From 699cecad4ff178c381f09a16cd01675ce0038ecd Mon Sep 17 00:00:00 2001
From: Valentin CZERYBA <valcze80@gmail.com>
Date: Wed, 28 Jun 2023 23:28:24 +0200
Subject: [PATCH] change with tmp files

---
 import_export_canalblog.py |  9 ++++---
 lib/WPChange.py            | 53 +++++++++++++++++++++++++++++++++++---
 2 files changed, 54 insertions(+), 8 deletions(-)

diff --git a/import_export_canalblog.py b/import_export_canalblog.py
index 3b14d59..a404d16 100644
--- a/import_export_canalblog.py
+++ b/import_export_canalblog.py
@@ -23,9 +23,9 @@ def errorRevert(logger, revert, tmp):
             logger.error("Error revert, because number files tmp is incompatible with parallel number")
             exit(1)
 
-def change(index, number, args, logger):
-    changeWp = WPChange(logger=logger, index_name=index, number_thread=number)
-    changeWp.fromDirectory(args.directory)
+def change(index, number, args, logger, tmp, revert):
+    changeWp = WPChange(logger=logger, index_name=index, number_thread=number, tmp=tmp)
+    changeWp.fromDirectory(args.directory, revert)
 
     del changeWp
 
@@ -333,7 +333,8 @@ if __name__ == '__main__':
         if len(args.directory) > 0:
             try:
                 with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
-                    wait_for = [ ex.submit(change, i, args.parallel, args, logger) for i in range(0, int(args.parallel)) ]
+                    errorRevert(logger, args.revert, args.tmp)
+                    wait_for = [ ex.submit(change, i, args.parallel, args, logger, args.tmp, args.revert) for i in range(0, int(args.parallel)) ]
             except Exception as err:
                 logger.error("Thread error for remove : {0}".format(err))
         if len(args.file) > 0:
diff --git a/lib/WPChange.py b/lib/WPChange.py
index 8d2b626..26ac449 100644
--- a/lib/WPChange.py
+++ b/lib/WPChange.py
@@ -4,11 +4,13 @@ import requests, os, logging, re, json
 
 class WPChange:
     # Constructor
-    def __init__(self, index_name=1, number_thread=1, logger=None, parser="html.parser"):
+    def __init__(self, index_name=1, number_thread=1, logger=None, parser="html.parser", tmp="/tmp/import_export_canablog"):
         self._name = "Thread-{0}".format(index_name)
         self._logger = logger
         self._number_thread = number_thread
         self._parser = parser
+        self._tmp = tmp
+        self._index_name = index_name
     
     # Destructor 
     def __del__(self):
@@ -37,19 +39,62 @@ class WPChange:
                 
 
     ## From directory
-    
-    def fromDirectory(self, directory="", number_thread=1, max_thread=1):
+
+
+    def fromDirectory(self, directory="", revert=False):
+        self._directory = directory
         directory = "{0}/archives".format(directory)
         directories = self._getDirectories([], "{0}".format(directory))
         if len(directories) > 0:
             files = self._getFiles(directories)
-            self.fromFile(files, number_thread, max_thread)
+            if revert is False:
+                self._tmpFiles(files=files, number_thread=self._index_name, max_thread=self._number_thread)
+            self._fromFileTmp()
         else:
             self._logger.error("{0} : No files for {1}".format(self._name, directory))
+    
+
+    def fromFile(self, files=[]):
+        for i in range(0, len(files)):
+            if os.path.exists(files[i]):
+                self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, len(files), files[i]))
+                self._change(files[i])
 
 
     # Private method
 
+    def _fromFileTmp(self):
+        try:
+            with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
+                files = json.loads(file.read())
+                self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(files)))
+                for i in range(0, len(files)):
+                    if os.path.exists(files[i]):
+                        self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, len(files), files[i]))
+                        self._change(files[i])
+        except Exception as ex:
+            self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
+
+
+    def _tmpFiles(self, files=[], number_thread=1, max_thread=1):
+        print()
+        divFiles = int(len(files) / int(max_thread))
+        currentRangeFiles = int(divFiles * (int(number_thread)+1))
+        firstRange = int(currentRangeFiles - divFiles)
+        self._logger.debug("{0} : index : {1}".format(self._name,number_thread))
+
+        self._logger.debug("{0} : first range : {1}".format(self._name,firstRange))
+        self._logger.debug("{0} : last range : {1}".format(self._name,currentRangeFiles))
+        webpage = []
+        for i in range(firstRange, currentRangeFiles):
+            webpage.append(files[i])
+        
+        try:
+            string_webpage = json.dumps(webpage)
+            open("{0}/{1}.json".format(self._tmp, self._name), "wt").write(string_webpage)
+        except Exception as ex:
+            self._logger.error("{0} : Error for writing webpage : {1}".format(self._name, ex))
+
     ## Get all files
 
     def _getFiles(self, item):