the lnum below the change.  When in one buffer
+/// lines have been inserted, in the other buffer df_lnum[] is the line below
+/// the insertion and df_count[] is zero.  When appending lines at the end of
+/// the buffer, df_lnum[] is one beyond the end!
+/// This is using a linked list, because the number of differences is expected
+/// to be reasonable small.  The list is sorted on lnum.
+/// Each diffblock also contains a cached list of inline diff of changes within
+/// the block, used for highlighting.
 typedef struct diffblock_S diff_T;
 struct diffblock_S {
   diff_T *df_next;
   linenr_T df_lnum[DB_COUNT];           // line number in buffer
   linenr_T df_count[DB_COUNT];          // nr of inserted/changed lines
   bool is_linematched;  // has the linematch algorithm ran on this diff hunk to divide it into
                         // smaller diff hunks?
+
+  bool has_changes;     ///< has cached list of inline changes
+  garray_T df_changes;  ///< list of inline changes (diffline_change_T)
+};
+
+/// Each entry stores a single inline change within a diff block. Line numbers
+/// are recorded as relative offsets, and columns are byte offsets, not
+/// character counts.
+/// Ranges are [start,end), with the end being exclusive.
+typedef struct diffline_change_S diffline_change_T;
+struct diffline_change_S {
+  colnr_T dc_start[DB_COUNT];       ///< byte offset of start of range in the line
+  colnr_T dc_end[DB_COUNT];         ///< 1 paste byte offset of end of range in line
+  int dc_start_lnum_off[DB_COUNT];  ///< starting line offset
+  int dc_end_lnum_off[DB_COUNT];    ///< end line offset
+};
+
+/// Describes a single line's list of inline changes. Use diff_change_parse() to
+/// parse this.
+typedef struct diffline_S diffline_T;
+struct diffline_S {
+  diffline_change_T *changes;
+  int num_changes;
+  int bufidx;
+  int lineoff;
 };
 
 #define SNAP_HELP_IDX   0

src/nvim/buffer_defs.h

@@ -240,6 +240,7 @@ static void changed_common(buf_T *buf, linenr_T lnum, colnr_T col, linenr_T lnum
   FOR_ALL_WINDOWS_IN_TAB(win, curtab) {
     if (win->w_buffer == buf && win->w_p_diff && diff_internal()) {
       curtab->tp_diff_update = true;
+      diff_update_line(lnum);
     }
   }