plint

commit f726bfffe714b25118092cb16b9e7360d7c12232
parent f57cc84f75061ee7ad554c555f600faf4b88db51
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Tue, 10 Jul 2012 21:58:08 +0200

add hiatus, improve forbidden reporting

Diffstat:
error.py
 | 
15
+++++++++++----
metric.py
 | 
20
+++++++++++++++-----
template.py
 | 
12
+++++++++---

3 files changed, 35 insertions(+), 12 deletions(-)
diff --git a/error.py b/error.py
@@ -43,12 +43,19 @@ class ErrorBadCharacters(Error):
         % ', '.join(["'" + a + "'" for a in self.characters]), short)
 
 class ErrorForbiddenPattern(Error):
-  def __init__(self):
-    # TODO give more info
-    pass
+  def __init__(self, pattern):
+    self.pattern = pattern
+
+  def report(self, short=False):
+    return Error.report(self, "Illegal ambiguous pattern: %s" % self.pattern,
+        short)
+
+class ErrorHiatus(Error):
+  def __init__(self, hiatus):
+    self.hiatus = hiatus
 
   def report(self, short=False):
-    return Error.report(self, "Illegal ambiguous pattern", short)
+    return Error.report(self, "Illegal hiatus: %s" % self.hiatus, short)
 
 class ErrorBadRhyme(Error):
   def __init__(self, expected, inferred):
diff --git a/metric.py b/metric.py
@@ -7,6 +7,9 @@ import vowels
 import haspirater
 
 
+no_hiatus = ["oui"]
+
+
 def annotate_aspirated(word):
   """Annotate aspirated 'h'"""
   if word[0] != 'h':
@@ -90,7 +93,7 @@ def feminine(align, verse, phon):
   return possible
 
 
-def parse(text, phon, bound, forbidden_ok, diaeresis):
+def parse(text, phon, bound, forbidden_ok, hiatus_ok, diaeresis):
   """Return possible aligns for text, bound is an upper bound on the align
   length to limit running time, phon is the pronunciation to help for gender,
   forbidden_ok is true if we allow classically forbidden patterns"""
@@ -145,7 +148,8 @@ def parse(text, phon, bound, forbidden_ok, diaeresis):
 
   pattern = re.compile('(['+consonants+'*-]*)', re.UNICODE)
 
-  forbidden = False
+  forbidden = None
+  hiatus = None
 
   # cut each word in chunks of vowels and consonants, with some specific
   # kludges
@@ -175,12 +179,16 @@ def parse(text, phon, bound, forbidden_ok, diaeresis):
             [1 for chunk in words[i-1] if is_vowels(chunk)]) > 1:
           words[i-1].pop(-1)
           words[i-1][-1] = words[i-1][-1]+"`"
+        if (is_vowels(words[i-1][-1]) and not words[i-1][-1][-1] == 'e'
+            and not (''.join(words[i]) in no_hiatus
+                and ''.join(words[i-1]) in no_hiatus)):
+          hiatus = words[i-1][-1] + ' ' + words[i][0]
       else:
         if words[i-1][-1] == 'ée' or words[i-1][-1] == 'ie':
-          forbidden = True
+          forbidden = words[i-1][-1]
       if words[i-1][-1] == 's' and len(words[i-1]):
         if words[i-1][-2] == 'ée' or words[i-1][-2] == 'ie':
-          forbidden = True
+          forbidden = words[i-1][-2]
         # TODO there are arcane rules for "aient"
       # case of "soient"
       # TODO there are a lot of "oient" in boileau and malherme
@@ -191,7 +199,9 @@ def parse(text, phon, bound, forbidden_ok, diaeresis):
       #       forbidden = True
 
   if forbidden and not forbidden_ok:
-    return None
+    return ("forbidden", forbidden)
+  if hiatus and not hiatus_ok:
+    return ("hiatus", hiatus)
 
   # group back words
   for word in words:
diff --git a/template.py b/template.py
@@ -46,6 +46,7 @@ class Template:
     self.template = []
     self.pattern_line_no = 0
     self.forbidden_ok = False
+    self.hiatus_ok = False
     self.normande_ok = True
     self.check_end_hemistiche = True
     self.check_occurrences = True
@@ -66,6 +67,8 @@ class Template:
       self.mergers.append(value)
     elif key == "forbidden_ok":
       self.forbidden_ok = str2bool(value)
+    elif key == "hiatus_ok":
+      self.hiatus_ok = str2bool(value)
     elif key == "normande_ok":
       self.normande_ok = str2bool(value)
     elif key == "diaeresis":
@@ -136,9 +139,12 @@ class Template:
 
     # compute alignments, check hemistiches, sort by score
     possible = parse(line, self.env[pattern.myid].phon, pattern.length + 2,
-        self.forbidden_ok, self.diaeresis)
-    if not possible:
-      errors.append(error.ErrorForbiddenPattern())
+        self.forbidden_ok, self.hiatus_ok, self.diaeresis)
+    if not isinstance(possible, list):
+      if possible[0] == "forbidden":
+        errors.append(error.ErrorForbiddenPattern(possible[1]))
+      elif possible[0] == "hiatus":
+        errors.append(error.ErrorHiatus(possible[1]))
       possible = []
       return errors, pattern
     possible = list(map((lambda p: (p[0], p[1],