plint

commit 63f02cd2424269acf84b0d7a63501e1f41ef1384
parent 7079f072bd1db3f28bc43ede33c6520d2abf2d6b
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sun,  6 Sep 2015 23:26:40 +0200

improve disjunction/elision handling

Diffstat:
verse.py
 | 
38
+++++++++++++++++++++++++++++++-------
versetest.py
 | 
50
++++++++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 81 insertions(+), 7 deletions(-)
diff --git a/verse.py b/verse.py
@@ -28,22 +28,44 @@ letters = {
     'z': 'zaide'
 }
 
-def elision(word):
-  if (word.startswith('y') and not word == 'y' and not word.startswith("yp") and
-      not word.startswith("yeu")):
-    return [False]
-  if word in ["oui", "ouis"] or word.startswith("ouistiti"):
+def elision(word, was_cap):
+  if word.startswith('y'):
+    if word == 'y':
+      return [True]
+    if was_cap:
+      if word == 'york':
+        return [False]
+      # Grevisse, Le Bon usage, 14th ed., paragraphs 49-50
+      # depends on whether it's French or foreign...
+      return [True, False]
+    else:
+      exc = ["york", "yeux", "yeuse", "ypérite"]
+      for w in exc:
+        if word.startswith(w):
+          return [True]
+      # otherwise, no elision
+      return [False]
+  if word in ["oui", "ouis"]:
     # elision for those words, but beware, no elision for "ouighour"
     # boileau : "Ont l'esprit mieux tourné que n'a l'homme ? Oui sans doute."
-    # so elission sometimes
+    # so elision sometimes
     return [True, False]
+  if word.startswith("ouistiti") or word.startswith("ouagadougou"):
+    return [False]
   # "un", "une" are non-elided as nouns ("cette une")
   if word in ["un", "une"]:
     return [True, False]
   # "onze" is not elided
   if word == "onze":
     return [False]
+  if word.startswith('ulul'):
+    return [False] # ululement, ululer, etc.
+  if word.startswith('uhlan'):
+    return [False] # uhlan
   if word[0] == 'h':
+    if word == "huis":
+      # special case, "huis" is elided but "huis clos" isn't
+      return [True, False]
     return list(map((lambda s: not s), haspirater.lookup(word)))
   if is_vowels(word[0]):
     return [True]
@@ -216,7 +238,9 @@ class Verse:
     # vowel elision problems
     for w in self.chunks:
       if 'elision' not in w[0].keys():
-        w[0]['elision'] = elision(''.join(x['text'] for x in w))
+        first_letter = common.rm_punct(w[0]['original'].strip())
+        w[0]['elision'] = elision(''.join(x['text'] for x in w),
+            first_letter == first_letter.upper())
 
     # case of 'y'
     ys_regexp = re.compile("(y*)")
diff --git a/versetest.py b/versetest.py
@@ -189,6 +189,56 @@ class PoemCounts(Counts):
     possible = self.runCount(self.v3, limit="6/6")
     self.assertTrue(self.achievesPossibility(possible, 12))
 
+class Disjunct(Counts):
+  # inspired by Grevisse, Le Bon usage, 14th ed., paragraphs 49-50
+  d = {
+      "hiérarchie": 4,
+      "yeux": 1,
+      "yeuse": 1,
+      "yodel": 3,
+      "yacht": 2,
+      "York": 1,
+      "yole": 2,
+      "Yourcenar": 4,
+      "Yvelines": 3,
+      "Ypres": 1,
+      "ypérite": 3,
+      "Ysaÿe": 3,
+      "Ionesco": 4,
+      "Yahvé": 3,
+      "Yungfrau": 3,
+      "yodler": 3,
+      "oui": 2,
+      "ouïe": 2,
+      "ouïr": 2,
+      "ouest": 1,
+      "Ouagadougou": 6,
+      "oisif": 2,
+      "huis": 2,
+      "huit": 2,
+      "huissier": 2,
+      "uhlan": 3,
+      "ululer": 4,
+      "ululement": 5,
+      "onze": 2,
+      "onzième": 3,
+      # both are possible for 'un' and 'une'
+      "Un": 2,
+      "un": 2,
+      "Une": 2,
+      "une": 1,
+      # too weird to figure out correct counts in poems
+      #"Yolande"
+      #"ouistiti"
+  }
+
+  def testDisjunct(self):
+    for k in self.d.keys():
+      v = self.d[k] + 1
+      vv = "belle " + k
+      possible = self.runCount(vv, limit=v)
+      self.assertTrue(self.achievesPossibility(possible, v))
+
 class SanityCheck(unittest.TestCase):
   def testSimple(self):
     text = "Patati patata patata tata vies"