text_wrap adds justify and use in format_table

Fixes #601

Author: rogerbinns

apsw/ext.py

@@ -2055,13 +2055,25 @@ def repl(s):
         )
 
     # break headers and cells into lines
-    def wrap(text: str, width: int) -> list[str]:
-        return list(apsw.unicode.text_wrap(text, width)) or [""]
+    def wrap(text: str, width: int, justify: apsw.unicode.Justify, hyphen: str) -> list[str]:
+        return list(apsw.unicode.text_wrap(text, width, justify=justify, hyphen=hyphen)) or [""]
+
+    # special formatting
+    formats = {
+        int: {"justify": apsw.unicode.Justify.RIGHT, "hyphen": ""},
+        float: {"justify": apsw.unicode.Justify.LEFT, "hyphen": ""},
+        "header": {"justify": apsw.unicode.Justify.CENTER, "hyphen": "-"},
+        "_": {"justify": apsw.unicode.Justify.LEFT, "hyphen": "-"},
+    }
+
+    colnames = [wrap(colnames[i], colwidths[i], **formats["header"]) for i in range(len(colwidths))]  # type: ignore
 
-    colnames = [wrap(colnames[i], colwidths[i]) for i in range(len(colwidths))]  # type: ignore
     for row in rows:
         for i, (text, t) in enumerate(row):  # type: ignore[misc]
-            row[i] = (wrap(text, colwidths[i]), t)  # type: ignore
+            row[i] = (
+                wrap(text, colwidths[i], **formats.get(t, formats["_"])),
+                t,
+            )  # type: ignore
 
     ## output
     # are any cells more than one line?
@@ -2085,29 +2097,22 @@ def do_bar(chars: str) -> None:
                 line += chars[2]
         out_lines.append(line)
 
-    def do_row(row, sep: str, *, centre: bool = False, header: bool = False) -> None:
+    def do_row(row, sep: str, *, header: bool = False) -> None:
         for n in range(max(len(cell[0]) for cell in row)):
             line = sep
             for i, (cell, t) in enumerate(row):
-                text = cell[n] if n < len(cell) else ""
-                text = " " + text.rstrip() + " "
-                lt = apsw.unicode.text_width(text)
-                extra = " " * max(colwidths[i] + 2 - lt, 0)
-                if centre:
-                    lpad = extra[: len(extra) // 2]
-                    rpad = extra[len(extra) // 2 :]
-                else:
-                    lpad = ""
-                    rpad = extra
+                text = cell[n] if n < len(cell) else " " * colwidths[i]
+                # we do a space on each side
+                text = " " + text + " "
                 if header:
-                    text = colour_wrap(lpad + text + rpad, None, header=True)
+                    text = colour_wrap(text, None, header=True)
                 else:
-                    text = lpad + colour_wrap(text, t) + rpad
+                    text = colour_wrap(text, t)
                 line += text + sep
             out_lines.append(line)
 
     do_bar("╭─┬╮" if use_unicode else "+-++")
-    do_row([(c, None) for c in colnames], "│" if use_unicode else "|", centre=True, header=True)
+    do_row([(c, None) for c in colnames], "│" if use_unicode else "|", header=True)
 
     # rows
     which = 0

apsw/tests/ftstests.py

@@ -2167,6 +2167,65 @@ def testIssue600(self):
             res = list(apsw.unicode.text_wrap(text, width, **args))
             self.assertEqual(res, expected)
 
+    def testWrapJustify(self):
+        "text_wrap justification"
+
+        longer = "a to the four verylongsingleword fivey ytrkes seven77 eightsse"
+
+        J = apsw.unicode.Justify
+        for text, justify, width, expected in (
+            ("hello", J.LEFT, 10, ["hello     "]),
+            ("hello", J.RIGHT, 10, ["     hello"]),
+            ("hello", J.CENTER, 10, ["  hello   "]),
+            (
+                longer,
+                J.LEFT,
+                10,
+                [
+                    "a to the  ",
+                    "four      ",
+                    "verylongs-",
+                    "ingleword ",
+                    "fivey     ",
+                    "ytrkes    ",
+                    "seven77   ",
+                    "eightsse  ",
+                ],
+            ),
+            (
+                longer,
+                J.RIGHT,
+                10,
+                [
+                    "  a to the",
+                    "      four",
+                    "verylongs-",
+                    " ingleword",
+                    "     fivey",
+                    "    ytrkes",
+                    "   seven77",
+                    "  eightsse",
+                ],
+            ),
+            (
+                longer,
+                J.CENTER,
+                10,
+                [
+                    " a to the ",
+                    "   four   ",
+                    "verylongs-",
+                    "ingleword ",
+                    "  fivey   ",
+                    "  ytrkes  ",
+                    " seven77  ",
+                    " eightsse ",
+                ],
+            ),
+        ):
+            res = list(apsw.unicode.text_wrap(text, width, justify=justify))
+            self.assertEqual(res, expected)
+
     def testVersion(self):
         # check we have it in the date list
         self.assertIn(apsw.unicode.unicode_version, apsw.unicode.version_dates)

apsw/unicode.py

@@ -114,6 +114,7 @@
 
 from typing import Iterator, Iterable, Any
 
+import enum
 import re
 
 ### BEGIN UNICODE UPDATE SECTION ###
@@ -559,6 +560,10 @@ def append_paragraph(p: list[str]) -> None:
     # turn back into newline as the expected delimiter
     return "\n".join(paragraphs) + "\n"
 
+class Justify(enum.Enum):
+    LEFT = 0
+    CENTER = 1
+    RIGHT = 2
 
 def text_wrap(
     text: str,
@@ -568,6 +573,7 @@ def text_wrap(
     hyphen: str = "-",
     combine_space: bool = True,
     invalid: str = "?",
+    justify: Justify = Justify.LEFT
 ) -> Iterator[str]:
     """Similar to :func:`textwrap.wrap` but Unicode grapheme cluster and line break aware
 
@@ -585,6 +591,7 @@ def text_wrap(
           off.
     :param invalid: If invalid codepoints are encountered such as control characters and surrogates
           then they are replaced with this.
+    :param justify: Where to align text within each line
 
     This yields one line of :class:`str` at a time, which will be
     exactly ``width`` when output to a terminal.  It will be right
@@ -597,6 +604,17 @@ def text_wrap(
 
     hyphen_width = text_width(hyphen)
 
+    def do_justify(line: str, n_spaces: int):
+        match justify:
+            case Justify.LEFT:
+                return line + " " * n_spaces
+            case Justify.RIGHT:
+                return " " * n_spaces + line
+            case Justify.CENTER:
+                l = n_spaces // 2
+                r = n_spaces - l
+                return " " * l + line + " " *r
+
     if hyphen_width:
         # we don't need a hyphen if the text already fits in one line
         # which avoids truncating indent unnecessarily
@@ -664,7 +682,7 @@ def text_wrap(
                     line_width = len(indent)
                     seg_width = text_width(segment)
                     continue
-                yield "".join(accumulated) + " " * (width - line_width)
+                yield do_justify("".join(accumulated), width - line_width)
                 if combine_space and segment[0] == " ":
                     # we added a space, but don't need it on new line
                     segment = segment[1:]
@@ -678,7 +696,7 @@ def text_wrap(
         if len(accumulated) != 1:
             # length 1 means it is only indent which we don't output
             # as last line
-            yield "".join(accumulated) + " " * (width - line_width)
+            yield do_justify("".join(accumulated), width - line_width)
 
 
 def codepoint_name(codepoint: int | str) -> str | None:

doc/changes.rst

@@ -16,6 +16,11 @@ APSW changes by version
 Fix wrapping, indents. and space for hyphens in
 :func:`apsw.unicode.text_wrap` under various conditions (:issue:`600`)
 
+Added option to :func:`apsw.unicode.text_wrap`  for justifying text.  This
+is used in :func:`apsw.ext.format_query_table` to centre column names,
+and right align integers.  That updates the :doc:`shell <shell>` output.
+(:issue:`601`)
+
 3.51.2.0
 ========