duire

commit d698e2137fa0ba51ffd8b54dac0252110794b20c
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Mon, 23 Mar 2015 00:57:40 +0100

initial

Diffstat:
inf.sh
 | 
4
++++
prefix.py
 | 
45
+++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 49 insertions(+), 0 deletions(-)
diff --git a/inf.sh b/inf.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+
+grep 'cat=v,@W' lefff-ext-3.2-utf.txt | cut -f 1 | sort | uniq
+
diff --git a/prefix.py b/prefix.py
@@ -0,0 +1,45 @@
+#!/usr/bin/python3
+# -*- encoding: utf-8
+
+import sys
+
+prefixes = ["dé", "re", "par", "ex", "sous", "sur", "in", "as", "bi", "em",
+"ac", "rec", "di", "su", "en"]
+words = set()
+interesting = {}
+threshold = 2
+
+for l in sys.stdin.readlines():
+    words.add(l.rstrip())
+
+for w in words:
+    for p in prefixes:
+        if w.startswith(p):
+            w2 = w[len(p):]
+            if w2 not in interesting.keys():
+                interesting[w2] = set()
+            interesting[w2].add(w)
+
+for wi, ws in interesting.items():
+    if wi in words:
+        continue
+    if len(ws) >= threshold:
+        print(wi, ws)
+
+#myprefs = {}
+#pthresh = 4
+#cons = "bcçdfghjklmnpqrstvwxz"
+# for wi, ws in interesting.items():
+#     if wi in words:
+#         continue
+#     for w2 in words:
+#         if w2.endswith(wi):
+#             mypref = w2[:len(wi)]
+#             if mypref not in myprefs.keys():
+#                 myprefs[mypref] = 0
+#             myprefs[mypref] += 1
+# 
+# for mypref, v in myprefs.items():
+#     if v >= pthresh:
+#         print (mypref, v)
+