fix: parse header and hunksize strictly before context

diffoscope/diff.py

@@ -661,7 +661,8 @@ class SideBySideDiff:
         """
         self.reset()
 
-        for l in diff_split_lines(self.unified_diff, False):
+        diff_lines = iter(diff_split_lines(self.unified_diff, False))
+        for l in diff_lines:
             self._bytes_processed += len(l) + 1
             m = re.match(r"^--- ([^\s]*)", l)
             if m:
@@ -689,8 +690,10 @@ class SideBySideDiff:
                     self.hunk_off2,
                     self.hunk_size2,
                 )
-                continue
+                break
 
+        for l in diff_lines:
+            self._bytes_processed += len(l) + 1
             if re.match(r"^\[", l):
                 yield from self.empty_buffer()
                 yield "C", l

tests/data/output_expected_363.html

+ class="diffoscope">
+<div class="difference">
+  <div class="diffheader">
+  <div class="diffcontrol">⊟</div>
+  <div><span class="diffsize">182 B</span></div>
+  <div id="top"><span class="source">example.sql</span>
+    <a class="anchor" href="#top">¶</a>
+  </div>
+  </div>
+<table class="diff">
+<colgroup><col class="colines"/><col class="coldiff"/>
+<col class="colines"/><col class="coldiff"/></colgroup>
+<tr style="display:none;"><td>&nbsp;</td><td>&nbsp;</td><td>&nbsp;</td><td>&nbsp;</td></tr>
+<tr class="diffhunk"><td colspan="2">Offset 1, 2 lines modified</td><td colspan="2">Offset 1, 2 lines modified</td></tr>
+<tr class="diffunmodified"><td class="diffline">1 </td><td class="diffpresent">--<span class="dp">·</span>unrelated<span class="dp">·</span>comment</td><td class="diffline">1 </td><td class="diffpresent">--<span class="dp">·</span>unrelated<span class="dp">·</span>comment</td></tr>
+<tr class="diffchanged"><td class="diffline">2 </td><td class="diffpresent">--<span class="dp">·</span>comment<span class="dp">·</span>containing<span class="dp">·</span>a<span class="dp">·</span>timestamp:<span class="dp">·</span>2024-02-12T02:0<del>7</del>:00.000Z</td><td class="diffline">2 </td><td class="diffpresent">--<span class="dp">·</span>comment<span class="dp">·</span>containing<span class="dp">·</span>a<span class="dp">·</span>timestamp:<span class="dp">·</span>2024-02-12T02:0<ins>8</ins>:00.000Z</td></tr>
+</table></div>

tests/data/salsa-issue-363.diff.json

+{
+  "diffoscope-json-version": 1,
+  "source1": "example.sql",
+  "source2": "example.sql",
+  "unified_diff": "@@ -1,2 +1,2 @@\n -- unrelated comment\n--- comment containing a timestamp: 2024-02-12T02:07:00.000Z\n+-- comment containing a timestamp: 2024-02-12T02:08:00.000Z\n"
+}

tests/test_presenters.py

@@ -200,6 +200,25 @@ def test_html_regression_875281(tmpdir, capsys):
         )
 
 
+def test_html_presentation_issue_363(tmpdir, capsys):
+    filename = "salsa-issue-363.diff.json"
+    diff_data = get_data(filename)
+    diff_path = str(tmpdir.join(filename))
+    with open(diff_path, "w") as fp:
+        fp.write(diff_data)
+    report_path = str(tmpdir.join("report.html"))
+    out = run(
+        capsys,
+        "--html",
+        report_path,
+        f"--load-existing-diff={diff_path}",
+    )
+    assert out == ""
+
+    with open(report_path, "r", encoding="utf-8") as f:
+        assert extract_body(f.read()) == get_data("output_expected_363.html")
+
+
 def test_limited_print():
     def fake(x):
         return None