From c09f3e9cda1ff230c5248fca5fbdc60b1c466f04 Mon Sep 17 00:00:00 2001
From: ZechyW <ZechyW@users.noreply.github.com>
Date: Tue, 11 Aug 2020 17:23:36 +0800
Subject: [PATCH] Use `wcswidth` instead of `len` to measure cell text width

- Fixes problems with cell alignment for CJK text within RST tables
---
 dashtable/dashutils/center_line.py             |  6 ++++--
 dashtable/dashutils/get_longest_line_length.py |  7 +++++--
 dashtable/data2md/get_column_width.py          |  5 ++++-
 dashtable/data2rst/cell/center_cell_text.py    |  7 +++++--
 .../data2rst/cell/get_longest_line_length.py   |  7 +++++--
 dashtable/data2rst/cell/v_center_cell_text.py  |  4 +++-
 dashtable/data2rst/make_cell.py                |  4 +++-
 .../restructify/converters/convert_h1.py       |  5 ++++-
 .../restructify/converters/convert_h2.py       |  5 ++++-
 .../restructify/converters/convert_h3.py       |  5 ++++-
 .../restructify/converters/convert_li.py       |  5 ++++-
 setup.py                                       |  1 +
 tests/static/cjk.html                          | 18 ++++++++++++++++++
 tests/static/cjk.md                            |  4 ++++
 tests/static/cjk.rst                           |  7 +++++++
 15 files changed, 75 insertions(+), 15 deletions(-)
 create mode 100644 tests/static/cjk.html
 create mode 100644 tests/static/cjk.md
 create mode 100644 tests/static/cjk.rst

diff --git a/dashtable/dashutils/center_line.py b/dashtable/dashutils/center_line.py
index 7a1bad2..8e35e25 100755
--- a/dashtable/dashutils/center_line.py
+++ b/dashtable/dashutils/center_line.py
@@ -1,5 +1,7 @@
 import math
 
+from wcwidth import wcswidth
+
 
 def center_line(space, line):
     """
@@ -21,8 +23,8 @@ def center_line(space, line):
     """
     line = line.strip()
 
-    left_length = math.floor((space - len(line)) / 2)
-    right_length = math.ceil((space - len(line)) / 2)
+    left_length = math.floor((space - wcswidth(line)) / 2)
+    right_length = math.ceil((space - wcswidth(line)) / 2)
 
     left_space = " " * int(left_length)
     right_space = " " * int(right_length)
diff --git a/dashtable/dashutils/get_longest_line_length.py b/dashtable/dashutils/get_longest_line_length.py
index 52dd3ec..16a16d9 100755
--- a/dashtable/dashutils/get_longest_line_length.py
+++ b/dashtable/dashutils/get_longest_line_length.py
@@ -1,10 +1,13 @@
+from wcwidth import wcswidth
+
+
 def get_longest_line_length(text):
     """Get the length longest line in a paragraph"""
     lines = text.split("\n")
     length = 0
 
     for i in range(len(lines)):
-        if len(lines[i]) > length:
-            length = len(lines[i])
+        if wcswidth(lines[i]) > length:
+            length = wcswidth(lines[i])
 
     return length
diff --git a/dashtable/data2md/get_column_width.py b/dashtable/data2md/get_column_width.py
index e150360..10678ca 100755
--- a/dashtable/data2md/get_column_width.py
+++ b/dashtable/data2md/get_column_width.py
@@ -1,3 +1,6 @@
+from wcwidth import wcswidth
+
+
 def get_column_width(column, table):
     """
     Get the character width of a column in a table
@@ -17,7 +20,7 @@ def get_column_width(column, table):
     width = 3
 
     for row in range(len(table)):
-        cell_width = len(table[row][column])
+        cell_width = wcswidth(table[row][column])
         if cell_width > width:
             width = cell_width
 
diff --git a/dashtable/data2rst/cell/center_cell_text.py b/dashtable/data2rst/cell/center_cell_text.py
index 7670f2c..caa947f 100755
--- a/dashtable/data2rst/cell/center_cell_text.py
+++ b/dashtable/data2rst/cell/center_cell_text.py
@@ -1,4 +1,7 @@
 import math
+
+from wcwidth import wcswidth
+
 from .get_longest_line_length import get_longest_line_length
 
 def center_cell_text(cell):
@@ -20,7 +23,7 @@ def center_cell_text(cell):
     cell : dashtable.data2rst.Cell
     """
     lines = cell.text.split('\n')
-    cell_width = len(lines[0]) - 2
+    cell_width = wcswidth(lines[0]) - 2
 
     truncated_lines = ['']
     for i in range(1, len(lines) - 1):
@@ -37,7 +40,7 @@ def center_cell_text(cell):
 
     for i in range(len(truncated_lines)):
         truncated_lines[i] = left_space + truncated_lines[i]
-        right_width = cell_width - len(truncated_lines[i])
+        right_width = cell_width - wcswidth(truncated_lines[i])
         truncated_lines[i] += right_width * ' '
 
     for i in range(1, len(lines) - 1):
diff --git a/dashtable/data2rst/cell/get_longest_line_length.py b/dashtable/data2rst/cell/get_longest_line_length.py
index 52dd3ec..16a16d9 100755
--- a/dashtable/data2rst/cell/get_longest_line_length.py
+++ b/dashtable/data2rst/cell/get_longest_line_length.py
@@ -1,10 +1,13 @@
+from wcwidth import wcswidth
+
+
 def get_longest_line_length(text):
     """Get the length longest line in a paragraph"""
     lines = text.split("\n")
     length = 0
 
     for i in range(len(lines)):
-        if len(lines[i]) > length:
-            length = len(lines[i])
+        if wcswidth(lines[i]) > length:
+            length = wcswidth(lines[i])
 
     return length
diff --git a/dashtable/data2rst/cell/v_center_cell_text.py b/dashtable/data2rst/cell/v_center_cell_text.py
index b0d35e7..67f4932 100755
--- a/dashtable/data2rst/cell/v_center_cell_text.py
+++ b/dashtable/data2rst/cell/v_center_cell_text.py
@@ -1,5 +1,7 @@
 import math
 
+from wcwidth import wcswidth
+
 
 def v_center_cell_text(cell):
     """
@@ -24,7 +26,7 @@ def v_center_cell_text(cell):
     cell : dashtable.data2rst.Cell
     """
     lines = cell.text.split('\n')
-    cell_width = len(lines[0]) - 2
+    cell_width = wcswidth(lines[0]) - 2
 
     truncated_lines = []
     for i in range(1, len(lines) - 1):
diff --git a/dashtable/data2rst/make_cell.py b/dashtable/data2rst/make_cell.py
index c595190..ebcc5b8 100755
--- a/dashtable/data2rst/make_cell.py
+++ b/dashtable/data2rst/make_cell.py
@@ -1,3 +1,5 @@
+from wcwidth import wcswidth
+
 from ..dashutils import get_span_char_width
 from ..dashutils import get_span_char_height
 from ..dashutils import get_span_row_count
@@ -35,7 +37,7 @@ def make_cell(table, span, widths, heights, use_headers):
 
     lines = text.split("\n")
     for i in range(len(lines)):
-        width_difference = width - len(lines[i])
+        width_difference = width - wcswidth(lines[i])
         lines[i] = ''.join([lines[i], " " * width_difference])
 
     height_difference = height - len(lines)
diff --git a/dashtable/html2data/restructify/converters/convert_h1.py b/dashtable/html2data/restructify/converters/convert_h1.py
index f780119..a72612c 100755
--- a/dashtable/html2data/restructify/converters/convert_h1.py
+++ b/dashtable/html2data/restructify/converters/convert_h1.py
@@ -1,7 +1,10 @@
+from wcwidth import wcswidth
+
+
 def convert_h1(element, text):
     """
     Add '=' to the bottom of text
     """
     if text:
-        text = text + '\n' + '=' * len(text)
+        text = text + '\n' + '=' * wcswidth(text)
     return text
diff --git a/dashtable/html2data/restructify/converters/convert_h2.py b/dashtable/html2data/restructify/converters/convert_h2.py
index fbf796d..fdd04d7 100755
--- a/dashtable/html2data/restructify/converters/convert_h2.py
+++ b/dashtable/html2data/restructify/converters/convert_h2.py
@@ -1,7 +1,10 @@
+from wcwidth import wcswidth
+
+
 def convert_h2(element, text):
     """
     Adds '-' to the bottom of the text
     """
     if text:
-        text = text + '\n' + '-' * len(text)
+        text = text + '\n' + '-' * wcswidth(text)
     return text
diff --git a/dashtable/html2data/restructify/converters/convert_h3.py b/dashtable/html2data/restructify/converters/convert_h3.py
index 2b3e5db..371f4c0 100755
--- a/dashtable/html2data/restructify/converters/convert_h3.py
+++ b/dashtable/html2data/restructify/converters/convert_h3.py
@@ -1,7 +1,10 @@
+from wcwidth import wcswidth
+
+
 def convert_h3(element, text):
     """
     Adds '~' to the bottom of the text
     """
     if text:
-        text = text + '\n' + '~' * len(text) + '\n'
+        text = text + '\n' + '~' * wcswidth(text) + '\n'
     return text
diff --git a/dashtable/html2data/restructify/converters/convert_li.py b/dashtable/html2data/restructify/converters/convert_li.py
index 758e18e..0dbee30 100755
--- a/dashtable/html2data/restructify/converters/convert_li.py
+++ b/dashtable/html2data/restructify/converters/convert_li.py
@@ -1,3 +1,6 @@
+from wcwidth import wcswidth
+
+
 def get_bullet_depth(element):
     depth = -1
     while element:
@@ -21,7 +24,7 @@ def convert_li(element, text):
         for i in range(len(children)):
             if parent.index(children[i]) == index:
                 number = str(i + 1)
-                interspace = (space - len(number)) * ' '
+                interspace = (space - wcswidth(number)) * ' '
                 text = number + '.' + interspace + text
 
         depth = get_bullet_depth(element)
diff --git a/setup.py b/setup.py
index cd1fcdd..2e47e1a 100755
--- a/setup.py
+++ b/setup.py
@@ -23,4 +23,5 @@
     url='https://github.com/doakey3/DashTable',
     download_url='https://github.com/doakey3/DashTable/tarball/1.4.5',
     license='MIT',
+    install_requires=['wcwidth'],
 )
diff --git a/tests/static/cjk.html b/tests/static/cjk.html
new file mode 100644
index 0000000..f042c63
--- /dev/null
+++ b/tests/static/cjk.html
@@ -0,0 +1,18 @@
+<table>
+    <caption>Testing tables with CJK text</caption>
+    <tr>
+        <td>This is</td>
+        <td>some text</td>
+        <td>in English.</td>
+    </tr>
+    <tr>
+        <td>This text</td>
+        <td>是用中文写的</td>
+        <td>instead.</td>
+    </tr>
+    <tr>
+        <td>This is</td>
+        <td>some more text</td>
+        <td>in English.</td>
+    </tr>
+</table>
diff --git a/tests/static/cjk.md b/tests/static/cjk.md
new file mode 100644
index 0000000..6263a7d
--- /dev/null
+++ b/tests/static/cjk.md
@@ -0,0 +1,4 @@
+|  This is  |   some text    | in English. |
+|-----------|----------------|-------------|
+| This text |  是用中文写的  |  instead.   |
+|  This is  | some more text | in English. |
\ No newline at end of file
diff --git a/tests/static/cjk.rst b/tests/static/cjk.rst
new file mode 100644
index 0000000..8cc1f1d
--- /dev/null
+++ b/tests/static/cjk.rst
@@ -0,0 +1,7 @@
++-----------+----------------+-------------+
+| This is   | some text      | in English. |
++-----------+----------------+-------------+
+| This text | 是用中文写的   | instead.    |
++-----------+----------------+-------------+
+| This is   | some more text | in English. |
++-----------+----------------+-------------+
\ No newline at end of file