Implement format_number utility and update various summaries to use it to separate 1000s with a thin space

Author: hossam26644

python/CHANGELOG.rst

@@ -20,6 +20,9 @@
 - Add ``TreeSequence.map_to_vcf_model`` method to return a mapping of
   the tree sequence to the VCF model. 
   (:user:`benjeffery`, :pr:`3163`)
+- Use a thin space as the thousands separator in HTML output,
+  and a comma in CLI output.
+  (:user:`hossam26644`, :pr:`3167`, :issue:`2951`)
 
 **Fixes**
 
@@ -38,6 +41,7 @@
   to the required pattern.
   (:user:`benjeffery`, :pr:`3163`)
 
+
 --------------------
 [0.6.3] - 2025-04-28
 --------------------

python/tests/test_highlevel.py

@@ -3764,19 +3764,19 @@ def test_str(self, ts_fixture):
             ╔═+╗\s*
             ║Tree.*?║\s*
             ╠═+╤═+╣\s*
-            ║Index.*?│\s*\d+║\s*
+            ║Index.*?│\s*[\d\u2009,]+║\s*
             ╟─+┼─+╢\s*
-            ║Interval.*?│\s*\d+-\d+\s*\(\d+\)║\s*
+            ║Interval.*?│\s*[\d\u2009,]+-[\d\u2009,]+\s*\([\d\u2009,]+\)║\s*
             ╟─+┼─+╢\s*
-            ║Roots.*?│\s*\d+║\s*
+            ║Roots.*?│\s*[\d\u2009,]+║\s*
             ╟─+┼─+╢\s*
-            ║Nodes.*?│\s*\d+║\s*
+            ║Nodes.*?│\s*[\d\u2009,]+║\s*
             ╟─+┼─+╢\s*
-            ║Sites.*?│\s*\d+║\s*
+            ║Sites.*?│\s*[\d\u2009,]+║\s*
             ╟─+┼─+╢\s*
-            ║Mutations.*?│\s*\d+║\s*
+            ║Mutations.*?│\s*[\d\u2009,]+║\s*
             ╟─+┼─+╢\s*
-            ║Total\s*Branch\s*Length.*?│\s*[\d,]+\.\d+║\s*
+            ║Total\s*Branch\s*Length.*?│\s*[\d\u2009,]+\.\d+║\s*
             ╚═+╧═+╝\s*
             """,
             re.VERBOSE | re.DOTALL,

python/tests/test_util.py

@@ -371,6 +371,19 @@ def test_naturalsize(value, expected):
         assert util.naturalsize(-value) == expected
 
 
+def test_format_number():
+    assert util.format_number(0) == "0"
+    assert util.format_number("1.23") == "1.23"
+    assert util.format_number(3216546.34) == "3 216 546.3"
+    assert util.format_number(3216546.34, 9) == "3 216 546.34"
+    assert util.format_number(-3456.23) == "-3 456.23"
+    assert util.format_number(-3456.23, sep=",") == "-3,456.23"
+
+    with pytest.raises(TypeError) as e_info:
+        util.format_number("bad")
+        assert str(e_info.value) == "The string cannot be converted to a number"
+
+
 @pytest.mark.parametrize(
     "obj, expected",
     [

python/tskit/genotypes.py

@@ -334,22 +334,25 @@ def __str__(self) -> str:
         Return a plain text summary of the contents of a variant.
         """
         try:
-            site_id = self.site.id
-            site_position = self.site.position
+            site_id = util.format_number(self.site.id, sep=",")
+            site_position = util.format_number(self.site.position, sep=",")
             counts = self.counts()
             freqs = self.frequencies()
+            samples = util.format_number(len(self.samples), sep=",")
+            num_alleles = util.format_number(self.num_alleles, sep=",")
             rows = (
                 [
-                    ["Site id", f"{site_id:,}"],
-                    ["Site position", f"{site_position:,}"],
-                    ["Number of samples", f"{len(self.samples):,}"],
-                    ["Number of alleles", f"{self.num_alleles:,}"],
+                    ["Site id", f"{site_id}"],
+                    ["Site position", f"{site_position}"],
+                    ["Number of samples", f"{samples}"],
+                    ["Number of alleles", f"{num_alleles}"],
                 ]
                 + [
                     [
                         f"Samples with allele "
                         f"""{'missing' if k is None else "'" + k + "'"}""",
-                        f"{counts[k]:,} ({freqs[k] * 100:.2g}%)",
+                        f"{util.format_number(counts[k], sep=',')} "
+                        f"({util.format_number(freqs[k] * 100, 2, sep=',')}%)",
                     ]
                     for k in self.alleles
                 ]

python/tskit/trees.py

@@ -2822,17 +2822,21 @@ def __str__(self):
         Return a plain text summary of a tree in a tree sequence
         """
         tree_rows = [
-            ["Index", f"{self.index:,}"],
+            ["Index", f"{util.format_number(self.index, sep=',')}"],
             [
                 "Interval",
-                f"{self.interval.left:,.8g}-{self.interval.right:,.8g}"
-                f"({self.span:,.8g})",
+                f"{util.format_number(self.interval.left, sep=',')}-"
+                f"{util.format_number(self.interval.right, sep=',')}"
+                f"({util.format_number(self.span, sep=',')})",
+            ],
+            ["Roots", f"{util.format_number(self.num_roots, sep=',')}"],
+            ["Nodes", f"{util.format_number(len(self.preorder()), sep=',')}"],
+            ["Sites", f"{util.format_number(self.num_sites, sep=',')}"],
+            ["Mutations", f"{util.format_number(self.num_mutations, sep=',')}"],
+            [
+                "Total Branch Length",
+                f"{util.format_number(self.total_branch_length, sep=',')}",
             ],
-            ["Roots", f"{self.num_roots:,}"],
-            ["Nodes", f"{len(self.preorder()):,}"],
-            ["Sites", f"{self.num_sites:,}"],
-            ["Mutations", f"{self.num_mutations:,}"],
-            ["Total Branch Length", f"{self.total_branch_length:,.8g}"],
         ]
         return util.unicode_table(tree_rows, title="Tree")
 
@@ -4399,17 +4403,10 @@ def __str__(self):
         Return a plain text summary of the contents of a tree sequence
         """
         ts_rows = [
-            ["Trees", str(self.num_trees)],
-            [
-                "Sequence Length",
-                str(
-                    int(self.sequence_length)
-                    if self.discrete_genome
-                    else self.sequence_length
-                ),
-            ],
+            ["Trees", util.format_number(self.num_trees, sep=",")],
+            ["Sequence Length", util.format_number(self.sequence_length, sep=",")],
             ["Time Units", self.time_units],
-            ["Sample Nodes", str(self.num_samples)],
+            ["Sample Nodes", util.format_number(self.num_samples, sep=",")],
             ["Total Size", util.naturalsize(self.nbytes)],
         ]
         header = ["Table", "Rows", "Size", "Has Metadata"]
@@ -4418,7 +4415,7 @@ def __str__(self):
             table_rows.append(
                 [
                     name.capitalize(),
-                    f"{table.num_rows:,}",
+                    f"{util.format_number(table.num_rows, sep=',')}",
                     util.naturalsize(table.nbytes),
                     (
                         "Yes"

python/tskit/util.py

@@ -480,6 +480,26 @@ def unicode_table(
     return "".join(out)
 
 
+def format_number(number, sig_digits=8, sep="\u2009"):
+    """
+    Format a number with with a separator to indicate thousands
+    and up to `sig_digits` significant digits using 'g' format.
+
+    number: int, float, or a numeric string.
+    sig_digits: int, number of significant digits to display.
+    sep: str, the separator to use for thousands, default is a thin space.
+    Returns a string.
+    """
+    if isinstance(number, str):
+        try:
+            number = float(number)
+        except ValueError:
+            raise TypeError("The string cannot be converted to a number")
+
+    fmt = f",.{sig_digits}g"
+    return format(number, fmt).replace(",", sep)
+
+
 def html_table(rows, *, header):
     headers = "".join(f"<th>{h}</th>" for h in header)
     rows = (
@@ -519,7 +539,7 @@ def tree_sequence_html(ts):
         f"""
             <tr>
                 <td>{name.capitalize()}</td>
-                <td>{table.num_rows:,}</td>
+                <td>{format_number(table.num_rows)}</td>
                 <td>{naturalsize(table.nbytes)}</td>
                 <td style="text-align: center;">
                     {'✅' if hasattr(table, "metadata") and len(table.metadata) > 0
@@ -599,10 +619,10 @@ def tree_sequence_html(ts):
                             </tr>
                         </thead>
                         <tbody>
-                            <tr><td>Trees</td><td>{ts.num_trees:,}</td></tr>
-                            <tr><td>Sequence Length</td><td>{ts.sequence_length:,}</td></tr>
+                            <tr><td>Trees</td><td>{format_number(ts.num_trees)}</td></tr>
+                            <tr><td>Sequence Length</td><td>{format_number(ts.sequence_length)}</td></tr>
                             <tr><td>Time Units</td><td>{ts.time_units}</td></tr>
-                            <tr><td>Sample Nodes</td><td>{ts.num_samples:,}</td></tr>
+                            <tr><td>Sample Nodes</td><td>{format_number(ts.num_samples)}</td></tr>
                             <tr><td>Total Size</td><td>{naturalsize(ts.nbytes)}</td></tr>
                             <tr>
                                 <td>Metadata</td><td style="text-align: left;">{md}</td>
@@ -671,13 +691,13 @@ def tree_html(tree):
                       </tr>
                     </thead>
                     <tbody>
-                      <tr><td>Index</td><td>{tree.index:,}</td></tr>
-                      <tr><td>Interval</td><td>{tree.interval.left:,.8g}-{tree.interval.right:,.8g} ({tree.span:,.8g})</td></tr>
-                      <tr><td>Roots</td><td>{tree.num_roots:,}</td></tr>
-                      <tr><td>Nodes</td><td>{len(tree.preorder()):,}</td></tr>
-                      <tr><td>Sites</td><td>{tree.num_sites:,}</td></tr>
-                      <tr><td>Mutations</td><td>{tree.num_mutations:,}</td></tr>
-                      <tr><td>Total Branch Length</td><td>{tree.total_branch_length:,.8g}</td></tr>
+                      <tr><td>Index</td><td>{format_number(tree.index)}</td></tr>
+                      <tr><td>Interval</td><td>{format_number(tree.interval.left)}-{format_number(tree.interval.right)} ({format_number(tree.span)})</td></tr>
+                      <tr><td>Roots</td><td>{format_number(tree.num_roots)}</td></tr>
+                      <tr><td>Nodes</td><td>{format_number(len(tree.preorder()))}</td></tr>
+                      <tr><td>Sites</td><td>{format_number(tree.num_sites)}</td></tr>
+                      <tr><td>Mutations</td><td>{format_number(tree.num_mutations)}</td></tr>
+                      <tr><td>Total Branch Length</td><td>{format_number(tree.total_branch_length)}</td></tr>
                     </tbody>
                   </table>
                 </div>
@@ -746,18 +766,18 @@ def variant_html(variant):
         return (
             html_body_head
             + f"""
-                <tr><td>Site Id</td><td>{site_id:,}</td></tr>
-                <tr><td>Site Position</td><td>{site_position:,.8g}</td></tr>
-                <tr><td>Number of Samples</td><td>{num_samples:,}</td></tr>
-                <tr><td>Number of Alleles</td><td>{num_alleles:,}</td></tr>
+                <tr><td>Site Id</td><td>{format_number(site_id)}</td></tr>
+                <tr><td>Site Position</td><td>{format_number(site_position)}</td></tr>
+                <tr><td>Number of Samples</td><td>{format_number(num_samples)}</td></tr>
+                <tr><td>Number of Alleles</td><td>{format_number(num_alleles)}</td></tr>
             """
             + "\n".join(
                 [
                     f"""<tr><td>Samples with Allele {'missing' if k is None
                                                      else "'" + k + "'"}</td><td>"""
-                    + f"{counts[k]:,}"
+                    + f"{format_number(counts[k])}"
                     + " "
-                    + f"({freqs[k] * 100:,.2g}%)"
+                    + f"({format_number(freqs[k] * 100, 2)}%)"
                     + "</td></tr>"
                     for k in variant.alleles
                 ]