From 52cb4793a2220b11ef17217560c6db7818371053 Mon Sep 17 00:00:00 2001
From: pat <pat@lucy.home>
Date: Mon, 26 Sep 2016 15:09:30 +0200
Subject: [PATCH] integration of the tind fime (for iteration) in TextFST
object
---
unitex/utils/formats.py | 11 ++---------
1 file changed, 2 insertions(+), 9 deletions(-)
diff --git a/unitex/utils/formats.py b/unitex/utils/formats.py
index 7a6abb7..9e0a6bc 100644
--- a/unitex/utils/formats.py
+++ b/unitex/utils/formats.py
@@ -458,8 +458,6 @@ class TextFST:
self.__tfst = None
self.__tind = None
- self.__size = 0
-
def __del__(self):
self.__tfst.close()
@@ -467,6 +465,8 @@ class TextFST:
return len(self.__tind)
def __getitem__(self, i):
+ if i >= len(self):
+ raise UnitexException("TextFST index out of range (size: %s)." % len(self))
position = self.__tind[i]
self.__tfst.seek(position)
@@ -606,13 +606,6 @@ class TextFST:
encoding = UnitexConstants.DEFAULT_ENCODING
self.__tfst = open(fst, "r", encoding=encoding)
-
- line = self.__tfst.readline()
- line = line.rstrip()
-
- # The number of sentence in the text fst (format: '000000000N')
- self.__size = int(line)
-
self.__tind = []
with open(index, "rb") as fin:
--
GitLab