Update handling traditional chinese autmatically

data/test_zh_t.txt

@@ -0,0 +1,74 @@
+呼嘯山莊
+
+-------------------------------------------------- ------------------------------
+
+第二章
+
+ 中文
+
+
+我想就在書房邊呼磨山霧了，想不想踩著雜草到。
+
+注意 注意 注意 吃 吃 飯 午飯 吃 並且 當作 所 房子 午 飯 的 附屬物 管家 , 一 位 的 這 位 的 這 位 的 這 位 的 這 位 並 開 意 的 太太 能 或 願 意 意 我 在 五 點 鐘 時 用 意 意 , 在 懷 懷 的 聯 系上一樓，邁一個進了屋子地上，有掃帚正在她她正在堆堆的煤渣搞一團散佈的立刻回頭了。場場初降鵝毛鵝毛大雪。
+
+在荒涼的涼涼裡，而而在山頂的樹叢中，踩著陸地的。到我的手指骨都痛了，狗狂吠起來。
+
+“倒霉的人至少就沒有禮了，我就直接為我家就這樣了。”我還在搖家裡孤零零地打電話。它。苦臉的約瑟夫從穀倉的一個圓窗裡探出頭來。
+
+“你幹嗎？”他大叫。 “主人在牛欄裡，你找他說話，就從這條路口繞過去。”
+
+“屋裡開門嗎？”我也叫起來。
+
+“除了老婆沒有別人。你就是騰到夜裡，她也不會開。”
+
+“為什麼？你就不能告訴她我是誰，呃，約瑟夫？”
+
+“別找我！我才不管這些閒事呢，”這個腦袋咕嚕著，又不見了。
+
+雪開始大 握住 門柄 門柄 又 沒有 一個 外衣 外衣 烤 ， ， 一草 耙 ， 在 院子 裡 他 招呼 我 著 走 走 一 穿過 一個 一個 和 一 鋪 一地 的 、 有 抽水機抽水機 抽水機 鴿籠 鴿籠 到 到 我 使 接待 過 的 那 個 的 、 屋子老婆以前從來沒有她的料。這麼一她我的等候，以為我會叫坐下。望望我，往的椅子一靠，不動，也不出。
+
+“天氣真！壞我我是老婆的老婆，說不定你的僕人大吃大喝，大勁才聽到，聽見了！我的”
+
+眼睛一瞪——也不開開。 總是一冷冷瞪她的神眸冷眼瞪住，使，而且不愉快
+
+“坐下吧，”那粗聲粗氣說，“他來了。”
+
+我咳了一聲，叫那惡狗朱諾。臨到第二次會面，它總算賞臉，搖起尾巴人埆礆。服從表柆示。衘礆。
+
+“好漂亮的狗我又開始說話了。”你是不是反對那些呢！
+
+“這一些我的女刺客克厲克民的回答，比希希親可愛的回答的主人公更冷冷的說那些人。”
+
+“，，我所愛的就是在一堆人身上！”
+
+“誰會愛這些東西那才怪呢！”她輕蔑地說。
+
+倒霉，原來又是一堆死兔子。我輕咳一聲，向火爐湊近，又把今晚天氣不好的話評論一通。
+
+“你本來就不會。。她說，站起來去拿”該台上的兩個茶葉罐。
+
+我沒有她的小官現在還沒有全身出現纖細的透明狀態，她的臉還沒有明顯的淡麗，而且她的身體還沒有明顯的淡麗。捲髮還不如說說，垂垂垂垂的她脖子上說說，輕而易舉地在與眼神能拒一些和說，要使人抗拒。一，而那張眼神 眼神 眼神 不是自然的。
+
+她夠不到。 動了 一動 ，她 猛 猛 猛 猛 地 向 像 像 守財奴 守財奴 他 他 的 金子 金子
+
+“我不要你幫忙，”她怒氣沖沖地說，“我自己得到了。”
+
+“對不起！”我連忙回答。
+
+“是請你來喫茶的嗎？”她問，把一條圍裙系在那乾淨的黑衣服上，就這樣站著，拿一盒茶葉〣邌候茶葉〣邌候茶葉〣邌
+
+“我很想喝杯茶。”我回答。
+
+“是請你來的嗎？”她又問。
+
+“沒有，”我說，勉強笑一笑。 “您請我喝茶。”
+
+
+
+
+目錄
+ 上一章
+ 下一章
+
+
+返回首頁

litbee/app.py

@@ -55,12 +55,13 @@ from pathlib import Path
 from types import SimpleNamespace
 from typing import Optional
 
-import debee
-import dzbee
-import ezbee
 import loguru
 import logzero
 import pandas as pd
+import ezbee
+import dzbee
+import debee
+
 import streamlit as st
 from loguru import logger as loggu
 from logzero import logger
@@ -68,25 +69,25 @@ from set_loglevel import set_loglevel
 from streamlit import session_state as state
 
 from litbee import __version__
-from litbee.home import home
-from litbee.info import info
-from litbee.multipage import Multipage
-from litbee.settings import settings
+# from litbee.options import options
 
 # from litbee.files2df import files2df
 # from litbee.utils import sb_front_cover, instructions, menu_items
 # from litbee.ezbee_page import ezbee_page
 # from litbee.dzbee_page import dzbee_page
 # from litbee.xbee_page import xbee_page
-from litbee.utils import menu_items, style_css
-
-# from litbee.options import options
+from litbee.utils import menu_items
 
+from litbee.multipage import Multipage
 
 # from litbee.fetch_upload import fetch_upload
 # from litbee.fetch_paste import fetch_paste
 # from litbee.fetch_urls import fetch_urls
 
+from litbee.home import home
+from litbee.settings import settings
+from litbee.info import info
+from litbee.utils import style_css
 
 # from ezbee import ezbee
 
@@ -134,6 +135,7 @@ _ = dict(
     beetype="ezbee",
     sourcetype="upload",
     sourcecount=2,
+    sent_ali=None,
     src_filename="",
     tgt_filename="",
     src_fileio=b"",

litbee/{fetch_upload.py → fetch_upload.py-}

@@ -281,6 +281,9 @@ def fetch_upload():  # noqa
     _ = st.expander("done aligned")
     with _:
         st.table(df_a.astype(str))
+        # st.markdown(df_a.astype(str).to_markdown())
+        # st.write("to_numpy().tolist()")
+        # st.write(df_a.astype(str).to_numpy().tolist())
 
     # insert seq no
     df_a.insert(0, "sn", range(len(df_a)))

litbee/files2df.py

@@ -14,7 +14,7 @@ def files2df(file1, file2):
     text1 = [_.strip() for _ in process_upload(file1).splitlines() if _.strip()]
 
     # if file2 is tempfile._TemporaryFileWrapper:
-    if isinstance(file2, tempfile._TemporaryFileWrapper):
+    if isinstance(file2, tempfile._TemporaryFileWrapper):  # pylint: disable=protected-access
         try:
             filename = file2.name
         except AttributeError:

litbee/home.py

@@ -2,13 +2,16 @@
 
 org ezbee_page.py.
 """
+# pylint: disable=invalid-name
+# pylint: disable=too-many-locals, too-many-return-statements, too-many-branches, too-many-statements
 import inspect
 
 # pylint: disable=invalid-name
 from functools import partial
 from itertools import zip_longest
-from time import perf_counter
+from about_time import about_time
 
+import hanzidentifier
 import logzero
 import numpy as np
 import pandas as pd
@@ -29,7 +32,7 @@ from st_aggrid import AgGrid, GridOptionsBuilder, GridUpdateMode
 # from st_aggrid.grid_options_builder import GridOptionsBuilder
 from streamlit import session_state as state
 
-# logzero.loglevel(set_loglevel())
+from litbee.t2s import t2s
 
 
 def home():  # noqa
@@ -114,26 +117,35 @@ def home():  # noqa
     # proceed when Submit is clicked
     msg1 = ""
     if filename1:
-        msg1 += f" file1: {filename1}"
+        msg1 += f" file1 {filename1}"
     msg2 = ""
     if filename2:
-        msg2 += f" file2: {filename2}"
+        msg2 += f" file2 {filename2}"
     glue = ""
     if filename1 and filename2:
         glue = ", "
 
-    st.write(f"  Submitted upload: {msg1}{glue}{msg2}")
+    upload_placeholder = st.empty()
+    prefix = f"  Submitted upload: {msg1}{glue}{msg2}"
+    upload_placeholder.write(prefix)
+
+    # st.write(f"  Submitted upload: {msg1}{glue}{msg2}")
     if not submitted:
         return None
 
     if not (filename1 or filename2):
-        st.write("|  no file uploaded")
+        # st.write("|  no file uploaded")
+        upload_placeholder.write(f"{prefix} no file uploaded")
         return None
-    elif not filename1:
-        st.write("|  file1 not ready")
+
+    if not filename1:
+        # st.write("|  file1 not ready")
+        upload_placeholder.write(f"{prefix}, file1 not ready")
         return None
-    elif not filename2:
-        st.write("|  file2 not ready")
+
+    if not filename2:
+        # st.write("|  file2 not ready")
+        upload_placeholder.write(f"{prefix}, file2 not ready")
         return None
 
     try:
@@ -218,28 +230,28 @@ def home():  # noqa
         # logger.debug("fn.__doc__: %s", fn.__doc__)
         logger.debug("fn.__name__: %s", fn.__name__)
 
-        from inspect import getabsfile
-
-        logger.debug("getabsfile(fn): %s", getabsfile(fn))
-
-        with st.spinner(" diggin..."):
-            then = perf_counter()
-            try:
-                # aset = ezbee/dzbee/debee
-                aset = globals()[state.ns.beetype](
-                    list1,
-                    list2,
-                    # eps=eps,
-                    # min_samples=min_samples,
-                )
-            except Exception as e:
-                # logger.error("aset = ezbee(...) exc: %s", e)
-                logger.exception("aset = globals()[state.ns.beetype](...) exc: %s", e)
-                aset = ""
-                # st.write(e)
-                st.write("Collecting inputs...")
-                return None
-        st.success(f"Done, took {perf_counter() - then:.2f} s")
+        # from inspect import getabsfile
+        # logger.debug("getabsfile(fn): %s", getabsfile(fn))
+
+        # convert to simplified chinese if is_tranditional
+        with about_time() as t:
+            with st.spinner(" diggin..."):
+                try:
+                    # aset = ezbee/dzbee/debee
+                    aset = globals()[state.ns.beetype](
+                        t2s(list1),  # t2s, handle trand.chinese
+                        t2s(list2),
+                        # eps=eps,
+                        # min_samples=min_samples,
+                    )
+                except Exception as e:
+                    # logger.error("aset = ezbee(...) exc: %s", e)
+                    logger.exception("aset = globals()[state.ns.beetype](...) exc: %s", e)
+                    aset = ""
+                    # st.write(e)
+                    st.write("Collecting inputs...")
+                    return None
+        st.success(f"Done, took {t.duration_human}")
 
     else:
         try:
@@ -278,6 +290,8 @@ def home():  # noqa
     _ = st.expander("done aligned")
     with _:
         st.table(df_a.astype(str))
+        # st.markdown(df_a.astype(str).to_markdown())
+        # st.markdown(df_a.astype(str).to_numpy().tolist())
 
     # insert seq no
     df_a.insert(0, "sn", range(len(df_a)))
@@ -308,3 +322,5 @@ def home():  # noqa
             # fit_columns_on_grid_load=True,
             update_mode=GridUpdateMode.MODEL_CHANGED,
         )
+
+    return None

poetry.lock

@@ -1719,6 +1719,17 @@ python-versions = ">=3.6"
 [package.dependencies]
 streamlit = ">=0.63"
 
+[[package]]
+name = "tabulate"
+version = "0.8.9"
+description = "Pretty-print tabular data"
+category = "main"
+optional = false
+python-versions = "*"
+
+[package.extras]
+widechars = ["wcwidth"]
+
 [[package]]
 name = "tenacity"
 version = "8.0.1"
@@ -2016,7 +2027,7 @@ testing = ["pytest (>=6)", "pytest-checkdocs (>=2.4)", "pytest-flake8", "pytest-
 [metadata]
 lock-version = "1.1"
 python-versions = "^3.8.3"
-content-hash = "eabc5b9c944b380d2a60b4ec1b1f218f4b1a3aea1426c7fb75fdc51d4889e57e"
+content-hash = "671c809c9707c753d7aab881b8319beb76c2311edd270ca629f40cfd10227faa"
 
 [metadata.files]
 about-time = [
@@ -3167,6 +3178,10 @@ streamlit-option-menu = [
     {file = "streamlit-option-menu-0.3.2.tar.gz", hash = "sha256:69d1aef6f30f83f29eda3dc9667733bc2e28cd640eb17b4b6ca315f633484c52"},
     {file = "streamlit_option_menu-0.3.2-py3-none-any.whl", hash = "sha256:0b7eae3ffdb0276c81d15750465c72957d57d2f766cb027c586d053519731178"},
 ]
+tabulate = [
+    {file = "tabulate-0.8.9-py3-none-any.whl", hash = "sha256:d7c013fe7abbc5e491394e10fa845f8f32fe54f8dc60c6622c6cf482d25d47e4"},
+    {file = "tabulate-0.8.9.tar.gz", hash = "sha256:eb1d13f25760052e8931f2ef80aaf6045a6cceb47514db8beab24cded16f13a7"},
+]
 tenacity = [
     {file = "tenacity-8.0.1-py3-none-any.whl", hash = "sha256:f78f4ea81b0fabc06728c11dc2a8c01277bfc5181b321a4770471902e3eb844a"},
     {file = "tenacity-8.0.1.tar.gz", hash = "sha256:43242a20e3e73291a28bcbcacfd6e000b02d3857a9a9fff56b297a27afdc932f"},

pyproject.toml

@@ -25,6 +25,7 @@ streamlit-option-menu = "^0.3.2"
 plotly = "^5.8.0"
 hanzidentifier = "^1.0.2"
 opencc-python-reimplemented = "^0.1.6"
+tabulate = "^0.8.9"
 
 [tool.poe.executor]
 type = "poetry"
@@ -32,6 +33,10 @@ type = "poetry"
 [tool.poe.tasks]
 memo = "echo poe test or poetry run poe test"
 test = "pytest tests"
+pyright = "pyright litbee"
+flake8 = "flake8 litbee --ignore F401,E501,F841"
+check = ["pyright", "flake8"]
+export = "poetry export --without-hashes -f requirements.txt -o requirements.txt"
 build = "poetry build"
 _publish = "poetry publish"
 release = ["test", "build", "_publish"]

requirements.txt

@@ -38,6 +38,7 @@ fasttext==0.9.2; python_version >= "3.8" and python_version < "4.0" and python_f
 gitdb==4.0.9; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
 gitpython==3.1.27; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
 grapheme==0.6.0; python_version >= "3.7" and python_version < "4" and python_full_version >= "3.8.3" and python_full_version < "4.0.0"
+hanzidentifier==1.0.2
 icecream==2.1.2
 idna==3.3; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.6"
 importlib-metadata==4.11.4; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
@@ -71,6 +72,7 @@ nest-asyncio==1.5.5; python_full_version >= "3.7.1" and python_version < "4.0" a
 nltk==3.7; python_full_version >= "3.8.3" and python_full_version < "4.0.0" and python_version >= "3.8" and python_version < "4.0"
 notebook==6.4.11; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
 numpy==1.22.4
+opencc-python-reimplemented==0.1.6
 packaging==21.3; python_full_version >= "3.8.3" and python_version < "4.0" and python_version >= "3.7" and python_full_version < "4.0.0"
 pandas==1.4.2; python_full_version >= "3.8.3" and python_version < "4.0" and python_version >= "3.8" and python_full_version < "4.0.0"
 pandocfilters==1.5.0; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
@@ -78,6 +80,7 @@ parso==0.8.3; python_full_version >= "3.7.1" and python_version < "4.0" and pyth
 pexpect==4.8.0; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.8" and sys_platform != "win32"
 pickleshare==0.7.5; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.8"
 pillow==9.1.1; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
+plotly==5.8.0; python_version >= "3.6"
 polyglot==16.7.4; python_full_version >= "3.8.3" and python_full_version < "4.0.0"
 prometheus-client==0.14.1; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
 prompt-toolkit==3.0.29; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.8"
@@ -121,6 +124,8 @@ streamlit-aggrid==0.2.3.post2; python_full_version >= "3.7.1" and python_version
 streamlit-multipage==0.0.18; python_version >= "3.6"
 streamlit-option-menu==0.3.2; python_version >= "3.6"
 streamlit==1.9.2; python_version >= "3.6"
+tabulate==0.8.9
+tenacity==8.0.1; python_version >= "3.6"
 terminado==0.15.0; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
 threadpoolctl==3.1.0; python_full_version >= "3.8.3" and python_full_version < "4.0.0" and python_version < "4.0" and python_version >= "3.8"
 tinycss2==1.1.1; python_full_version >= "3.7.1" and python_version < "4.0" and python_version >= "3.7"
@@ -143,4 +148,5 @@ widgetsnbextension==3.6.0; python_full_version >= "3.7.1" and python_version < "
 win32-setctime==1.1.0; sys_platform == "win32" and python_version >= "3.5"
 word2word==1.0.0; python_full_version >= "3.8.3" and python_full_version < "4.0.0" and python_version >= "3.6"
 xlsxwriter==3.0.3; python_full_version >= "3.8.3" and python_full_version < "4.0.0" and python_version >= "3.4"
+zhon==1.1.5
 zipp==3.8.0; python_full_version >= "3.7.1" and python_version < "3.9" and python_version >= "3.7"

run-poetry-export-requirements.sh

@@ -0,0 +1 @@
+poetry export --without-hashes -f requirements.txt -o requirements.txt