balkian/senpy
1
0
Fork 0
mirror of https://github.com/gsi-upm/senpy synced 2025-08-23 18:12:20 +00:00
Code Issues Releases Wiki Activity

New schema for parameters

Browse Source 
* Improve extra requirement handling
* New mechanism to handle parameters beforehand in chained
  calls, and the ability to get help on available parameters in chained
  calls (through `?help`).
* Redefined Analysis, to reflect the new ontology
* Add parameters as an entity in the schema
* Update examples to include analyses and parameters
* Add processing plugins, with an interface similar to analysis plugins
* Update tests
* Avoid duplication in split plugin

Closes #51

Squashed commit of the following:

commit d145a852e7
commit 6a1069780b
commit ca69bddc17
commit aa35e62a27
This commit is contained in:
J. Fernando Sánchez
2018-12-07 18:30:05 +01:00
parent 41aa142ce0
commit 4ba30304a4
30 changed files with 717 additions and 353 deletions
Download Patch File Download Diff File Expand all files Collapse all files

158
tests/test_api.py
View File

@@ -3,8 +3,9 @@ import logging
logger = logging.getLogger(__name__)
from unittest import TestCase
from senpy.api import parse_params, API_PARAMS, NIF_PARAMS, WEB_PARAMS
from senpy.models import Error
from senpy.api import (boolean, parse_params, get_extra_params, parse_analysis,
API_PARAMS, NIF_PARAMS, WEB_PARAMS)
from senpy.models import Error, Plugin
class APITest(TestCase):
@@ -89,3 +90,156 @@ class APITest(TestCase):
assert "Dummy" in p['algorithm']
assert 'input' in p
assert p['input'] == 'Aloha my friend'
def test_parse_analysis(self):
'''The API should parse user parameters and return them in a format that plugins can use'''
plugins = [
Plugin({
'name': 'plugin1',
'extra_params': {
# Incompatible parameter
'param0': {
'aliases': ['p1', 'parameter1'],
'options': ['option1', 'option2'],
'default': 'option1',
'required': True
},
'param1': {
'aliases': ['p1', 'parameter1'],
'options': ['en', 'es'],
'default': 'en',
'required': False
},
'param2': {
'aliases': ['p2', 'parameter2'],
'required': False,
'options': ['value2_1', 'value2_2', 'value3_3']
}
}
}), Plugin({
'name': 'plugin2',
'extra_params': {
'param0': {
'aliases': ['parameter1'],
'options': ['new option', 'new option2'],
'default': 'new option',
'required': False
},
'param1': {
'aliases': ['myparam1', 'p1'],
'options': ['en', 'de', 'auto'],
'default': 'de',
'required': True
},
'param3': {
'aliases': ['p3', 'parameter3'],
'options': boolean,
'default': True
}
}
})
]
call = {
'param1': 'en',
'0.param0': 'option1',
'0.param1': 'en',
'param2': 'value2_1',
'param0': 'new option',
'1.param1': 'de',
'param3': False,
}
expected = [
{
'param0': 'option1',
'param1': 'en',
'param2': 'value2_1',
}, {
'param0': 'new option',
'param1': 'de',
'param3': False,
}
]
p = parse_analysis(call, plugins)
for i, arg in enumerate(expected):
params = p[i].params
for k, v in arg.items():
assert params[k] == v
def test_get_extra_params(self):
'''The API should return the list of valid parameters for a set of plugins'''
plugins = [
Plugin({
'name': 'plugin1',
'extra_params': {
# Incompatible parameter
'param0': {
'aliases': ['p1', 'parameter1'],
'options': ['option1', 'option2'],
'default': 'option1',
'required': True
},
'param1': {
'aliases': ['p1', 'parameter1'],
'options': ['en', 'es'],
'default': 'en',
'required': False
},
'param2': {
'aliases': ['p2', 'parameter2'],
'required': False,
'options': ['value2_1', 'value2_2', 'value3_3']
}
}
}), Plugin({
'name': 'plugin2',
'extra_params': {
'param0': {
'aliases': ['parameter1'],
'options': ['new option', 'new option2'],
'default': 'new option',
'required': False
},
'param1': {
'aliases': ['myparam1', 'p1'],
'options': ['en', 'de', 'auto'],
'default': 'de',
'required': True
},
'param3': {
'aliases': ['p3', 'parameter3'],
'options': boolean,
'default': True
}
}
})
]
expected = {
# Overlapping parameters
'plugin1.param0': plugins[0]['extra_params']['param0'],
'plugin1.param1': plugins[0]['extra_params']['param1'],
'plugin2.param0': plugins[1]['extra_params']['param0'],
'plugin2.param1': plugins[1]['extra_params']['param1'],
# Non-overlapping parameters
'param2': plugins[0]['extra_params']['param2'],
'param3': plugins[1]['extra_params']['param3'],
# Intersection of overlapping parameters
'param1': {
'aliases': ['p1'],
'options': ['en'],
'default': None,
'required': True
}
}
result = get_extra_params(plugins)
for ik, iv in expected.items():
assert ik in result
for jk, jv in iv.items():
assert jk in result[ik]
assert expected[ik][jk] == result[ik][jk]

69
tests/test_blueprints.py
View File

@@ -26,8 +26,7 @@ class BlueprintsTest(TestCase):
cls.senpy.init_app(cls.app)
cls.dir = os.path.join(os.path.dirname(__file__), "..")
cls.senpy.add_folder(cls.dir)
cls.senpy.activate_plugin("Dummy", sync=True)
cls.senpy.activate_plugin("DummyRequired", sync=True)
cls.senpy.activate_all()
cls.senpy.default_plugin = 'Dummy'
def setUp(self):
@@ -107,6 +106,7 @@ class BlueprintsTest(TestCase):
assert isinstance(js, models.Error)
resp = self.client.get("/api/?i=My aloha mohame&algo=DummyRequired&example=notvalid")
self.assertCode(resp, 400)
self.app.config['TESTING'] = True
resp = self.client.get("/api/?i=My aloha mohame&algo=DummyRequired&example=a")
self.assertCode(resp, 200)
@@ -138,16 +138,77 @@ class BlueprintsTest(TestCase):
# Calling dummy twice, should return the same string
self.assertCode(resp, 200)
js = parse_resp(resp)
assert len(js['analysis']) == 1
assert len(js['analysis']) == 2
assert js['entries'][0]['nif:isString'] == 'My aloha mohame'
resp = self.client.get("/api/Dummy+Dummy?i=My aloha mohame")
# Same with pluses instead of slashes
self.assertCode(resp, 200)
js = parse_resp(resp)
assert len(js['analysis']) == 1
assert len(js['analysis']) == 2
assert js['entries'][0]['nif:isString'] == 'My aloha mohame'
def test_analysis_chain_required(self):
"""
If a parameter is required and duplicated (because two plugins require it), specifying
it once should suffice
"""
resp = self.client.get("/api/DummyRequired/DummyRequired?i=My aloha mohame&example=a")
js = parse_resp(resp)
assert len(js['analysis']) == 2
assert js['entries'][0]['nif:isString'] == 'My aloha mohame'
assert js['entries'][0]['reversed'] == 2
def test_requirements_chain_help(self):
'''The extra parameters of each plugin should be merged if they are in a chain '''
resp = self.client.get("/api/split/DummyRequired?help=true")
self.assertCode(resp, 200)
js = parse_resp(resp)
assert 'valid_parameters' in js
vp = js['valid_parameters']
assert 'example' in vp
assert 'delimiter' in vp
def test_requirements_chain_repeat_help(self):
'''
If a plugin appears several times in a chain, there should be a way to set different
parameters for each.
'''
resp = self.client.get("/api/split/split?help=true")
self.assertCode(resp, 200)
js = parse_resp(resp)
assert 'valid_parameters' in js
vp = js['valid_parameters']
assert 'delimiter' in vp
resp = self.client.get("/api/split/split?help=true&verbose=false")
js = parse_resp(resp)
vp = js['valid_parameters']
assert len(vp.keys()) == 1
def test_requirements_chain(self):
"""
It should be possible to specify different parameters for each step in the chain.
"""
# First, we split by sentence twice. Each call should generate 3 additional entries
# (one per sentence in the original).
resp = self.client.get('/api/split/split?i=The first sentence. The second sentence.'
'\nA new paragraph&delimiter=sentence')
js = parse_resp(resp)
assert len(js['analysis']) == 2
assert len(js['entries']) == 7
# Now, we split by sentence. This produces 3 additional entries.
# Then, we split by paragraph. This should create 2 additional entries (One per paragraph
# in the original text)
resp = self.client.get('/api/split/split?i=The first sentence. The second sentence.'
'\nA new paragraph&0.delimiter=sentence&1.delimiter=paragraph')
# Calling dummy twice, should return the same string
self.assertCode(resp, 200)
js = parse_resp(resp)
assert len(js['analysis']) == 2
assert len(js['entries']) == 6
def test_error(self):
"""
The dummy plugin returns an empty response,\

48
tests/test_extensions.py
View File

@@ -3,10 +3,6 @@ import os
from copy import deepcopy
import logging
try:
from unittest import mock
except ImportError:
import mock
from functools import partial
from senpy.extensions import Senpy
@@ -18,7 +14,8 @@ from unittest import TestCase
def analyse(instance, **kwargs):
request = api.parse_call(kwargs)
basic = api.parse_params(kwargs, api.API_PARAMS)
request = api.parse_call(basic)
return instance.analyse(request)
@@ -49,9 +46,9 @@ class ExtensionsTest(TestCase):
'''Should be able to add and delete new plugins. '''
new = plugins.Analysis(name='new', description='new', version=0)
self.senpy.add_plugin(new)
assert new in self.senpy.plugins()
assert new in self.senpy.plugins(is_activated=False)
self.senpy.delete_plugin(new)
assert new not in self.senpy.plugins()
assert new not in self.senpy.plugins(is_activated=False)
def test_adding_folder(self):
""" It should be possible for senpy to look for plugins in more folders. """
@@ -60,7 +57,7 @@ class ExtensionsTest(TestCase):
default_plugins=False)
assert not senpy.analysis_plugins
senpy.add_folder(self.examples_dir)
assert senpy.analysis_plugins
assert senpy.plugins(plugin_type=plugins.AnalysisPlugin, is_activated=False)
self.assertRaises(AttributeError, senpy.add_folder, 'DOES NOT EXIST')
def test_installing(self):
@@ -121,8 +118,8 @@ class ExtensionsTest(TestCase):
# Leaf (defaultdict with __setattr__ and __getattr__.
r1 = analyse(self.senpy, algorithm="Dummy", input="tupni", output="tuptuo")
r2 = analyse(self.senpy, input="tupni", output="tuptuo")
assert r1.analysis[0].id == "endpoint:plugins/Dummy_0.1"
assert r2.analysis[0].id == "endpoint:plugins/Dummy_0.1"
assert r1.analysis[0].algorithm == "endpoint:plugins/Dummy_0.1"
assert r2.analysis[0].algorithm == "endpoint:plugins/Dummy_0.1"
assert r1.entries[0]['nif:isString'] == 'input'
def test_analyse_empty(self):
@@ -130,7 +127,7 @@ class ExtensionsTest(TestCase):
senpy = Senpy(plugin_folder=None,
app=self.app,
default_plugins=False)
self.assertRaises(Error, senpy.analyse, Results())
self.assertRaises(Error, senpy.analyse, Results(), [])
def test_analyse_wrong(self):
""" Trying to analyse with a non-existent plugin should raise an error."""
@@ -156,29 +153,32 @@ class ExtensionsTest(TestCase):
r2 = analyse(self.senpy,
input="tupni",
output="tuptuo")
assert r1.analysis[0].id == "endpoint:plugins/Dummy_0.1"
assert r2.analysis[0].id == "endpoint:plugins/Dummy_0.1"
assert r1.analysis[0].algorithm == "endpoint:plugins/Dummy_0.1"
assert r2.analysis[0].algorithm == "endpoint:plugins/Dummy_0.1"
assert r1.entries[0]['nif:isString'] == 'input'
def test_analyse_error(self):
mm = mock.MagicMock()
mm.id = 'magic_mock'
mm.name = 'mock'
mm.is_activated = True
mm.process.side_effect = Error('error in analysis', status=500)
self.senpy.add_plugin(mm)
class ErrorPlugin(plugins.Analysis):
author = 'nobody'
version = 0
ex = Error()
def process(self, *args, **kwargs):
raise self.ex
m = ErrorPlugin(ex=Error('error in analysis', status=500))
self.senpy.add_plugin(m)
try:
analyse(self.senpy, input='nothing', algorithm='MOCK')
analyse(self.senpy, input='nothing', algorithm='ErrorPlugin')
assert False
except Error as ex:
assert 'error in analysis' in ex['message']
assert ex['status'] == 500
ex = Exception('generic exception on analysis')
mm.process.side_effect = ex
m.ex = Exception('generic exception on analysis')
try:
analyse(self.senpy, input='nothing', algorithm='MOCK')
analyse(self.senpy, input='nothing', algorithm='ErrorPlugin')
assert False
except Exception as ex:
assert 'generic exception on analysis' in str(ex)
@@ -194,7 +194,7 @@ class ExtensionsTest(TestCase):
def test_load_default_plugins(self):
senpy = Senpy(plugin_folder=self.examples_dir, default_plugins=True)
assert len(senpy.plugins()) > 1
assert len(senpy.plugins(is_activated=False)) > 1
def test_convert_emotions(self):
self.senpy.activate_all(sync=True)

18
tests/test_models.py
View File

@@ -5,7 +5,8 @@ import jsonschema
import json
import rdflib
from unittest import TestCase
from senpy.models import (Emotion,
from senpy.models import (Analysis,
Emotion,
EmotionAnalysis,
EmotionSet,
Entry,
@@ -61,7 +62,7 @@ class ModelsTest(TestCase):
def test_id(self):
""" Adding the id after creation should overwrite the automatic ID
"""
r = Entry()
r = Entry(_auto_id=True)
j = r.jsonld()
assert '@id' in j
r.id = "test"
@@ -189,6 +190,19 @@ class ModelsTest(TestCase):
assert isinstance(js['plugins'], list)
assert js['plugins'][0]['@type'] == 'sentimentPlugin'
def test_parameters(self):
'''An Analysis should contain the algorithm and the list of parameters to be used'''
a = Analysis()
a.params = {'param1': 1, 'param2': 2}
assert len(a.parameters) == 2
for param in a.parameters:
if param.name == 'param1':
assert param.value == 1
elif param.name == 'param2':
assert param.value == 2
else:
raise Exception('Unknown value %s' % param)
def test_from_string(self):
results = {
'@type': 'results',