A bug in the text dumper caused み to be replaced with a narrow
non-breaking space and the actual nbsp character in the unicode private
use area to go untranslated.
Only affected OR/AS.
Fix with a simple text replace on the affected entries.
import csv
in_ = open("../pokedex/data/csv/pokemon_species_flavor_text.csv", "r")
out = open("fixed.csv", "w")
r = csv.reader(in_, lineterminator='\n')
w = csv.writer(out, lineterminator='\n')
for row in r:
text = row[3]
if row[1] in {'25', '26'}:
text = text.replace("\u202f", "\u307f") # nbsp => mi
text = text.replace("\ue07f", "\u202f") # e07f => nbsp
w.writerow((row[0], row[1], row[2], text))
out.close()
Fixes#308
The initial import of data from S/M and US/UM only included held items for Moon
and Ultra Sun. (Presumably those were the only games eevee had access to?)
I now have all four games and I've verified that the data file which
contains held items is completely identical between Sun and Moon, and between
Ultra Sun and Ultra Moon.
So we can just copy the held items from the other two games.
pokedex=> insert into pokemon_items (select pokemon_id, 27, item_id, rarity from pokemon_items where version_id=28);
INSERT 0 356
Time: 220.560 ms
pokedex=> insert into pokemon_items (select pokemon_id, 30, item_id, rarity from pokemon_items where version_id=29);
INSERT 0 362
Time: 102.063 ms
- Add exception handling for SQLite, PostgreSQL, and MySQL.
Sqlalchemy gives different exceptions for each database engine so I have
seperated each exception for each engine.
The error message passed by each engine is also different so that is
checked as well.
- Once the function gets one of those exceptions the following message
is outputed and the program ends: `ERROR: Table 'TABLENAME' already
exists in the database. Did you mean to use 'pokedex load -D'`
- If the error is not that a table already exists, then the output will
just be the full error from sqlalchemy. That way if someone runs into
that error it can be reported as an issue and whoever can help would see
the full error.
---
Resolves part of issue #16
Filtering on pokemon name was broken. Not sure why, since the search CLI
was added way after the i18n stuff was added. The error is related to
AssociationProxy, which figures because nothing about association
proxies ever seems to work right. I don't know enough about SQLAlchemy
internals to know if what it was trying to do was supposed to
work, or how to fix it if so. So, fix it by using the same boring
join-based filtering that spline-pokedex uses.
Running `pokedex search --name=gloom` caused the following error with
PostgreSQL and SQLAlchemy 1.3.5:
sqlalchemy.exc.ProgrammingError: (psycopg2.ProgrammingError) can't adapt type 'ColumnAssociationProxyInstance'
...and this error with SQLite.
sqlalchemy.exc.InterfaceError: (sqlite3.InterfaceError) Error binding parameter 4 - probably unsupported type.
[parameters: (9, 9, 9, 9, ColumnAssociationProxyInstance(AssociationProxy('names_local', 'name')), 'gloom')]
Additionaly, the following error would happen with PostgreSQL and SQLAlchemy 0.9.7,
but that's probably unrelated:
sqlalchemy.exc.ProgrammingError: (ProgrammingError) subquery in FROM must have an alias
LINE 4: FROM pokemon_species, (SELECT pokemon_species_names.name AS ...
^
Fixes#296
Run git-diff after dumping the database to make sure that there are no
changes to the csv files.
We often get PRs from people who edit stuff by hand and don't know to do
a load/dump of the database afterwards. Usually this means they were
adding game data by hand, which we want to discourage, but even if it's
done for valid reasons it's good to canonicalize the csv files as soon
as possible to avoid interfering with future commits.
The help text for pokedex dump -l says
-l LANGS, --langs LANGS
comma-separated list of language codes to load,
'none', or 'all' (default: en)
However, the values 'none' and 'all' don't work (they are treated as
language names) and the default is to dump all languages, not just
English.
So `dump -l all` actually dumps no text at all because there is no
language called 'all'.
This is about to cause trouble for Travis CI because the next commit
will start checking that load/dump successfully round-trips... which
definitely isn't the case right now because the dump is missing
all text from translated tables.
The dump command and its documentation should definitely be fixed at
some point, but for now just stop using -l.
`sudo: false` used to be the way to opt in to their container-based
infrastructure. That system was retired in 2018 and so the `sudo` option
no longer does anything. Remove it.
Travi CI supports multiple OSes now. Add `os: linux` to be explicit
about which OS we want to test on.
Gen III contest effects were part of the initial data import from brown
veekun and so have probably never gotten a proper rip.
This commit re-rips contest effect stats, types, combos, and the mapping
between contest effects and moves.
None of the stats changed (jam & appeal); a few moves got assigned
different contest effects (specifically: smokescreen, flash, spite, and
shadow ball); and several combos changed.
For example, Hail was listed as comboing into the following moves
(this is also what Bulbapedia says):
Ice Beam, Blizzard, Aurora Beam, Haze, Powder Snow, Icy Wind,
Ice Ball, Weather Ball, Sheer Cold, Icicle Spear
However, the actual list is much shorter:
Blizzard, Powder Snow, Weather Ball
I verified on my Ruby cart that Hail->Ice Beam and Hail->Icy Wind are
not real combos, and that Hail->Blizzard is.
n.b. There are a bunch of unused contest effects in the game: out of 48
effects, 15 are not used by any move. The pokedex doesn't currently
include these. I elected not to add them, partly because i didn't want
to have to rip flavor text for them, but mostly because if i were going
to go that route i would probably reassign contest_effects.id to match
their internal ids, which would have made for uglier diffs.
Fixes#280
- link to nonexistant `[]{move:shield}` in Color Change (changed to protect)
- invalid syntax `[move:sandstorm]` in Primordial Sea
- invalid syntax `[move:sandstorm]` in Desolate Land
- invalid syntax `[move:sandstorm]` in Delta Stream
Found by test_markdown_values, which is not run by default because it is
slow. It can be run manually with
py.test pokedex/tests/test_strings.py --all
Pikachu's Cap forms and Cosplay forms had overlapping orders because i
forgot about Cosplay Pikachu when i set the form_order for the cap
forms.
??? Arceus had form_order=3 so it would go after all the other forms,
which had form_order=2. These were assigned distinct form_orders 2
through 17 at some point during a recent dump, but ??? Arceus was
missed.
Fixes#282
This commit adds Korean translations for egg groups, generations, languages, regions, stats, and version names. Also the ??? and shadow types.
It also fills in some missing foreign version names for S/M and US/UM (from Bulbapedia).
Roserade and Roselia were mistakenly sharing dex number 72. Presumably
an error from the original X/Y spreadsheet. I guess we never reripped
these after 3DS rom dumping became possible.
Regardless of how it happened, add a test and some database constraints
to ensure that it doesn't happen again.
- Add a test for gaps in pokedex numbers
- Add uniqueness constraints to pokemon_dex_numbers. A species can
only appear once per pokedex, and a number cannot be used more than
once per pokedex.
Cosplay Pikachu was listed as being encountered in contest-hall, but
that's a Sinnoh location. In OR/AS, the contest halls are not a
first-class location but rather just treated as part of the town or city
they are in. Cosplay Pikachu is given to the player after they
participate in their first contest, so its location can be any of the
four cities with a contest hall.
The Sinnoh starter that the player obtains in OR/AS is on Hoenn Route
101, not Sinnoh Route 201 (probably a copy/paste error).
Add a test to make sure that encounter regions always match the
region(s) that their game takes place in.
Add Japanese translations of the generation names, thanks to @SethETaron. I think this marks our first unofficial Japanese translation. 🎉
Also, fix some issues with Gen VII:
* generation names were missing for every language
* the identifier was incorrect — it was sun-moon instead of generation-vii
Fixes#250, #251, and #252.
Fossil pokémon don't fit with the rest of the gift pokemon: rather than
being given a pokémon directly, you get an item which can be exchanged
for a pokémon somewhere else ("revived"). To handle them properly we'd
have to add an item location table. I don't want to figure that out
right now, so just ignore the fossils for now.
Python 3 doesn't allow the ur'' prefix for raw unicode strings.
We want a unicode string here so that we get a unicode regexp,
so drop the r and do the escaping explicitly (the regexp in question
is pretty simple, fortunately).
Unova Routes 19-23 were added in B/W 2 and i forgot to prefix them when
we added the locations. Kalos Victory Road used to have a prefix but it
got dropped when i re-ripped X/Y locations. Kalos Pokemon League gets a
prefix too, since both Sinnoh and Alola also have Pokemon Leagues.
Add a test to ensure that we don't forget again in the future.
Also require location area identifiers to be unique to their location.
I almost added a duplicate pokemon-league location when adding S/M
locations. This should help.
The previous commit added a nullable subtitle field to location_names.
This caused a test in test_schema.py to fail because the name field
wasn't also nullable.
A comment above the test says, "If there's more than one text column in a
translation table they have to be nullable, to support missing
translations", but i don't think that logic holds in this case.
The idea is that we might have a translation for the subtitle, but not
the name, or vice versa, so both need to be nullable in case one or the
other is missing. But in this particular case that doesn't make sense:
if you don't have a name, you don't have a location; it may or may not
have a subtitle, but a location will always have a name.
Therefore, add an exception to the test.
Using the following steps:
* Unroll the previous commit with bulk CSV changes (b7500e92f0)
* remove and re-initialize the database
* follow the steps in that commit
* Only select location_area_prose in English
(there's a bunch of czech translations tracked separately in pokedex/data/csv/translations/cs.csv)
This change implies no functionality, but it was easiest to just
make it - sqlite insists on quoting spaces, though it's not needed.
Not rolling it back, since theoretically someone could run into the
same thing again in the future.
Andrew Ekstedt