Added literal mark (#2593)

I'm not sure this is a direction we want to go or if this is a good
implementation, but for editors where we don't have hats it would be
nice if you could just speak the literal text instead.
`"take hello world"`
`"chuck state get value"`

In this implementation I'm only looking in the viewport and I'm solving
ambiguity by selecting the match closest to a cursor selection.

## Checklist

- [x] I have added
[tests](https://www.cursorless.org/docs/contributing/test-case-recorder/)
- [/] I have updated the
[docs](https://github.com/cursorless-dev/cursorless/tree/main/docs) and
[cheatsheet](https://github.com/cursorless-dev/cursorless/tree/main/cursorless-talon/src/cheatsheet)
- [/] I have not broken the cheatsheet

- [x] Add list for literal prefix/keyword. Empty by default. test that
this behaves properly
- [x] Tag to remove prefix and have literal as a replacement for hats.
Check dfa compilation
- [x] Tag to default to preferred scope instead of containing scope

---------

Co-authored-by: pre-commit-ci-lite[bot] <117423508+pre-commit-ci-lite[bot]@users.noreply.github.com>
Co-authored-by: Phil Cohen <[email protected]>

Author: 3 people

src/marks/literal_mark.py

@@ -0,0 +1,50 @@
+from talon import Context, Module
+
+from .mark_types import LiteralMark
+
+mod = Module()
+
+mod.list("private_cursorless_literal_mark", desc="Cursorless literal mark")
+
+# This is a private tag and should not be used by non Cursorless developers
+mod.tag(
+    "private_cursorless_literal_mark_no_prefix",
+    desc="Tag for enabling literal mark without prefix",
+)
+
+ctx_no_prefix = Context()
+ctx_no_prefix.matches = r"""
+tag: user.private_cursorless_literal_mark_no_prefix
+"""
+
+
+# NB: <phrase> is used over <user.text> for DFA performance reasons
+# (we intend to replace this with a dynamic list of document contents eventually)
+@mod.capture(rule="{user.private_cursorless_literal_mark} <phrase>")
+def cursorless_literal_mark(m) -> LiteralMark:
+    return construct_mark(str(m.phrase))
+
+
+@ctx_no_prefix.capture("user.cursorless_literal_mark", rule="<phrase>")
+def cursorless_literal_mark_no_prefix(m) -> LiteralMark:
+    return construct_mark(str(m.phrase))
+
+
+def construct_mark(text: str) -> LiteralMark:
+    return {
+        "type": "literal",
+        "modifier": {
+            "type": "preferredScope",
+            "scopeType": {
+                "type": "customRegex",
+                "regex": construct_fuzzy_regex(text),
+                "flags": "gui",
+            },
+        },
+    }
+
+
+def construct_fuzzy_regex(text: str) -> str:
+    parts = text.split(" ")
+    # Between each word there can be any number of non-alpha symbols (including escape characters: \t\r\n). No separator at all is also valid -- for example, when searching for a camelCase identifier.
+    return r"([^a-zA-Z]|\\[trn])*".join(parts)

src/marks/mark.py

@@ -8,6 +8,7 @@
 @mod.capture(
     rule=(
         "<user.cursorless_decorated_symbol> | "
+        "<user.cursorless_literal_mark> | "
         "<user.cursorless_simple_mark> |"
         "<user.cursorless_line_number>"  # row (ie absolute mod 100), up, down
     )

src/marks/mark_types.py

@@ -7,6 +7,11 @@ class DecoratedSymbol(TypedDict):
     character: str
 
 
+class LiteralMark(TypedDict):
+    type: Literal["literal"]
+    modifier: dict
+
+
 SimpleMark = dict[Literal["type"], str]
 
 LineNumberType = Literal["modulo100", "relative"]
@@ -28,4 +33,4 @@ class LineNumberRange(TypedDict):
 
 LineNumber = Union[LineNumberMark, LineNumberRange]
 
-Mark = Union[DecoratedSymbol, SimpleMark, LineNumber]
+Mark = Union[DecoratedSymbol, LiteralMark, SimpleMark, LineNumber]

src/modifiers/simple_scope_modifier.py

@@ -1,6 +1,6 @@
 from typing import Any
 
-from talon import Module
+from talon import Module, settings
 
 mod = Module()
 
@@ -13,6 +13,13 @@
     desc="Cursorless ancestor scope modifiers",
 )
 
+# This is a private setting and should not be used by non Cursorless developers
+mod.setting(
+    "private_cursorless_use_preferred_scope",
+    bool,
+    desc="Use preferred scope instead of containing scope for all scopes by default (EXPERIMENTAL)",
+)
+
 
 @mod.capture(
     rule=(
@@ -35,6 +42,12 @@ def cursorless_simple_scope_modifier(m) -> dict[str, Any]:
             "ancestorIndex": 1,
         }
 
+    if settings.get("user.private_cursorless_use_preferred_scope", False):
+        return {
+            "type": "preferredScope",
+            "scopeType": m.cursorless_scope_type,
+        }
+
     return {
         "type": "containingScope",
         "scopeType": m.cursorless_scope_type,

src/targets/primitive_target.py

@@ -11,7 +11,15 @@
     )
 )
 def cursorless_primitive_target(m) -> PrimitiveTarget:
-    return PrimitiveTarget(
-        getattr(m, "cursorless_mark", None),
-        getattr(m, "cursorless_modifier_list", None),
-    )
+    mark = getattr(m, "cursorless_mark", None)
+    modifiers = getattr(m, "cursorless_modifier_list", None)
+
+    # for grammar performance reasons, the literal modifier is exposed to Talon as a mark,
+    # but is converted to a modifier in the engine.
+    if mark is not None and mark["type"] == "literal":
+        if modifiers is None:
+            modifiers = []
+        modifiers.append(mark["modifier"])
+        mark = None
+
+    return PrimitiveTarget(mark, modifiers)