yhchen
/
ptt-sententree
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

哈哈

Browse Source
dev/addPush
Zovjsra 4 years ago
parent 24f92cd6f0
commit efdd3ec965
11 changed files with 171 additions and 37 deletions
Show Stats Download Patch File Download Diff File

BIN
__pycache__/dataHandlerPTT.cpython-37.pyc
View File

Binary file not shown.

2
app.py
Unescape View File

@ -67,7 +67,7 @@ def ptt_push():
@app.route('/ptt_push/init', methods=['POST'])
def pttPushInit():
author = 'yuetsu'
author = 'a58461351'
pushes = pttPush.findAuthorPush(
author=[author], stopwords=pttPush.defaultStopWords)
result: dict = {

6
dataHandlerPTT.py
Unescape View File

@ -31,6 +31,8 @@ with open('/home/vis/pttDatabase/PTTData/Gossiping/content/content.pck', 'rb') a
defaultStopWords = []
data = PTTData('Gossiping', '/home/vis/pttDatabase/PTTData')
sentence_length = 100
use_push_count = False
with open('resource/stopWords.txt', 'r', encoding='UTF-8') as file:
for word in file.readlines():
@ -75,7 +77,7 @@ def contentProcess(content, text):
result = []
for i in cutted:
result.append(i)
if (len(result) >= 50):
if (len(result) >= sentence_length):
sentenses.append(result.copy())
result = []
if (result != []):
@ -324,7 +326,7 @@ def getDefault(startDate, endDate):
'date': datetime.strptime(i['date'], '%Y%m%d%H%M%S').strftime("%a %b %d %H:%M:%S %Y"),
'part': i['content'][max(0, cut[0] - 20): min(len(i['content']), cut[1])].replace('\n', '')
}
result.append([len(result), seq, 1000 + i['pushes']])
result.append([len(result), seq, (1000 + i['pushes'])if use_push_count else 3000])
fileString = io.StringIO()
writer = csv.writer(fileString, delimiter='\t')

5
resource/stopWords.txt
Unescape View File

@ -68,3 +68,8 @@
只
所以
講
記者
看
者
媒體
和

32
static/css/main.css
Unescape View File

@ -112,6 +112,24 @@ html {
transition-duration: 0.5s;
}
#pttPageWindowContent {
display: block;
position: fixed;
background-color: #FFF;
left: 50%;
top: 50%;
transform: translate(-50%, -50%);
width: 90%;
height: 90%;
border: lightgray;
border-width: 1px;
border-style: solid;
border-radius: 20px;
padding: 26px;
z-index: 99;
box-shadow: 0px 5px 20px rgba(0, 0, 0, .3);
}
.deleteListElement {
position: absolute;
right: 0;
@ -141,6 +159,18 @@ li a {
}
.info {
background-color: rgba(255, 255, 255, 0.6);
animation: fadeIn 0.2s;
animation-fill-mode: forwards;
position: fixed;
display: flex;
top: 0;
width: 100%;
height: 100%;
z-index: 90;
}
.info99 {
background-color: rgba(255, 255, 255, 0.6);
animation: fadeIn 0.2s;
animation-fill-mode: forwards;
@ -189,7 +219,7 @@ li a {
height: auto;
border-radius: 15px;
padding: 10px 15px;
z-index: 99;
z-index: 90;
align-content: center;
}

94
static/js/ptt.js
Unescape View File

@ -76,7 +76,41 @@ function init() {
hideIdfEditor()
}
})
$('#pttPageWindow').click(function(e) {
if ($('#pttPageWindow').is(e.target)) {
hidePTTPage()
}
})
changeMode(0)
destroyCurrentGraph()
buildSentetree()
}
function loadTemplate(num) {
templates = [{
startDate: '2020-12-01',
endDate: '2020-12-31',
keyword: '',
mode: 1
},
{
startDate: '2020-01-01',
endDate: '2020-03-01',
keyword: '衛生紙',
mode: 2
},
{
startDate: '2020-01-11',
endDate: '2020-01-12',
keyword: '',
mode: 2
}
]
chosenTemp = templates[num]
setDate(chosenTemp.startDate, chosenTemp.endDate)
$('#keywordBox').val(chosenTemp.keyword)
changeMode(chosenTemp.mode)
sendRequest()
}
function clearStopWord() {
@ -117,8 +151,6 @@ function changeMode(_mode) {
}
}
mode = _mode
destroyCurrentGraph()
buildSentetree()
}
function scrollIdfList() {
@ -170,8 +202,12 @@ function showIdfEditor() {
.append($('<th>')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('單詞'))
.append($('<th>'))
.append($('<th>').attr('class', 'w3-right-align')
.append($('<th>')
.attr('class', 'w3-center-align')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('操作'))
.append($('<th>')
.attr('class', 'w3-right-align')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('單詞頻率')
)
@ -259,6 +295,15 @@ function hideIdfEditor() {
$('#idfEditorLayer').addClass('hidden')
}
function showPTTPage(url) {
$('#pttPageWindowContent iframe').attr('src', url)
$('#pttPageWindow').removeClass('hidden')
}
function hidePTTPage() {
$('#pttPageWindow').addClass('hidden')
}
function updateIdfTable() {
let wordList = $("#ieContainer").find("tr").slice(1).map((_, a) => {
return [
@ -315,6 +360,23 @@ function closeEventListner() {
}
function sendRequest() {
content = getContent()
startDate = $('#startDate').val()
endDate = $('#endDate').val()
console.log(content)
$.ajax({
type: 'POST',
url: '/addRequest',
data: content,
contentType: 'application/json',
success: function(data) {
console.log(data)
changeGraph(data.Result)
}
})
}
function getContent() {
content = JSON.stringify({
startDate: $('#startDate').val(),
endDate: $('#endDate').val(),
@ -330,19 +392,7 @@ function sendRequest() {
other: $('#other').is(':checked')
}
})
startDate = $('#startDate').val()
endDate = $('#endDate').val()
console.log(content)
$.ajax({
type: 'POST',
url: '/addRequest',
data: content,
contentType: 'application/json',
success: function(data) {
console.log(data)
changeGraph(data.Result)
}
})
return content
}
function changeGraph(data) {
@ -466,8 +516,9 @@ function buildSentetree() {
$('#titleListContainer').empty()
for (i of titleList) {
$('#titleListContainer').append(
$('<li>').attr('class', 'w3-panel').append(
$('<a>').attr('href', i.url).attr('target', '_blank').append(
$('<li>').attr('class', 'w3-panel')
.css('cursor', 'pointer').append(
$('<p>').attr('target', '_blank').append(
$('<h4>').html(i.title)
).append(
$('<span>').attr('style', 'margin: 0px 10px').html(i.author)
@ -476,7 +527,10 @@ function buildSentetree() {
).append(
$('<span>').attr('style', 'margin: 0px 10px').html('推文數：' + i.pushes)
)
)
).click(function() {
let indx = $(this).index()
showPTTPage((titleList[indx].url).replace('www.ptt.cc', 'www.pttweb.cc'))
})
)
}
})

31
static/js/pttPush.js
Unescape View File

@ -25,6 +25,7 @@ function init() {
console.log(wordPushList)
$('#idBox').val(data.Result.author)
buildSentetree(tsvString)
changeMode(2)
}
})
$(document).ready(function() {
@ -66,6 +67,31 @@ function init() {
})
}
function loadTemplate(num) {
templates = [{
userId: '',
aid: '1Vv4iFY6',
keyword: '',
mode: 0,
}, {
userId: '',
aid: '1VyJ2vP_',
keyword: '',
mode: 0
}, {
userId: 'xetherz3',
aid: '',
keyword: '',
mode: 0
}]
let chosenTemplate = templates[num]
$('#idBox').val(chosenTemplate.userId)
$('#titleBox').val(chosenTemplate.aid)
$('#keywordBox').val(chosenTemplate.keyword)
changeMode(0)
sendRequest()
}
function clearStopWord() {
stopwords = []
$('#sweContainer').html('')
@ -138,7 +164,10 @@ function showIdfEditor() {
.append($('<th>')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('單詞'))
.append($('<th>'))
.append($('<th>')
.attr('class', 'w3-center-align')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('操作'))
.append($('<th>').attr('class', 'w3-right-align')
.attr('style', 'position: sticky; top: 0; background: white;')
.append('單詞頻率')

2
static/node_modules/sententree/dist/sententree-standalone.js generated vendored
Unescape View File

@ -19158,7 +19158,7 @@
maxc = value
}
} else {
if (value < maxSupport && (value * Math.pow(distRatio, dist - 1)) > maxc && stopwords.indexOf(itemset[w]) < 0 && dist == 1) {
if (value < maxSupport && value > maxc && dist < 2 && stopwords.indexOf(itemset[w]) < 0 && dist == 1) {
maxw = +w
maxc = value
console.log(maxc)

3
templates/generalTxt.html
Unescape View File

@ -41,7 +41,8 @@
<button class="general-button" type="button" id="editSWButton" style="margin:10px 0px;" onclick="showStopwordEditor()">編輯停用詞</button>
</div>
<div id='rawText' class=''>
<textarea id='rawTextBox' rows=25></textarea>
<textarea id='rawTextBox' rows=25 placeholder="輸入要視覺化的文字
換行為斷句"></textarea>
<button class='general-button' style='margin: 10px 0px' onclick="submit()">提交</button>
</div>
<div>

25
templates/ptt.html
Unescape View File

@ -15,6 +15,11 @@
<button class="w3-button w3-teal" type="button" onclick="location.href='/generalTxt'">泛用文字視覺化工具</button>
</div>
<div id='pttPageWindow' class='info99 hidden'>
<div id='pttPageWindowContent'>
<iframe src="http://ptt.cc" style='width:100%; height:100%' frameborder=0></iframe>
</div>
</div>
<div id="nodeTitle" class="nodeTitle hidden">
<div id="nodeTitleContent"></div>
</div>
@ -66,17 +71,19 @@
<div id="heading">
<h2>{{title}}</h2>
<p>SentenTree <a href="https://github.com/twitter/SentenTree">https://github.com/twitter/SentenTree</a></p>
<p id="comment">同時使用關鍵詞和詞性搜尋的時候，必須選擇所設關鍵詞本身的詞性，否則會搜尋不到結果。</p>
<p id="comment">點選圖上的單詞可以查看單詞的資訊，觀看原始文章，也會快速切換關鍵字。</p>
<p id="comment">若搜尋到的文章超過50篇，圖表僅會顯示推文數最多的前50篇文章。</p>
<p id="comment">停用詞的處理改為將不直接忽略停用詞，但是停用詞不會被設為主要單詞，並且大小會比其他單詞更小。</p>
<p>可直接從下方範例搜尋條件中選擇</p>
<div>
<button class='w3-button w3-green' onclick='loadTemplate(0)'>範例1</button>
<button class='w3-button w3-green' onclick='loadTemplate(1)'>範例2</button>
<button class='w3-button w3-green' onclick='loadTemplate(2)'>範例3</button>
</div>
<div id="searchingTarget">
<form name="form" enctype="multipart/form-data">
<span>搜尋日期範圍 從</span>
<input id="startDate" name="startDate" type="date">
<span>到</span>
<input id="endDate" name="endDate" type="date">
<button class="general-button" type="button" id="resetDateButton" onclick="setDate(defaultStartDate, defaultEndDate)">預設日期</button>
<button class="general-button" type="button" id="resetDateButton" onclick="setDate(defaultStartDate, defaultEndDate)">預設日期</button><br>
<input id="keywordBox" type="text" name="message" placeholder="輸入關鍵詞">
<button class="general-button" type="button" id="sendButton" onclick="sendRequest()">搜尋關鍵字</button>
<button class="general-button" type="button" id="resetButton" onclick="document.getElementById('keywordBox').value=''">清除關鍵字</button>
@ -97,10 +104,10 @@
</form>
</div>
<div id='modeSelector'>
<button type="button" id='mode0' class='w3-button' onclick="changeMode(0)">相鄰詞模式</button>
<button type="button" id='mode1' class='w3-button ' onclick="changeMode(1)">Idf模式</button>
<button type="button" id='mode2' class='w3-button ' onclick="changeMode(2)">停用詞模式</button>
<button type="button" id='mode3' class='w3-button ' onclick="changeMode(3)">根單詞停用模式</button>
<button type="button" id='mode0' class='w3-button' onclick="changeMode(0);destroyCurrentGraph();buildSentetree()">相鄰詞模式</button>
<button type="button" id='mode1' class='w3-button ' onclick="changeMode(1);destroyCurrentGraph();buildSentetree()">Idf模式</button>
<button type="button" id='mode2' class='w3-button ' onclick="changeMode(2);destroyCurrentGraph();buildSentetree()">停用詞模式</button>
<button type="button" id='mode3' class='w3-button ' onclick="changeMode(3);destroyCurrentGraph();buildSentetree()">根單詞停用模式</button>
</div>
</div>
<div id="graphInfo"></div>

6
templates/ptt_push.html
Unescape View File

@ -67,6 +67,12 @@
<p>SentenTree <a href="https://github.com/twitter/SentenTree">https://github.com/twitter/SentenTree</a></p>
<p id="comment">同時使用關鍵詞和詞性搜尋的時候，必須選擇所設關鍵詞本身的詞性，否則會搜尋不到結果。</p>
<p id="comment">鄉民ID和文章aid之中必須至少設定一個搜尋條件。</p>
<p>可直接從下方範例搜尋條件中選擇</p>
<div>
<button class='w3-button w3-green' onclick='loadTemplate(0)'>範例1</button>
<button class='w3-button w3-green' onclick='loadTemplate(1)'>範例2</button>
<button class='w3-button w3-green' onclick='loadTemplate(2)'>範例3</button>
</div>
<div id="searchingTarget">
<form name="form" enctype="multipart/form-data">
<input id="idBox" class='searchBox' type="text" name="message" placeholder="輸入鄉民ID">

Write Preview
Loading…
Cancel
Save