Game meta, data validation and schema

.gitignore

@@ -1,4 +1,3 @@
-/_build/
+_build/
+.env/
 *.pyc
-.env
-

.travis.yml

@@ -7,8 +7,7 @@ virtualenv:
 addons:
   ssh_known_hosts: sirius.solovyov.net
 install:
- - pip install cyrax
- - pip install pyyaml
+ - pip install -r requirements.txt
 script:
  - cyrax
  - test 6000 -lt $(cat _build/index.html | wc -l)

Makefile

@@ -5,6 +5,9 @@ CYRAX ?= cyrax
 build:
 	$(CYRAX)
 
+install:
+	pip install -r requirements.txt
+
 update:
 	git pull
 	$(CYRAX)

README.md

@@ -4,32 +4,26 @@ This is a source of [http://osgameclones.com](http://osgameclones.com). Feel
 free to submit pull requests to add new games and improve information about
 those already in the database.
 
-## How
+## Games database
 
-Check `games.yaml` out. All information is inside, and you should more or less
+Check [`games.yaml`][games] out. All information is inside, and you should more or less
 understand what's going on by reading it. Sorting is alphabetical, with an
 exception of ScummVM, just because it's so many games at once.
 
-Simplest way to contribute:
-[edit games.yaml](https://github.com/piranha/osgameclones/edit/master/games.yaml),
-and then your changes will be submitted as a pull request.
+## Add a game
 
-Please refer to [this template](https://github.com/piranha/osgameclones/blob/master/.github/ISSUE_TEMPLATE.md).
+Simplest way to contribute is to edit the [`games.yaml`][games] file. Your
+changes will be submitted as a pull request.
 
 - `name`/`names`: Name of the original game
   - If the game goes under multiple names, or if the clone is inspired by multiple related games, use `names`
   - A Wikipedia link is created for the name; if the article link is different, use the syntax `[Name, Name of Wikipedia article]`
 - `clones`/`reimplementations`: List the clones/reimplementations under this heading. Multiple clones can be listed.
-  - `name`: Name of the clone
-  - `url`: URL of clone main page
-  - `repo`: (optional) if the clone has an online code repository (e.g. GitHub), link here
-  - `info`: free text, but try to use terms already used in the list. Should include code details. HTML is supported
-  - `status`: (string) status of the project (active, halted)
-  - `license`: (string) project license
-  - `lang`: (string|list) programming languages used in the project
-  - `framework`: (string|list) programming frameworks and/or engines used in the project
-  - `added`: (optional) date when this clone was first added. Newly added clones are highlighted
-  - `media`: (optional) list of screenshots or videos for the clone
-    - `image`: URL of an image to display, preferrably a thumbnail
-    - `url`: (optional) link to a bigger image, when the user clicks the thumbnail
-    - `raw`: to embed raw HTML, e.g. to embed a video, use this tag followed by raw HTML
+
+Please refer to the [template][template] and the [`schema.yaml`][schema] file
+when adding new games.
+
+
+[games]: https://github.com/piranha/osgameclones/edit/master/games.yaml
+[schema]: https://github.com/piranha/osgameclones/edit/master/schema.yaml
+[template]: https://github.com/piranha/osgameclones/blob/master/.github/ISSUE_TEMPLATE.md

_ext.py

@@ -3,9 +3,11 @@
 import os.path as op
 from datetime import date, timedelta
 from collections import OrderedDict
+from functools import partial
 
 import yaml
 from cyrax import events
+from pykwalify.core import Core
 
 
 def abort(msg):
@@ -35,8 +37,7 @@ def parse_tag(tag):
     return tag.replace(' ', '-').lower()
 
 
-def parse_tags(entry):
-    fields = ['status', 'license', 'lang', 'framework']
+def parse_tags(entry, fields):
     tags = []
 
     for field in fields:
@@ -49,44 +50,71 @@ def parse_tags(entry):
     return tags
 
 
-def parse_global_tags(site, games, tag):
-    for game in games:
-        if tag in game:
-            if not getattr(site, tag, False):
-                setattr(site, tag, {})
+def parse_global_tags(site, item, tag):
+    if tag in item:
+        if not getattr(site, tag, False):
+            setattr(site, tag, {})
 
-            if isinstance(game[tag], basestring):
-                game[tag] = [game[tag]]
+        if isinstance(item[tag], basestring):
+            item[tag] = [item[tag]]
 
-            for t in game[tag]:
-                tagObj = getattr(site, tag, False)
-                if not tagObj.get(t, False):
-                    tagObj[t] = {'tag_count': 0}
-                tagObj[t]['tag_count'] += 1
+        for t in item[tag]:
+            tagObj = getattr(site, tag, False)
+            if not tagObj.get(t, False):
+                tagObj[t] = {'tag_count': 0}
+            tagObj[t]['tag_count'] += 1
 
     setattr(site, tag, OrderedDict(sorted(getattr(site, tag, {}).items())))
 
 
-def parse_item(entry):
+def parse_item(entry, entry_tags=[], meta={}, meta_tags=[]):
     added = entry.get('added') or date(1970, 1, 1)
     return dict(entry,
                 new=(date.today() - added) < timedelta(days=30),
-                tags=parse_tags(entry))
+                tags=parse_tags(entry, entry_tags) + parse_tags(meta, meta_tags))
 
 
 def parse_items(site, item, key):
     if key in item and validate(item, key):
         if not getattr(site, key, False):
             setattr(site, key, [])
 
-        parse_global_tags(site, item[key], 'lang')
-        getattr(site, key).append((names(item), map(parse_item, item[key])))
+        meta = item.get('meta', {})
+        meta_tags = ['genre', 'subgenre', 'theme']
+        game_tags = ['status', 'development', 'license', 'lang', 'framework']
+        parse_fn = partial(parse_item, entry_tags=game_tags, meta=meta, meta_tags=meta_tags)
+
+        for game in item[key]:
+            parse_global_tags(site, game, 'lang')
+        getattr(site, key).append((names(item), meta, map(parse_fn, item[key])))
+
+
+def show_validation_errors(data, errors):
+    print('\n')
+    for error in errors:
+        path = error.path.split('/')
+        game = data[int(path[1])]
+        name = game.get('name') or game.get('names')
+        print('\033[91m' + '  ' + str(name) + '\033[0m')
+        print('    ' + error.__repr__())
+    print('\n  ' + str(len(errors)) + ' errors\n')
+    sys.exit(1)
 
 
 def parse_data(site):
     data = yaml.load(file(op.join(op.dirname(__file__), 'games.yaml')))
 
+    try:
+        core = Core(source_data=data, schema_files=['schema.yaml'])
+        core.validate(raise_exception=True)
+    except Exception as error:
+        if len(core.errors) > 0:
+            show_validation_errors(data, core.errors)
+        else:
+            raise error
+
     for item in data:
+        parse_global_tags(site, item.get('meta', {}), 'genre')
         parse_items(site, item, 'clones')
         parse_items(site, item, 'reimplementations')