FYI : This has been generated by AI
Use a named module-level logger instead of root logger in h4d.py
Summary
html4docx/h4d.py emits all its warnings via the module-level logging functions (logging.warning(...)), which write to the root logger. As a library, this is problematic: consumers cannot raise the verbosity or silence these messages without affecting their entire application logging, because there is no dedicated logger name to target in their logging config.
Why this matters
In a Django (or any framework) LOGGING config, you normally silence a noisy dependency with:
"loggers": {
"html4docx": {"level": "ERROR", "propagate": False},
},This currently has no effect, because the records are created on the root logger, not on an html4docx-namespaced logger. In production we get hundreds of lines like:
WARNING ... Unrecognized paragraph style 'letter-spacing', will be skipped.
WARNING ... Unrecognized paragraph style 'margin', will be skipped.
WARNING ... Unrecognized paragraph style 'padding', will be skipped.
for every HTML→DOCX conversion (these CSS properties are very common), and there is no clean way to filter them out short of attaching a custom logging.Filter on a handler and matching record.module == 'h4d' — which is fragile.
Affected version
1.1.5
Proposed fix
Define a module-level named logger and route every logging.<level>(...) call through it.
1. Near the top of html4docx/h4d.py (after the import logging on line 2):
logger = logging.getLogger(__name__)
This gives the logger the name html4docx.h4d, so consumers can configure html4docx (parent) or html4docx.h4d directly.
2. Replace the 14 logging.<level>(...) calls with logger.<level>(...). Affected lines in h4d.py (v1.1.5):
| Line |
Current |
| 531 |
logging.warning(f"Warning: Unrecognized style '{style_name}', will be skipped.") |
| 551 |
logging.warning(f"Warning: Unrecognized paragraph style '{style_name}', will be skipped.") |
| 712 |
logging.warning(f"Warning: Could not parse font-size '{font_size}': {e}") |
| 761 |
logging.warning(f"Warning: Could not apply font-family '{font_family}': {e}") |
| 788 |
logging.warning(f"Could not apply color '{color_value}': {e}") |
| 826 |
logging.warning(f"Warning: Unsupported text transform '{text_transform}'") |
| 829 |
logging.warning(f"Warning: Could not apply text-transform '{text_transform}': {e}") |
| 923 |
logging.warning(f"Warning: Unsupported text decoration '{text_decoration_line}'") |
| 945 |
logging.warning(f"Warning: Style not recognized'{text_decoration_style}', defaulting to single line.") |
| 976 |
logging.warning(f"Warning: Unsupported background color '{background_color}'") |
| 996 |
logging.warning(f"Could not apply background-color to paragraph: {e}") |
| 1005 |
logging.warning(f"Warning: Unsupported background color '{background_color}'") |
| 1027 |
logging.warning(f"Could not apply background-color to run: {e}") |
| 1513 |
logging.warning(f"Warning: Custom style '{custom_style}' not found in document, Ignoring style.") |
i.e. logging.warning(...) → logger.warning(...). A single search-and-replace of logging.warning( → logger.warning( within h4d.py covers all of them.
Notes
- This is fully backwards-compatible: by default the records still propagate to the root logger, so existing behaviour is unchanged. It only adds the ability to configure them.
- The CLI entry point (
argparse/__main__) can keep a logging.basicConfig(...) call so command-line users still see output.
- Optionally, since "unrecognized style" messages are expected for any non-trivial CSS, consider downgrading the recurring ones (lines 531/551) from
warning to debug — they describe normal, non-actionable skips rather than problems.
FYI : This has been generated by AI
Use a named module-level logger instead of root logger in
h4d.pySummary
html4docx/h4d.pyemits all its warnings via the module-level logging functions (logging.warning(...)), which write to the root logger. As a library, this is problematic: consumers cannot raise the verbosity or silence these messages without affecting their entire application logging, because there is no dedicated logger name to target in their logging config.Why this matters
In a Django (or any framework)
LOGGINGconfig, you normally silence a noisy dependency with:This currently has no effect, because the records are created on the root logger, not on an
html4docx-namespaced logger. In production we get hundreds of lines like:for every HTML→DOCX conversion (these CSS properties are very common), and there is no clean way to filter them out short of attaching a custom
logging.Filteron a handler and matchingrecord.module == 'h4d'— which is fragile.Affected version
1.1.5Proposed fix
Define a module-level named logger and route every
logging.<level>(...)call through it.1. Near the top of
html4docx/h4d.py(after theimport loggingon line 2):This gives the logger the name
html4docx.h4d, so consumers can configurehtml4docx(parent) orhtml4docx.h4ddirectly.2. Replace the 14
logging.<level>(...)calls withlogger.<level>(...). Affected lines inh4d.py(v1.1.5):logging.warning(f"Warning: Unrecognized style '{style_name}', will be skipped.")logging.warning(f"Warning: Unrecognized paragraph style '{style_name}', will be skipped.")logging.warning(f"Warning: Could not parse font-size '{font_size}': {e}")logging.warning(f"Warning: Could not apply font-family '{font_family}': {e}")logging.warning(f"Could not apply color '{color_value}': {e}")logging.warning(f"Warning: Unsupported text transform '{text_transform}'")logging.warning(f"Warning: Could not apply text-transform '{text_transform}': {e}")logging.warning(f"Warning: Unsupported text decoration '{text_decoration_line}'")logging.warning(f"Warning: Style not recognized'{text_decoration_style}', defaulting to single line.")logging.warning(f"Warning: Unsupported background color '{background_color}'")logging.warning(f"Could not apply background-color to paragraph: {e}")logging.warning(f"Warning: Unsupported background color '{background_color}'")logging.warning(f"Could not apply background-color to run: {e}")logging.warning(f"Warning: Custom style '{custom_style}' not found in document, Ignoring style.")i.e.
logging.warning(...)→logger.warning(...). A single search-and-replace oflogging.warning(→logger.warning(withinh4d.pycovers all of them.Notes
argparse/__main__) can keep alogging.basicConfig(...)call so command-line users still see output.warningtodebug— they describe normal, non-actionable skips rather than problems.