From 8349ca9a0016b8bda44a439712c91981bd662640 Mon Sep 17 00:00:00 2001
From: remittor <remittor@gmail.com>
Date: Sat, 24 Sep 2022 12:35:13 +0300
Subject: [PATCH] Add support translate many HTML files (generate sed patch)

---
 data/lang_install.sh   |   4 +
 data/lang_patch.sh     |   4 +
 data/lang_uninstall.sh |   1 +
 install_lang.py        | 161 +++++++++++++++++++++++++++++++++++++++++
 4 files changed, 170 insertions(+)

diff --git a/data/lang_install.sh b/data/lang_install.sh
index b2acd49..fe1e6e4 100644
--- a/data/lang_install.sh
+++ b/data/lang_install.sh
@@ -28,6 +28,10 @@ fi
 mv -f /tmp/base.*.lmo $DIR_PATCH/
 mv -f /tmp/lang_patch.sh $DIR_PATCH/
 chmod +x $DIR_PATCH/lang_patch.sh
+if [ -e "/tmp/lang_patch1.sh" ]; then
+	mv -f /tmp/lang_patch1.sh $DIR_PATCH/
+	chmod +x $DIR_PATCH/lang_patch1.sh
+fi
 
 INSTALL_METHOD=1
 if [ -e "/usr/lib/os-release" ]; then
diff --git a/data/lang_patch.sh b/data/lang_patch.sh
index 1c32f5d..2385c13 100644
--- a/data/lang_patch.sh
+++ b/data/lang_patch.sh
@@ -35,6 +35,10 @@ sed -i 's/ and features\["system"\]\["i18n"\] == "1" //' /usr/lib/lua/luci/view/
 # unlock change luci.main.lang
 sed -i "s/option CHANNEL 'stable'/option CHANNEL 'release'/g" /usr/share/xiaoqiang/xiaoqiang_version
 
+if [ -e "$DIR_PATCH/lang_patch1.sh" ]; then
+	sh $DIR_PATCH/lang_patch1.sh
+fi
+
 echo "lang patched" > /tmp/lang_patch.log
 
 MAIN_LANG=$( uci -q get luci.main.lang )
diff --git a/data/lang_uninstall.sh b/data/lang_uninstall.sh
index 4edc35e..12b9d4b 100644
--- a/data/lang_uninstall.sh
+++ b/data/lang_uninstall.sh
@@ -19,6 +19,7 @@ mv /etc/crontabs/root.new /etc/crontabs/root
 rm -f /etc/rc.lang
 rm -f /etc/lang_patch.sh
 rm -f $DIR_PATCH/lang_patch.sh
+rm -f $DIR_PATCH/lang_patch1.sh
 rm -f $DIR_PATCH/base.*.lmo
 rm -f /tmp/lang_patch.log
 
diff --git a/install_lang.py b/install_lang.py
index 80c40ea..853709e 100644
--- a/install_lang.py
+++ b/install_lang.py
@@ -3,12 +3,122 @@
 
 import os
 import sys
+import types
+import re
 
 sys.path.append(os.path.dirname(os.path.abspath(__file__)))
+import ssh2
 import gateway
 from gateway import die
 
 
+class www_lmo():
+  fn_local = None
+  fn_remote = None
+  data = None
+  out = []
+  
+  def __init__(w, fn_local = None, fn_remote = None):
+    w.fn_local = fn_local
+    w.fn_remote = fn_remote
+    w.data = None
+    w.out = []    
+
+  def load_file(w):
+    with open(w.fn_local, "r", encoding="utf-8") as file:
+      w.data = file.read()
+  
+  def parse(w):
+    if not w.data:
+      w.load_file()
+    p = re.compile(r'[^%][>]([^><]*?[\u4e00-\u9fff][^><]*?)[<][^%]')
+    w.parse1(p)
+    p = re.compile(r'[\']([^><\n\']*?[\u4e00-\u9fff][^><\n\']*?)[\']')
+    w.parse1(p)
+    p = re.compile(r'["]([^><\n"]*?[\u4e00-\u9fff][^><\n"]*?)["]')
+    w.parse1(p)
+
+  def parse1(w, template):
+    for m in template.finditer(w.data):
+      txt = m.group()
+      if txt.find('<%') > 0 or txt.find('%>') > 0:
+        continue
+      if txt.find('// ') > 0:  # skip comments
+        continue
+      if txt.find('if ') > 0:  # skip code
+        continue
+      if txt.find('{') > 0:    # skip code
+        continue
+      if txt.find(' = ') > 0:  # skip code
+        continue
+      b = m.start()
+      t1 = m.start(1)
+      t2 = m.end(1)
+      e = m.end()
+      #print(b, t1, t2, e)
+      prefix = w.data[b:t1]
+      string = w.data[t1:t2]
+      postfix = w.data[t2:e]
+      if len(prefix) == 2 and prefix[1] == '>':
+        prefix = prefix[1]
+        txt = txt[1:]
+      if len(postfix) == 2 and postfix[0] == '<':
+        postfix = postfix[0]
+        txt = txt[:-1]
+      #print('"{}" "{}" "{}"'.format(prefix, string, postfix))
+      s = string.strip()
+      if s == string:
+        out = prefix + '<%:' + string + '%>' + postfix
+        msg = string
+      else:
+        p = string.find(s)
+        if p < 0:
+          continue  # fixme
+        out = prefix + string[:p] + '<%:' + s + '%>' + string[p+len(s):] + postfix
+        msg = s
+      #print(b, out)
+      # check for dup
+      dup = 0
+      for i, v in enumerate(w.out):
+        if v.txt_orig == txt:
+          dup = 1
+          break
+      if dup == 0:
+        v = types.SimpleNamespace()
+        v.pos = b
+        v.txt_orig = txt
+        v.txt_new = out
+        v.sed = ""
+        v.msg = msg
+        w.out.append(v)
+
+  def sed_escape(w, txt):
+    txt = txt.replace('\\', '\\\\')
+    txt = txt.replace('\n', r'\n')
+    txt = txt.replace('\r', r'\r')
+    txt = txt.replace('\t', r'\t')
+    txt = txt.replace("'", r"'\''")
+    txt = txt.replace('[', r'\[')
+    txt = txt.replace(']', r'\]')
+    txt = txt.replace('$', r'\$')
+    txt = txt.replace('*', r'\*')
+    txt = txt.replace('.', r'\.')
+    txt = txt.replace('$', r'\$')
+    txt = txt.replace('^', r'\^')
+    txt = txt.replace('/', r'\/')
+    return txt
+
+  def gen_sed(w):
+    for i, v in enumerate(w.out):
+      #print(v.pos, v.txt_new)
+      orig = w.sed_escape(v.txt_orig)
+      new  = w.sed_escape(v.txt_new)
+      prefix = ''
+      if v.txt_orig.find('\n') > 0:
+        prefix = ':a;N;$!ba;'  # see: https://stackoverflow.com/questions/1251999/how-can-i-replace-each-newline-n-with-a-space-using-sed
+      v.sed = "sed -i '{}s/{}/{}/g' {}".format(prefix, orig, new, w.fn_remote) 
+
+
 gw = gateway.Gateway()
 
 fn_dir      = 'data/'
@@ -18,6 +128,8 @@ fn_local_i  = 'data/lang_install.sh'
 fn_remote_i = '/tmp/lang_install.sh'
 fn_local_u  = 'data/lang_uninstall.sh'
 fn_remote_u = '/tmp/lang_uninstall.sh'
+fn_www_local  = 'data/lang_patch1.sh'
+fn_www_remote = '/tmp/lang_patch1.sh'
 
 action = 'install'
 if len(sys.argv) > 1:
@@ -30,6 +142,19 @@ if action == 'install':
 
 gw.upload(fn_local_u, fn_remote_u)
 
+if action == 'install':
+  patch1_installed = True
+  fn = 'data/lang_patch.log'
+  if os.path.exists(fn):
+    os.remove(fn)
+  try:
+    gw.download('/tmp/lang_patch.log', fn, verbose = 0)
+  except ssh2.exceptions.SCPProtocolError:
+    patch1_installed = False
+  if patch1_installed:
+    print("Uninstall lang_patch...")
+    gw.run_cmd("sh " + fn_remote_u)
+
 if action == 'install':
   import po2lmo
   for filename in [fn for fn in os.listdir(fn_dir) if fn.split(".")[-1] in ['po']]:
@@ -42,6 +167,41 @@ if action == 'install':
     lmo.save_to_bin(fn_dir + lmo_fname)
     gw.upload(fn_dir + lmo_fname, '/tmp/' + lmo_fname)
 
+  dn_www = "tmp/www"
+  os.makedirs(dn_www, exist_ok = True)
+  wwwlst = [ "/usr/lib/lua/luci/view/web/index.htm",
+             "/usr/lib/lua/luci/view/web/apindex.htm",
+             "/usr/lib/lua/luci/view/web/inc/g.js.htm",
+             "/usr/lib/lua/luci/view/web/inc/header.htm",
+             "/usr/lib/lua/luci/view/web/inc/sysinfo.htm",
+             "/usr/lib/lua/luci/view/web/inc/wanCheck.js.htm",
+             "/usr/lib/lua/luci/view/web/setting/iptv.htm",
+           ]  
+  www = []
+  for i, www_remote in enumerate(wwwlst):
+    www_local = dn_www + '/' + www_remote.replace('/', '_')
+    try:
+      gw.download(www_remote, www_local, verbose = 0)
+    except ssh2.exceptions.SCPProtocolError:
+      print('WARN: file "{}" not found'.format(www_remote))
+      continue
+    w = www_lmo(www_local, www_remote)
+    www.append(w)
+  if os.path.exists(fn_www_local):
+    os.remove(fn_www_local)  
+  file = open(fn_www_local, "wt", encoding='UTF-8', newline = "\n")
+  file.write('#!/bin/sh\n')
+  for i, w in enumerate(www):
+    #print("===== FILE:", w.fn_remote)
+    w.parse()
+    w.gen_sed()
+    file.write('# ======= FILE: {} ======= \n'.format(w.fn_remote))
+    for i, v in enumerate(w.out):
+      if v.sed:
+        file.write(v.sed + '\n')
+  file.close()
+  gw.upload(fn_www_local, fn_www_remote)
+  
 print("All files uploaded!")
 
 print("Run scripts...")
@@ -53,5 +213,6 @@ else:
 gw.run_cmd("rm -f " + fn_remote)
 gw.run_cmd("rm -f " + fn_remote_i)
 gw.run_cmd("rm -f " + fn_remote_u)
+gw.run_cmd("rm -f " + fn_www_remote)
 
 print("Ready! The language files are installed.")