plint

commit c1a68b108950f1281d4b8d2d6748035ccf92dde6
parent eeae0d2afab46d6ae86169cb34803bff4f8dbd29
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Mon, 12 Aug 2019 01:26:00 +0200

code to count syllables

Diffstat:
lexique_comparison/count_syllables_lexique.py
 | 
25
+++++++++++++++++++++++++
lexique_comparison/count_syllables_plint.py
 | 
34
++++++++++++++++++++++++++++++++++

2 files changed, 59 insertions(+), 0 deletions(-)
diff --git a/lexique_comparison/count_syllables_lexique.py b/lexique_comparison/count_syllables_lexique.py
@@ -0,0 +1,25 @@
+#!/usr/bin/python3
+
+# count the number of syllables of words according to lexique
+
+import sys
+
+vowels = "ae$E2@#u)9ioO(y"
+consonants = "dpgmtRwszlbkZjknfvSNJx8"
+
+for l in sys.stdin.readlines():
+    f = l.strip().split("\t")
+    nsyl = 0
+    for a in f[1]:
+        if a in vowels:
+            nsyl += 1
+        elif a in consonants:
+            pass
+        else:
+            print("unknown phoneme %s" % a, file=sys.stderr)
+            sys.exit(1)
+    # workaround bug in lexique
+    if f[1].endswith("@") and f[0] != "afin de":
+        nsyl -= 1
+    print("%s\t%d" % (f[0], nsyl))
+
diff --git a/lexique_comparison/count_syllables_plint.py b/lexique_comparison/count_syllables_plint.py
@@ -0,0 +1,34 @@
+#!/usr/bin/python3
+
+import os
+import sys
+
+# modules are in the parent folder
+sys.path.insert(1, os.path.join(sys.path[0], '..'))
+
+import verse
+import rhyme
+import template
+import diaeresis
+from pprint import pprint
+
+diaeresis.load_diaeresis("diaeresis.json")
+
+templateobj = template.Template()
+patternobj = template.Pattern("12")
+
+for l in sys.stdin.readlines():
+    w = (l.strip().split("\t"))[0]
+    v = verse.Verse(w, templateobj, patternobj)
+    rhymeobj = rhyme.Rhyme(v.normalized,
+              patternobj.constraint, templateobj.mergers, templateobj.options)
+    v.phon = rhymeobj.phon
+    v.annotate()
+    mx = 0
+    mn = 0
+    for c in v.chunks:
+        if 'weights' in c.keys():
+            mn += min(c['weights'])
+            mx += max(c['weights'])
+    print("%s\t%d\t%d" % (w, mn, mx))
+