dkpro-jwktl
dkpro-jwktl copied to clipboard
Pictures in Wiktionary cause incorrect parsing
Pictures on Wiktionary pages cause parsing errors
Pictures result in the page source having [[File: ...]]
or [[Image: ... ]]
at some location in the page source. The class WiktionaryEntryParser
has code for removing the latter, but not the former. Depending on the position of this markup in the page source (it ought to be located immediately beneath the language header eg ==English==
but sometimes is placed under a different header), this can cause issues. See examples below.
Unless the [[File: ...]]
markup is used elsewhere by the parser (I haven't done a thorough check), I'd propose amending WiktionaryEntryParser to also delete this markup.
Example 1
The page for word Trump has [[File: ...
located under the ===Pronunciation===
markup in the page source (see image below), and as a result the first IWiktionaryEntry
when parsing this page, let's call it entry
, returns null
upon calling entry.getPronunciations()
, instead of the actual pronunciations which are present.
Example 2
The page for word Obama has [[File: ...
located under the ===Etymology 1===
markup in the page source, and as a result the first IWiktionaryEntry
when parsing this page, let's call it entry
, returns [[File:President Barack Obama.jpg|thumb|Barack Obama ''(sense 2)'']]
upon calling `entry.getWordEtymology().getText()', instead of the actual etymology written on the page.