add

project-polymorph · Jan 22, 2025 · e194b4f · e194b4f
1 parent 7dbe8b7
commit e194b4f
Show file tree

Hide file tree

Showing 4 changed files with 6 additions and 1 deletion.
diff --git a/.gitmodules b/.gitmodules
@@ -1,3 +1,6 @@
 [submodule "webpage_archive"]
 	path = webpage_archive
 	url = https://github.com/transTerminus/webpage_archive
+[submodule "content_archive"]
+	path = content_archive
+	url = https://github.com/project-polymorph/content_archive
diff --git a/content_archive b/content_archive
diff --git a/download/Makefile b/download/Makefile
@@ -36,6 +36,7 @@ random:
 random_new:
 	python .github/downloader/download/download.py --output-dir .github/downloader/webpage_archive/new_all_results/20250121 --download-type webpage --pattern ".*" --order random
 	python .github/downloader/download/download.py --output-dir .github/downloader/webpage_archive/new_all_results/20250122 --download-type webpage --pattern ".*sohu.*|.*sina.*|.*ifeng.*|.*163.*|.*chinanews.*|.*thepaper.*|.*qq.*" --order random
+	python .github/downloader/download/download.py --output-dir .github/downloader/webpage_archive/new_all_results/20250123_res/downloads --download-type webpage --pattern ".*" --order random
 
 
 jina:

diff --git a/webpage_archive b/webpage_archive