diff options
Diffstat (limited to 'src/modal.zig')
| -rw-r--r-- | src/modal.zig | 474 |
1 files changed, 360 insertions, 114 deletions
diff --git a/src/modal.zig b/src/modal.zig index 3827d11f..7e10a877 100644 --- a/src/modal.zig +++ b/src/modal.zig @@ -15,6 +15,8 @@ pub const Normal = struct { prefix_replace, prefix_next, prefix_prev, + prefix_register, + prefix_view_sticky, goto_file_start, goto_last_line, @@ -27,15 +29,19 @@ pub const Normal = struct { goto_view_top, goto_view_center, goto_view_bottom, + goto_last_modification, + goto_word, goto_definition, goto_declaration, goto_type_definition, goto_implementation, goto_references, + goto_callstack, view_top, view_center, view_bottom, + view_middle, view_scroll_down, view_scroll_up, @@ -49,6 +55,13 @@ pub const Normal = struct { add_newline, goto_diagnostic, goto_diagnostic_end, + goto_function, + goto_class, + goto_parameter, + goto_comment, + goto_test, + goto_entry, + goto_xml_element, move_left, move_right, @@ -96,6 +109,17 @@ pub const Normal = struct { trim_sels, select_regex, split_regex, + keep_regex, + remove_regex, + ensure_forward, + expand_node, + shrink_node, + next_sibling, + prev_sibling, + all_siblings, + all_children, + parent_node_end, + parent_node_start, delete, delete_noyank, @@ -108,6 +132,10 @@ pub const Normal = struct { to_lowercase, to_uppercase, join_lines, + join_select_space, + rotate_contents_fwd, + rotate_contents_back, + align_selections, indent, unindent, format, @@ -123,13 +151,19 @@ pub const Normal = struct { pipe_selection_to, insert_output, append_output, + keep_pipe, + macro_record, + macro_replay, + repeat_insert, search, search_next, search_prev, + search_selection, + search_selection_word, }; pub const Input = struct { - roles: std.EnumSet(Role) = .initEmpty(), + roles: std.EnumSet(Role) = .empty, cp: u21, ctrl: bool = false, alt: bool = false, @@ -156,6 +190,7 @@ pub const Normal = struct { replace, next, prev, + register, }; pub const MatchSub = enum(u8) { @@ -172,9 +207,14 @@ pub const Normal = struct { prefix: Prefix = .none, match_sub: MatchSub = .none, held_char: u21 = 0, + /// `"<reg>`: the register the next command names, 0 for none + register: u21 = 0, + /// `Z`: the view keys stay armed until Esc (helix's sticky view) + sticky: bool = false, + /// Drop what is pending; sticky view mode stays armed through it. pub fn clear(state: *State) void { - state.* = .{}; + state.* = if (state.sticky) .{ .sticky = true, .prefix = .view } else .{}; } }; @@ -188,13 +228,22 @@ pub const Normal = struct { insert, /// `A-!` — no stdin, and the output is appended AFTER each selection. append, + /// `$` — stdin is the selection; the output is discarded, and only the + /// selections the command exits 0 on are kept. + keep, /// Do the selections become stdin? helix's `pipe` flag. pub fn pipes(b: PipeBehavior) bool { - return b == .replace or b == .ignore; + return b == .replace or b == .ignore or b == .keep; } }; + /// helix's tree-sitter selection keys: where a range goes in the tree + /// helix's tree-sitter textobjects, by what their queries call them + pub const Object = enum { function, class, parameter, comment, @"test", entry, xml_element }; + pub const NodeWalk = enum { expand, shrink, next_sibling, prev_sibling, all_siblings, all_children, parent_end, parent_start }; + /// what the s/S/K/Alt-K prompt does with its pattern's matches + pub const RegexMode = enum { select, split, keep, remove }; pub const Scope = enum { once, per_selection }; pub const Direction = enum { backward, forward }; pub const Motion = enum { @@ -221,8 +270,9 @@ pub const Normal = struct { view_top, view_center, view_bottom, + last_modification, }; - pub const View = enum { top, center, bottom, scroll_down, scroll_up }; + pub const View = enum { top, center, bottom, middle, scroll_down, scroll_up }; pub const Find = enum { forward, backward, till_forward, till_backward }; pub const Line = enum { start, end, first_nonws }; pub const Page = enum { half_down, half_up, down, up }; @@ -247,6 +297,7 @@ pub const Normal = struct { merge, merge_consecutive, trim, + ensure_forward, }; pub const Edit = enum { delete, @@ -260,13 +311,17 @@ pub const Normal = struct { lowercase, uppercase, join_lines, + join_select_space, + rotate_contents_forward, + rotate_contents_backward, + align_selections, indent, unindent, comment_toggle, undo, redo, }; - pub const Lsp = enum { definition, declaration, type_definition, implementation, references, format }; + pub const Lsp = enum { definition, declaration, type_definition, implementation, references, callstack, format }; pub const Counted = struct { count: u32, @@ -295,7 +350,7 @@ pub const Normal = struct { insert: struct { kind: Insert, count: u32 }, select: struct { kind: Select, count: u32 }, multi: struct { kind: Multi, count: u32 }, - select_regex: bool, // false = select, true = split + select_regex: RegexMode, edit: struct { kind: Edit, count: u32 }, lsp: Lsp, adjust_number: i64, @@ -303,7 +358,14 @@ pub const Normal = struct { command_line, pipe_selection: PipeBehavior, search, - search_step: Direction, + search_step: struct { direction: Direction, count: u32 }, + search_selection: bool, + goto_word, + macro_record, + macro_replay: u32, + node: NodeWalk, + object_jump: struct { object: Object, direction: Direction, count: u32 }, + repeat_insert: u32, pub fn scope(value: Action) Scope { return switch (value) { @@ -315,9 +377,23 @@ pub const Normal = struct { .pipe_selection, .search, .search_step, + .search_selection, + .macro_record, + .macro_replay, + .node, + .goto_word, + .repeat_insert, => .once, .edit => |edit| switch (edit.kind) { - .comment_toggle, .undo, .redo => .once, + .join_lines, + .join_select_space, + .rotate_contents_forward, + .rotate_contents_backward, + .align_selections, + .comment_toggle, + .undo, + .redo, + => .once, else => .per_selection, }, else => .per_selection, @@ -354,7 +430,10 @@ pub const Normal = struct { pub fn parse(state: *State, key: Input) Result { if (key.has(.escape)) { - state.clear(); + // Esc out of sticky view mode leaves that mode and nothing else + const sticky = state.sticky; + state.* = .{}; + if (sticky) return .ignored; return resultAction(.escape); } @@ -375,27 +454,18 @@ pub const Normal = struct { switch (state.prefix) { .goto => { state.prefix = .none; - if (key.has(.goto_file_start)) return resultAction(.{ .goto = .{ .target = .file_start, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_last_line)) return resultAction(.{ .goto = .{ .target = .last_line, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_line_start)) return resultAction(.{ .goto = .{ .target = .line_start, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_line_end)) return resultAction(.{ .goto = .{ .target = .line_end, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_first_nonws)) return resultAction(.{ .goto = .{ .target = .first_nonws, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_line_down)) return resultAction(.{ .goto = .{ .target = .line_down, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_line_up)) return resultAction(.{ .goto = .{ .target = .line_up, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_column)) return resultAction(.{ .goto = .{ .target = .column, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_view_top)) return resultAction(.{ .goto = .{ .target = .view_top, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_view_center)) return resultAction(.{ .goto = .{ .target = .view_center, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_view_bottom)) return resultAction(.{ .goto = .{ .target = .view_bottom, .count = count, .explicit_count = counted.explicit } }); - if (key.has(.goto_definition)) return resultAction(.{ .lsp = .definition }); - if (key.has(.goto_declaration)) return resultAction(.{ .lsp = .declaration }); - if (key.has(.goto_type_definition)) return resultAction(.{ .lsp = .type_definition }); - if (key.has(.goto_implementation)) return resultAction(.{ .lsp = .implementation }); - if (key.has(.goto_references)) return resultAction(.{ .lsp = .references }); + // each target's role is its name after `goto_` + inline for (comptime std.enums.values(Goto)) |target| + if (key.has(@field(Role, "goto_" ++ @tagName(target)))) return resultAction(.{ .goto = .{ .target = target, .count = count, .explicit_count = counted.explicit } }); + if (key.has(.goto_word)) return resultAction(.goto_word); + inline for (.{ .definition, .declaration, .type_definition, .implementation, .references, .callstack }) |lsp| + if (key.has(@field(Role, "goto_" ++ @tagName(lsp)))) return resultAction(.{ .lsp = lsp }); return .ignored; }, .view => { - state.prefix = .none; + if (!state.sticky) state.prefix = .none; if (key.has(.view_top)) return resultAction(.{ .view = .top }); + if (key.has(.view_middle)) return resultAction(.{ .view = .middle }); if (key.has(.view_center)) return resultAction(.{ .view = .center }); if (key.has(.view_bottom)) return resultAction(.{ .view = .bottom }); if (key.has(.view_scroll_down)) return resultAction(.{ .view = .scroll_down }); @@ -473,14 +543,28 @@ pub const Normal = struct { if (key.has(.add_newline)) return resultAction(.{ .add_newline = .{ .direction = direction, .count = count } }); if (key.has(.goto_diagnostic)) return resultAction(.{ .diagnostic = .{ .direction = direction, .endpoint = false } }); if (key.has(.goto_diagnostic_end)) return resultAction(.{ .diagnostic = .{ .direction = direction, .endpoint = true } }); + inline for (.{ .{ .goto_function, .function }, .{ .goto_class, .class }, .{ .goto_parameter, .parameter }, .{ .goto_comment, .comment }, .{ .goto_test, .@"test" }, .{ .goto_entry, .entry }, .{ .goto_xml_element, .xml_element } }) |pair| + if (key.has(pair[0])) return resultAction(.{ .object_jump = .{ .object = pair[1], .direction = direction, .count = count } }); return .ignored; }, + .register => { + state.prefix = .none; + const char = key.literal() orelse return .ignored; + state.register = char; + // the count typed before `"` is still the command's + state.count = if (counted.explicit) count else 0; + return .pending; + }, .none => {}, } // Prefix setters retain the count for their continuation. if (key.has(.prefix_goto)) return armPrefix(state, .goto, if (counted.explicit) count else 0); if (key.has(.prefix_view)) return armPrefix(state, .view, if (counted.explicit) count else 0); + if (key.has(.prefix_view_sticky)) { + state.sticky = true; + return armPrefix(state, .view, 0); + } if (key.has(.prefix_find_fwd)) return armPrefix(state, .find_fwd, if (counted.explicit) count else 0); if (key.has(.prefix_find_back)) return armPrefix(state, .find_back, if (counted.explicit) count else 0); if (key.has(.prefix_till_fwd)) return armPrefix(state, .till_fwd, if (counted.explicit) count else 0); @@ -489,6 +573,7 @@ pub const Normal = struct { if (key.has(.prefix_next)) return armPrefix(state, .next, if (counted.explicit) count else 0); if (key.has(.prefix_prev)) return armPrefix(state, .prev, if (counted.explicit) count else 0); if (key.has(.prefix_match)) return armPrefix(state, .match, 0); + if (key.has(.prefix_register)) return armPrefix(state, .register, if (counted.explicit) count else 0); if (key.has(.move_left)) return resultAction(.{ .move = .{ .motion = .left, .count = count } }); if (key.has(.move_right)) return resultAction(.{ .move = .{ .motion = .right, .count = count } }); @@ -510,20 +595,11 @@ pub const Normal = struct { if (key.has(.page_down)) return resultAction(.{ .page = .{ .kind = .down, .count = count } }); if (key.has(.page_up)) return resultAction(.{ .page = .{ .kind = .up, .count = count } }); - if (key.has(.insert)) return resultAction(.{ .insert = .{ .kind = .at, .count = count } }); - if (key.has(.append)) return resultAction(.{ .insert = .{ .kind = .append, .count = count } }); - if (key.has(.insert_line_start)) return resultAction(.{ .insert = .{ .kind = .line_start, .count = count } }); - if (key.has(.insert_line_end)) return resultAction(.{ .insert = .{ .kind = .line_end, .count = count } }); - if (key.has(.open_below)) return resultAction(.{ .insert = .{ .kind = .open_below, .count = count } }); - if (key.has(.open_above)) return resultAction(.{ .insert = .{ .kind = .open_above, .count = count } }); + inline for (.{ .{ .insert, .at }, .{ .append, .append }, .{ .insert_line_start, .line_start }, .{ .insert_line_end, .line_end }, .{ .open_below, .open_below }, .{ .open_above, .open_above } }) |pair| + if (key.has(pair[0])) return resultAction(.{ .insert = .{ .kind = pair[1], .count = count } }); - if (key.has(.select_mode)) return resultAction(.{ .select = .{ .kind = .mode, .count = count } }); - if (key.has(.select_line)) return resultAction(.{ .select = .{ .kind = .line, .count = count } }); - if (key.has(.select_line_bounds)) return resultAction(.{ .select = .{ .kind = .line_bounds, .count = count } }); - if (key.has(.shrink_to_line_bounds)) return resultAction(.{ .select = .{ .kind = .shrink_to_line_bounds, .count = count } }); - if (key.has(.collapse_selection)) return resultAction(.{ .select = .{ .kind = .collapse, .count = count } }); - if (key.has(.flip_selection)) return resultAction(.{ .select = .{ .kind = .flip, .count = count } }); - if (key.has(.select_all)) return resultAction(.{ .select = .{ .kind = .all, .count = count } }); + inline for (.{ .{ .select_mode, .mode }, .{ .select_line, .line }, .{ .select_line_bounds, .line_bounds }, .{ .shrink_to_line_bounds, .shrink_to_line_bounds }, .{ .collapse_selection, .collapse }, .{ .flip_selection, .flip }, .{ .select_all, .all } }) |pair| + if (key.has(pair[0])) return resultAction(.{ .select = .{ .kind = pair[1], .count = count } }); if (key.has(.copy_sel_below)) return resultAction(.{ .multi = .{ .kind = .copy_below, .count = count } }); if (key.has(.copy_sel_above)) return resultAction(.{ .multi = .{ .kind = .copy_above, .count = count } }); @@ -535,8 +611,19 @@ pub const Normal = struct { if (key.has(.merge_sels)) return resultAction(.{ .multi = .{ .kind = .merge, .count = count } }); if (key.has(.merge_consecutive_sels)) return resultAction(.{ .multi = .{ .kind = .merge_consecutive, .count = count } }); if (key.has(.trim_sels)) return resultAction(.{ .multi = .{ .kind = .trim, .count = count } }); - if (key.has(.select_regex)) return resultAction(.{ .select_regex = false }); - if (key.has(.split_regex)) return resultAction(.{ .select_regex = true }); + if (key.has(.select_regex)) return resultAction(.{ .select_regex = .select }); + if (key.has(.split_regex)) return resultAction(.{ .select_regex = .split }); + if (key.has(.keep_regex)) return resultAction(.{ .select_regex = .keep }); + if (key.has(.remove_regex)) return resultAction(.{ .select_regex = .remove }); + if (key.has(.ensure_forward)) return resultAction(.{ .multi = .{ .kind = .ensure_forward, .count = count } }); + if (key.has(.expand_node)) return resultAction(.{ .node = .expand }); + if (key.has(.shrink_node)) return resultAction(.{ .node = .shrink }); + if (key.has(.next_sibling)) return resultAction(.{ .node = .next_sibling }); + if (key.has(.prev_sibling)) return resultAction(.{ .node = .prev_sibling }); + if (key.has(.all_siblings)) return resultAction(.{ .node = .all_siblings }); + if (key.has(.all_children)) return resultAction(.{ .node = .all_children }); + if (key.has(.parent_node_end)) return resultAction(.{ .node = .parent_end }); + if (key.has(.parent_node_start)) return resultAction(.{ .node = .parent_start }); if (key.has(.delete)) return resultAction(.{ .edit = .{ .kind = .delete, .count = count } }); if (key.has(.delete_noyank)) return resultAction(.{ .edit = .{ .kind = .delete_noyank, .count = count } }); @@ -549,6 +636,10 @@ pub const Normal = struct { if (key.has(.to_lowercase)) return resultAction(.{ .edit = .{ .kind = .lowercase, .count = count } }); if (key.has(.to_uppercase)) return resultAction(.{ .edit = .{ .kind = .uppercase, .count = count } }); if (key.has(.join_lines)) return resultAction(.{ .edit = .{ .kind = .join_lines, .count = count } }); + if (key.has(.join_select_space)) return resultAction(.{ .edit = .{ .kind = .join_select_space, .count = count } }); + if (key.has(.rotate_contents_fwd)) return resultAction(.{ .edit = .{ .kind = .rotate_contents_forward, .count = count } }); + if (key.has(.rotate_contents_back)) return resultAction(.{ .edit = .{ .kind = .rotate_contents_backward, .count = count } }); + if (key.has(.align_selections)) return resultAction(.{ .edit = .{ .kind = .align_selections, .count = count } }); if (key.has(.indent)) return resultAction(.{ .edit = .{ .kind = .indent, .count = count } }); if (key.has(.unindent)) return resultAction(.{ .edit = .{ .kind = .unindent, .count = count } }); if (key.has(.format)) return resultAction(.{ .lsp = .format }); @@ -564,9 +655,15 @@ pub const Normal = struct { if (key.has(.pipe_selection_to)) return resultAction(.{ .pipe_selection = .ignore }); if (key.has(.insert_output)) return resultAction(.{ .pipe_selection = .insert }); if (key.has(.append_output)) return resultAction(.{ .pipe_selection = .append }); + if (key.has(.keep_pipe)) return resultAction(.{ .pipe_selection = .keep }); + if (key.has(.macro_record)) return resultAction(.macro_record); + if (key.has(.macro_replay)) return resultAction(.{ .macro_replay = count }); + if (key.has(.repeat_insert)) return resultAction(.{ .repeat_insert = count }); if (key.has(.search)) return resultAction(.search); - if (key.has(.search_next)) return resultAction(.{ .search_step = .forward }); - if (key.has(.search_prev)) return resultAction(.{ .search_step = .backward }); + if (key.has(.search_selection)) return resultAction(.{ .search_selection = false }); + if (key.has(.search_selection_word)) return resultAction(.{ .search_selection = true }); + if (key.has(.search_next)) return resultAction(.{ .search_step = .{ .direction = .forward, .count = count } }); + if (key.has(.search_prev)) return resultAction(.{ .search_step = .{ .direction = .backward, .count = count } }); return .unbound; } @@ -669,7 +766,7 @@ fn codepointAt(text: []const u8, off: usize) u21 { return std.unicode.utf8Decode(text[off .. off + n]) catch 0xFFFD; } -fn isUnicodeWhitespace(cp: u21) bool { +pub fn isUnicodeWhitespace(cp: u21) bool { if (cp == ' ' or (cp >= '\t' and cp <= '\r') or cp == 0x85) return true; return switch (uucode.get(.general_category, cp)) { .separator_space, .separator_line, .separator_paragraph => true, @@ -738,13 +835,6 @@ pub fn firstNonWs(line: []const u8) usize { return i; } -pub fn firstNonWsOf(lines: []const []const u8, c: Cursor) Cursor { - // the row can sit past the content (mouse click below a short pane's - // last line) — out of range reads as an empty line, like lineLenOf - if (c.row >= lines.len) return .{ .row = c.row, .col = 0 }; - return .{ .row = c.row, .col = firstNonWs(lines[c.row]) }; -} - pub fn charLeft(lines: []const []const u8, c: Cursor) Cursor { if (c.row >= lines.len) return .{ .row = c.row, .col = 0 }; return .{ .row = c.row, .col = prevGrapheme(lines[c.row], @min(c.col, lines[c.row].len)) }; @@ -870,13 +960,77 @@ pub fn lineCount(content: []const u8) usize { // byte offset of the start of line `row` (0-based). row may == lineCount() // (== content.len, the end). pub fn lineStartOffset(content: []const u8, row: usize) usize { + // Newlines counted 32 bytes at a time, the last block byte by byte: one + // indexOfScalarPos per line was a call per line, 7000 of them to find + // line 7000 of a source, for every cursor move, selection and language + // request there. + const block = 32; + const Block = @Vector(block, u8); var off: usize = 0; - var r: usize = 0; - while (r < row) : (r += 1) { - const nl = std.mem.indexOfScalarPos(u8, content, off, '\n') orelse return content.len; - off = nl + 1; + var left = row; + if (left == 0) return 0; + while (off + block <= content.len) : (off += block) { + const chunk: Block = content[off..][0..block].*; + const found: usize = @popCount(@as(u32, @bitCast(chunk == @as(Block, @splat('\n'))))); + if (found >= left) break; + left -= found; + } + while (off < content.len) : (off += 1) if (content[off] == '\n') { + left -= 1; + if (left == 0) return off + 1; + }; + return content.len; +} + +test "lineStartOffset by blocks and offsetAt agree with a line at a time" { + var prng: std.Random.DefaultPrng = .init(0x9e3779b9); + const random = prng.random(); + var buf: [300]u8 = undefined; + for (0..2000) |_| { + const len = random.uintLessThan(usize, buf.len); + for (buf[0..len]) |*b| b.* = if (random.uintLessThan(u8, 4) == 0) '\n' else 'a'; + const text = buf[0..len]; + for (0..len + 3) |row| { + var off: usize = 0; + var r: usize = 0; + const want = while (r < row) : (r += 1) { + off = (std.mem.indexOfScalarPos(u8, text, off, '\n') orelse break text.len) + 1; + } else off; + try std.testing.expectEqual(want, lineStartOffset(text, row)); + const cursor: Cursor = .{ .row = row, .col = random.uintLessThan(usize, 6) }; + const clamped = lineStartOffset(text, @min(row, cursorLineCount(text) - 1)); + const end = std.mem.indexOfScalarPos(u8, text, clamped, '\n') orelse text.len; + try std.testing.expectEqual(clamped + graphemeStart(text[clamped..end], @min(cursor.col, end - clamped)), offsetAt(text, cursor)); + // lineSpan, as it was: a line at a time, null past the last. + const span: ?LineSpan = span: { + if (text.len == 0) break :span null; + var start: usize = 0; + for (0..row) |_| start = (std.mem.indexOfScalarPos(u8, text, start, '\n') orelse break :span null) + 1; + if (start == text.len and !(row == 0 or text[text.len - 1] == '\n')) break :span null; + break :span .{ .start = start, .end = std.mem.indexOfScalarPos(u8, text, start, '\n') orelse text.len }; + }; + try std.testing.expectEqual(span, lineSpan(text, row)); + // rangeBytes, as it was: each line looked for from the top, twice. + const other: Cursor = .{ .row = random.uintLessThan(usize, len + 3), .col = random.uintLessThan(usize, 6) }; + var lo = cursor; + var hi = other; + if (hi.row < lo.row or (hi.row == lo.row and hi.col < lo.col)) std.mem.swap(Cursor, &lo, &hi); + const lo_line = lineSlice(text, lo.row); + const hi_line = lineSlice(text, hi.row); + const s0 = lineStartOffset(text, lo.row) + graphemeStart(lo_line, @min(lo.col, lo_line.len)); + var e0 = lineStartOffset(text, hi.row) + graphemeStart(hi_line, @min(hi.col, hi_line.len)); + if (e0 < text.len) e0 = nextGrapheme(text, e0); + const got = rangeBytes(text, cursor, other); + try std.testing.expectEqual(s0, got.s); + try std.testing.expectEqual(@max(s0, e0), got.e); + } } - return off; +} + +/// The line that starts at `start` (lineSlice's, once its start is known). +fn lineFrom(content: []const u8, start: usize) []const u8 { + if (start >= content.len) return ""; + return content[start .. std.mem.indexOfScalarPos(u8, content, start, '\n') orelse content.len]; } // the text of line `row` (no terminator), a slice into `content`. @@ -892,23 +1046,29 @@ pub fn lineSlice(content: []const u8, row: usize) []const u8 { /// A bounded scan of one line; null if the row does not exist. pub const LineSpan = struct { start: usize, end: usize }; +/// Just past the `row`-th newline; null if the text has fewer. +pub fn lineStartExact(content: []const u8, row: usize) ?usize { + const at = lineStartOffset(content, row); + // At the end: either the row-th newline was the last byte, or there were fewer. + if (at == content.len and std.mem.count(u8, content, "\n") < row) return null; + return at; +} + pub fn lineSpan(content: []const u8, row: usize) ?LineSpan { // Match lineCount: empty content has no lines; a trailing newline adds one. if (content.len == 0) return null; - var start: usize = 0; - var r: usize = 0; - while (r < row) : (r += 1) { - const nl = std.mem.indexOfScalarPos(u8, content, start, '\n') orelse return null; - start = nl + 1; - } - if (start > content.len) return null; + const start = lineStartExact(content, row) orelse return null; if (start == content.len and !(row == 0 or content.len == 0 or content[content.len - 1] == '\n')) return null; const end = std.mem.indexOfScalarPos(u8, content, start, '\n') orelse content.len; return .{ .start = start, .end = end }; } -fn spliceAlloc(alloc: std.mem.Allocator, content: []const u8, start: usize, end: usize, replacement: []const u8) ![]u8 { - const out = try alloc.alloc(u8, content.len - (end - start) + replacement.len); +pub fn spliceAlloc(alloc: std.mem.Allocator, content: []const u8, start: usize, end: usize, replacement: []const u8) ![]u8 { + // Raw: every byte is written below, where `alloc` fills a safe build's + // new memory first, a whole file's worth per keystroke. + const len = content.len - (end - start) + replacement.len; + if (len == 0) return alloc.alloc(u8, 0); + const out = (alloc.rawAlloc(len, .of(u8), @returnAddress()) orelse return error.OutOfMemory)[0..len]; @memcpy(out[0..start], content[0..start]); @memcpy(out[start..][0..replacement.len], replacement); @memcpy(out[start + replacement.len ..], content[end..]); @@ -917,13 +1077,18 @@ fn spliceAlloc(alloc: std.mem.Allocator, content: []const u8, start: usize, end: /// insert `text` at (row, col). col is clamped to the line length. pub fn insertAt(alloc: std.mem.Allocator, content: []const u8, c: Cursor, text: []const u8) ![]u8 { + const off = insertOffset(content, c); + return spliceAlloc(alloc, content, off, off, text); +} + +/// Where insertAt puts its text: col clamped to the line, no further. +pub fn insertOffset(content: []const u8, c: Cursor) usize { // Only an out-of-range row requires counting the whole document. const span = lineSpan(content, c.row) orelse blk: { const last = lineCount(content) -| 1; break :blk lineSpan(content, last) orelse LineSpan{ .start = content.len, .end = content.len }; }; - const off = span.start + @min(c.col, span.end - span.start); - return spliceAlloc(alloc, content, off, off, text); + return span.start + @min(c.col, span.end - span.start); } // delete the character at (row, col). no-op if col is past the line end. @@ -986,12 +1151,15 @@ fn rangeBytes(content: []const u8, a: Cursor, b: Cursor) struct { s: usize, e: u lo = b; hi = a; } - const lo_line = lineSlice(content, lo.row); - const hi_line = lineSlice(content, hi.row); + // Each line found once, the second from the first. + const lo_start = lineStartOffset(content, lo.row); + const hi_start = if (hi.row == lo.row) lo_start else lo_start + lineStartOffset(content[lo_start..], hi.row - lo.row); + const lo_line = lineFrom(content, lo_start); + const hi_line = lineFrom(content, hi_start); const lo_col = graphemeStart(lo_line, @min(lo.col, lo_line.len)); const hi_col = graphemeStart(hi_line, @min(hi.col, hi_line.len)); - const s = lineStartOffset(content, lo.row) + lo_col; - var e = lineStartOffset(content, hi.row) + hi_col; + const s = lo_start + lo_col; + var e = hi_start + hi_col; if (e < content.len) e = nextGrapheme(content, e); // include the grapheme under the head return .{ .s = s, .e = @max(s, e) }; } @@ -1019,12 +1187,12 @@ pub fn clearLine(alloc: std.mem.Allocator, content: []const u8, row: usize) ![]u // paste `text` as a new line BELOW `row`. Multiline `text` becomes several lines. pub fn pasteLineBelow(alloc: std.mem.Allocator, content: []const u8, row: usize, text: []const u8) ![]u8 { - const n = lineCount(content); - const off = if (row + 1 < n) lineStartOffset(content, row + 1) else content.len; + // The next line's start, or the end: no whole-text line count to know which. + const off = lineStartExact(content, row + 1) orelse content.len; const suffix_newline = off < content.len or (content.len > 0 and content[content.len - 1] == '\n'); const prefix_newline = !suffix_newline and off > 0; const extra: usize = @intFromBool(suffix_newline or prefix_newline); - const out = try alloc.alloc(u8, content.len + text.len + extra); + const out = try @import("memory.zig").unfilled(alloc, content.len + text.len + extra); var write: usize = 0; @memcpy(out[write..][0..off], content[0..off]); write += off; @@ -1211,6 +1379,58 @@ pub fn lineEndOffset(text: []const u8, line: usize) usize { } /// gap offset -> (row, col) cell +/// The start of the UTF-8 rune `off` falls in. Addresses (the 9P `addr`, +/// `dot`, `#n`, `line:col`, a search's match) count runes, as sam's and +/// acme's do, not grapheme clusters: a lone combining mark or a `\r` is a +/// place of its own, whatever a terminal draws in one cell. +pub fn runeStart(text: []const u8, off: usize) usize { + const o = @min(off, text.len); + if (o == text.len or text[o] & 0xC0 != 0x80) return o; + var s = o; + var steps: u8 = 0; + while (s > 0 and text[s] & 0xC0 == 0x80 and steps < 3) : (steps += 1) s -= 1; + // Only a lead byte whose sequence reaches `o`, its bytes all + // continuations, makes `o` the inside of a rune; a stray continuation + // byte (invalid UTF-8) is a place of its own. + const n = std.unicode.utf8ByteSequenceLength(text[s]) catch return o; + if (s + n <= o or s + n > text.len) return o; + for (text[s + 1 .. s + n]) |c| if (c & 0xC0 != 0x80) return o; + return s; +} + +/// `off`, or the end of the rune it falls inside. +pub fn runeEnd(text: []const u8, off: usize) usize { + const o = @min(off, text.len); + const s = runeStart(text, o); + if (s == o) return o; + return s + (std.unicode.utf8ByteSequenceLength(text[s]) catch 1); +} + +/// The rune boundary after the one at `off`. +pub fn nextRune(text: []const u8, off: usize) usize { + return if (off >= text.len) text.len else runeEnd(text, off + 1); +} + +/// The rune boundary before `off`. +pub fn prevRune(text: []const u8, off: usize) usize { + return if (off == 0) 0 else runeStart(text, @min(off, text.len) - 1); +} + +/// `positionAt` snapped to a rune rather than a grapheme cluster. +pub fn runePositionAt(text: []const u8, off: usize) Cursor { + const bounded = runeStart(text, off); + const s = if (std.mem.lastIndexOfScalar(u8, text[0..bounded], '\n')) |nl| nl + 1 else 0; + return .{ .row = lineAtOffset(text, bounded), .col = bounded - s }; +} + +/// `offsetAt` snapped to a rune rather than a grapheme cluster. +pub fn runeOffsetAt(text: []const u8, c: Cursor) usize { + const row = @min(c.row, cursorLineCount(text) - 1); + const s = lineStartOffset(text, row); + const e = std.mem.indexOfScalarPos(u8, text, s, '\n') orelse text.len; + return s + runeStart(text[s..e], @min(c.col, e - s)); +} + pub fn positionAt(text: []const u8, off: usize) Cursor { const bounded = @min(off, text.len); // Find the line start backward without a second scan of the document prefix. @@ -1222,8 +1442,10 @@ pub fn positionAt(text: []const u8, off: usize) Cursor { /// (row, col) -> clamped gap offset; col == line length lands ON the '\n' pub fn offsetAt(text: []const u8, c: Cursor) usize { - const row = @min(c.row, cursorLineCount(text) - 1); - const s = lineStartOffset(text, row); + // A row past the last is the last: found back from the end, where + // counting every line of the text to clamp it cost a pass over all of it. + var s = lineStartOffset(text, c.row); + if (s == text.len) s = if (std.mem.lastIndexOfScalar(u8, text, '\n')) |nl| nl + 1 else 0; // lineEndOffset(text, row) inlined: it starts by walking to `row` again, // and we are already standing there const e = std.mem.indexOfScalarPos(u8, text, s, '\n') orelse text.len; @@ -1257,6 +1479,17 @@ fn wordClassAt(text: []const u8, off: usize) WordClass { }; } +pub fn searchWordAt(text: []const u8, off: usize) bool { + return off < text.len and wordClassAt(text, off) == .word; +} + +pub fn searchWordBefore(text: []const u8, off: usize) bool { + if (off == 0) return false; + var previous = off - 1; + while (previous > 0 and text[previous] & 0xc0 == 0x80) previous -= 1; + return searchWordAt(text, previous); +} + fn isWordWhitespace(c: WordClass) bool { // Rust char::is_whitespace (includes line endings) return c == .ws or c == .eol; } @@ -1287,6 +1520,53 @@ fn wmIsPrev(t: WordTarget) bool { } /// w/b/e/W/B/E: helix word_move — each step selects the traversed span. +/// helix jump_to_word's candidates: the words of two or more word +/// characters between `start` and `end`, nearest the cursor first, taking one +/// forward and one back in turn, each over its word pointing the way it was +/// found, as many as `out` holds. +pub fn jumpWords(text: []const u8, cursor: usize, start: usize, end: usize, out: []Selection) usize { + var fwd: Selection = .{ .anchor = cursor, .head = cursor }; + var rev = fwd; + if (cursor < text.len and !isWordWhitespace(wordClassAt(text, cursor))) { + const word_end = moveWord(text, fwd, 1, .next_word_end); + if (word_end.anchor == cursor) fwd = word_end; // a word of one grapheme + const word_start = moveWord(text, rev, 1, .prev_word_start); + if (word_start.anchor == nextGrapheme(text, cursor)) rev = word_start; + } + var n: usize = 0; + while (n < out.len) { + var changed = false; + while (fwd.head < end) { + fwd = moveWord(text, fwd, 1, .next_word_end); + // the two graphemes before the head are word characters + const one = prevGrapheme(text, fwd.head); + if (fwd.head == 0 or one == 0 or wordClassAt(text, one) != .word or wordClassAt(text, prevGrapheme(text, one)) != .word) continue; + changed = true; + var from = fwd.anchor; + while (from < fwd.head and wordClassAt(text, from) != .word) from = nextGrapheme(text, from); + fwd.anchor = from; + out[n] = fwd; + n += 1; + break; + } + if (n == out.len) break; + while (rev.head > start) { + rev = moveWord(text, rev, 1, .prev_word_start); + // the two graphemes from the head are word characters + if (wordClassAt(text, rev.head) != .word or wordClassAt(text, nextGrapheme(text, rev.head)) != .word) continue; + changed = true; + var to = rev.anchor; + while (to > rev.head and wordClassAt(text, prevGrapheme(text, to)) != .word) to = prevGrapheme(text, to); + rev.anchor = to; + out[n] = rev; + n += 1; + break; + } + if (!changed) break; + } + return n; +} + pub fn moveWord(text: []const u8, r0: Selection, count: usize, target: WordTarget) Selection { const is_prev = wmIsPrev(target); if ((is_prev and r0.head == 0) or (!is_prev and r0.head == text.len)) return r0; @@ -1408,18 +1688,6 @@ pub fn moveParagraph(text: []const u8, r: Selection, count: usize, fwd: bool, ex return .{ .anchor = anchor, .head = head }; } -/// j/k target: helix move_vertically — goal_col clamps to the line's content -/// length, i.e. the cursor may land ON the '\n' of a shorter line. -pub fn verticalTarget(text: []const u8, pos: usize, down: bool, count: usize, goal_col: usize) usize { - const nlines = cursorLineCount(text); - const line = lineAtOffset(text, pos); - const nline = if (down) @min(line + @max(1, count), nlines - 1) else line -| @max(1, count); - const s = lineStartOffset(text, nline); - // lineEndOffset(text, nline) without its second walk to nline (see offsetAt) - const e = std.mem.indexOfScalarPos(u8, text, s, '\n') orelse text.len; - return s + graphemeStart(text[s..e], @min(goal_col, e - s)); -} - /// f/F/t/T target cell. helix find_char: the exclusive (till) search starts /// one further out so repeats make progress; not-found = null (no move). pub fn findTarget(text: []const u8, cursor: usize, ch: u21, fwd: bool, till: bool, count: usize) ?usize { @@ -1554,21 +1822,10 @@ test "selection textobjects distinguish words and paragraph boundaries" { try std.testing.expectEqual(@as(usize, 7), r.head); } -/// leading-whitespace visual width (tab -> next multiple of INDENT_W) -pub fn indentWidth(line: []const u8) usize { - var w: usize = 0; - for (line) |ch| { - if (ch == ' ') w += 1 else if (ch == '\t') w = (w / INDENT_W + 1) * INDENT_W else break; - } - return w; -} - -/// full indent LEVELS of a line as spaces (helix indent_level_for_line: -/// partial levels round down) — what o/O/insert-newline copy. +/// A line's indent as it is, tabs and spaces both, as acme's autoindent +/// copies it: what o/O/insert-newline give the new line. pub fn indentText(line: []const u8) []const u8 { - const level = indentWidth(line) / INDENT_W; - const max = " "; // 8 levels is plenty (ponytail) - return max[0..@min(level * INDENT_W, max.len)]; + return line[0 .. std.mem.indexOfNone(u8, line, " \t") orelse line.len]; } /// Copy full indent levels and add one after (, [, {, or ), without parsing. @@ -1678,14 +1935,6 @@ test "paragraph selections cross blank lines" { try std.testing.expectEqual(@as(usize, 7), r.head); } -test "vertical target clamps the goal column onto the newline cell" { - const t = "abcdef\nab\nabcdef\n"; - // from (0,5) down: line "ab" clamps to its '\n' at offset 9 - try std.testing.expectEqual(@as(usize, 9), verticalTarget(t, 5, true, 1, 5)); - // two down with the same goal restores col 5 - try std.testing.expectEqual(@as(usize, 15), verticalTarget(t, 9, true, 1, 5)); -} - test "find targets count matches and skip adjacent till targets" { const t = "abcabc\n"; try std.testing.expectEqual(@as(usize, 2), findTarget(t, 0, 'c', true, false, 1).?); @@ -1850,14 +2099,11 @@ test "codepoint classes distinguish words punctuation and whitespace" { try std.testing.expectEqual(Kind.ws, kindOfCodepoint('\n')); } -test "insert cursor steps characters and skips indentation" { +test "insert cursor steps characters" { const lines = [_][]const u8{ "alpha beta", " two words", "x" }; const c = Cursor{ .row = 0, .col = 5 }; try std.testing.expectEqual(Cursor{ .row = 0, .col = 4 }, charLeft(&.{"hello"}, c)); try std.testing.expectEqual(Cursor{ .row = 0, .col = 6 }, charRight(&lines, c)); - try std.testing.expectEqual(Cursor{ .row = 1, .col = 2 }, firstNonWsOf(&lines, Cursor{ .row = 1, .col = 0 })); - // cursor row past the content (mouse click below a short pane): no panic - try std.testing.expectEqual(Cursor{ .row = 24, .col = 0 }, firstNonWsOf(&lines, Cursor{ .row = 24, .col = 3 })); } test "lineSpan agrees with the whole-document scans it replaces" { |
