feat: didChange text document notification

src/malls/lsp/classes.py

@@ -1,4 +1,15 @@
-from tree_sitter import Tree
+import logging
+
+import tree_sitter_mal as ts_mal
+from tree_sitter import Language, Parser, Point, Tree
+
+from ..ts.utils import lsp_to_tree_sitter_position
+from .models import Position
+
+MAL_LANGUAGE = Language(ts_mal.language())
+PARSER = Parser(MAL_LANGUAGE)
+
+log = logging.getLogger(__name__)
 
 
 class Document:
@@ -11,3 +22,84 @@ class Document:
     def __init__(self, tree: Tree, text: str):
         self.text = text  # text is in bytes
         self.tree = tree
+
+    def _pos_to_byte(self, point: Point):
+        line, column = point.row, point.column
+        lines = self.text.split(b"\n")
+
+        byte_offset = 0
+
+        for i in range(line):
+            byte_offset += len(lines[i]) + 1  # +1 for the newline character
+
+        # Add the column position in the current line
+        byte_offset += column
+
+        return byte_offset
+
+    def _change_text(self, start: Point, end: Point, new_text: str) -> (int, int):
+        """
+        Auxiliary method to simply edit the text corresponding to
+        the document
+        """
+        # we need to convert from position to byte offset
+        start_byte, end_byte = self._pos_to_byte(start), self._pos_to_byte(end)
+        old_final_end_byte = self.tree.root_node.end_byte
+
+        # we also need to check if the end byte is bigger than the current
+        # text length
+        if old_final_end_byte < end_byte:
+            # replace end of file
+            self.text = self.text[:start_byte] + new_text
+            # therefore, the end of the file was the last changed byte
+            old_end_byte = old_final_end_byte
+        else:
+            # otherwise, we are changing in the middle of the file
+            self.text = self.text[:start_byte] + new_text + self.text[end_byte:]
+            # therefore, we only replaced things
+            old_end_byte = end_byte
+
+        return (start_byte, end_byte, old_end_byte)
+
+    def execute_changes(self, change_range: dict, text: str) -> None:
+        """
+        This function will process changes to files. To do this, the source
+        code must be edited, TreeSitter alerted of the location of the changes
+        and finally the code must be reparsed.
+        """
+
+        # start by converting the range to TreeSitter positions
+        start_position_lsp = Position(
+            line=change_range.start.line, character=change_range.start.character
+        )
+        end_position_lsp = Position(
+            line=change_range.end.line, character=change_range.end.character
+        )
+
+        start_position = lsp_to_tree_sitter_position(self.text, start_position_lsp, text)
+        end_position = lsp_to_tree_sitter_position(self.text, end_position_lsp, text)
+
+        # change text
+        start_byte, end_byte, old_end_byte = self._change_text(start_position, end_position, text)
+
+        # alert Treesitter of changes
+        self.tree.edit(
+            start_byte,
+            old_end_byte,
+            end_byte,
+            start_position,
+            end_position if old_end_byte == end_byte else self.tree.root_node.end_point,
+            end_position,
+        )
+
+        # reparse
+        self.tree = PARSER.parse(self.text, self.tree)
+
+    def change_whole_file(self, text: str) -> None:
+        """
+        Since we have a completely new file, we can simply replace the text and
+        reparse, as this is not a change to the tree - it's a new tree
+        """
+
+        self.text = text
+        self.tree = PARSER.parse(self.text)

src/malls/lsp/models.py

@@ -2615,3 +2615,37 @@ class DidOpenTextDocumentParams(BaseModel):
     """
 
     textDocument: TextDocumentItem
+
+
+class WholeFileChange(BaseModel):
+    """
+    Class to represent a change to the whole file
+    """
+
+    text: str
+
+
+class TextDocumentContentChangeEvent(BaseModel):
+    """
+    https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#textDocumentContentChangeEvent
+    """
+
+    range: Range
+
+    range_length: int | None = None
+
+    text: str | WholeFileChange
+
+    model_config = base_config
+
+
+class DidChangeTextDocumentParams(BaseModel):
+    """
+    https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#didChangeTextDocumentParams
+    """
+
+    text_document: VersionedTextDocumentIdentifier
+
+    content_changes: list[TextDocumentContentChangeEvent]
+
+    model_config = base_config

src/malls/mal_lsp.py

@@ -114,6 +114,7 @@ def state(self) -> LifecycleFSM:
     def trace_value(self) -> TraceValue:
         return self.__trace_value
 
+    @property
     def files(self) -> dict:
         return self.__files
 
@@ -281,3 +282,25 @@ def m_text_document__did_open(self, **params: dict | None) -> None:
 
         # with the opened file and included files parsed, we are done
         return
+
+    def m_text_document__did_change(self, **params: dict | None) -> None:
+        """
+        This function will process changes to files. To do this, the source
+        code must be edited, TreeSitter alerted of the location of the changes
+        and finally the code must be reparsed.
+        """
+
+        # Obtain text document
+        textDocument = models.DidChangeTextDocumentParams(**params)
+        doc_uri = uri_to_path(textDocument.text_document.uri)
+        document = self.__files[doc_uri]
+
+        # There could be various changes, so we need to iterate over them
+        for change in textDocument.content_changes:
+            changed_range = change.range
+            if type(change.text) is str:
+                text = change.text.encode()
+                document.execute_changes(changed_range, text)
+            else:
+                text = change.text.text.encode()  # whole file change
+                document.change_whole_file(text)

src/malls/ts/utils.py

@@ -156,16 +156,21 @@ def find_current_scope(cursor: TreeCursor, point: Point):
     return owner
 
 
-def lsp_to_tree_sitter_position(text: str, pos: Position) -> Point:
+def lsp_to_tree_sitter_position(text: str, pos: Position, new_text: str = None) -> Point:
     """
     Converts an LSP position (UTF-16 character index) to a Tree-sitter position (UTF-8 byte offset).
     """
     lsp_line, lsp_char = pos.line, pos.character
 
     lines = text.splitlines(keepends=True)
 
-    # Get correct line
-    line_text = lines[lsp_line]
+    if lsp_line >= len(lines):
+        # there is an extension to the text itself (didChange)
+        # so we have to consider the line being written
+        line_text = new_text.splitlines(keepends=True)[lsp_line - len(lines)]
+    else:
+        # Get correct line
+        line_text = lines[lsp_line]
 
     # The idea is to conver the string to UTF-16.
     # Since UTF-16 characters correspond to 2 bytes,
@@ -175,7 +180,7 @@ def lsp_to_tree_sitter_position(text: str, pos: Position) -> Point:
 
     # Convert to UTF-16 (each UTF-16 code unit is 2 bytes)
     # https://en.wikipedia.org/wiki/UTF-16#Byte-order_encoding_schemes
-    line_utf16 = line_text.encode("utf-16")
+    line_utf16 = line_text.decode().encode("utf-16")
 
     # check for BOM
     bom_size = 0

tests/conftest.py

@@ -8,6 +8,11 @@
 
 from .util import (
     BASE_OPEN_FILE,
+    CHANGE_FILE_1,
+    CHANGE_FILE_2,
+    CHANGE_FILE_3,
+    CHANGE_FILE_4,
+    CHANGE_FILE_5,
     OPEN_FILE_WITH_FAKE_INCLUDE,
     OPEN_FILE_WITH_INCLUDED_FILE,
     build_payload,
@@ -66,6 +71,11 @@ def template() -> typing.BinaryIO:
                 ([BASE_OPEN_FILE], fixture_name + "_base_open_file"),
                 ([OPEN_FILE_WITH_INCLUDED_FILE], fixture_name + "_with_included_file"),
                 ([OPEN_FILE_WITH_FAKE_INCLUDE], fixture_name + "_with_fake_include"),
+                ([BASE_OPEN_FILE, CHANGE_FILE_1], "change_middle_of_file_single_line"),
+                ([BASE_OPEN_FILE, CHANGE_FILE_2], "change_middle_of_file_multiple_lines"),
+                ([BASE_OPEN_FILE, CHANGE_FILE_3], "change_end_of_file"),
+                ([BASE_OPEN_FILE, CHANGE_FILE_4], "change_middle_of_file_twice"),
+                ([BASE_OPEN_FILE, CHANGE_FILE_5], "change_whole_file"),
             ]
             for payload, new_name in payloads:
                 fixture = pytest.fixture(

tests/fixtures/writeable_fixtures/did_change_notif.in.lsp

@@ -0,0 +1,6 @@
+Content-Length: 72
+
+{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"trace":"off"}}
+Content-Length: 41
+
+{"jsonrpc":"2.0","method":"initialized"}