As far as I can tell this will only search from the beginning of a string. Is there a way to search within a string as DDG does? In other words, what "trick" do we have to pull to do what DDG is able to do?

In response to the /face/stats/ URI, there is a typo in the humanized_time_difference function.
This prevents the response to show the uptime week and shows day instead.

And yeah, libface is amazing. @dhruvbird ⚡

Hello !
I do not know what license do you want for this project. Can you add a license to it ?
Cheers !

The input file might not always be mmap()able. For example, the case of streaming input or input from hdfs.

This might help reduce the running time. Might result in a constant factor improvement, but nothing asymptotically.

Consider using a flex based scanner for more maintainability.
The current scanner is hand-written, but allows us to use mmap(). With a flex based scanner, we need to investigate if we can track the file offset. Also benchmark the flex based scanner to see if it's fast enough (I think it should be).

Hello
It seems that one can guess search queries by listening to auto-complete session and record response lengths, even if the connection is encrypted.
I've made a POC:
https://github.com/ohadcn/bingo
a youtube video - shows the attack in action:
https://www.youtube.com/watch?v=XhbHFUHmvjo
and a presentation - explaining the attack flow (includes why DDG is easier to attack than other search engines):
https://prezi.com/pu5ilmqi-hyh/search-incognito/

#7 doesn't seem to be completely fixed. My query url is: "http://localhost:6767/face/suggest/?q=rules+to+old+maid" and it returns no results. If I manually encode the space instead to "%20" then I get results. e.g. "http://localhost:6767/face/suggest/?q=rules%20to%20old%20maid". I am running cpp-libface behind a reverse proxy in nginx. I noticed DDG passes the url as "+". Even a workaround would be nice; whether it be in nginx or cpp-libface.

Thanks!

The O(1) RMQ promises to be faster, with the same memory footprint.

TODO: Show that the space requirement is the same (or better).

I've been experimenting with lib-face. I imported a set of data that included terms containing spaces, for example:

14  API Design  api-design
3   API Security    api-security

Running queries for "api%20des" or "api+des" do not return any results:

http://localhost:6767/face/suggest/?n=10&q=api%20des
http://localhost:6767/face/suggest/?n=10&q=api+des

Is this intended behaviour? The example in the quick-start guide includes terms that have spaces in them: https://github.com/duckduckgo/cpp-libface/wiki/Quick-Start-Guide

hello, i am trying out this application to achieve autocompletion. But it gives me weird results
Here is my entry :

25 hey
21 how do you do?
15 how I met your mother
10 hello world
9 great Days ahead

and my query : http://localhost:6767/face/suggest/?q=h&n=3

and the result :

"[ { "phrase": "hey", "score": 25 },
{ "phrase": "how do you do?", "score": 21 },
{ "phrase": "hello world", "score": 10 }]

Why is it not, hey/ how do you do / how i met, which are the most frequent ones??
thank you

I get this error when using make

g++ -o lib-face src/main.cpp src/httpserver.o deps/libuv/libuv.a deps/http-parser/http_parser.o -I . -I deps -Wall -D_FILE_OFFSET_BITS=64 -O2 -lm -lrt -pthread
In file included from src/main.cpp:21:
./include/benderrmq.hpp: In member function 'uint_t LookupTables::query_max(uint_t, uint_t, uint_t)':
./include/benderrmq.hpp:250: warning: format '%d' expects type 'int', but argument 4 has type 'size_t'
./include/benderrmq.hpp:251: warning: format '%d' expects type 'int', but argument 4 has type 'size_t'
./include/benderrmq.hpp:252: warning: format '%d' expects type 'int', but argument 4 has type 'size_t'
./include/benderrmq.hpp: In member function 'void BenderRMQ::initialize(const vui_t&)':
./include/benderrmq.hpp:353: warning: format '%d' expects type 'int', but argument 3 has type 'size_t'
./include/benderrmq.hpp:353: warning: format '%d' expects type 'int', but argument 4 has type 'size_t'
./include/benderrmq.hpp:354: warning: format '%d' expects type 'int', but argument 3 has type 'size_t'
./include/benderrmq.hpp:354: warning: format '%d' expects type 'int', but argument 4 has type 'size_t'
src/main.cpp: In function 'int do_import(std::string, uint_t, int&, int&)':
src/main.cpp:516: warning: format '%llu' expects type 'long long unsigned int', but argument 3 has type 'off_t'
src/main.cpp: In function 'void handle_stats(client_t*, parsed_url_t&)':
src/main.cpp:715: warning: format '%d' expects type 'int', but argument 3 has type 'size_t'

any idea, http server loads but says page not found

The output should pass a simple JSON Lint test. We have our Content-Type set at "application/json; charset=UTF-8"

Here's my output:
[ { "phrase": "maid card game
", "score": 12 }
]

There are a couple things wrong:

1. There should be no '\n' in the results
2. There should also not be any unnecessary spaces (for instance a space between "[" and "{")
3. The output should pass a simple JSON Lint test. The output above results in an error from jsonlint.com of:

"Parse error on line 3:
...{ "phrase": "maid card game",
----------------------^
Expecting 'STRING', 'NUMBER', 'NULL', 'TRUE', 'FALSE', '{', '['"