Fix template search path by ddi-acassidy · Pull Request #473 · wireviz/WireViz

ddi-acassidy · 2025-07-11T22:52:32Z

Addresses #472

the source file path is now forwarded to the Harness object and on to the HTML generator code. Its the last argument with a default to keep API compatibility. Ive also kept the output-relative search location to avoid hurting anyone relying on that behavior

the source file path is now forwarded to the Harness object and on to the HTML generator code. Its the last argument to keep API compatibility. Ive also kept the output-relative search location to avoid anyone relying on that behavior

kvid · 2025-07-25T17:29:39Z

Thank you @ddi-acassidy for this suggested fix for #472. However, I think we need to discuss the implementation a bit.

It seems to fix the issue for simple cases where the template name is specified in the main YAML input file, but not if the template name is specified in a prepend YAML file located in a different folder than the main YAML input file.
The problem described above is very similar to searching for images with a relative path, and I think we can reuse the image_paths list that was created to resolve the same issue for images with a relative path. If this can be verified, then I suggest renaming it to input_paths and use it for all relative paths (to images, templates, and any other separate input files specified by a relative path).
If my suggestion is accepted and verified, then PR Add support for a custom template directory path #444 also needs to be adjusted accordingly.
Please argue against my suggestion, and search for a use case where my suggestion will not be a good solution.
Please remember running isort and black on Python files before committing your changes - as described together with other recommendations in CONTRIBUTING.md.

ddi-acassidy · 2025-07-25T17:57:26Z

I hadnt thought of that case. If template is specified in the include file then the search path needs to be different from when its in the main file. Fixing that requires parsing the include file separately instead of just concatenating it with the other file.

kvid · 2025-07-25T20:56:11Z

I hadnt thought of that case. If template is specified in the include file then the search path needs to be different from when its in the main file. Fixing that requires parsing the include file separately instead of just concatenating it with the other file.

I understand your reasoning, and the same argument was used when discussing solutions to the similar image relative path issue. However, a pragmatic approach was finally selected that adds folders of all input YAML files into a list of paths, by searching relative image paths relative to all folders in this list, and use the first one found. I suggest using the same approach when searching templates for consistency.

ddi-acassidy · 2025-07-25T21:02:56Z

Got it, I wasnt aware of how those two bullet points fit together to solve the problem. I'll look into updating my PR with those changes when I have time

Relatedly, I put together support for Jinja templates as an alternative to the ersatz html comment template system available at the moment. It allows much more flexibility with HTML templates, like variable sized revision blocks and arbitrarily structured metadata, any interest in me cleaning that up and PRing it?

kvid · 2025-07-26T15:52:48Z

Got it, I wasnt aware of how those two bullet points fit together to solve the problem. I'll look into updating my PR with those changes when I have time

Please verify that reusing the existing image_paths variable will resolve the issue before (in a separate commit) renaming the variable to a generic name.

Relatedly, I put together support for Jinja templates as an alternative to the ersatz html comment template system available at the moment. It allows much more flexibility with HTML templates, like variable sized revision blocks and arbitrarily structured metadata, any interest in me cleaning that up and PRing it?

See also the existing PR #382 for Jinja.

kvid

I have in this review detailed my suggestions to reuse the existing image_paths list that already has collected all input folders to resolve the issue with more than one input YAML file at different folders. If you agree to this (after testing several use cases), then as a separate commit, renaming image_paths to input_folders might be a good change to reflect the more generic usage.

kvid · 2025-07-27T03:00:14Z

src/wireviz/Harness.py

    metadata: Metadata
    options: Options
    tweak: Tweak
+    source_path: Path


Suggested change

source_path: Path

input_folders: Union[str, Path, List]

kvid · 2025-07-27T03:07:20Z

src/wireviz/Harness.py

        # HTML output
        if "html" in fmt:
-            generate_html_output(filename, bomlist, self.metadata, self.options)
+            generate_html_output(filename, bomlist, self.metadata, self.options, self.source_path)


Suggested change

generate_html_output(filename, bomlist, self.metadata, self.options, self.source_path)

generate_html_output(

filename, bomlist, self.metadata, self.options, self.input_folders

)

The splitting into multiple lines is due to black formatting.

kvid · 2025-07-27T03:07:57Z

src/wireviz/wireviz.py

    output_dir: Union[str, Path] = None,
    output_name: Union[None, str] = None,
    image_paths: Union[Path, str, List] = [],
+    source_path = None,


Suggested change

source_path = None,

kvid · 2025-07-27T03:12:45Z

src/wireviz/wireviz.py

        metadata=Metadata(**yaml_data.get("metadata", {})),
        options=Options(**yaml_data.get("options", {})),
        tweak=Tweak(**yaml_data.get("tweak", {})),
+        source_path=source_path


I suggest reusing the existing image_paths that already has collected all input folders:

Suggested change

source_path=source_path

input_folders=image_paths,

kvid · 2025-07-27T03:14:42Z

src/wireviz/wv_cli.py

            output_dir=_output_dir,
            output_name=_output_name,
            image_paths=list(image_paths),
+            source_path=file


Suggested change

source_path=file

kvid · 2025-07-27T03:17:23Z

src/wireviz/wv_html.py

    bom_list: List[List[str]],
    metadata: Metadata,
    options: Options,
+    source: Union[str, Path] = None,


Suggested change

source: Union[str, Path] = None,

input_folders: Union[str, Path, List] = [],

kvid · 2025-07-27T03:18:36Z

src/wireviz/wv_html.py

+    template_search_paths = [ Path(filename).parent, Path(__file__).parent / "templates"]
+
+    if source is not None:
+        template_search_paths.insert(0, Path(source).parent)
+


Suggested change

template_search_paths = [ Path(filename).parent, Path(__file__).parent / "templates"]

if source is not None:

template_search_paths.insert(0, Path(source).parent)

kvid · 2025-07-27T03:27:40Z

src/wireviz/wv_html.py

        templatefile = smart_file_resolve(
-            f"{templatename}.html",
-            [Path(filename).parent, Path(__file__).parent / "templates"],
+            f"{templatename}.html", template_search_paths


Remember inserting from wireviz.wv_bom import make_list in the import section above and run isort *.py to format and sort them. (Github don't allow me to directly suggest changes at lines more than a few lines away from existing PR changes.)

Suggested change

f"{templatename}.html", template_search_paths

f"{templatename}.html",

make_list(input_folders) + [Path(__file__).parent / "templates"],

My suggestion here does not add the output folder in the list when different because I don't see any use case where that might be useful, but please argue against my suggestion.

tbornon-sts · 2026-02-12T07:20:29Z

Hi,
I'm interested by that change. Is there anything holding it back from being merged ?

kvid · 2026-02-13T11:09:51Z

@tbornon-sts wrote:

Hi, I'm interested by that change. Is there anything holding it back from being merged ?

As you can see above, I have reviewed this suggested change and verified that the simple cases are resolved, but I suggested a few more changes to also resolve the more general cases.

Feel free to download/install the suggested changes from this branch yourself to test if your use case is resolved.

If the author doesn't find time to do the work, then we probably should consider merging it as is and add the additional changes on top.

Fix template search path

37d2394

the source file path is now forwarded to the Harness object and on to the HTML generator code. Its the last argument to keep API compatibility. Ive also kept the output-relative search location to avoid anyone relying on that behavior

kvid requested changes Jul 27, 2025

View reviewed changes

kvid mentioned this pull request Nov 9, 2025

templates are searched relative to the html file path, not the yaml like the docs claim #472

Open

	source: Union[str, Path] = None,
	input_folders: Union[str, Path, List] = [],

	f"{templatename}.html", template_search_paths
	f"{templatename}.html",
	make_list(input_folders) + [Path(__file__).parent / "templates"],

Conversation

ddi-acassidy commented Jul 11, 2025

Uh oh!

kvid commented Jul 25, 2025

Uh oh!

ddi-acassidy commented Jul 25, 2025

Uh oh!

kvid commented Jul 25, 2025

Uh oh!

ddi-acassidy commented Jul 25, 2025

Uh oh!

kvid commented Jul 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kvid left a comment

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025

Choose a reason for hiding this comment

Uh oh!

kvid Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tbornon-sts commented Feb 12, 2026

Uh oh!

kvid commented Feb 13, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

kvid commented Jul 26, 2025 •

edited

Loading

kvid Jul 27, 2025 •

edited

Loading

kvid Jul 27, 2025 •

edited

Loading