style changes, better parsing

2019-01-30 21:33:38 -08:00
parent ccc04d2735
commit bd77bc4ddf
4 changed files with 38 additions and 25 deletions
--- a/cgi-bin/proxy.cgi
+++ b/cgi-bin/proxy.cgi
@@ -4,17 +4,27 @@ import cgi
 import cgitb
 import subprocess

+from goose3 import Goose
+
 cgitb.enable()

-def fetch_site(url):
-    data = subprocess.check_output(['w3m', '-dump', url])
-    return data.decode()
+goose = Goose({'enable_image_fetching': True})

-def format_output(url, data):
+def fetch_site(url):
+    return goose.extract(url=url)
+
+def format_output(article):
    with open('template.html', 'r') as f:
        template = f.read()

-    return template.format(url, data)
+    extra = ""
+    if article.top_image:
+        extra += """<img src="{}" />""".format(article.top_image.src)
+
+    return template.format(
+        title=article.title,
+        body=article.cleaned_text,
+        extra=extra)

 def print_headers():
    print('Content-Type: text/html; charset=utf8\r\n\r\n')
@@ -31,7 +41,7 @@ def main():
    data = fetch_site(url)

    print_headers()
-    print(format_output(url, data))
+    print(format_output(data))

 if __name__ == '__main__':
    main()
--- a/index.html
+++ b/index.html
@@ -3,13 +3,12 @@
  <head>
    <meta name="viewport" content="width=device-width, initial-scale=1">
    <title>Text Proxy</title>
-    <link href="https://fonts.googleapis.com/css?family=Nanum+Gothic+Coding|Nanum+Gothic:800" rel="stylesheet">
    <link href="site.css" rel="stylesheet">
  </head>
  <body>
    <h1>Text Proxy</h1>
    <div class="main">
-      <form action="/proxy.cgi" method="get">
+      <form action="/cgi-bin/proxy.cgi" method="get">
        <label for="u">Site</label>
        <input name="u" id="u" type="url" autofocus required>
        <input type="submit" value="Go">
--- a/site.css
+++ b/site.css
@@ -1,9 +1,3 @@
-html,
-body,
-pre {
-	font-family: 'Nanum Gothic Coding', monospace;
-}
-
 html,
 body {
 	background-color: #2e3440;
@@ -13,7 +7,6 @@ body {
 }

 h1 {
-	font-family: 'Nanum Gothic', sans-serif;
 	font-size: 28px;
 	font-weight: 800;
 	text-align: center;
@@ -26,7 +19,19 @@ h1 {
 	padding: 32px;
 }

-pre,
+.text {
+	font-size: 16px;
+	line-height: 22px;
+	margin: 0;
+	max-width: 40em;
+	white-space: pre-line;
+}
+
+img {
+	margin: 16px 0;
+	max-width: 100%;
+}
+
 label,
 input {
 	font-size: 16px;
@@ -34,10 +39,6 @@ input {
 	margin: 0;
 }

-pre {
-	max-width: 100%;
-}
-
 form {
 	display: flex;
 	flex: 1 1 auto;
@@ -55,6 +56,10 @@ input {
 	box-sizing: border-box;
 }

+input[type=submit] {
+	cursor: pointer;
+}
+
 input[type=url] {
 	padding: 4px 22px;
 	margin: 22px 0;
--- a/template.html
+++ b/template.html
@@ -2,14 +2,13 @@
 <html>
  <head>
    <meta name="viewport" content="width=device-width, initial-scale=1">
-    <title>{0} - TextProxy</title>
-    <link href="https://fonts.googleapis.com/css?family=Nanum+Gothic+Coding|Nanum+Gothic:800" rel="stylesheet">
-    <link href="site.css" rel="stylesheet">
+    <title>{title} - TextProxy</title>
+    <link href="/site.css" rel="stylesheet">
  </head>
  <body>
-    <h1>{0}</h1>
+    <h1>{title}</h1>
    <div class="main">
-      <pre>{1}</pre>
+      <div class="text">{extra}{body}</div>
    </div>
  </body>
 </html>