Trickybrain Creating the LineReader file

python/selfie-lib/selfie_lib/LineReader.py

@@ -0,0 +1,59 @@
+from typing import Optional, Union
+import io
+
+class LineTerminatorReader(io.BufferedReader):
+    """Overrides read operations to detect carriage returns."""
+    def __init__(self, reader: io.TextIOWrapper) -> None:
+        super().__init__(reader.buffer)
+        self._unix_newlines = True
+
+    def read(self, size: int = -1) -> bytes:
+        chunk = super().read(size)
+        if b'\r' in chunk:
+            self._unix_newlines = False
+        return chunk
+
+    def unix_newlines(self) -> bool:
+        """Check if the newlines are Unix style."""
+        return self._unix_newlines
+
+class LineTerminatorAware(io.TextIOWrapper):
+    """Keeps track of the first line to determine newline style."""
+    def __init__(self, reader: LineTerminatorReader) -> None:
+        super().__init__(reader, encoding='utf-8')
+        self._first_line: Optional[str] = self.readline()
+
+    def readline(self, limit: int = -1) -> str:
+        if self._first_line is not None:
+            result, self._first_line = self._first_line, None
+            return result
+        return super().readline(limit)
+
+class LineReader:
+    """A reader that is aware of line terminators and line numbers."""
+    def __init__(self, reader: Union[io.StringIO, io.BufferedReader]) -> None:
+        self._reader = LineTerminatorAware(LineTerminatorReader(reader))
+
+    @staticmethod
+    def for_string(content: str) -> 'LineReader':
+        """Create a LineReader for string content."""
+        return LineReader(io.StringIO(content))
+
+    @staticmethod
+    def for_binary(content: bytes) -> 'LineReader':
+        """Create a LineReader for binary content."""
+        return LineReader(io.BufferedReader(io.BytesIO(content)))
+
+    def get_line_number(self) -> int:
+        """Get the current line number."""
+        # Assuming a way to track line numbers or using a wrapper that does.
+        # This is a placeholder as Python's io does not provide a direct lineno attribute.
+        return 0
+
+    def read_line(self) -> Optional[str]:
+        """Read the next line from the reader."""
+        return self._reader.readline()
+
+    def unix_newlines(self) -> bool:
+        """Check if the reader uses Unix newlines."""
+        return self._reader.unix_newlines()

python/selfie-lib/selfie_lib/__init__.py

@@ -3,4 +3,6 @@
 from .selina import get_interesting_fact as get_interesting_fact
 from .harvir import silly_addition as silly_addition
 from .edwin import simple_subtraction as simple_subtraction
+from .LineReader import LineReader as LineReader
 from .Slice import Slice as Slice
+

python/selfie-lib/tests/lineReaderTest.py

@@ -0,0 +1,42 @@
+from selfie_lib import LineReader
+
+def test_should_find_unix_separator_from_binary():
+    reader = LineReader.for_binary(b"This is a new line\n")
+    assert reader.unix_newlines() == True
+    assert reader.read_line() == "This is a new line"
+
+def test_should_find_windows_separator_from_binary():
+    reader = LineReader.for_binary(b"This is a new line\r\n")
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "This is a new line"
+
+def test_should_find_unix_separator_from_string():
+    reader = LineReader.for_string("This is a new line\n")
+    assert reader.unix_newlines() == True
+    assert reader.read_line() == "This is a new line"
+
+def test_should_find_windows_separator_from_string():
+    reader = LineReader.for_string("This is a new line\r\n")
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "This is a new line"
+
+def test_should_get_unix_line_separator_when_there_is_none():
+    reader = LineReader.for_binary(b"This is a new line")
+    assert reader.unix_newlines() == True
+    assert reader.read_line() == "This is a new line"
+
+def test_should_read_next_line_without_problem():
+    reader = LineReader.for_binary(b"First\r\nSecond\r\n")
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "First"
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "Second"
+    assert reader.unix_newlines() == False
+
+def test_should_use_first_line_separator_and_ignore_next():
+    reader = LineReader.for_binary(b"First\r\nAnother separator\n")
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "First"
+    assert reader.unix_newlines() == False
+    assert reader.read_line() == "Another separator"
+    assert reader.unix_newlines() == False