z

utils/script.py

@@ -1,7 +1,7 @@
 def process_text(text, nazwisko_user, nazwisko_system):
-    lines = text.split('\n')
+    lines = text.split('\n' )
     result=[]
-    for line in lines:
+    for line in lines :
         if nazwisko_system in line:
             result.append("system " + line.split(nazwisko_system, 1)[1].strip())
         elif nazwisko_user in line:
@@ -9,55 +9,50 @@ def process_text(text, nazwisko_user, nazwisko_system):
         else:
             result.append(line)
     return '\n'.join(result)
-def join_lines(text):
+def join_lines(text ):
-    lines = text.split("\n")
+    lines = text.split("\n" )
-    joined_lines = [lines[0]]
+    joined_lines = [lines[0] ]
-
+    for line in lines[1:] :
-    for line in lines[1:]:
+        if line.startswith("[") :
-        if line.startswith("["):
+            joined_lines.append(line )
-            joined_lines.append(line)
+        else :
-        else:
             joined_lines[-1] += " " +  line
-
+    return "\n".join(joined_lines )
-    return "\n".join(joined_lines)
 def remove_empty_lines(text):
-    lines = text.split('\n')
+    lines = text.split('\n' )
-    non_empty_lines = []
+    non_empty_lines=[]
-    for line in lines:
+    for line in lines :
         if line.strip():
             non_empty_lines.append(line)
-
     return '\n'.join(non_empty_lines)
-
+def read_tsv(file_path ):
-def read_tsv(file_path):
+    with open(file_path, 'r', encoding='utf-8') as tsv_file :
-    with open(file_path, 'r', encoding='utf-8') as tsv_file:
+        content =tsv_file.read()
-        content = tsv_file.read()
     return  content
-
+def write_tsv(file_path, content ):
-def write_tsv(file_path, content):
+    with open(file_path, 'w', encoding='utf-8') as tsv_file :
-    with open(file_path, 'w', encoding='utf-8') as tsv_file:
+        tsv_file.write(content )
-        tsv_file.write(content)
+def replace_first_space_with_tab(text) :
-def replace_first_space_with_tab(text):
+    lines = text.split("\n" )
-    lines = text.split("\n")
+    new_lines =[]
-    new_lines = []
+    for line in lines :
-    for line in lines:
+        first_space_index = line.find(" " )
-        first_space_index = line.find(" ")
+        if first_space_index != -1 :
-        if first_space_index != -1:
+            new_line = line[:first_space_index] + "\t" + line[first_space_index + 1: ]
-            new_line = line[:first_space_index] + "\t" + line[first_space_index + 1:]
+            new_lines.append(new_line )
-            new_lines.append(new_line)
+        else :
-        else:
+            new_lines.append(line )
-            new_lines.append(line)
+    return "\n".join(new_lines )
-    return "\n".join(new_lines)
+if __name__ == '__main__' :
-if __name__ == '__main__':
+    # wypelnic przed uruchomieniemm
-    # wypelnic przed uruchomieniem
+    nazwisko_user =""
-    nazwisko_user = ""
+    nazwisko_system= ""
-    nazwisko_system = ""
+    numer_user =""
-    numer_user = ""
+    numer_system= ""
-    numer_system = ""
+    file_path =f"J:\PycharmProjects\systemy_dialogowe\data\dialog-{numer_system}-{numer_user}-01.tsv"
-    file_path = f"J:\PycharmProjects\systemy_dialogowe\data\dialog-{numer_system}-{numer_user}-01.tsv"
+    text= read_tsv(file_path )
-    text = read_tsv(file_path )
+    text= remove_empty_lines(text )
-    text = remove_empty_lines(text )
+    text= join_lines(text )
-    text = join_lines(text )
+    text= process_text(text, nazwisko_user, nazwisko_system )
-    text = process_text(text, nazwisko_user, nazwisko_system )
+    text= replace_first_space_with_tab(text )
-    text = replace_first_space_with_tab(text )
     write_tsv(file_path, text  )