summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* stricter argument parsingHEADmasterSathyanarayanan Gunasekaran2012-09-28
| | | | | | * check lengths of country, search, lookup parameters * make error_msg more modular * refactor
* make fingerprint, hashed_fingerprint case sensitiveSathyanarayanan Gunasekaran2012-09-28
| | | | | Now that we have a lookup column which is case insensitive, we can make fingerprint and hashed_fingerprint case sensitive.
* Use the same LIKE-based approach for lookup parameter.Karsten Loesing2012-09-28
| | | | | | | | | | | | There were two problems with requests containing the lookup parameter: - We currently mix AND and OR conditions in the SELECT statement without correctly parethesizing them. This could lead to surprising results when combining the lookup parameter with other parameters. - Lookup requests took 1.6 seconds on my local machine compared to 20 milliseconds for other requests. Using the same LIKE-based approach for the lookup parameter that we also used for the search parameter fixes both problems.
* Merge remote-tracking branch 'gsathya/bug_6954'Karsten Loesing2012-09-27
|\
| * fix bug in thread invocationSathyanarayanan Gunasekaran2012-09-27
| |
* | Merge remote-tracking branch 'gsathya/bug_6953'Karsten Loesing2012-09-27
|\ \
| * | Add get_timestamp()Sathyanarayanan Gunasekaran2012-09-27
| |/ | | | | | | | | | | get_timestamp() queries the database for the latest know published relay consensus and bridge network status document and returns a tuple containing these 2 values.
* | make lookup and country parameters case insensitiveSathyanarayanan Gunasekaran2012-09-27
| |
* | Make search column case insensitiveSathyanarayanan Gunasekaran2012-09-27
| |
* | Use "text" type in database schemaSathyanarayanan Gunasekaran2012-09-27
|/
* lookup fingerprint as wellSathyanarayanan Gunasekaran2012-09-26
|
* rename hex_fingerprint to hashed_fingerprintSathyanarayanan Gunasekaran2012-09-26
|
* Make lookup use hex_fingerprintSathyanarayanan Gunasekaran2012-09-26
|
* Fix searchSathyanarayanan Gunasekaran2012-09-26
| | | | | Exlude '$' if prefixed. 'filter' is a python keyword, lets use 'search_string' instead.
* Search hashed fingerprint alsoSathyanarayanan Gunasekaran2012-09-26
|
* Fix typo in error message.Karsten Loesing2012-09-25
|
* Merge remote-tracking branch 'gsathya/bug_6955'Karsten Loesing2012-09-25
|\
| * Raise proper exceptions, cleanup codeSathyanarayanan Gunasekaran2012-09-25
| |
* | Handle bridges without any flags in summary file.Karsten Loesing2012-09-24
|/ | | | | | | Bridges without any flags have two subsequent spaces in the summary file. Pass a value for sep when calling split(), so that "consecutive delimiters are not grouped together and are deemed to delimit empty strings" (Python documentation).
* Find a better hack for OFFSET without LIMIT.Karsten Loesing2012-09-24
| | | | | | | SQLite doesn't understand OFFSET without LIMIT, which is why the current code used a local counter to request TOTAL_ROWS-offset_value as the LIMIT. However, that code is not transaction-safe. A better solution is to pass -1 as the LIMIT, which has the same effect.
* Only search for beginning of nickname etc.Karsten Loesing2012-09-24
|
* Implement search filterSathyanarayanan Gunasekaran2012-09-23
|
* Remove addresses, flags table; Add get_router_tupleSathyanarayanan Gunasekaran2012-09-23
| | | | | | | | Use summary table to handle all the data. Add a search, flags, and addresses column to the summary table. get_router_tuple(fields) returns the values corresponding to the fields
* Make OFFSET work without a LIMITSathyanarayanan Gunasekaran2012-09-22
| | | | | | sqlite doesn't understand OFFSET without a LIMIT clause. Use TOTAL_ROWS-offset_value to find the LIMIT.
* Fix country parameterSathyanarayanan Gunasekaran2012-09-21
|
* Refactor database.pySathyanarayanan Gunasekaran2012-09-21
| | | | | | | | | | | | | | | | This is a pretty big commit with lots of refactoring: - Remove certain global vars, use config file - Remove freshen_database() - It's now a part of update_databases() - get_database() is now renamed to get_database_conn(), and removed the unused conn.row_factory variable - update_database() is now renamed to update_databases() - Add more logging info - Make modular insertion statements - get_summary_routers() returns relays/bridges which are now Router objects - Refactor code * Remove unnecessary return statements and whitespace * Make multiline statements more pythonic
* Merge remote-tracking branch 'gsathya/bug_6707'Karsten Loesing2012-09-17
|\
| * Add missing fields in /summary?type=fooSathyanarayanan Gunasekaran2012-09-16
| |
* | Remove cookie_secret.py and related codeSathyanarayanan Gunasekaran2012-09-16
|/
* Use pyonionoo.conf to specify location of summary file.Norman Danner2012-08-23
| | | | | | | | Probably most paths should be specified in pyonionoo.conf; this is a quick hack just to make the codebase marginally more portable. pyonionoo.conf must be edited with correct paths. Probably command-line options could be used as well, but I haven't looked into this.
* Replaced Python 2.7 string formatting with 2.6 C-style %-formatting.Norman Danner2012-08-13
| | | | Pyonionoo needs to run on Python 2.6.
* Changed the database update interval to 1 minute from 30 minutes.Norman Danner2012-08-10
| | | | In retrospect, a 30-minute interval was silly.
* Application is now non-blocking, using threads.Norman Danner2012-08-10
| | | | | | Each request is handled by a separate thread for querying the database and assembling the results. Also fixed the (incorrect) regular expressions that define the URLs we respond to.
* Deleted some testing files that were accidentally added to the repo.Norman Danner2012-08-10
|
* SQLite database updated automatically according to a specified interval.Norman Danner2012-08-10
| | | | | | | | We use a Timer object that updates the database every database.DB_UPDATE_INTERVAL seconds. As a Timer, this is not performed in the main thread, so doesn't block incoming connections. Happily, the database update is automatically a single transaction, so the update is not visible to clients until it is complete.
* SQLite database now created at application start, and updated when outNorman Danner2012-08-09
| | | | | | | | | of date. We have not tested whether transactions are doing the "right thing," allowing us to update the database while allowing other threads to continue reading the "not-yet-updated" database. And see comments at beginning of database.py related to using an in-memory database.
* Make the database connection use sqlite3.Row for the row factory.Norman Danner2012-08-09
| | | | | | | | | Having done so, we can now treat each row returned by a SELECT as a dictionary, with keys equal to the table field names. This simplifies the summary handler. Note that there are also a few comments related to implementing the 'search' parameter, which still hasn't been done yet.
* Forgot to delete ARGUMENTS in database.py.Norman Danner2012-08-09
|
* Externalized GET request parameter parsing.Norman Danner2012-08-09
| | | | | | | | | Added handlers/arguments.py, which has a single function parse. parse takes the GET request parameter dictionary as an argument, and returns a dictionary mapping strings to values. That dictionary can be used as the keyword arguments dictionary for the database module functions, so now the database module does not need to understand anything about GET requests.
* Deleted remnant global variable CURSOR in database.py.Norman Danner2012-08-09
|
* In database.py, do all filtering, ordering, etc. through the selectNorman Danner2012-08-09
| | | | | | | | | | statement. Made a few other minor refactorings, like having database() return the connection in order to eliminate the global CURSOR variable. Main tasks still to do: externalize request parameter handling to the handlers module; handle 'search' parameter; have another thread to create database; transactions on the database.
* Consensus weight ordering and additional tables of Summary database have ↵Megan2012-08-08
| | | | | | | | | | been implemented. 'flags' and 'addresses' tables have been added to the Summary database in addition to the 'summary' table. Each flag and address have their own row in their respective tables, but they have corresponding row ids with their respective router in the 'summary' table. User can now also order the requested routers by consensus_weight.
* Added sqlite database functionality:Megan2012-08-07
| | | | | | | | | | | Created a database to store the information of each router according to the different fields. One table (summary) contains all the fields except for 'flags' and 'addresses'. A database and the summary table is created using database(). get_summary_routers() takes in the arguments from SummaryHandler, reads the summary table, and appends the necessary routers to a tuple of lists that will be passed back to the SummaryHandler. 'flags' and 'addresses' will be in their own separate tables. The next step will be to add these two tables to the database.
* Created new branch database for Phase 2.Megan2012-08-03
|
* Phase 1 Complete:Megan2012-08-01
| | | | | | | Ensured SummaryHandler is returning JSON to client. get_router.py has been renamed to database.py and it reads the summary document upon each request. Next phase entails starting database work and storing the contents in an in-memory sqlite database.
* Fixed a few bugs.Megan2012-07-27
|
* Refactored code and moved filtering functionality to get_router.py.Megan2012-07-26
| | | | | | | | Code that goes through handler's arguments and filters the routers is now in get_router.py. As a result, the handlers will now focus solely on creating their respective documents. Timestamps for relays and bridges have been added, but more work needs to be done for the bandwidth handler, as well as some of the fields in the details handler.
* Moved summary.py to handlers.Megan2012-07-20
|
* Added DetailHandler, as well as added more functionality to pyonionoo.Megan2012-07-20
| | | | | | | | | Detail Handler has been added and is working properly along with the Summary handler. Users can now provide the following parameters for both handlers: type, running, search, lookup, country. The last three parameters have not been implemented yet: order, offset, limit. Bandwidth handler was started but is not finished yet. Note some of the relay information is incomplete in the detail handler.
* Reorganized & separated get router functionality from summary.pyMegan2012-07-18
| | | | | | Added new module get_router.py, and created a handlers package which will contain all handlers including summary, details, and bandwidth. Summary.py has been moved to this package.
generated by cgit v1.2.1 (git 2.18.0) at 2024-02-02 10:18:32 +0000