更新雅俗通字典

config_dev_env.py

@@ -22,4 +22,5 @@
 WAIT_TIME = 5  # seconds
 
 # Database
-DATABASE = '.\\Kong_Un.db'
+# DATABASE = '.\\Kong_Un.db'
+DATABASE = '.\\Nga_Siok_Thong_Ji_Tian.db'

m300_查字典標注音.py

@@ -2,7 +2,8 @@
 import sys
 
 import settings
-from p100_tsa_ji_tian import main_run as tsa_ji_tian_tshue_tsu_im
+# from p100_tsa_ji_tian import main_run as tsa_ji_tian_tshue_tsu_im
+from p100_tsa_nga_siok_thong_ji_tian import main_run as tsa_ji_tian_tshue_tsu_im
 
 
 def myfunc(argv):

p100_tsa_nga_siok_thong_ji_tian.py

@@ -0,0 +1,196 @@
+import re
+
+import sqlite3
+import xlwings as xw
+
+# 專案全域常數
+from config_dev_env import DATABASE
+
+def main_run(CONVERT_FILE_NAME):
+    # ==========================================================
+    # 在「漢字注音表」B欄已有台羅拼音，需將之拆分成聲母、韻母、調號
+    # 聲母、韻母、調號，分別存放在 C、D、E 欄
+    # ==========================================================
+
+    # 指定提供來源的【檔案】
+    file_path = CONVERT_FILE_NAME
+    wb = xw.Book(file_path)
+
+    # 指定提供來源的【工作表】；及【總列數】
+    source_sheet = wb.sheets["漢字注音表"]
+    end_of_row_no = (
+        source_sheet.range("A" + str(source_sheet.cells.last_cell.row)).end("up").row
+    )
+    end_of_row_no = int(end_of_row_no) - 1
+    print(f"end_row = {end_of_row_no}")
+
+    # ==========================================================
+    # 備妥程式需使用之工作表
+    # ==========================================================
+    sheet_name_list = [
+        "缺字表",
+        "字庫表",
+    ]
+    # ----------------------------------------------------------
+    # 檢查工作表是否已存在？
+    # 若已存在，則清除工作表內容；
+    # 若不存在，則新增工作表
+    # ----------------------------------------------------------
+    for sheet_name in sheet_name_list:
+        sheet = wb.sheets[sheet_name]
+        try:
+            sheet.select()
+            sheet.clear()
+            continue
+        except Exception as e:
+            # CommandError 的 Exception 發生日，表工作表不存在
+            # 新增程式需使用之工作表
+            print(e)
+            wb.sheets.add(name=sheet_name)
+
+    khiam_ji_piau = wb.sheets["缺字表"]
+    ji_khoo_piau = wb.sheets["字庫表"]
+
+    # ==========================================================
+    # 在「漢字注音表」B欄已有台羅拼音，需將之拆分成聲母、韻母、調號
+    # 聲母、韻母、調號，分別存放在 C、D、E 欄
+    # ==========================================================
+    han_ji_tsu_im_piau = wb.sheets["漢字注音表"]
+    han_ji_tsu_im_piau.select()
+
+    # =========================================================="
+    # 資料庫",
+    # =========================================================="
+    conn = sqlite3.connect(DATABASE)
+    db_cursor = conn.cursor()
+    source_index = 1  # index for source sheet
+    target_index = 1
+    ji_khoo_index = 1
+    khiam_ji_index = 1
+
+    while source_index <= end_of_row_no:
+        print(f"row = {source_index}")
+        # 自 source_sheet 取出一個「欲查注音的漢字」(beh_tshue_tsu_im_e_ji)
+        beh_tshue_tsu_im_e_ji = str(
+            source_sheet.range("A" + str(source_index)).value
+        ).strip()
+
+        # =========================================================
+        # 如是空白或換行，處理換行
+        # =========================================================
+        if beh_tshue_tsu_im_e_ji == " " or beh_tshue_tsu_im_e_ji == "":
+            target_index += 1
+            source_index += 1
+            continue
+        elif beh_tshue_tsu_im_e_ji == "\n":
+            han_ji_tsu_im_piau.range("A" + str(target_index)).value = "\n"
+            target_index += 1
+            source_index += 1
+            continue
+
+        # =========================================================
+        # 若取出之字為標點符號，則跳過，並繼續取下一個漢字。
+        # =========================================================
+        piau_tiam_1 = r"[，、：；．。？！（）「」【】《》“]"
+        piau_tiam_2 = r"[\uFF0C\uFF08-\uFF09\u2013-\u2014\u2026\\u2018-\u201D\u3000\u3001-\u303F\uFE50-\uFE5E]"  # noqa: E501
+        # piau_tiam = r"[\u2013-\u2026\u3000-\u303F\uFE50-\uFF20]"
+        piau_tiam = f"{piau_tiam_1}|{piau_tiam_2}"
+        is_piau_tiam = re.search(piau_tiam, beh_tshue_tsu_im_e_ji, re.M | re.I)
+        if is_piau_tiam:
+            target_index += 1
+            source_index += 1
+            continue
+
+        # =========================================================
+        # 在【字庫】資料庫查找【注音碼】
+        # SQL 查詢指令：自字庫查找某漢字之注音碼
+        # =========================================================
+        # sql = select id, han_ji, tl_im, freq, siann, un, tiau
+        #           from han_ji
+        #           where han_ji='{search_han_ji}'
+        sql = (
+            "SELECT 識別號, 漢字, 切音, 字韻, 聲調, 原始拼音, 舒促聲, 聲, 韻, 調, 拼音碼, 雅俗通標音, 十五音標音, 常用度 "
+            "FROM 雅俗通字庫 "
+            f"WHERE 漢字='{beh_tshue_tsu_im_e_ji}' "
+            "ORDER BY 常用度 DESC;"
+        )
+        db_cursor.execute(sql)
+        ji_e_piau_im = db_cursor.fetchall()
+
+        # =========================================================
+        # 若是查不到漢字的注音碼，在【缺字表】做記錄
+        # =========================================================
+        if not ji_e_piau_im:
+            print(f"Can not find 【{beh_tshue_tsu_im_e_ji}】in Han-Ji-Khoo!!")
+            # 記錄【缺字表】的【列號】
+            khiam_ji_piau.range("A" + str(khiam_ji_index)).value = khiam_ji_index
+            # 記錄【缺字表】的【漢字】
+            khiam_ji_piau.range("B" + str(khiam_ji_index)).value = beh_tshue_tsu_im_e_ji
+            # 記錄【漢字注音表】的【列號】
+            khiam_ji_piau.range("C" + str(khiam_ji_index)).value = source_index
+            khiam_ji_index += 1
+            target_index += 1
+            source_index += 1
+            continue
+
+        # =========================================================
+        # 自【字庫】查到的【漢字】，取出：聲母、韻母、調號
+        # =========================================================
+        piau_im_tsong_soo = len(ji_e_piau_im)
+        han_ji_id = ji_e_piau_im[0][0]
+        tsu_im = ji_e_piau_im[0][10]
+        siann_bu = ji_e_piau_im[0][7]
+        un_bu = ji_e_piau_im[0][8]
+        tiau_ho = ji_e_piau_im[0][9]
+        freq = ji_e_piau_im[0][13]
+
+        # =========================================================
+        # 寫入：【漢字注音表】
+        # =========================================================
+        han_ji_tsu_im_piau.range("B" + str(target_index)).value = tsu_im
+        han_ji_tsu_im_piau.range("C" + str(target_index)).value = siann_bu
+        han_ji_tsu_im_piau.range("D" + str(target_index)).value = un_bu
+        han_ji_tsu_im_piau.range("E" + str(target_index)).value = tiau_ho
+        han_ji_tsu_im_piau.range("F" + str(target_index)).value = piau_im_tsong_soo
+        han_ji_tsu_im_piau.range("G" + str(target_index)).value = freq
+
+        # =========================================================
+        # 若是查到漢字有一個以上的注音碼，在【字庫表】做記錄
+        # ji_khoo_sheet  = wb.sheets["字庫表"]
+        # =========================================================
+        if piau_im_tsong_soo > 1:
+            for piau_im_index in range(piau_im_tsong_soo):
+                han_ji_id = ji_e_piau_im[piau_im_index][0]
+                tsu_im = ji_e_piau_im[piau_im_index][2]
+                freq = ji_e_piau_im[piau_im_index][3]
+                siann_bu = ji_e_piau_im[piau_im_index][4]
+                un_bu = ji_e_piau_im[piau_im_index][5]
+                tiau_ho = ji_e_piau_im[piau_im_index][6]
+
+                # 記錄對映至【漢字注音表】的【列號】
+                ji_khoo_piau.range("A" + str(ji_khoo_index)).value = source_index
+
+                # 記錄【字庫】資料庫的【紀錄識別碼（Record ID of Table）】
+                ji_khoo_piau.range("B" + str(ji_khoo_index)).value = han_ji_id
+
+                ji_khoo_piau.range(
+                    "C" + str(ji_khoo_index)
+                ).value = beh_tshue_tsu_im_e_ji
+                ji_khoo_piau.range("D" + str(ji_khoo_index)).value = tsu_im
+                ji_khoo_piau.range("E" + str(ji_khoo_index)).value = siann_bu
+                ji_khoo_piau.range("F" + str(ji_khoo_index)).value = un_bu
+                ji_khoo_piau.range("G" + str(ji_khoo_index)).value = tiau_ho
+                ji_khoo_piau.range("H" + str(ji_khoo_index)).value = freq
+
+                ji_khoo_index += 1
+
+        # =========================================================
+        # 調整讀取來源；寫入標的各手標
+        # =========================================================
+        target_index += 1
+        source_index += 1
+
+    # ==========================================================
+    # 關閉資料庫
+    # ==========================================================
+    conn.close()

tools/SQLite/Nga-Siok-Thong/聲母對照表.csv

@@ -0,0 +1,85 @@
+識別號,韻母碼,國際音標,白話字,台羅,閩拼,方音,十五音,十五音序,舒促聲,十五音識別碼
+1,un,un,un,un,un,ㄨㄣ,君,1,舒聲,君舒
+2,ut,ut̚,ut,ut,ut,ㄨㆵ,君,1,促聲,君促
+3,ian,ian,ian,ian,ian,ㄧㄢ,堅,2,舒聲,堅舒
+4,iat,iat̚,iat,iat,iat,ㄧㄚㆵ,堅,2,促聲,堅促
+5,im,im,im,im,im,ㄧㆬ,金,3,舒聲,金舒
+6,ip,ip̚,ip,ip,ip,一ㆴ,金,3,促聲,金促
+7,ui,ui,ui,ui,ui,ㄨㄧ,規,4,舒聲,規舒
+8,ee,ɛ,e,ee,e,ㄝ,嘉,5,舒聲,嘉舒
+9,eeh,ɛ?,eeh,eeh,,ㄝㆷ,嘉,5,促聲,嘉促
+10,an,an,an,an,an,ㄢ,干,6,舒聲,干舒
+11,at,at̚,at,at,at,ㄚㆵ,干,6,促聲,干促
+12,ong,ɔŋ,ong,ong,ong,ㆲ,公,7,舒聲,公舒
+13,ok,ɔk̚,ok,ok,ok,ㆦㆻ,公,7,促聲,公促
+14,uai,uai,oai,uai,uai,ㄨㄞ,乖,8,舒聲,乖舒
+15,uaih,uai?,oaih,uaih,uaih,ㄨㄞㆷ,乖,8,促聲,乖促
+16,ing,iŋ,eng,ing,ing,ㄧㄥ,經,9,舒聲,經舒
+17,ik,ik̚,ek,ik,ik,ㄧㆻ,經,9,促聲,經促
+18,uan,uan,oan,uan,uan,ㄨㄢ,觀,10,舒聲,觀舒
+19,uat,uat̚,oat,uat,uat,ㄨㄚㆵ,觀,10,促聲,觀促
+20,oo,ɔu,o͘,oo,oo,ㆦ,沽,11,舒聲,沽舒
+21,iau,iau,iau,iau,iao,ㄧㄠ,嬌,12,舒聲,嬌舒
+22,iauh,iau?,iauh,iauh,iaoh,ㄧㄠㆷ,嬌,12,促聲,嬌促
+23,ei,ei,e,e,e,ㆤ,稽,13,舒聲,稽舒
+24,iong,iɔŋ,iong,iong,iong,ㄧㆲ,恭,14,舒聲,恭舒
+25,iok,iɔk̚,iok,iok,iok,ㄧㆦㆻ,恭,14,促聲,恭促
+26,o,ə,o,o,o,ㄜ,高,15,舒聲,高舒
+27,oh,ə?,oh,oh,oh,ㄜㆷ,高,15,促聲,高促
+28,ai,ai,ai,ai,ai,ㄞ,皆,16,舒聲,皆舒
+29,in,in,in,in,in,ㄧㄣ,巾,17,舒聲,巾舒
+30,it,it̚,it,it,it,ㄧㆵ,巾,17,促聲,巾促
+31,iang,iaŋ,iang,iang,iang,ㄧㄤ,姜,18,舒聲,姜舒
+32,iak,iak̚,iak,iak,iak,ㄧㄚㆻ,姜,18,促聲,姜促
+33,am,am,am,am,am,ㆰ,甘,19,舒聲,甘舒
+34,ap,ap̚,ap,ap,ap,ㄚㆴ,甘,19,促聲,甘促
+35,ua,ua,oa,ua,ua,ㄨㄚ,瓜,20,舒聲,瓜舒
+36,uah,ua?,oah,uah,uah,ㄨㄚㆷ,瓜,20,促聲,瓜促
+37,ang,aŋ,ang,ang,ang,ㄤ,江,21,舒聲,江舒
+38,ak,ak̚,ak,ak,ak,ㄚㆻ,江,21,促聲,江促
+39,iam,iam,iam,iam,iam,ㄧㆰ,兼,22,舒聲,兼舒
+40,iap,iap̚,iap,iap,iap,ㄧㄚㆴ,兼,22,促聲,兼促
+41,au,au,au,au,ao,ㄠ,交,23,舒聲,交舒
+42,auh,au?,auh,auh,aoh,ㄠㆷ,交,23,促聲,交促
+43,ia,ia,ia,ia,ia,ㄧㄚ,迦,24,舒聲,迦舒
+44,iah,ia?,iah,iah,iah,ㄧㄚㆷ,迦,24,促聲,迦促
+45,ue,ue,oe,ue,ue,ㄨㆤ,檜,25,舒聲,檜舒
+46,ueh,ue?,oeh,ueh,ueh,ㄨㆤㆷ,檜,25,促聲,檜促
+47,ann,ã,aⁿ,ann,na,ㆩ,監,26,舒聲,監舒
+48,ahnn,ã?,aⁿh,annh,nah,ㆩㆷ,監,26,促聲,監促
+49,u,u,u,u,u,ㄨ,艍,27,舒聲,艍舒
+50,uh,u?,uh,uh,uh,ㄨㆷ,艍,27,促聲,艍促
+51,a,a,a,a,a,ㄚ,膠,28,舒聲,膠舒
+52,ah,a?,ah,ah,ah,ㄚㆷ,膠,28,促聲,膠促
+53,i,i,i,i,i,ㄧ,居,29,舒聲,居舒
+54,ih,i?,ih,ih,ih,ㄧㆷ,居,29,促聲,居促
+55,iu,iu,iu,iu,iu,ㄧㄨ,丩,30,舒聲,丩舒
+56,enn,ẽ,eⁿ,enn,ne,ㆥ,更,31,舒聲,更舒
+57,ehnn,ẽ?,eⁿh,ennh,neh,ㆥㆷ,更,31,促聲,更促
+58,uinn,uĩ,uiⁿ,uinn,nui,ㄨㆪ,褌,32,舒聲,褌舒
+59,io,iə,io,io,io,ㄧㄜ,茄,33,舒聲,茄舒
+60,ioh,iə?,ioh,ioh,ioh,ㄧㄜㆷ,茄,33,促聲,茄促
+61,inn,ĩ,iⁿ,inn,ni,ㆪ,梔,34,舒聲,梔舒
+62,ihnn,ĩ?,iⁿh,innh,nih,ㆪㆷ,梔,34,促聲,梔促
+63,ionn,ĩɔ̃,ioⁿ,ionn,nioo,ㄧㆧ,薑,35,舒聲,薑舒
+64,iann,iã,iaⁿ,iannh,nia,ㄧㆩ,驚,36,舒聲,驚舒
+65,uann,ũã,oaⁿ,uann,nua,ㄨㆩ,官,37,舒聲,官舒
+66,ng,ŋ̍,ng,ng,ng,ㆭ,鋼,38,舒聲,鋼舒
+67,e,e,e,e,e,ㆤ,伽,39,舒聲,伽舒
+68,eh,e?,eh,eh,eh,ㆤㆷ,伽,39,促聲,伽促
+69,ainn,ãĩ,aiⁿ,ainn,nai,ㆮ,閒,40,舒聲,閒舒
+70,oonn,ɔ̃ũ,oⁿ,onn,noo,ㆧ,姑,41,舒聲,姑舒
+71,m,m̩,m,m,m,ㆬ,姆,42,舒聲,姆舒
+72,uang,uaŋ,oang,uang,uang,ㄨㄤ,光,43,舒聲,光舒
+73,uak,uak̚,oak,uak,uak,ㄨㄚㆻ,光,43,促聲,光促
+74,uainn,uãĩ,oaiⁿ,uainn,nuai,ㄨㆮ,閂,44,舒聲,閂舒
+75,uaihnn,uãĩ?,oaiⁿh,uainnh,nuaih,ㄨㆮㆷ,閂,44,促聲,閂促
+76,uenn,uẽ,oeⁿ,uenn,nue,ㄨㆥ,糜,45,舒聲,糜舒
+77,iaunn,ĩãũ,iauⁿ,iaunn,niao,ㄧㆯ,嘄,46,舒聲,嘄舒
+78,iauhnn,ĩãũ?,iauⁿh,iaunnh,niaoh,ㄧㆯㆷ,嘄,46,促聲,嘄促
+79,om,ɔm,om,om,om,ㆱ,箴,47,舒聲,箴舒
+80,op,ɔp̚,op,op,op,ㆦㆴ,箴,47,促聲,箴促
+81,aunn,ãũ,auⁿ,aunn,nao,ㆯ,爻,48,舒聲,爻舒
+82,onn,õ,oⁿ,onn,noo,ㆧ,扛,49,舒聲,扛舒
+83,ohnn,õh,oⁿh,onnh,nooh,ㆧㆷ,扛,49,促聲,扛促
+84,iunn,iũ,iuⁿ,iunn,niu,ㄧㆫ,牛,50,舒聲,牛舒