- Only pass processor config arg required by configure_processor. Not
the unused full config object
- Type arguments passed to methods configure processors
- Import json for use by conversation processor to load logs
- Results get priority screen real estate
- Allows quick speed key based traversal of results as cursor
on switching to buffer is at top level heading
- E.g C-x o n n o 2 jumps to entry in actual file of second result
- Unlike before when it is at the #+STARTUP org buffer customization
settings
- Only allow adding files with appropriate file extension for each search type
- e.g .org for org-mode search, directory for image search
- Extract file paths added to config and enablement state of each search type
- This extracted state will be used to populate the khoj.yml config file
- Simplifies the configure screen layout and allows it to be of constant width
- It was buggy, the configure screen would dynamically expand but not
restore back to original size on disabling search type after enable
- Follow convention, two hyphens indicate variable private to library
- Defcustom are user configurable variables. So they should have single -
- Use khoj-results-count variable directly in code
- Make config_file an optional arg. It defaults to default khoj config dir
- Return args.config as None if no config_file explicitly passed by user
- Parent can use args.config = None as signal to trigger first run experience
- Fix regression since moving to use `which-key-show-full-keymap~
- The above function reads user keypress, so eats up 1 keypress
before starting to enter query
- No way to pass no-paging config via the external function to the
internally used which-key--show-keymap function that does allow
setting no-paging to not read user keypress
- So use the internal function instead and set no-paging arg to t
- The keybindings to select search types was previously confusing as
it only highlighted the final symbol to press (the C-x was shown but
it wasn't made apparent that it had to be pressed before)
- Previously some keybindings unrelated to khoj were also being shown
in the which-key popup. Now only the khoj keybindings are visible
- Search is being reconfigured multiple times in /regenerate and
n/reload. More appropriate name is configure_ rather than initialize_
for it
- Standardize name of methods under configure.py
- Main.py was becoming too big to manage. It had both
controllers/routers and component configurations (search, processors)
in it
- Now that the native app GUI code is also getting added to the main
path, good time to split/modularize/clean main.py
- Put global state into a separate file to share across modules
- Run FastAPI server in a separate thread.
- This allows starting both the server and gui in parallel
- Create System Tray for Khoj
- Contains menu items that open search or config pages in browser
- Rearrange code to have only the code required to start Backend and
GUI in the run() method
- Move the backend setup code into a separate method
- More generally, this allows configuring the khoj search anytime
while in khoj minibuffer window
- Earlier could only configure search type at the start of the search
- What
- Default to last used search type, when no search type specified
- Allow user to change search type before they enter query (and
after they've called khoj), if they want
- Why
- Reduce time from intent to results by using reasonable defaults
- Make interactions smoother, more intuitive
- Test invalid config file path throws. Remove redundant cli test
- Simplify cli parser code
- Do not need to explicitly check if args.config_file set.
argparser checks for positional arguments automatically
- Use standard semantics for cli args
- All positional args are required. Non positional args are optional
- Improve command line --help description
- Add custom validator to throw if neither input_filter or
input_<files|directories> are specified
- Set field expecting paths to type Path
- Now that default_config isn't used in code. We can update
fields in rawconfig to specify whether they're required or not.
This lets pydantic validate config file and throw appropriate error
- Reason
- Simplifies code. No merge_dict required
- 1 place for user to see all configurables, defaults and required values
- Details
- Remove default_config from code. Set defaults in khoj_sample.yml itself
- Keep fields required to be set by user as empty in khoj_sample to YAML
- Set defaults for fields not requiring configuration by user
- Do not want browsers to use the small, grainy favicons
- Firefox for Android does use the bigger icon, when it's the only one available
- Update svg to match the 144x144 ratio just for consistency
Currently only get into this state when debug breakpoints on backend
are keeping the connection open and user exits khoj search from Emacs
Results in a number of open connections that slow khoj down.
- Most concretely right now,
it eliminates the re-rank latency hit
on re-rank triggered on user hitting enter
after re-rank is already done on user idle
in the emacs interface
- Improves search latency of (incremental) search
- Makes it easier to fold/unfold, traverse and read results
- This 2 level nesting is already being used on the web interface
- Previously we were using the original nesting depth of the entry.
This was aimed at providing more of the orginal context of the
results. But currently this additional information does not provide
as much, for the decreased legibility of the results
- Improve code layout by ensuring all web interface specific code
under the src/interface/web directory
- Rename config API to more specifi /config instead of /ui
- Rename config data GET, POST api to /config/data instead of /config
- Previously we were statically populating types dropdown field in the web interface with all available search types
- This change populates the type dropdown field with only search types that are enabled/configured
- It queries the `/config` backend API to see which of the available search types are configured
- Populate via `.then` after enabled search types in dropdown are
populated
- Call to `/config` API is async and will usually complete after the value of type field is set from url
- So value of type field would earlier be overridden when search types
dropdown is populated after the call to `/config` API completes
- Get /config API and check config for which available search types is
populated. This gives us the list of enabled search types
- Dynamically populate search type field with enabled search types only
- Setting query value to default option when query param wasn't
passed via URL was overriding placeholder text in query field
- We wanted placeholder text in field, not the query field to actually
be populated by placeholder text
- This clears field when user starts typing query into the query field,
instead of them having to manually delete the default text populated
- Setting up default compressed-jsonl, embeddings-file was only required
for org search_type, while org-files and org-filter were allowed to be
passed as command line argument
- This avoided having to set compressed-jsonl and embeddings-file via
command line argument as well for org search type
- Now that all search types are only configurable via config file, We
can default all search types to None. The default config for the
rest of the search types wasn't being used anyway
- Previously org-files were configurable via cmdline args.
Where as none of the other search types are
- This is an artifact of how the application grew
- It can be removed for better consistency and
equal preference given all search types
Having org-mode result headings change size based on their depth in
the source document makes is a confusing UI experience.
Improve font-size, line-spacing and margins of results to make
delineation between entries, and differntiating between entry heading
and it's body easier to visually infer.
Do not white-space: pre-line. Improves rendering of Markdown results
## Support Incremental Search on Khoj Web Interface
- Use default, fast path to query /search API while user is typing
- Upgrade to cross-encoder re-ranked results once user hits enter on search box
## Improve Render of Org Results on Web Interface
- We were previously just wrapping results from /search API into a pre formatted div field. This was not easy to read
- Use [org.js](https://mooz.github.io/org-js/) to render results from Khoj `/search` API as proper HTML
- Improve org.js to render all task states, stylize task tags and make org-mode results look more like original content
Closes#42#41
In current state:
- Rerank results:
- If user idles while entering query OR
- exits normally
- Do not rerank results:
- If user exits abnormally, e.g via C-g from query
- Rename functions to more standard, descriptive names
- Keep known, required code for incremental search
- E.g Do not set buffer local flag in hooks on minibuffer setup
- Only query when user in khoj minibuffer
- Use active-minibuffer-window and track khoj minibuffer
- (minibuffer-prompt) is not useful for our use-case here
- (For now) Run re-rank only if user idle while querying
- Do not run rerank on teardown/completion
- The reranking lag (~2s) is annoying; hit enter,
wait to see results
- Also triggered when user exits abnormally,
so C-g also results in rerank which is even more annoying
- Emacs will still hang if re-ranking gets triggered on idle but
that's better than always getting triggered. And better than not
having mechanism to get results re-ranked via cross-encoder at all
- Update khoj-simple to work cross-encoder re-ranked results like before
- Increment major version as incremental search considered a breaking
change and a major update to search capability
- Improve search speed by ~10x
Tested on corpus of 125K lines, 12.5K entries
- Allow cross-encoder to re-rank results by settings &?r=true when querying /search API
- It's an optional param that default to False
- Earlier all results were re-ranked by cross-encoder
- Making this configurable allows for much faster results, if desired
but for lower accuracy
- Formalize filters into class with can_filter() and filter() methods
- Use can_filter() method to decide whether to apply filter and
create deep copies of entries and embeddings for it
- Improve search speed for queries with no filters
as deep copying entries, embeddings takes the most time
after cross-encodes scoring when calling the /search API
Earlier we would create deep copies of entries, embeddings
even if the query did not contain any filter keywords
- Reason:
Allow natural search on markdown based notes, documentation,
websites etc
- Details:
- Create markdown processor to extract Markdown entries (identified by
Heading) into standard jsonl format required by text_search
- Update API, Configs to support interfacing with new markdown type
- Update Emacs, Web clients to support interfacing with new markdown
type via API
- Update Readme to mentiond markdown is also supported
Closes#35
- The code for both the text search types were mostly the same
It was earlier done this way for expedience while experimenting
- The minor differences were reconciled and merged into a single
text_search type
- This simplifies the app and making it easier to process other
text types
Now that the logic to compile entries is in the processor layer, the
extract_entries method is standard across (text) search_types
Extract the load_jsonl method as a utility helper method.
Use it in (a)symmetric search types
- The logic for compiling a beancount entry (for later encoding) now
completely resides in the org-to-jsonl processor layer
- This allows symmetric search to be generic and not be aware of
beancount specific properties that were extracted by the
beancount-to-jsonl processor layer
- Now symmetric search just expects the jsonl to (at least) have the
'compiled' and 'raw' keys for each entry. What original text the
entry was compiled from is irrelevant to it. The original text
could be location, transaction, chat etc, it doesn't have to care
- The logic for compiling an org-mode entry (for later encoding) now
completely resides in the org-to-jsonl processor layer
- This allows asymmetric search to be generic and not be aware of
org-mode specific properties that were extracted by the org-to-jsonl
processor layer
- Now asymmetric search just expects the jsonl to (at least) have the
'compiled' and 'raw' keys for each entry. What original text the
entry was compiled from is irrelevant to it. The original text
could be mail, chat, markdown, org-mode etc, it doesn't have to care
- Pass Scheduled, Closed Dates of Entries to Include in Embeddings
- The (new?) model seems to understand dates. So can give more
relevant entries if date in natural language mentioned in query
- E.g "Went Surfing with Friends" vs "Went Surfing with Friends in 1984"
will give different results, with the second prioritizing entries
mentioning any entries with closed, scheduled dates from 1984
- While it's true those strings are going to be used to generated
embeddings, the more generic term allows them to be used elsewhere as
well
- Their main property is that they are processed, compiled for
usage by semantic search
- Unlike the 'raw' string which contains the external representation
of the data, as is
- Had already made some progress on this earlier by updating the image
search responses. But needed to update the text search responses to
use lowercase entry and score
- Update khoj.el to consume the updated json response keys for text
search
- Image order returned by glob is OS dependent
- This prevented sharing image embeddings across machines running different OS
- A stable sort order for processed images allows sharing embeddings
across machines.
- Use case:
A more powerful, always on machine actually computes the image embeddings regularly
The client machine just load these periodically to provide semantic search functionality
- Handle case where current image batch smaller than batch_size
- Handle case where no XMP metadata for current image
- return empty strings in such a scenario instead of ". "
Issue:
- Had different schema of extracted entries for symmetric_ledger vs asymmetric
- Entry extraction for asymmetric was dirty, relying on cryptic
indices to store raw entry vs cleaned entry meant to be passed to embeddings
- This was pushing the load of figuring out what property to extract
from each entry to downstream processes like the filters
- This limited the filters to only work for asymmetric search, not for
symmetric_ledger
- Fix
- Use consistent format for extracted entries
{
'embed': entry_string_meant_to_be_passed_to_model_and_get_embeddings,
'raw' : raw_entry_string_meant_to_be_passed_to_use
}
- Result
- Now filters can be applied across search types, and the specific
field they should be applied on can be configured by each search
type
- The all-MiniLM-L6-v2 is more accurate
- The exact previous model isn't benchmarked but based on the
performance of the closest model to it. Seems like the new model
maybe similar in speed and size
- On very preliminary evaluation of the model, the new model seems
faster, with pretty decent results
- The multi-qa-MiniLM-L6-cos-v1 is more extensively benchmarked[1]
- It has the right mix of model query speed, size and performance on benchmarks
- On hugging face it has way more downloads and likes than the msmarco model[2]
- On very preliminary evaluation of the model
- It doubles the encoding speed of all entries (down from ~8min to 4mins)
- It gave more entries that stay relevant to the query (3/5 vs 1/5 earlier)
[1]: https://www.sbert.net/docs/pretrained_models.html
[2]: https://huggingface.co/sentence-transformers
- Avoids having to click the query input box
- Just open page, type whatever and hit enter to do image search
- For other search types select appropriate type from dropdown
- Use shr to render image response from html in result buffer
Earlier was using org-mode. But rendering HTML with shr seems cleaner
- Use Headings to Add highlights
- Use Random to Force fetch of Image. Similar to what was done for Web interface
- Remove trailing elisp brackets from response
- Show query match scores by image model for each image in results
- Metadata match score were consistently giving higher scores by a
factor of ~3x wrt to image match score. This was resulting in all
results being from the metadata match with query and none from the
image match with query.
- Scaling the metadata match scores down by scaling factor seems to
give more consistently give a blend of results from both image and
metadata matches
