import html

This commit is contained in:
life
2016-12-28 15:25:27 +08:00
parent 9e969142f0
commit 1bc8dc21d9
4 changed files with 443 additions and 68 deletions

100
node_modules/file.js generated vendored
View File

@@ -75,80 +75,44 @@ var File = {
callback && callback(attach);
});
}
/*
// var buf = new Buffer(data, 'utf-8');
// 读出来
var err = fs.writeFile(filePath, data, 'ascii');
if(err) {
console.log(err);
callback(false);
return;
}
*/
/*
var exec = require('child_process').exec,
last = exec('"/Users/life/Documents/kuaipan/go/go-study/bin/file2" "' + txtPath + '" "' + filePath + '"');
// last = exec('ls /');
// last.stdout.on('data', function (data) {
// console.log('标准输出:' + data);
// });
last.on('exit', function (code) {
console.log('子进程已关闭,代码:' + code);
fs.unlink(txtPath);
if(!code) {
if(isImage) {
me._addImage(Common.objectId(), filePath, function(newImg) {
newImg.IsImage = true;
callback && callback(newImg);
});
} else {
me._addAttach(filePath, fileTitle, callback)
}
} else {
callback(false);
}
});
*/
/*
// console.log(txtPath);
// console.log(filePath);
// 先写到txt中去, 然后调用命令将.txt -> img
var txtPath = filePath + '.txt';
var err = fs.writeFileSync(txtPath, data);
Common.cmd([txtPath, filePath], function(code) {
// console.log('子进程已关闭,代码:' + code);
// fs.unlink(txtPath);
if(!code) {
try {
if(isImage) {
me._addImage(Common.objectId(), filePath, function(newImg) {
newImg.IsImage = true;
callback && callback(newImg);
});
} else {
me._addAttach(filePath, fileTitle, callback)
}
} catch(e) {
console.log(e);
callback(false);
}
} else {
callback(false);
}
});
*/
// return callback && callback({FileId: Common.objectId(), IsImage: true});
} catch(e) {
console.log("error!!!");
console.error(e);
}
},
copyFile: function(originPath, isImage, callback) {
var me = this;
var basePath;
if(isImage) {
basePath = User.getCurUserImagesPath();
} else {
basePath = User.getCurUserAttachsPath();
}
var filePathAttr = Common.splitFile(originPath);
var fileId = Common.objectId();
var newFilename = fileId + '_html_' + '.' + filePathAttr.ext;
var newFilePath = basePath + '/' + newFilename;
Common.copyFile(originPath, newFilePath, function(ret) {
if(ret) {
if(isImage) {
me._addImage(fileId, newFilePath, function(newImg) {
newImg.IsImage = true;
callback(newImg);
});
} else {
me._addAttach(newFilePath, '', function(attach) {
callback(attach);
});
}
} else {
callback(false);
}
});
},
getFileBase64: function(filePath) {
try {
// read binary data

View File

@@ -4,6 +4,7 @@ var Config = {
"template",
"import_leanote",
"import_evernote",
"import_html",
"export_pdf",
"export_html",
"export_leanote",

View File

@@ -0,0 +1,185 @@
var fs = require('fs');
var Evt = require('evt');
var File = require('file');
var Note = require('note');
var Web = require('web');
var Tag = require('tag');
var async = require('async');
var Common = require('common');
var path = require('path');
var resanitize = require('resanitize');
var Import = {
/*
*/
// callback 是全局的
// eachFileCallback是每一个文件的
// eachNoteFileCallback是每一个笔记的
// filePaths = []
importFromHTML: function(notebookId, filePaths, callback,
eachFileCallback,
eachNoteCallback) {
var me = this;
// var filePaths = filePaths.split(';');
//
var filePaths = filePaths || [];
async.eachSeries(filePaths, function(path, cb) {
try {
var data = fs.readFileSync(path, 'utf-8');
me.parseHTML(notebookId, data, path, function(ret) {
// 单个文件完成
eachFileCallback(ret, path);
cb();
},
// 单个笔记
function(ret) {
eachNoteCallback(ret);
});
} catch(e) {
cb();
return false;
}
}, function() {
// 全部完成
callback(true);
});
},
fixContent: function (content) {
// srip unsage attrs
var unsafeAttrs = ['id', , /on\w+/i, /data-\w+/i, 'clear', 'target'];
content = content.replace(/<([^ >]+?) [^>]*?>/g, resanitize.filterTag(resanitize.stripAttrs(unsafeAttrs)));
// strip unsafe tags
content = resanitize.stripUnsafeTags(content,
['wbr','style', 'comment', 'plaintext', 'xmp', 'listing',
'applet','base','basefont','bgsound','blink','body','button','dir','embed','fieldset','frameset','head',
'html','iframe','ilayer','input','isindex','layer','legend','link','marquee','menu','meta','noframes',
'noscript','object','optgroup','option','param','plaintext','script','select','textarea','xml']
);
return content;
},
getTitle: function (htmlData, filename) {
// 1. 在<title></title> header中找
try {
var title = htmlData.match(/<title[^>]*>([\s\S]*?)<\/title>/i)[1];
if (title && title.replace(/(^\s*)|(\s*$)/g, "")) {
return title;
}
} catch(e) {
}
// 2. 通过文件名获取
var pathInfo = Api.commonService.splitFile(filename);
return pathInfo.nameNotExt;
},
parseHTML: function (notebookId, htmlData, filename, callback, eachCallback) {
var me = this;
try {
var body = htmlData.match(/<body[^>]*>([\s\S]*?)<\/body>/i)[1];
if (!body) {
console.log('没有body');
return callback(false);
}
var title = this.getTitle(htmlData, filename) || getMsg('Untitled');
// 解析里面的图片
var reg = /<img[^>]*?( src=(?:["'])([^>]+?)(?:["']))[^>]*?>/gi;
var ret;
var imagePaths = [];
while(ret = reg.exec(body)) {
console.log(ret);
/*
"<img id="wiz_todo_1429010269089_521433" class="wiz-todo-img wiz-img-cannot-drag" state="unchecked" _src="wiz测试笔记_files/unchecked.png" src="wiz测试笔记_files/unchecked.png" />"
" src="wiz测试笔记_files/unchecked.png""
"wiz测试笔记_files/unchecked.png"
*/
var imagePath = ret[2];
var imagePathLower = imagePath.toLowerCase();
// 是http图片,skip
if (imagePathLower.indexOf('http://') < 0 && imagePathLower.indexOf('https://') < 0) {
imagePaths.push(imagePath);
}
}
var dirname = path.dirname(filename);
var imagePath2ImageInfo = {}; // imagePath => imageInfo
async.eachSeries(imagePaths, function(imagePath, cb) {
// 重复图片
if (imagePath2ImageInfo[imagePath]) {
return cb();
}
var absImagePath;
// 绝对路径
if (imagePath.indexOf('file://') == 0) {
absImagePath = imagePath.substr('file://'.length);
} else {
absImagePath = path.join(dirname, imagePath);
}
if (fs.existsSync(absImagePath)) {
File.copyFile(absImagePath, true, function (imageInfo) {
if (imageInfo) {
imagePath2ImageInfo[imagePath] = imageInfo;
}
cb();
});
} else {
return cb();
}
}, function () {
// 替换图片
console.log(imagePath2ImageInfo);
while(ret = reg.exec(body)) {
// console.log('=====');
// console.log(ret);
var imagePath = ret[2];
var imageInfo = imagePath2ImageInfo[imagePath];
if (imageInfo) {
// console.log('有的');
var imageLink = ' src="leanote://file/getImage?fileId=' + imageInfo.FileId + '"';
body = body.replace( ret[0], ret[0].replace(ret[1], imageLink) );
}
}
// console.log(body);
// OK了
// 添加到数据库中
var jsonNote = {
Title: title,
Content: me.fixContent(body),
Tags: [],
CreatedTime: new Date(),
UpdatedTime: new Date(),
IsMarkdown: false,
Attachs: null,
NotebookId: notebookId,
Desc: '',
NoteId: Common.objectId(),
IsNew: true
};
jsonNote._id = jsonNote.NoteId;
Note.updateNoteOrContent(jsonNote, function(insertedNote) {
eachCallback && eachCallback(insertedNote);
callback(true);
});
});
} catch(e) {
console.error(e);
callback(false);
}
},
};
module.exports = Import;

View File

@@ -0,0 +1,225 @@
/**
* 导入html, 重构
* @author life@leanote.com
* @date 2015/04/09
*/
define(function() {
var importService; // = nodeRequire('./public/plugins/import_html/import');
var html = {
langs: {
'en-us': {
'importHTML': 'Import HTML',
},
'de-de': {
'importHTML': 'HTML Datei importieren',
'Choose HTML files(.html)': 'HTML Dateien (.html) auswählen',
'Close': "Schliessen",
'Import to': "Importiere in Notizbuch",
"Done! %s notes imported!": "Abgeschlossen! Es wurden %s Notizen importiert!",
"Import file: %s Success!": "Datei importieren: %s erfolgreich!",
"Import file: %s Failure, is html file ?": "Datei importieren: %s fehlgeschlagen! Ist das eine HTML Datei?",
"Import: %s Success!": "Import: %s erfolgreich!"
},
'zh-cn': {
'importHTML': '导入HTML/为知笔记',
'Choose HTML files(.html)': '选择HTML文件(.html)',
'Close': "关闭",
'Import to': "导入至",
"Done! %s notes imported!": "完成, 成功导入 %s 个笔记!",
"Import file: %s Success!": "文件 %s 导入成功!",
"Import file: %s Failure, is html file ?": "文件 %s 导入失败! 是HTML文件?",
"Import: %s Success!": "导入笔记: %s 成功!"
},
'zh-hk': {
'importHTML': '導入HTML',
'Choose HTML files(.html)': '選擇HTML文件(.html)',
'Close': "關閉",
"Import to": "導入至",
"Done! %s notes imported!": "完成, 成功導入 %s 個筆記!",
"Import file: %s Success!": "文件 %s 導入成功!",
"Import file: %s Failure, is html file ?": "文件 %s 導入失敗! 是HTML文件?",
"Import: %s Success!": "導入筆記: %s 成功!"
}
},
_tpl: `
<style>
#importHTMLDialog .tab-pane {
text-align: center;
padding: 10px;
padding-top: 20px;
}
#importHTMLDialog .alert {
margin-top: 10px;
padding: 0;
border: none;
}
</style>
<div class="modal fade bs-modal-sm" id="importHTMLDialog" tabindex="-1" role="dialog" aria-labelledby="mySmallModalLabel">
<div class="modal-dialog modal-sm">
<div class="modal-content">
<div class="modal-header">
<button type="button" class="close" data-dismiss="modal" aria-hidden="true">&times;</button>
<h4 class="modal-title" class="modalTitle"><span class="lang">Import to</span> <span id="importDialogNotebook"></span></h4>
</div>
<div class="modal-body" id="">
<div role="tabpanel">
<!-- Tab panes -->
<div class="tab-content">
<div role="tabpanel" class="tab-pane active" id="htmlTab">
<!-- import -->
<a id="chooseHTMLFile" class="btn btn-success btn-choose-file">
<i class="fa fa-upload"></i>
<span class="lang">Choose HTML files(.html)</span>
</a>
<!-- 消息 -->
<div id="importHTMLMsg" class="alert alert-info">
<div class="curImportFile"></div>
<div class="curImportNote"></div>
<div class="allImport"></div>
</div>
</div>
<div role="tabpanel" class="tab-pane" id="youdaoTab">
<!-- 文件选择框 -->
<input id="importHTMLInput" type="file" nwsaveas="" accept=".html" multiple style="" style="display: none"/>
</div>
</div>
</div>
</div>
<div class="modal-footer ">
<button type="button" class="btn btn-default upgrade-cancel-btn lang" data-dismiss="modal">Close</button>
</div>
</div><!-- /.modal-content -->
</div><!-- /.modal-dialog -->
</div><!-- /.modal -->
`,
_importDialog: null,
_curNotebook: null,
_inited: false,
getMsg: function(txt, data) {
return Api.getMsg(txt, 'plugin.import_html', data)
},
init: function() {
var me = this;
me._inited = true;
$('body').append(me._tpl);
me._importDialog = $("#importHTMLDialog");
me._importDialog.find('.lang').each(function() {
var txt = $.trim($(this).text());
$(this).text(me.getMsg(txt));
});
// 导入, 选择文件
$('#chooseHTMLFile').click(function() {
Api.gui.dialog.showOpenDialog(Api.gui.getCurrentWindow(),
{
properties: ['openFile', 'multiSelections'],
filters: [
{ name: 'HTML', extensions: ['html'] }
]
},
function(paths) {
if(!paths) {
return;
}
var notebookId = me._curNotebook.NotebookId;
var n = 0;
me.clear();
if (!importService) {
importService = nodeRequire('./public/plugins/import_html/import');
}
importService.importFromHTML(notebookId, paths,
// 全局
function(ok) {
// $('#importHTMLMsg .curImportFile').html("");
// $('#importHTMLMsg .curImportNote').html("");
setTimeout(function() {
$('#importHTMLMsg .allImport').html(me.getMsg('Done! %s notes imported!', n));
}, 500);
},
// 单个文件
function(ok, filename) {
if(ok) {
$('#importHTMLMsg .curImportFile').html(me.getMsg("Import file: %s Success!", filename));
} else {
$('#importHTMLMsg .curImportFile').html(me.getMsg("Import file: %s Failure, is html file ?", filename));
}
},
// 单个笔记
function(note) {
if(note) {
n++;
$('#importHTMLMsg .curImportNote').html(me.getMsg("Import: %s Success!", note.Title));
// 不要是新的, 不然切换笔记时又会保存一次
note.IsNew = false;
// 插入到当前笔记中
Note.addSync([note]);
}
}
);
}
);
});
},
clear: function() {
$('#importHTMLMsg .curImportFile').html("");
$('#importHTMLMsg .curImportNote').html("");
$('#importHTMLMsg .allImport').html('');
},
open: function(notebook) {
var me = this;
if(!notebook) {
return;
}
if(!me._inited) {
me.init();
}
me.clear();
$('#importDialogNotebook').html(notebook.Title);
me._curNotebook = notebook;
var notebookId = notebook.NotebookId;
me._importDialog.modal('show');
},
// 打开前要执行的
onOpen: function() {
var me = this;
var gui = Api.gui;
Api.addImportMenu({
label: Api.getMsg('plugin.import_html.importHTML'),
click: (function() {
return function(notebook) {
me.open(notebook);
};
})()
});
},
// 打开后
onOpenAfter: function() {
},
// 关闭时需要运行的
onClose: function() {
}
};
return html;
});