for indexing, index empty/invalid length as '-' status code
for reading, serve as 204 no content.
ensure that StatusAndHeaders has a valid statusline when serving
if http content-length is valid,, limit stream to that content-length
as well as record content-length (whichever is smaller)
replace content-length when buffering
supports warc.gz, arc.gz, warc, arc and optional sorting
outputs cdx 11 but possible to extend to other formats
(additional edge case testing needed)
DecompressingBufferedReader refactoring to support multi-member gzip
Unit tests for indexer, addtional unit tests for bufferedreaders and loaders,
and recordloaders
memento support enabled by default, togglable via 'enable_memento' config property
supporting timegate and memento apis, no timemap yet
supporting pattern 2.3 for archival and pattern 1.3 for proxy modes
also:
simplify exception hierarchy a bit more, move down to utils
make WbRequest and WbResponse extensible with mixins (eg for memento)
custom processing ops, of which perms is a specific type
add lazy_ops test to ensure all cdx processing ops are lazy
perms: set up a 'perms policy' factory and perms policy implementation
perms policy setting results in a custom processing op
update tests to work with new config
IndexReader handles both cdx server + perms policy
create 'framework' subpackage for general purpose components!
contains routing, request/response, exceptions and wsgi wrappers
update framework package for pep8
dsrules: using load_config_yaml() (pushed to utils)
to init default config
archivalrouter: support empty collection, with and without SCRIPT_NAME
cdx: remove cdx source test, including access denied
replay: when content-type present, limit the decompressed stream to content-length
(this ensures last 4 bytes in warc/arc record are not read)
integration tests for identity replay
- head insert callback passed in with rule, up to template
to handle additional inserts based on rule properties
- ability to pass in custom rules config to both cdx server
and content rewriter
- move canonicalize to utils pkg
- add wombat, modify wb.js to remove wombat-related settings
contains configs for cdx canon, fuzzy matching and rewriting!
rewriting: ability to add custom regexs per domain
also, ability to toggle js rewriting and custom rewriting file
(default is wombat.js)
instead of strptime to automatically clamp timestamp to allowed
range (instead of erroring) on invalid timestamps.
returns datetime.datetime as advertised instead of struct_time as well
split binsearch further into binsearch and linearsearch components
reading blocks one at a time currently, due to zlib decompress limitations
fix bufferedreader.readline() and fileloader bugs
remove max_len from DecompressingBufferedReader as it applied to
the compressed size, not original size.
Add integration test for verifying content length of larger file
move to distinct packages: pywb.utils, pywb.cdx, pywb.warc, pywb.util, pywb.rewrite!
each package will have its own README and tests
shared sample_data and install
