diff --git a/allzweckmesser/meters.py b/allzweckmesser/meters.py
index fdfcc9d4c042f527f0e1bc8b18ec833ec0b3f4e0..0ffa84c072d36e2caa82ed001cad6feb65326447 100644
--- a/allzweckmesser/meters.py
+++ b/allzweckmesser/meters.py
@@ -3,104 +3,144 @@
import re
+from .features import ReadingMeterFeatures
from .model import Reading, Position
-def caesurae_together(position_specs, reward):
- def get_reward(meter: Meter, reading: Reading):
- for spec in position_specs:
- position = Position.after(spec[0], reading, spec[1], meter)
- if not position.word_boundary:
- return 0
- else:
- return reward
- return get_reward
-
-
-def bridge(position_spec, reward):
- def get_reward(meter: Meter, reading: Reading):
+def bridge(position_spec, feature):
+ def get_feature(meter: Meter, reading: Reading):
position = Position.after(position_spec[0], reading, meter,
position_spec[1])
if position.word_boundary:
- return 0
+ return None
else:
- return reward
- return get_reward
+ return feature
+ return get_feature
class Meter:
- def __init__(self, name: str, schema: str, conditions: list = None,
- short_name: str = None):
+ def __init__(self, name: str, schema: str, breaks: list = None,
+ conditions: list = None, short_name: str = None,
+ id: int = None):
self.name = name
self.schema = schema
+ self.break_specs = breaks
# Convert condition functions to instance-bound methods.
self.conditions = ([cond.__get__(self) for cond in conditions]
if conditions else [])
self.short_name = short_name
+ self.id = id
def match_reading(self, reading: Reading):
return re.match(self.schema, reading.get_schema())
- def get_rewards(self, reading: Reading):
- return sum(cond(reading) for cond in self.conditions)
+ def collect_condition_features(self, reading: Reading):
+ features = []
+ for cond in self.conditions:
+ feature = condition(reading)
+ if feature:
+ features.append(feature)
+ return features
+
+ def reading_has_usual_breaks(self, reading: Reading):
+ if self.break_specs:
+ for breaks in self.break_specs:
+ satisfied = True
+ for b in breaks:
+ position = Position.after(b[0], reading, b[1], self)
+ if not (hasattr(position, 'word_boundary')
+ and position.word_boundary):
+ satisfied = False
+ break
+ if satisfied:
+ return True
+ else:
+ return False
+ else:
+ return True
AEOLIC_BASE = r'(?:(–)(–)|(–)(â‘)|(â‘)(–))'
ALL_METERS = {
-
'Catalectic Dactylic Hexameter': Meter(
'Catalectic Dactylic Hexameter',
r'(–)(â‘â‘|–)(–)(â‘â‘|–)(–)(â‘â‘|–)(–)(â‘â‘|–)(–)(â‘â‘|–)(â‘|–)',
- conditions={
- caesurae_together([('mora', 6, 'Trithemimeral'),
- ('mora', 14, 'Hephthemimeral')], 2),
- caesurae_together([('mora', 10, 'Penthemimeral')], 2),
- caesurae_together([('mora', 16, 'Bucolic Diaeresis')], 1),
- bridge(('mora', 15, 'Hermann’s Bridge'), 1)
- },
- short_name='6da‸'
+ conditions=[
+ bridge(('mora', 15, 'Hermann’s Bridge'),
+ ReadingMeterFeatures.HEXAMETER_BRIDGE_VIOLATED)
+ ],
+ breaks=[
+ [('mora', 6, 'Trithemimeral'), ('mora', 14, 'Hephthemimeral')],
+ [('mora', 10, 'Penthemimeral')],
+ [('mora', 16, 'Bucolic Diaeresis')]
+ ],
+ short_name='6da‸',
+ id=0
),
'Dactylic Pentameter': Meter(
'Dactylic Pentameter',
r'(–)(â‘â‘|–)(–)(â‘â‘|–)(–)(–)(â‘â‘)(–)(â‘â‘)(â‘|–)',
- conditions={
- caesurae_together([('mora', 5, 'Middle diaresis')], 2)
- },
- short_name='3da‸3da‸'
+ breaks=[[('mora', 5, 'Middle diaeresis')]],
+ short_name='3da‸3da‸',
+ id=1
),
'Iambic Trimeter': Meter(
'Iambic Trimeter',
r'(â‘|â‘â‘|–)(â‘â‘|–)(â‘)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘)(â‘|–)',
- conditions={
- caesurae_together([('element', 4, 'After fourth element')], 1),
- caesurae_together([('element', 8, 'After eighth element')], 1),
- },
- short_name='3ia'
+ breaks=[
+ [('element', 4, 'After fourth element')]
+ [('element', 8, 'After eighth element')]
+ ],
+ short_name='3ia',
+ id=2
),
'Iambic Senarius': Meter(
'Iambic Senarius',
r'(â‘|â‘â‘|–)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘|â‘â‘|–)(â‘â‘|–)(â‘)(â‘|–)',
- short_name='6ia'
+ short_name='6ia',
+ id=3
),
'Sapphic Hendecasyllable': Meter(
'Sapphic Hendecasyllable',
r'(–)(–|â‘)(–)(–|â‘)(–)(â‘)(â‘)(–)(â‘)(–)(â‘|–)',
conditions={},
- short_name='sap hen'
+ short_name='sap hen',
+ id=4
),
'Adoneus': Meter(
'Adoneus',
r'(–)(â‘â‘)(–)(â‘|–)',
short_name='adoneus',
+ id=5
),
'Phalaecian Hendecasyllable': Meter(
'Phalaecian Hendecasyllable',
AEOLIC_BASE + r'(–)(â‘)(â‘)(–)(â‘)(–)(â‘)(–)(â‘|–)',
- conditions={
- caesurae_together([('element', 6, 'After sixth element')], 1)
- },
- short_name='hen'
+ breaks=[[('element', 6, 'After sixth element')]],
+ short_name='hen',
+ id=6
),
}
+
+
+def get_reading_meter_combinations(readings, meters=ALL_METERS):
+ reading_meter_rmfeatures = [
+ [reading, meter, {}]
+ for reading, meter
+ in itertools.product(readings, meters)
+ ]
+ for reading, meter, rmfeatures in reading_meter_rmfeatures:
+ rmfeatures[ReadingMeterFeatures.DOES_NOT_FIT_METER] = (
+ meter.match_reading(reading) is None)
+
+ # XXX: Implement this.
+ rmfeatures[ReadingMeterFeatures.NECESSARY_CHANGES_TO_MAKE_IT_FIT] = 0
+
+ rmfeatures[ReadingMeterFeatures.METER] = meter.id
+ rmfeatures[ReadingMeterFeatures.NO_USUAL_BREAK_PRESENT] = int(
+ meter.reading_has_usual_breaks(reading))
+ for feature in meter.collect_condition_features(reading):
+ rmfeatures[feature] = 1
+ return reading_meter_rmfeatures
diff --git a/allzweckmesser/model.py b/allzweckmesser/model.py
index fe2edfbf97f26aa26ca9e4af7c2371abe5b0444d..f86b31429fb6e312fc01846da4f1fffba243e0c8 100644
--- a/allzweckmesser/model.py
+++ b/allzweckmesser/model.py
@@ -1,6 +1,7 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
+from collections import defaultdict
import itertools
import json
import os
@@ -360,6 +361,7 @@ class Reading:
def __init__(self, tokens: List[Token] = None, phenomena: dict = None):
self.tokens = tokens or list()
self.phenomena = phenomena or dict()
+ self.features = defaultdict(lambda: 0)
@classmethod
def from_json(cls, json_file):
diff --git a/allzweckmesser/scanner.py b/allzweckmesser/scanner.py
index d074a282204a0f2e6c2b34f05f87aa02bf58cdbe..393ebb258d043e645fd6f50149c81355a8124078 100644
--- a/allzweckmesser/scanner.py
+++ b/allzweckmesser/scanner.py
@@ -7,6 +7,7 @@ from itertools import product
from .db import FormAnalysis
from .model import Reading, Syllable, Token, Verse, Phenomenon
+from .features import ReadingFeature
from .wordlist import WordList
CLITICS = ['que', 'qve', 'ue', 've', 'ne']
@@ -468,6 +469,7 @@ def generate_synizesis(reading):
syllable.syllable_length = 2
syllable.vowel_length = 2
syllable.phenomena['synizesis'] = Phenomenon(chars=syn_dict[syllable.id][3])
+ reading.features[ReadingFeature.SYNIZESIS] += 1
for s in token.syllables[j+2:]:
s.id -= 1
@@ -635,6 +637,7 @@ def parse_verse(verse):
(s.phenomena['positional lengthening']
.overruled_by) = 'muta cum liquida'
elif blueprint[syll_id] == '2':
+ reading.features[ReadingFeature.MCL_TRIGGERS_PL] += 1
s.syllable_length = 2
syll_id += 1
diff --git a/features.py b/features.py
new file mode 100644
index 0000000000000000000000000000000000000000..e6a80e88c31c3f786be5cc5d132c0234e9fa7a0d
--- /dev/null
+++ b/features.py
@@ -0,0 +1,28 @@
+# -*- coding: utf-8 -*-
+
+from enum import Enum
+
+
+class ReadingFeature(Enum):
+ MCL_TRIGGERS_PL = 0
+ SYNIZESIS = 1
+ S_ELISION = 2
+ HIAT = 3
+
+
+class ReadingMeterFeatures(Enum):
+ DOES_NOT_FIT_METER = 10
+ NECESSARY_CHANGES_TO_MAKE_IT_FIT = 11
+ METER = 12
+ NO_USUAL_BREAK_PRESENT = 13
+ HEXAMETER_BRIDGE_VIOLATED = 14
+
+
+class CombinedFeatures(Enum):
+ MCL_TRIGGERS_PL = 0
+ SYNIZESIS = 1
+ S_ELISION = 2
+ DOES_NOT_FIT_METER = 3
+ NECESSARY_CHANGES_TO_MAKE_IT_FIT = 4
+ NO_USUAL_BREAK_PRESENT = 5
+ BRIDGES_VIOLATED = 6
diff --git a/scripts/extract_meters.py b/scripts/extract_meters.py
index 29ab5de9f03b1dad2a192702cf78c5c33a2b8804..1c2d583be11809bd0d4801f1eb5724b8b4b36350 100644
--- a/scripts/extract_meters.py
+++ b/scripts/extract_meters.py
@@ -41,7 +41,7 @@ def main(hypotactic_dir, outfile):
pair[0] = list(pair[0])
with open(outfile, 'w') as f:
obj = {'poem_meters': poem_meters,
- 'line_meters': line_meters}
+ 'line_meters': line_meters}
json.dump(obj, f, indent=2)