René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"XML translation drops numbers from translation","articleBody":"When translating XML documents, I have problems with numbers being dropped in the translated result, especially in tables.\n\nInput:\n\n\t\t\t\t\t\t\t\u003crow\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003e22\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003eVentilations- och defrostermunstycken SB styrhytt\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\u003c/row\u003e\n\t\t\t\t\t\t\t\u003crow\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003e23\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003eLuft från intagsaggregatet till ventilationsmunstycken i styrhytt\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\u003c/row\u003e\n\t\t\t\t\t\t\t\u003crow\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003e1100\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\t\u003centry\u003e\n\t\t\t\t\t\t\t\t\t\u003cpara\u003eLuft från intagsaggregatet till ventilationsmunstycken i styrhytt\u003c/para\u003e\n\t\t\t\t\t\t\t\t\u003c/entry\u003e\n\t\t\t\t\t\t\t\u003c/row\u003e\n\n In the table below, you can see how numbers are being dropped from the translated content. This is serious as it is not always easy to spot.\n\n                           \u003crow\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003e22\u003c/para\u003e\n                                \u003c/entry\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003eVentilation and defroster nozzles SB control cabin\u003c/para\u003e\n                                \u003c/entry\u003e\n                            \u003c/row\u003e\n                            \u003crow\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003e2\u003c/para\u003e\n                                \u003c/entry\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003eAir from intake unit to ventilation nozzles in control cabin\u003c/para\u003e\n                                \u003c/entry\u003e\n                            \u003c/row\u003e\n\t\t\t    \u003crow\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003e11\u003c/para\u003e\n                                \u003c/entry\u003e\n                                \u003centry\u003e\n                                    \u003cpara\u003eAir from intake unit to ventilation nozzles in wheelhouse\u003c/para\u003e\n                                \u003c/entry\u003e\n                            \u003c/row\u003e\n\nThe code I am using:\n\n    result = deepl_client.translate_text(\n        text,\n        tag_handling=\"xml\",\n        source_lang=\"SV\",\n        target_lang=\"EN-GB\",\n        model_type=\"prefer_quality_optimized\",\n        non_splitting_tags=\"div\",\n        split_sentences=\"nonewlines\"\n    )\n\nI have a cumbersome workaround for this particular issue, but it's not safe. Pre-processing the content input file as a string with regular expressions that add a 'fake' element around particular numbers and then use the 'ignore_tags' option on this fake element.\n\nThis will however not find all all instances of the problem.\n\nIs there anything else that can be done?\n","author":{"url":"https://github.com/matsbert","@type":"Person","name":"matsbert"},"datePublished":"2025-09-01T13:14:03.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/145/deepl-python/issues/145"}