MediaWiki Dump
MediaWiki XML 傾印檔包含 wiki 的內容 (wiki 頁面及其所有修訂版本)，但不包含網站相關資料。 XML 傾印檔不會建立 wiki 資料庫的完整備份，傾印檔不包含使用者帳戶、圖片、編輯紀錄等。
本節說明如何將 MediaWiki XML 傾印檔載入為可供下游使用的文件格式。
它使用 mediawiki-utilities 中的 mwxml 進行傾印，並使用 earwig 中的 mwparserfromhell 來剖析 MediaWiki wikicode。
可以使用 dumpBackup.php 或在特殊頁面取得傾印檔案:統計資料的 Wiki。
# mediawiki-utilities supports XML schema 0.11 in unmerged branches
%pip install --upgrade --quiet git+https://github.com/mediawiki-utilities/python-mwtypes@updates_schema_0.11
# mediawiki-utilities mwxml has a bug, fix PR pending
%pip install --upgrade --quiet git+https://github.com/gdedrouas/python-mwxml@xml_format_0.11
%pip install --upgrade --quiet mwparserfromhell
from langchain_community.document_loaders import MWDumpLoader
API 參考：MWDumpLoader
loader = MWDumpLoader(
file_path="example_data/testmw_pages_current.xml",
encoding="utf8",
# namespaces = [0,2,3] Optional list to load only specific namespaces. Loads all namespaces by default.
skip_redirects=True, # will skip over pages that just redirect to other pages (or not if False)
stop_on_error=False, # will skip over pages that cause parsing errors (or not if False)
)
documents = loader.load()
print(f"You have {len(documents)} document(s) in your data ")
You have 177 document(s) in your data
documents[:5]
[Document(page_content='\t\n\t\n\tArtist\n\tReleased\n\tRecorded\n\tLength\n\tLabel\n\tProducer', metadata={'source': 'Album'}),
Document(page_content='{| class="article-table plainlinks" style="width:100%;"\n|- style="font-size:18px;"\n! style="padding:0px;" | Template documentation\n|-\n| Note: portions of the template sample may not be visible without values provided.\n|-\n| View or edit this documentation. (About template documentation)\n|-\n| Editors can experiment in this template\'s [ sandbox] and [ test case] pages.\n|}Category:Documentation templates', metadata={'source': 'Documentation'}),
Document(page_content='Description\nThis template is used to insert descriptions on template pages.\n\nSyntax\nAdd <noinclude></noinclude> at the end of the template page.\n\nAdd <noinclude></noinclude> to transclude an alternative page from the /doc subpage.\n\nUsage\n\nOn the Template page\nThis is the normal format when used:\n\nTEMPLATE CODE\n<includeonly>Any categories to be inserted into articles by the template</includeonly>\n<noinclude>{{Documentation}}</noinclude>\n\nIf your template is not a completed div or table, you may need to close the tags just before {{Documentation}} is inserted (within the noinclude tags).\n\nA line break right before {{Documentation}} can also be useful as it helps prevent the documentation template "running into" previous code.\n\nOn the documentation page\nThe documentation page is usually located on the /doc subpage for a template, but a different page can be specified with the first parameter of the template (see Syntax).\n\nNormally, you will want to write something like the following on the documentation page:\n\n==Description==\nThis template is used to do something.\n\n==Syntax==\nType <code>{{t|templatename}}</code> somewhere.\n\n==Samples==\n<code><nowiki>{{templatename|input}}</nowiki></code> \n\nresults in...\n\n{{templatename|input}}\n\n<includeonly>Any categories for the template itself</includeonly>\n<noinclude>[[Category:Template documentation]]</noinclude>\n\nUse any or all of the above description/syntax/sample output sections. You may also want to add "see also" or other sections.\n\nNote that the above example also uses the Template:T template.\n\nCategory:Documentation templatesCategory:Template documentation', metadata={'source': 'Documentation/doc'}),
Document(page_content='Description\nA template link with a variable number of parameters (0-20).\n\nSyntax\n \n\nSource\nImproved version not needing t/piece subtemplate developed on Templates wiki see the list of authors. Copied here via CC-By-SA 3.0 license.\n\nExample\n\nCategory:General wiki templates\nCategory:Template documentation', metadata={'source': 'T/doc'}),
Document(page_content='\t\n\t\t \n\t\n\t\t Aliases\n\t Relatives\n\t Affiliation\n Occupation\n \n Biographical information\n Marital status\n \tDate of birth\n Place of birth\n Date of death\n Place of death\n \n Physical description\n Species\n Gender\n Height\n Weight\n Eye color\n\t\n Appearances\n Portrayed by\n Appears in\n Debut\n ', metadata={'source': 'Character'})]