Mercurial Repositories > eric / file comparison
comparison: eric6/WebBrowser/Bookmarks/NsHtmlReader.py
eric6/WebBrowser/Bookmarks/NsHtmlReader.py
- changeset 7775
- 4a1db75550bd
- parent 7360
- 9190402e4505
- child 7923
- 91e843545d9a
equal
deleted
inserted
replaced
| 7774:9eed155411f0 | 7775:4a1db75550bd |
|---|---|
| 5 | 5 |
| 6 """ | 6 """ |
| 7 Module implementing a class to read Netscape HTML bookmark files. | 7 Module implementing a class to read Netscape HTML bookmark files. |
| 8 """ | 8 """ |
| 9 | 9 |
| 10 import re | |
| 10 | 11 |
| 11 from PyQt5.QtCore import QObject, QIODevice, QFile, QRegExp, Qt, QDateTime | 12 from PyQt5.QtCore import QObject, QIODevice, QFile, QDateTime |
| 12 | 13 |
| 13 from .BookmarkNode import BookmarkNode | 14 from .BookmarkNode import BookmarkNode |
| 14 | 15 |
| 15 import Utilities | 16 import Utilities |
| 16 | 17 |
| 25 """ | 26 """ |
| 26 Constructor | 27 Constructor |
| 27 """ | 28 """ |
| 28 super(NsHtmlReader, self).__init__() | 29 super(NsHtmlReader, self).__init__() |
| 29 | 30 |
| 30 self.__folderRx = QRegExp("<DT><H3(.*)>(.*)</H3>", Qt.CaseInsensitive) | 31 self.__folderRx = re.compile("<DT><H3(.*?)>(.*?)</H3>", re.IGNORECASE) |
| 31 self.__folderRx.setMinimal(True) | 32 self.__endFolderRx = re.compile("</DL>", re.IGNORECASE) |
| 32 | 33 self.__bookmarkRx = re.compile("<DT><A(.*?)>(.*?)</A>", re.IGNORECASE) |
| 33 self.__endFolderRx = QRegExp("</DL>", Qt.CaseInsensitive) | 34 self.__descRx = re.compile("<DD>(.*)", re.IGNORECASE) |
| 34 | 35 self.__separatorRx = re.compile("<HR>", re.IGNORECASE) |
| 35 self.__bookmarkRx = QRegExp("<DT><A(.*)>(.*)</A>", Qt.CaseInsensitive) | 36 self.__urlRx = re.compile('HREF="(.*?)"', re.IGNORECASE) |
| 36 self.__bookmarkRx.setMinimal(True) | 37 self.__addedRx = re.compile(r'ADD_DATE="(\d*?)"', re.IGNORECASE) |
| 37 | 38 self.__modifiedRx = re.compile(r'LAST_MODIFIED="(\d*?)"', |
| 38 self.__descRx = QRegExp("<DD>(.*)", Qt.CaseInsensitive) | 39 re.IGNORECASE) |
| 39 | 40 self.__visitedRx = re.compile(r'LAST_VISIT="(\d*?)"', re.IGNORECASE) |
| 40 self.__separatorRx = QRegExp("<HR>", Qt.CaseInsensitive) | 41 self.__foldedRx = re.compile("FOLDED", re.IGNORECASE) |
| 41 | |
| 42 self.__urlRx = QRegExp('HREF="(.*)"', Qt.CaseInsensitive) | |
| 43 self.__urlRx.setMinimal(True) | |
| 44 | |
| 45 self.__addedRx = QRegExp(r'ADD_DATE="(\d*)"', Qt.CaseInsensitive) | |
| 46 self.__addedRx.setMinimal(True) | |
| 47 | |
| 48 self.__modifiedRx = QRegExp( | |
| 49 r'LAST_MODIFIED="(\d*)"', Qt.CaseInsensitive) | |
| 50 self.__modifiedRx.setMinimal(True) | |
| 51 | |
| 52 self.__visitedRx = QRegExp(r'LAST_VISIT="(\d*)"', Qt.CaseInsensitive) | |
| 53 self.__visitedRx.setMinimal(True) | |
| 54 | |
| 55 self.__foldedRx = QRegExp("FOLDED", Qt.CaseInsensitive) | |
| 56 | 42 |
| 57 def read(self, fileNameOrDevice): | 43 def read(self, fileNameOrDevice): |
| 58 """ | 44 """ |
| 59 Public method to read a Netscape HTML bookmark file. | 45 Public method to read a Netscape HTML bookmark file. |
| 60 | 46 |
| 77 root = BookmarkNode(BookmarkNode.Root) | 63 root = BookmarkNode(BookmarkNode.Root) |
| 78 folders.append(root) | 64 folders.append(root) |
| 79 | 65 |
| 80 while not dev.atEnd(): | 66 while not dev.atEnd(): |
| 81 line = str(dev.readLine(), encoding="utf-8").rstrip() | 67 line = str(dev.readLine(), encoding="utf-8").rstrip() |
| 82 if self.__folderRx.indexIn(line) != -1: | 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: | |
| 83 # folder definition | 79 # folder definition |
| 84 arguments = self.__folderRx.cap(1) | 80 arguments = match.group(1) |
| 85 name = self.__folderRx.cap(2) | 81 name = match.group(2) |
| 86 node = BookmarkNode(BookmarkNode.Folder, folders[-1]) | 82 node = BookmarkNode(BookmarkNode.Folder, folders[-1]) |
| 87 node.title = Utilities.html_udecode(name) | 83 node.title = Utilities.html_udecode(name) |
| 88 node.expanded = self.__foldedRx.indexIn(arguments) == -1 | 84 node.expanded = self.__foldedRx.search(arguments) is None |
| 89 if self.__addedRx.indexIn(arguments) != -1: | 85 addedMatch = self.__addedRx.search(arguments) |
| 86 if addedMatch is not None: | |
| 90 node.added = QDateTime.fromTime_t( | 87 node.added = QDateTime.fromTime_t( |
| 91 int(self.__addedRx.cap(1))) | 88 int(addedMatch.group(1))) |
| 92 folders.append(node) | 89 folders.append(node) |
| 93 lastNode = node | 90 lastNode = node |
| 94 | 91 |
| 95 elif self.__endFolderRx.indexIn(line) != -1: | 92 elif match.re is self.__endFolderRx: |
| 96 # end of folder definition | 93 # end of folder definition |
| 97 folders.pop() | 94 folders.pop() |
| 98 | 95 |
| 99 elif self.__bookmarkRx.indexIn(line) != -1: | 96 elif match.re is self.__bookmarkRx: |
| 100 # bookmark definition | 97 # bookmark definition |
| 101 arguments = self.__bookmarkRx.cap(1) | 98 arguments = match.group(1) |
| 102 name = self.__bookmarkRx.cap(2) | 99 name = match.group(2) |
| 103 node = BookmarkNode(BookmarkNode.Bookmark, folders[-1]) | 100 node = BookmarkNode(BookmarkNode.Bookmark, folders[-1]) |
| 104 node.title = Utilities.html_udecode(name) | 101 node.title = Utilities.html_udecode(name) |
| 105 if self.__urlRx.indexIn(arguments) != -1: | 102 match1 = self.__urlRx.search(arguments) |
| 106 node.url = self.__urlRx.cap(1) | 103 if match1 is not None: |
| 107 if self.__addedRx.indexIn(arguments) != -1: | 104 node.url = match1.group(1) |
| 105 match1 = self.__addedRx.search(arguments) | |
| 106 if match1 is not None: | |
| 108 node.added = QDateTime.fromTime_t( | 107 node.added = QDateTime.fromTime_t( |
| 109 int(self.__addedRx.cap(1))) | 108 int(match1.group(1))) |
| 110 if self.__modifiedRx.indexIn(arguments) != -1: | 109 match1 = self.__modifiedRx.search(arguments) |
| 110 if match1 is not None: | |
| 111 node.modified = QDateTime.fromTime_t( | 111 node.modified = QDateTime.fromTime_t( |
| 112 int(self.__modifiedRx.cap(1))) | 112 int(match1.group(1))) |
| 113 if self.__visitedRx.indexIn(arguments) != -1: | 113 match1 = self.__visitedRx.search(arguments) |
| 114 if match1 is not None: | |
| 114 node.visited = QDateTime.fromTime_t( | 115 node.visited = QDateTime.fromTime_t( |
| 115 int(self.__visitedRx.cap(1))) | 116 int(match1.group(1))) |
| 116 lastNode = node | 117 lastNode = node |
| 117 | 118 |
| 118 elif self.__descRx.indexIn(line) != -1: | 119 elif match.re is self.__descRx: |
| 119 # description | 120 # description |
| 120 if lastNode: | 121 if lastNode: |
| 121 lastNode.desc = Utilities.html_udecode( | 122 lastNode.desc = Utilities.html_udecode( |
| 122 self.__descRx.cap(1)) | 123 match.group(1)) |
| 123 | 124 |
| 124 elif self.__separatorRx.indexIn(line) != -1: | 125 elif match.re is self.__separatorRx: |
| 125 # separator definition | 126 # separator definition |
| 126 BookmarkNode(BookmarkNode.Separator, folders[-1]) | 127 BookmarkNode(BookmarkNode.Separator, folders[-1]) |
| 127 | 128 |
| 128 return root | 129 return root |
