Mercurial Repositories > eric / file comparison
comparison: src/eric7/WebBrowser/Bookmarks/NsHtmlReader.py
src/eric7/WebBrowser/Bookmarks/NsHtmlReader.py
- branch
- eric7
- changeset 9209
- b99e7fd55fd3
- parent 8881
- 54e42bc2437a
- child 9221
- bf71ee032bb4
equal
deleted
inserted
replaced
| 9208:3fc8dfeb6ebe | 9209:b99e7fd55fd3 |
|---|---|
| 1 # -*- coding: utf-8 -*- | |
| 2 | |
| 3 # Copyright (c) 2012 - 2022 Detlev Offenbach <detlev@die-offenbachs.de> | |
| 4 # | |
| 5 | |
| 6 """ | |
| 7 Module implementing a class to read Netscape HTML bookmark files. | |
| 8 """ | |
| 9 | |
| 10 import re | |
| 11 | |
| 12 from PyQt6.QtCore import QObject, QIODevice, QFile, QDateTime | |
| 13 | |
| 14 from .BookmarkNode import BookmarkNode | |
| 15 | |
| 16 import Utilities | |
| 17 | |
| 18 | |
| 19 class NsHtmlReader(QObject): | |
| 20 """ | |
| 21 Class implementing a reader object for Netscape HTML bookmark files. | |
| 22 """ | |
| 23 indentSize = 4 | |
| 24 | |
| 25 def __init__(self): | |
| 26 """ | |
| 27 Constructor | |
| 28 """ | |
| 29 super().__init__() | |
| 30 | |
| 31 self.__folderRx = re.compile("<DT><H3(.*?)>(.*?)</H3>", re.IGNORECASE) | |
| 32 self.__endFolderRx = re.compile("</DL>", re.IGNORECASE) | |
| 33 self.__bookmarkRx = re.compile("<DT><A(.*?)>(.*?)</A>", re.IGNORECASE) | |
| 34 self.__descRx = re.compile("<DD>(.*)", re.IGNORECASE) | |
| 35 self.__separatorRx = re.compile("<HR>", re.IGNORECASE) | |
| 36 self.__urlRx = re.compile('HREF="(.*?)"', re.IGNORECASE) | |
| 37 self.__addedRx = re.compile(r'ADD_DATE="(\d*?)"', re.IGNORECASE) | |
| 38 self.__modifiedRx = re.compile(r'LAST_MODIFIED="(\d*?)"', | |
| 39 re.IGNORECASE) | |
| 40 self.__visitedRx = re.compile(r'LAST_VISIT="(\d*?)"', re.IGNORECASE) | |
| 41 self.__foldedRx = re.compile("FOLDED", re.IGNORECASE) | |
| 42 | |
| 43 def read(self, fileNameOrDevice): | |
| 44 """ | |
| 45 Public method to read a Netscape HTML bookmark file. | |
| 46 | |
| 47 @param fileNameOrDevice name of the file to read (string) | |
| 48 or reference to the device to read (QIODevice) | |
| 49 @return reference to the root node (BookmarkNode) | |
| 50 """ | |
| 51 if isinstance(fileNameOrDevice, QIODevice): | |
| 52 dev = fileNameOrDevice | |
| 53 else: | |
| 54 f = QFile(fileNameOrDevice) | |
| 55 if not f.exists(): | |
| 56 return BookmarkNode(BookmarkNode.Root) | |
| 57 f.open(QIODevice.OpenModeFlag.ReadOnly) | |
| 58 dev = f | |
| 59 | |
| 60 folders = [] | |
| 61 lastNode = None | |
| 62 | |
| 63 root = BookmarkNode(BookmarkNode.Root) | |
| 64 folders.append(root) | |
| 65 | |
| 66 while not dev.atEnd(): | |
| 67 line = str(dev.readLine(), encoding="utf-8").rstrip() | |
| 68 match = ( | |
| 69 self.__folderRx.search(line) or | |
| 70 self.__endFolderRx.search(line) or | |
| 71 self.__bookmarkRx.search(line) or | |
| 72 self.__descRx.search(line) or | |
| 73 self.__separatorRx.search(line) | |
| 74 ) | |
| 75 if match is None: | |
| 76 continue | |
| 77 | |
| 78 if match.re is self.__folderRx: | |
| 79 # folder definition | |
| 80 arguments = match.group(1) | |
| 81 name = match.group(2) | |
| 82 node = BookmarkNode(BookmarkNode.Folder, folders[-1]) | |
| 83 node.title = Utilities.html_udecode(name) | |
| 84 node.expanded = self.__foldedRx.search(arguments) is None | |
| 85 addedMatch = self.__addedRx.search(arguments) | |
| 86 if addedMatch is not None: | |
| 87 node.added = QDateTime.fromSecsSinceEpoch( | |
| 88 int(addedMatch.group(1))) | |
| 89 folders.append(node) | |
| 90 lastNode = node | |
| 91 | |
| 92 elif match.re is self.__endFolderRx: | |
| 93 # end of folder definition | |
| 94 folders.pop() | |
| 95 | |
| 96 elif match.re is self.__bookmarkRx: | |
| 97 # bookmark definition | |
| 98 arguments = match.group(1) | |
| 99 name = match.group(2) | |
| 100 node = BookmarkNode(BookmarkNode.Bookmark, folders[-1]) | |
| 101 node.title = Utilities.html_udecode(name) | |
| 102 match1 = self.__urlRx.search(arguments) | |
| 103 if match1 is not None: | |
| 104 node.url = match1.group(1) | |
| 105 match1 = self.__addedRx.search(arguments) | |
| 106 if match1 is not None: | |
| 107 node.added = QDateTime.fromSecsSinceEpoch( | |
| 108 int(match1.group(1))) | |
| 109 match1 = self.__modifiedRx.search(arguments) | |
| 110 if match1 is not None: | |
| 111 node.modified = QDateTime.fromSecsSinceEpoch( | |
| 112 int(match1.group(1))) | |
| 113 match1 = self.__visitedRx.search(arguments) | |
| 114 if match1 is not None: | |
| 115 node.visited = QDateTime.fromSecsSinceEpoch( | |
| 116 int(match1.group(1))) | |
| 117 lastNode = node | |
| 118 | |
| 119 elif match.re is self.__descRx: | |
| 120 # description | |
| 121 if lastNode: | |
| 122 lastNode.desc = Utilities.html_udecode( | |
| 123 match.group(1)) | |
| 124 | |
| 125 elif match.re is self.__separatorRx: | |
| 126 # separator definition | |
| 127 BookmarkNode(BookmarkNode.Separator, folders[-1]) | |
| 128 | |
| 129 return root |
