victormiller
commited on
Commit
•
beddb3a
1
Parent(s):
89ba804
Update curated.py
Browse files- curated.py +31 -4
curated.py
CHANGED
@@ -485,7 +485,34 @@ wiki_examples = Div(
|
|
485 |
),
|
486 |
)
|
487 |
|
|
|
|
|
488 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
489 |
|
490 |
filtering_process = Div(
|
491 |
Section(
|
@@ -695,10 +722,10 @@ filtering_process = Div(
|
|
695 |
Li("Local dedup was done within freelaw itself which removed 90%+ duplicates"),
|
696 |
),
|
697 |
table_div_freelaw,
|
698 |
-
|
699 |
-
|
700 |
-
|
701 |
-
|
702 |
|
703 |
),
|
704 |
),
|
|
|
485 |
),
|
486 |
)
|
487 |
|
488 |
+
def get_freelaw_data(data_source: str = "Freelaw", doc_id: int = 3, target: str = "foo"):
|
489 |
+
doc_id = max(0, min(int(doc_id), 9))
|
490 |
|
491 |
+
if data_source == "Freelaw":
|
492 |
+
raw_sample_doc = json.load(open("data/curated_samples/freelaw_raw.json"))
|
493 |
+
extracted_sample_doc = json.load(
|
494 |
+
open("data/curated_samples/freelaw_extract.json")
|
495 |
+
)
|
496 |
+
else:
|
497 |
+
raw_sample_doc = extracted_sample_doc = [{} for _ in range(10)]
|
498 |
+
|
499 |
+
raw_json = raw_sample_doc[doc_id]
|
500 |
+
extracted_json = extracted_sample_doc[doc_id]
|
501 |
+
return view_data(
|
502 |
+
raw_json,
|
503 |
+
extracted_json,
|
504 |
+
doc_id=doc_id,
|
505 |
+
data_source="Freelaw",
|
506 |
+
data_sources="Freelaw",
|
507 |
+
target=target,
|
508 |
+
)
|
509 |
+
|
510 |
+
freelaw_examples = Div(
|
511 |
+
Div(
|
512 |
+
get_freelaw_data(target=gen_random_id()),
|
513 |
+
style="border: 1px solid #ccc; padding: 20px;",
|
514 |
+
),
|
515 |
+
)
|
516 |
|
517 |
filtering_process = Div(
|
518 |
Section(
|
|
|
722 |
Li("Local dedup was done within freelaw itself which removed 90%+ duplicates"),
|
723 |
),
|
724 |
table_div_freelaw,
|
725 |
+
Details(
|
726 |
+
Summary("FreeLaw Filtering Examples"),
|
727 |
+
freelaw_examples,
|
728 |
+
),
|
729 |
|
730 |
),
|
731 |
),
|