Spaces:
Running
Running
Update pipeline.py
Browse files- pipeline.py +9 -8
pipeline.py
CHANGED
@@ -464,17 +464,17 @@ def pipeline_with_gemini(accessions,stop_flag=None, niche_cases=None, save_df=No
|
|
464 |
links = unique_preserve_order(links)
|
465 |
acc_score["source"] = links
|
466 |
else:
|
467 |
-
print("
|
468 |
-
print("chunk filename: ", str(chunks_filename))
|
469 |
|
470 |
try:
|
471 |
temp_source = False
|
472 |
if save_df is not None and not save_df.empty:
|
473 |
print("save df not none")
|
474 |
-
print(str(chunk_filename))
|
475 |
-
print(str(all_filename))
|
476 |
-
if
|
477 |
-
link = save_df.loc[save_df["file_chunk"]==
|
478 |
#link = row["Sources"].iloc[0]
|
479 |
if "http" in link:
|
480 |
print("yeah http in save df source")
|
@@ -483,8 +483,8 @@ def pipeline_with_gemini(accessions,stop_flag=None, niche_cases=None, save_df=No
|
|
483 |
print("tempo source")
|
484 |
#acc_score["source"] = [str(all_filename), str(chunks_filename)]
|
485 |
temp_source = True
|
486 |
-
elif
|
487 |
-
link = save_df.loc[save_df["file_all_output"]==
|
488 |
#link = row["Sources"].iloc[0]
|
489 |
print(link)
|
490 |
print("list of link")
|
@@ -505,6 +505,7 @@ def pipeline_with_gemini(accessions,stop_flag=None, niche_cases=None, save_df=No
|
|
505 |
#acc_score["source"] = [str(file_all_path), str(file_chunk_path)]
|
506 |
temp_source = True
|
507 |
if temp_source:
|
|
|
508 |
if doi != "unknown":
|
509 |
link = 'https://doi.org/' + doi
|
510 |
# get the file to create listOfFile for each id
|
|
|
464 |
links = unique_preserve_order(links)
|
465 |
acc_score["source"] = links
|
466 |
else:
|
467 |
+
print("inside the try of reusing chunk or all output")
|
468 |
+
#print("chunk filename: ", str(chunks_filename))
|
469 |
|
470 |
try:
|
471 |
temp_source = False
|
472 |
if save_df is not None and not save_df.empty:
|
473 |
print("save df not none")
|
474 |
+
print("chunk file name: ",str(chunk_filename))
|
475 |
+
print("all filename: ",str(all_filename))
|
476 |
+
if acc_score["file_chunk"]:
|
477 |
+
link = save_df.loc[save_df["file_chunk"]==acc_score["file_chunk"],"Sources"].iloc[0]
|
478 |
#link = row["Sources"].iloc[0]
|
479 |
if "http" in link:
|
480 |
print("yeah http in save df source")
|
|
|
483 |
print("tempo source")
|
484 |
#acc_score["source"] = [str(all_filename), str(chunks_filename)]
|
485 |
temp_source = True
|
486 |
+
elif acc_score["file_all_output"]:
|
487 |
+
link = save_df.loc[save_df["file_all_output"]==acc_score["file_all_output"],"Sources"].iloc[0]
|
488 |
#link = row["Sources"].iloc[0]
|
489 |
print(link)
|
490 |
print("list of link")
|
|
|
505 |
#acc_score["source"] = [str(file_all_path), str(file_chunk_path)]
|
506 |
temp_source = True
|
507 |
if temp_source:
|
508 |
+
print("temp source is true so have to try again search link")
|
509 |
if doi != "unknown":
|
510 |
link = 'https://doi.org/' + doi
|
511 |
# get the file to create listOfFile for each id
|