eg. /[from]*[to]/[url] or /[from]-[to]/[url], with both from and to optional, closes#130
exposes lower and upper bound timestamps in timeutils, pad_timestamp
data. This is necessary to override the `wsgi.input` to allow the post data to be read again via a fallback handler, even
after reading POST query data in replay handler, addresses #117
to avoid charset warning if head insert exceeds 1024 bytes (#86)
also encode head insert with detected charset, if possible
chunkeddatareader: add read() function to ensure read will read upto specified
length across chunks
for canonicalzation, treat urns as is, already canonical
for wburl, don't add http:// prefix if urn: prefix is present
add example-wpull warc for testing
showNumPages=true returns json indicating num pages, page=N can be set to page number 0-numPages - 1
loaders: add read_last_line() to read last line of a seekable file, used to read last line of index file when
at end
tests: additional test for binsearch boundary conditions
zipnum: secondary index output supports json also
option to index and merge instead of reindexing whole dir, #74
additional testing for recursive indexing, index merge
timeutils: add timestamp20_now() function
frames: add request_ts to wbinfo and use that as the timestamp in the top-frame. for exact replay, request_ts == timestamp
for latest replay / no timestamp / memento timegate, redirect to current time instead of time of last capture, while serving
last capture.
timeutils: add timestamp_now() function to return timestamp of current datetime
Add extra tests for this mode
Tracked via #72
toggleable via 'enable_cache' in config -- regular replayview does not
need any cache info
move add_range() components to statusandheaders from wbrequestresponse
add x-pywb-noredirect' header which disables date related redirect
video replay works w/o cache if supported by frontend (nginx)
defaults include no-cache and p3p policy (needed for IE default settings)
fix link generation for proxy_select page, better exception handling of ssl errors
record_loader can optionally parse 'request' records
archiveindexer has -a flag to write all records ('request' included),
-p flag to append post query
post-test.warc.gz and cdx
POST redirects using 307
* BufferedReader defaults to no decompression
* DecompressingBufferedReader defaults to gzip decomp
* ChunkedDataReader defaults to no gzip decomp, but decomp
can be set later via set_decomp().
This allow chunked responses to be de-chunked but not decompressed
(eg for non-text responses)
(due to more data being needed for gzip decomp), keep reading more blocks until there is data
or last block is reached (or error). Ensure a single read() call will return some data if there is any
