* CardDB: merge all card properties in a new structure
* Pre Json parser changes
* Cockatrice: use qt's builtin json support
* Move qt-json src dir from cockatrice to oracle
* Add dummy cockatricexml4 parser (yet to be implemented)
* Implement a new parser and xml format
* cockatricexml4: new xml parser following the "generic properties hash" pattern;
* oracleimporter: refactor the parsing code to better adapt to cockatricexml4; rewrote split cards parsing
* carddb: change "colors" from a stringlist to a string
* carddb: move the getMainCardType() method to the cockatricexml3 parser
*
* CardInfo: show all properties (stil missing: nice name + translation)
* Rework the "add related card" feature so that it doesn't change the card name in the carddb
Also, fix token count display
* Picture loader: Added support for transform cards
* Fix side information for flip cards
Mtgjson uses side a/b for flip cards, while scryfall doesn't
* Pictureloader: dynamic tag resolution from card properties
Examples old => new
* !cardid! => !set:muid!
* !uuid! => !set:uuid!
* !collectornumber! => !set:num!
New examples:
* !prop:type!
* !prop:manacost!
* Start moving mtg-related property names to a specific file
* Clangify
* Fix tests
* Make gcc an happy puppy
* Revert "Make gcc an happy puppy"
This reverts commit 446ec5f27516c4d3b32dbfc79557f4827c5c5bdf.
* Some gcc fixes
* Share set list between different db parsers, so they won't overwrite one each other
* All glory to the hypnoclangifier!
* Fix test compilation
* Cleanup edited files in the prior PR. (#3519)
* Cleanup edited files in the prior PR.
Signed-off-by: Zach Halpern <ZaHalpern+github@gmail.com>
* Fix includes
Signed-off-by: Zach Halpern <ZaHalpern+github@gmail.com>
* Update carddatabase.h
* improve loading from plain text
Fixes the loadFromStream_Plain function that is used to load plain text
decklists.
The rewritten function uses more regexes and is a bit cleaner.
This fixes multiple bugs with loading the various sources of decklists.
Note that the new function still has a few issues that are shared with
the original version like creating duplicate cards.
* clang format comments
apparently clang-format even complains about the spacing in your
comments
* refactor loading_from_clipboard tests
Remove all heap allocation and use references.
Use std::pair and std::string so gtest will show the cardnames in error messages.
(note that using QPair or QString does not work with gtest)
Improve the last two testcases to include weird names; and name and
comments testing.
Remove empty header file.
* fix compatibility with more formats
skip "sideboard" line
include everything in mainboard when there are multiple empty lines
add removal of the mwdeck cardversion selector in round braces
add replacal of lowercase ae combination that should never occur
Set cardname to lowercase as apparently our checks are hardcoded to only
accept lowercase.
* remove bugged test
The current load from plain is simply broken, removed checking the
comments for correct contents.
* rework load_from_clipboard tests again
rework the test to have less code duplication
add more tests and more special cases
note that text is still all lowercase
* improve loading from plain text
Fixes the loadFromStream_Plain function that is used to load plain text
decklists.
The rewritten function uses more regexes and is a bit cleaner.
This fixes multiple bugs with loading the various sources of decklists.
Note that the new function still has a few issues that are shared with
the original version like creating duplicate cards.
* clang format comments
apparently clang-format even complains about the spacing in your
comments
* refactor loading_from_clipboard tests
Remove all heap allocation and use references.
Use std::pair and std::string so gtest will show the cardnames in error messages.
(note that using QPair or QString does not work with gtest)
Improve the last two testcases to include weird names; and name and
comments testing.
Remove empty header file.
* fix compatibility with more formats
skip "sideboard" line
include everything in mainboard when there are multiple empty lines
add removal of the mwdeck cardversion selector in round braces
add replacal of lowercase ae combination that should never occur
Set cardname to lowercase as apparently our checks are hardcoded to only
accept lowercase.
* remove bugged test
The current load from plain is simply broken, removed checking the
comments for correct contents.
* rework load_from_clipboard tests again
rework the test to have less code duplication
add more tests and more special cases
note that text is still all lowercase
* remove forcing of lowercase cardnames
Cardnames in DeckList::loadFromStream_Plain will no longer be forced
lowercase if they aren't found in the database.
Empty lines in the comments of plaintext decklists will not be skipped.
The loading_from_clipboard_test gets its functions declared in a
separate header "clipboard_testing.h".
Add more edgecase tests.
Refactor code.
* add old QHash version support
QT 5.5 does not support using initializer lists for QHash.
Implement a preprocessor version check for conditionally using inserts
instead of a const with initializer list.
* add old QHash version support
QT 5.5 does not support using initializer lists for QHash.
Implement a preprocessor version check for conditionally using [] access
instead of a const with initializer list.
* add qHash on QRegularExpression below QT 5.6
Apparently QRegularExpression can't be hashed in lower QT versions,
so we add our own hash function, and everyone lived happily ever after,
and none the wiser.
* add header guards to clipboard_testing.h
* Added failing test to demonstrate issue with dec loading
* Prevents empty lines between comments from being interpreted as the start of the sideboard
* oracle now can be run in spoiler or normal mode
* tests for travis
* only run on relaunch
* spoilers in client (not oracle now) and tray icon shows when done
* spoiler status will be checked before downloading spoiler file
* only download if they care about spoilers
* reload db on spoiler download
* manual update button, code cleanup, and fix enabling sets when new
* cleanup, nullchecks, and fixes to spoiler
* reload DB even if not in spoiler season; necessary as we have a check elsewhere to prevent the reload if spoiler check happens
* Implement changes from 2991#issuecomment-356169374
* Change implicit nullptrs, alert on file deletion, minor changes
* make reload thread safe and minor changes from 2991#issuecomment-356450302
* Fix locking
* Disable update now button while process running
