feat: issue #32 markdown editor for TUI

lib/bds/ui/code_editor/overlay.ex

@@ -0,0 +1,225 @@
+defmodule BDS.UI.CodeEditor.Overlay do
+  @moduledoc """
+  Re-styles `%ExRatatui.Text.Line{}` spans for the syntaxes the base
+  syntect lexer does not know: post macros (`[[gallery names="x"]]`) and
+  Liquid tags (`{{ … }}` / `{% … %}`).
+
+  Ranges are byte-based `{start, length, %Style{}}` triples applied with
+  patch semantics — the overlay's non-nil fields override the base span
+  style, so the theme background survives. Overlay colours mirror the
+  GUI Monaco theme (`assets/js/monaco/theme.js`): macro accents are the
+  custom `keyword.macro` / `attribute.name` / `attribute.value` rules,
+  Liquid accents are the vs-dark keyword/string/number base tokens.
+  """
+
+  alias ExRatatui.Style
+  alias ExRatatui.Text.Line
+  alias ExRatatui.Text.Span
+
+  @type styled_range :: {non_neg_integer(), non_neg_integer(), Style.t()}
+
+  @macro_style %Style{fg: {:rgb, 197, 134, 192}, modifiers: [:bold]}
+  @macro_attr_style %Style{fg: {:rgb, 156, 220, 254}}
+  @macro_value_style %Style{fg: {:rgb, 206, 145, 120}}
+
+  @liquid_keyword_style %Style{fg: {:rgb, 86, 156, 214}}
+  @liquid_string_style %Style{fg: {:rgb, 206, 145, 120}}
+  @liquid_number_style %Style{fg: {:rgb, 181, 206, 168}}
+
+  @macro_pattern ~r/\[\[\s*[a-zA-Z][\w-]*[^]]*\]\]/
+  @macro_opener ~r/\A\[\[\s*[a-zA-Z][\w-]*/
+  @macro_attr_name ~r/[a-zA-Z][\w-]*(?=\s*=)/
+  @macro_quoted_value ~r/"[^"]*"/
+  @macro_bare_value ~r/[^\s=\[\]"]++(?!\s*=)/
+
+  @liquid_tag ~r/\{\{-?.*?-?\}\}|\{%-?.*?-?%\}/
+  @liquid_open ~r/\A\{[\{%]-?/
+  @liquid_close ~r/-?[\}%][\}%]\z/
+  @liquid_keywords ~r/\b(?:assign|capture|case|comment|cycle|decrement|echo|elsif|else|endcase|endcapture|endif|endfor|endunless|endcomment|for|if|include|increment|liquid|paginate|raw|render|tablerow|unless|when|true|false|nil|blank|empty|contains)\b/
+  @liquid_filter ~r/\|\s*[a-zA-Z_][\w-]*/
+  @liquid_string ~r/"[^"]*"|'[^']*'/
+  @liquid_number ~r/\b\d+(?:\.\d+)?\b/
+
+  @doc """
+  Recolours every `[[macro]]` on the line, preserving the base styling
+  of the surrounding markdown.
+  """
+  @spec markdown_macros(Line.t()) :: Line.t()
+  def markdown_macros(%Line{} = line) do
+    apply_ranges(line, markdown_macro_ranges(line_text(line)))
+  end
+
+  @doc """
+  Recolours every Liquid output/tag expression on the line, preserving
+  the base styling of the surrounding HTML.
+  """
+  @spec liquid(Line.t()) :: Line.t()
+  def liquid(%Line{} = line) do
+    apply_ranges(line, liquid_ranges(line_text(line)))
+  end
+
+  @doc """
+  Byte ranges of every macro on `text`: `[[` + name and `]]` in the
+  macro style, attribute names and values in their accent styles.
+  """
+  @spec markdown_macro_ranges(String.t()) :: [styled_range()]
+  def markdown_macro_ranges(text) do
+    text
+    |> scan(@macro_pattern)
+    |> Enum.flat_map(fn {start, len} ->
+      macro = binary_part(text, start, len)
+
+      macro
+      |> macro_inner_ranges()
+      |> Enum.map(fn {inner_start, inner_len, style} ->
+        {start + inner_start, inner_len, style}
+      end)
+    end)
+  end
+
+  @doc """
+  Byte ranges (relative to the start of `macro`, including its `[[` and
+  `]]`) of the styled sub-parts of a single macro match. Quoted values
+  are masked before scanning for attribute names and bare values so
+  nothing matches inside a quoted string.
+  """
+  @spec macro_inner_ranges(String.t()) :: [styled_range()]
+  def macro_inner_ranges(macro) do
+    [{0, opener_len}] = Regex.run(@macro_opener, macro, return: :index)
+    inner = binary_part(macro, opener_len, byte_size(macro) - opener_len - 2)
+    quoted = scan(inner, @macro_quoted_value)
+    masked = mask(inner, quoted)
+
+    inner_ranges =
+      with_style(quoted, @macro_value_style) ++
+        (masked |> scan(@macro_attr_name) |> with_style(@macro_attr_style)) ++
+        (masked |> scan(@macro_bare_value) |> with_style(@macro_value_style))
+
+    [{0, opener_len, @macro_style}, {byte_size(macro) - 2, 2, @macro_style}] ++
+      offset_ranges(inner_ranges, opener_len)
+  end
+
+  @doc """
+  Byte ranges of every Liquid expression on `text`: the `{{`/`}}` and
+  `{%`/`%}` delimiters, tag keywords and filters in the keyword style,
+  strings and numbers in their accent styles. Strings are masked before
+  scanning so keywords/numbers inside them keep the string style.
+  """
+  @spec liquid_ranges(String.t()) :: [styled_range()]
+  def liquid_ranges(text) do
+    text
+    |> scan(@liquid_tag)
+    |> Enum.flat_map(fn {start, len} ->
+      tag = binary_part(text, start, len)
+      [{0, open_len}] = Regex.run(@liquid_open, tag, return: :index)
+      [{close_start, close_len}] = Regex.run(@liquid_close, tag, return: :index)
+      inner = binary_part(tag, open_len, close_start - open_len)
+      strings = scan(inner, @liquid_string)
+      masked = mask(inner, strings)
+
+      inner_ranges =
+        with_style(strings, @liquid_string_style) ++
+          (masked |> scan(@liquid_keywords) |> with_style(@liquid_keyword_style)) ++
+          (masked |> scan(@liquid_filter) |> with_style(@liquid_keyword_style)) ++
+          (masked |> scan(@liquid_number) |> with_style(@liquid_number_style))
+
+      [{start, open_len, @liquid_keyword_style}, {start + close_start, close_len, @liquid_keyword_style}] ++
+        offset_ranges(inner_ranges, start + open_len)
+    end)
+  end
+
+  @doc """
+  Re-splits a line's spans at the given byte ranges, patch-merging each
+  range's style over the base span style it covers. Overlapping ranges
+  resolve to the one starting earliest (strings/quotes therefore win
+  over keywords matched inside them).
+  """
+  @spec apply_ranges(Line.t(), [styled_range()]) :: Line.t()
+  def apply_ranges(%Line{} = line, []), do: line
+
+  def apply_ranges(%Line{spans: spans} = line, ranges) do
+    ranges = Enum.sort_by(ranges, &elem(&1, 0))
+
+    {new_spans, _offset} =
+      Enum.map_reduce(spans, 0, fn span, offset ->
+        {split_span(span, offset, ranges), offset + byte_size(span.content)}
+      end)
+
+    %{line | spans: new_spans |> List.flatten() |> Enum.reject(&(&1.content == ""))}
+  end
+
+  @doc """
+  Patch-merges `patch` over `base`: non-nil colours override, modifiers
+  union. Used for overlay accents, the cursor cell and the current-line
+  highlight so the theme background/foreground survives underneath.
+  """
+  @spec merge_style(Style.t(), Style.t()) :: Style.t()
+  def merge_style(%Style{} = base, %Style{} = patch) do
+    %Style{
+      fg: patch.fg || base.fg,
+      bg: patch.bg || base.bg,
+      underline_color: patch.underline_color || base.underline_color,
+      modifiers: Enum.uniq(base.modifiers ++ patch.modifiers)
+    }
+  end
+
+  defp split_span(%Span{} = span, span_start, ranges) do
+    span_end = span_start + byte_size(span.content)
+
+    cuts =
+      ranges
+      |> Enum.flat_map(fn {range_start, range_len, _style} ->
+        range_end = range_start + range_len
+
+        if range_end > span_start and range_start < span_end do
+          [max(range_start, span_start), min(range_end, span_end)]
+        else
+          []
+        end
+      end)
+      |> Enum.filter(&(&1 > span_start and &1 < span_end))
+      |> Enum.uniq()
+      |> Enum.sort()
+
+    [span_start | cuts ++ [span_end]]
+    |> Enum.chunk_every(2, 1, :discard)
+    |> Enum.map(fn [segment_start, segment_end] ->
+      content = binary_part(span.content, segment_start - span_start, segment_end - segment_start)
+
+      style =
+        case Enum.find(ranges, &covers?(&1, segment_start, segment_end)) do
+          nil -> span.style
+          {_start, _len, patch} -> merge_style(span.style, patch)
+        end
+
+      %Span{content: content, style: style}
+    end)
+  end
+
+  defp covers?({range_start, range_len, _style}, segment_start, segment_end) do
+    segment_start >= range_start and segment_end <= range_start + range_len
+  end
+
+  defp line_text(line), do: Enum.map_join(line.spans, & &1.content)
+
+  defp scan(text, pattern) do
+    pattern
+    |> Regex.scan(text, return: :index)
+    |> Enum.map(fn [{start, len}] -> {start, len} end)
+  end
+
+  defp with_style(ranges, style), do: Enum.map(ranges, fn {start, len} -> {start, len, style} end)
+
+  defp offset_ranges(ranges, offset),
+    do: Enum.map(ranges, fn {start, len, style} -> {start + offset, len, style} end)
+
+  # Blanks out the given byte ranges with spaces so later scans cannot
+  # match inside them (offsets stay intact).
+  defp mask(text, ranges) do
+    Enum.reduce(ranges, text, fn {start, len}, acc ->
+      binary_part(acc, 0, start) <>
+        String.duplicate(" ", len) <>
+        binary_part(acc, start + len, byte_size(acc) - start - len)
+    end)
+  end
+end