Chemical component cache poisoning #703

Activity

1. sbliven commented on Sep 4, 2017

   

   sbliven

   on Sep 4, 2017

   Member

   More actions

   A simple check would be to re-download files if they are too small.

   The smallest PDB file I could get BioJava to read is 79 bytes, which compresses to 47B with gzip.

   ATOM      1  K     K K   1           1       1       1     1     1           K
   

   The smallest CIF file I made is 384B, compressed 176B

   data_
   _atom_site.group_PDB ATOM
   _atom_site.id 1
   _atom_site.type_symbol K
   _atom_site.label_atom_id K
   _atom_site.label_alt_id .
   _atom_site.label_comp_id K
   _atom_site.label_asym_id K
   _atom_site.pdbx_PDB_ins_code ?
   _atom_site.Cartn_x 1
   _atom_site.Cartn_y 1
   _atom_site.Cartn_z 1
   _atom_site.occupancy 1
   _atom_site.B_iso_or_equiv 1
   _atom_site.auth_seq_id 1
   _atom_site.pdbx_PDB_model_num 1
   

   An empty file compresses to 20B (just the header). Maybe set the threshold to about 40B to be safe?
2. sbliven

   self-assigned this

   on Sep 5, 2017
3. josemduarte commented on Sep 5, 2017

   

   josemduarte

   on Sep 5, 2017

   Last edited by josemduarte

   ContributorAuthor

   More actions

   It seems that there's been a redirect in place at RCSB that did:

   http://rcsb.org/pdb/files/ligand/HEM.cif  ----> https://files.org/ligands/HEM.cif
   

   So it looks like biojava wasn't handling URL redirects correctly for this case. That'd be another fix we need to correct the caching problems.
4. sbliven commented on Sep 5, 2017

   

   sbliven

   on Sep 5, 2017

   Member

   More actions

   @josemduarte Sounds like an opportunity to move FileDownloadUtils to nio.
5. josemduarte commented on Sep 5, 2017

   

   josemduarte

   on Sep 5, 2017

   ContributorAuthor

   More actions

   It seems that after all the redirect thing is not a bug but a feature. See https://stackoverflow.com/a/1884427/3914327

   We will change the redirects at RCSB not to cross protocols for maximum compatibility.
6. josemduarte commented on Sep 7, 2017

   

   josemduarte

   on Sep 7, 2017

   ContributorAuthor

   More actions

   Since Wednesday the redirect from rcsb.org is from http to http. That should fix problems with any version before 4.2.8.
7. josemduarte

   added a commit that references this issue on Sep 8, 2017

   Solving biojava#704 and adding some tests for biojava#703

   3123589
8. sbliven

   added a commit that references this issue on Jan 15, 2018

   Re-download empty structure or chemical component structures

   ...

   3f1232f
9. lafita

   added

   bugBugs and bugfixesBugs and bugfixes

   on Mar 11, 2018
10. lafita

    added this to the BioJava 5.0.1This milestone has been deleted milestone on Mar 11, 2018
11. sbliven

    added 3 commits that reference this issue on Jun 8, 2018

    Add test for biojava#703

    4785d57

    Make tests for biojava#703 fail with an assertion

    ...

    b207d34

    Re-download empty structure or chemical component structures

    ...

    6ea9ae7
12. sbliven

    added a commit that references this issue on Jun 18, 2018

    Merge pull request #774 from sbliven/fix703

    ...

    This issue will close once commit 5a4a68c is merged into the 'master' branch.

    5a4a68c
13. sbliven

    closed this as completedin 5a4a68con Jan 2, 2019
14. aalhossary

    added a commit that references this issue on Jun 16, 2021

    Fix biojava#933 URLConnection to invalid HTTP URL is not detected as …

    ...

    0a66b4e
15. jnbdz

    added a commit that references this issue on Aug 2, 2026

    Fix biojava#933 URLConnection to invalid HTTP URL is not detected as …

    ...

    df586e8