From 87dc396bb6bc877c52e5b99083980e4d23aa8848 Mon Sep 17 00:00:00 2001
From: Alexis Filipozzi <alexis.filipozzi@gmail.com>
Date: Thu, 16 Jun 2016 20:24:49 +0200
Subject: [PATCH] small correction in ul translation
---
data.py | 2 ++
html_parser.py | 19 ++++++++++---------
main.py | 3 +--
3 files changed, 13 insertions(+), 11 deletions(-)
diff --git a/data.py b/data.py
index f600fd7..4055f40 100644
--- a/data.py
+++ b/data.py
@@ -89,6 +89,8 @@ class TextContent(ParagraphContent):
def get_string(self, clean = True):
text = self._text
+ if not text:
+ text = ""
if clean:
text = text.strip()
return text
diff --git a/html_parser.py b/html_parser.py
index 91a1e69..22f2156 100644
--- a/html_parser.py
+++ b/html_parser.py
@@ -2,11 +2,11 @@ from bs4 import BeautifulSoup
from data import *
def iterable(a):
- try:
- (x for x in a)
- return True
- except TypeError:
- return False
+ try:
+ (x for x in a)
+ return True
+ except TypeError:
+ return False
def iterable_and_not_string(a):
return iterable(a) and not isinstance(a, basestring)
@@ -88,10 +88,11 @@ class Parser:
self._data._body.append(Paragraph(indent))
else:
self.parse_paragraph(c)
- if self._data._body[-1]._content:
- last_content = self._data._body[-1]._content[-1]
- if isinstance(last_content, TextContent):
- last_content._new_line = True
+ if content.name == "li":
+ if self._data._body[-1]._content:
+ last_content = self._data._body[-1]._content[-1]
+ if isinstance(last_content, TextContent):
+ last_content._new_line = True
def parse_image(self, image_content):
self._data._body.append(self.return_parsed_image(image_content))
diff --git a/main.py b/main.py
index 5e65628..bd203ad 100644
--- a/main.py
+++ b/main.py
@@ -32,12 +32,11 @@ def parse_and_format(dir_path, filename, dest):
for sub_file in os.listdir(path_to_file):
parse_and_format(path_to_file, sub_file, path_to_dst)
elif os.path.isfile(path_to_file):
- parse_format_and_write_file(path_to_file, path_to_dst)
+ parse_format_and_write_file(path_to_file, os.path.join(dest, filename.replace(".html", ".protoxml")))
else:
print "Unhandled file type (probably symlink) with file " + str(path_to_file)
def parse_format_and_write_file(filename, dest):
- print "parse file"
has_been_parsed = False
parser = None
formatter = None
--
GitLab