html export: improve custom url handling (#411)

Co-authored-by: viseshrp <[email protected]>

Author: viseshrp

src/ansys/dynamicreporting/core/serverless/html_exporter.py

@@ -124,6 +124,7 @@ def _wrap_full_document(self, html_fragment_or_doc: str) -> str:
         text = html_fragment_or_doc.lstrip()
 
         # Already a full document? Return as-is.
+        # only look at the first ~2KB to avoid scanning huge strings.
         lowered = text[:2000].lower()
         if lowered.startswith("<!doctype") or "<html" in lowered:
             return html_fragment_or_doc
@@ -168,14 +169,10 @@ def _replace_files(self, text: str, inline: bool = False, size_check: bool = Fal
 
         patterns = []
         if ver:
-            patterns.append(f"{self._static_url}ansys{ver}/")  # custom
-            patterns.append(f"/static/ansys{ver}/")  # legacy literal
-
-        # custom first, then legacy literals
-        patterns.extend([self._static_url, self._media_url, "/static/", "/media/"])
-
+            patterns.append(f"{self._static_url}ansys{ver}/")
+        patterns.extend([self._static_url, self._media_url])
         if ver:
-            patterns.append(f"/ansys{ver}/")  # server-root style
+            patterns.append(f"/ansys{ver}/")
 
         while True:
             # Find the next match using the legacy priority order
@@ -361,15 +358,10 @@ def _process_file(self, path_in_html: str, pathname: str, inline: bool = False)
         # Resolve source file location based on the raw pathname (no normalization)
         ver = str(self._ansys_version) if self._ansys_version is not None else ""
 
-        # Source resolution: custom first, then legacy literals
         if pathname.startswith(self._media_url):
             source_file = self._media_dir / pathname.replace(self._media_url, "", 1)
         elif pathname.startswith(self._static_url):
             source_file = self._static_dir / pathname.replace(self._static_url, "", 1)
-        elif pathname.startswith("/media/"):  # legacy literal
-            source_file = self._media_dir / pathname.replace("/media/", "", 1)
-        elif pathname.startswith("/static/"):  # legacy literal
-            source_file = self._static_dir / pathname.replace("/static/", "", 1)
         elif ver and pathname.startswith(f"/ansys{ver}/"):
             source_file = self._static_dir / pathname.lstrip("/")
         else:
@@ -417,19 +409,11 @@ def _process_file(self, path_in_html: str, pathname: str, inline: bool = False)
         else:
             content = self._fix_viewer_component_paths(basename, content)
 
-        # Output path:
-        # keep ansys tree if the input path came from /static/ansys{ver}/ (custom OR legacy literal)
-        if ver and (
-            pathname.startswith(f"{self._static_url}ansys{ver}/")
-            or pathname.startswith(f"/static/ansys{ver}/")
-        ):
-            local_pathname = os.path.dirname(pathname)
-            # normalize either custom or legacy /static/ to './'
-            if local_pathname.startswith(self._static_url):
-                local_pathname = local_pathname.replace(self._static_url, "./", 1)
-            elif local_pathname.startswith("/static/"):
-                local_pathname = local_pathname.replace("/static/", "./", 1)
-
+        # Output path (exact legacy behavior):
+        # - If /static/ansys{ver}/ -> keep ansys tree, remove '/static/' -> './ansys{ver}/.../<basename>'
+        # - Else -> './media/<basename>'
+        if ver and pathname.startswith(f"{self._static_url}ansys{ver}/"):
+            local_pathname = os.path.dirname(pathname).replace(self._static_url, "./", 1)
             result = f"{local_pathname}/{basename}"
             target_file = self._output_dir / local_pathname.lstrip("./") / basename
         else:
@@ -445,12 +429,12 @@ def _process_file(self, path_in_html: str, pathname: str, inline: bool = False)
     def _find_block(self, text: str, start: int, prefix: str, suffix: str) -> tuple[int, int, str]:
         """
         Legacy-compatible: return the next [prefix ... suffix] block that contains at least
-        one asset-like reference. Accept both the literal legacy prefixes and the configured
-        custom prefixes. Also accept any '/ansys<ver>/' or generic '/ansys<digits>/'.
+        one asset-like reference. Accept the configured custom prefixes and the dynamic
+        /ansys<ver>/ root. No hard-coded legacy literals.
         """
-        # Normalize known prefixes (custom URLs may differ from /static/ and /media/)
         custom_static = (self._static_url or "").strip()
         custom_media = (self._media_url or "").strip()
+        ver = str(self._ansys_version) if self._ansys_version is not None else ""
 
         while True:
             try:
@@ -462,14 +446,13 @@ def _find_block(self, text: str, start: int, prefix: str, suffix: str) -> tuple[
             except ValueError:
                 return -1, -1, ""
             idx2 += len(suffix)
+
             block = text[idx1:idx2]
 
             if (
-                ("/static/" in block)
-                or ("/media/" in block)
-                or (custom_static and custom_static in block)
+                (custom_static and custom_static in block)
                 or (custom_media and custom_media in block)
-                or re.search(r"/ansys\d+/", block) is not None
+                or (ver and f"/ansys{ver}/" in block)
             ):
                 return idx1, idx2, block
 

tests/serverless/test_adr.py

@@ -1044,18 +1044,19 @@ def test_export_report_as_html(adr_serverless, tmp_path, monkeypatch):
 
     adr_serverless.create_template(BasicLayout, name="TestExportReport", parent=None)
 
-    # Return a fragment (not a full <html>) so wrapper kicks in
+    # Use whatever ADR is actually configured to serve as STATIC_URL
+    href = f"{adr_serverless.static_url}website/content/site.css"
+
     def mock_render_report(self, name, **kwargs):
         # Use the provided static asset path that is known to exist.
         return (
             '<div class="body-content">'
-            '<link rel="stylesheet" type="text/css" href="/static/website/content/site.css"/>'
+            f'<link rel="stylesheet" type="text/css" href="{href}"/>'
             "</div>"
         )
 
     monkeypatch.setattr(ADR, "render_report", mock_render_report)
 
-    # Act: Call the method under test. The full exporter logic will run.
     output_path = adr_serverless.export_report_as_html(
         output_directory=tmp_path,
         name="TestExportReport",
@@ -1064,7 +1065,7 @@ def mock_render_report(self, name, **kwargs):
     assert output_path == tmp_path / "index.html"
     assert output_path.exists()
 
-    # Since we flatten unknown /static/* (non-ansys) into ./media/
+    # Unknown /static/* (i.e., non-ansys path) should flatten to ./media/<basename>
     assert (tmp_path / "media" / "site.css").exists()
 
     html = output_path.read_text(encoding="utf-8")

tests/serverless/test_html_exporter.py

@@ -20,7 +20,7 @@ def _write(p: Path, data: bytes | str):
         p.write_text(data, encoding="utf-8")
 
 
-def _fragment_with_static_ref(path: str = "/static/website/content/site.css") -> str:
+def _fragment_with_static_ref(path: str) -> str:
     # pure fragment (no <html>), to force the exporter to wrap into a full document
     return f'<div class="body-content m-1"><link rel="stylesheet" href="{path}"/></div>'
 
@@ -40,8 +40,10 @@ def test_wraps_fragment_and_emits_title_and_favicon(adr_serverless, tmp_path: Pa
     _write(static_dir / "website/images/favicon.png", b"\x89PNG\x00")
     _write(static_dir / "website/content/site.css", "body{overflow:hidden;}")
 
+    href = f"{adr_serverless.static_url}website/content/site.css"
+
     exporter = ServerlessReportExporter(
-        html_content=_fragment_with_static_ref(),
+        html_content=_fragment_with_static_ref(href),
         output_dir=tmp_path / "export1",
         static_dir=static_dir,
         media_dir=media_dir,
@@ -76,9 +78,12 @@ def test_static_is_flattened_media_and_ansys_tree_preserved(adr_serverless, tmp_
     # favicon for wrapper
     _write(static_dir / "website/images/favicon.png", b"\x89PNG\x00")
 
-    html = _fragment_with_static_ref("/static/website/scripts/plotly.min.js") + textwrap.dedent(
+    plotly_href = f"{adr_serverless.static_url}website/scripts/plotly.min.js"
+    ansys_util_src = f"{adr_serverless.static_url}ansys{ver}/nexus/utils/js-test.js"
+
+    html = _fragment_with_static_ref(plotly_href) + textwrap.dedent(
         f"""
-        <script src="/static/ansys{ver}/nexus/utils/js-test.js"></script>
+        <script src="{ansys_util_src}"></script>
         """
     )
 
@@ -112,8 +117,10 @@ def test_favicon_png_is_duplicated_as_ico(adr_serverless, tmp_path: Path):
     _write(static_dir / "website/images/favicon.png", b"PNGDATA")
     _write(static_dir / "website/content/site.css", "h2{}")
 
+    href = f"{adr_serverless.static_url}website/content/site.css"
+
     exporter = ServerlessReportExporter(
-        html_content=_fragment_with_static_ref(),
+        html_content=_fragment_with_static_ref(href),
         output_dir=tmp_path / "export4",
         static_dir=static_dir,
         media_dir=media_dir,
@@ -140,10 +147,13 @@ def test_inline_viewer_size_exception_sets_proxy_only(adr_serverless, tmp_path:
     _write(media_dir / "preview.png", b"P")
     _write(static_dir / "website/images/favicon.png", b"P")
 
+    media_src = f"{adr_serverless.media_url}bigfile.stl"
+    media_preview = f"{adr_serverless.media_url}preview.png"
+
     html = textwrap.dedent(
-        """
+        f"""
         <div>
-          <ansys-nexus-viewer src="/media/bigfile.stl" proxy_img="/media/preview.png"></ansys-nexus-viewer>
+          <ansys-nexus-viewer src="{media_src}" proxy_img="{media_preview}"></ansys-nexus-viewer>
         </div>
         """
     )
@@ -174,11 +184,13 @@ def test_scene_js_inlines_binary_block_and_namespaces_filename(adr_serverless, t
     ver = str(adr_serverless.ansys_version)
 
     # Babylon scene referencing a binary block; exporter will inline it
-    _write(static_dir / "website/scenes/guid123.scene.js", "load_binary_block('/media/blob.bin');")
+    media_blob = f"{adr_serverless.media_url}blob.bin"
+    _write(static_dir / "website/scenes/guid123.scene.js", f"load_binary_block('{media_blob}');")
     _write(media_dir / "blob.bin", b"\x00\x01\x02\x03")
     _write(static_dir / "website/images/favicon.png", b"P")
 
-    html = '<div><script src="/static/website/scenes/guid123.scene.js"></script></div>'
+    script_src = f"{adr_serverless.static_url}website/scenes/guid123.scene.js"
+    html = f'<div><script src="{script_src}"></script></div>'
 
     exporter = ServerlessReportExporter(
         html_content=html,
@@ -239,7 +251,8 @@ def test_no_inline_flag_forces_copy_not_data_uri(adr_serverless, tmp_path: Path)
     _write(static_dir / "website/images/favicon.png", b"P")
     _write(media_dir / "tiny.bin", b"ABCD")
 
-    html = '<div><a href="/media/tiny.bin">dl</a></div>'
+    href = f"{adr_serverless.media_url}tiny.bin"
+    html = f'<div><a href="{href}">dl</a></div>'
 
     exporter = ServerlessReportExporter(
         html_content=html,
@@ -268,7 +281,8 @@ def test_missing_source_file_keeps_original_ref(adr_serverless, tmp_path: Path):
     _write(static_dir / "website/images/favicon.png", b"P")
 
     # Refer to a media file that doesn't exist
-    html = '<div><img src="/media/does_not_exist.xyz"/></div>'
+    missing_href = f"{adr_serverless.media_url}does_not_exist.xyz"
+    html = f'<div><img src="{missing_href}"/></div>'
 
     exporter = ServerlessReportExporter(
         html_content=html,
@@ -283,4 +297,4 @@ def test_missing_source_file_keeps_original_ref(adr_serverless, tmp_path: Path):
 
     out = (tmp_path / "export9" / "index.html").read_text(encoding="utf-8")
     # Exporter leaves the original path when it can't find a local file
-    assert "/media/does_not_exist.xyz" in out
+    assert missing_href in out