mirror of
https://github.com/ArchiveBox/ArchiveBox
synced 2024-11-10 06:34:16 +00:00
minor url fixes and refactoring
This commit is contained in:
Nick Sweeting
parent
0df2bfe4c8
commit
881de8adbe
6 changed files with 46 additions and 22 deletions
|
|
@ -22,6 +22,7 @@
|
|||
<DT><A HREF="https://duckduckgo.com/?q=archive+firefox+bookmarks&t=ffab&ia=web" ADD_DATE="1497562974" LAST_MODIFIED="1497562974" ICON_URI="https://duckduckgo.com/favicon.ico" ICON="data:image/png;base64,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">archive firefox bookmarks at DuckDuckGo</A>
|
||||
<DT><A HREF="https://github.com/nodiscc" ADD_DATE="1497562974" LAST_MODIFIED="1497562974" ICON_URI="https://assets-cdn.github.com/favicon.ico" ICON="data:image/png;base64,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">nodiscc (nodiscc) · GitHub</A>
|
||||
<DT><A HREF="https://github.com/pirate/bookmark-archiver#troubleshooting" ADD_DATE="1497562975" LAST_MODIFIED="1497562975" ICON_URI="https://assets-cdn.github.com/favicon.ico" ICON="data:image/png;base64,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">pirate/bookmark-archiver · Github</A>
|
||||
<DT><A HREF="http://www.cs.unc.edu/~fabian/papers/foniks-oak11.pdf" ADD_DATE="1497562976" LAST_MODIFIED="1497562976" ICON_URI="https://assets-cdn.github.com/favicon.ico" ICON="data:image/png;base64,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">Phonotactic Reconstruction of Encrypted VoIP Conversations</A>
|
||||
<DT><A HREF="https://www.ghacks.net/2009/07/23/firefox-bookmarks-archiver/" ADD_DATE="1497562974" LAST_MODIFIED="1497562974" ICON_URI="https://www.ghacks.net/wp-content/uploads/2005/10/favicon.ico" ICON="data:image/png;base64,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">Firefox Bookmarks Archiver - gHacks Tech News</A>
|
||||
</DL><p>
|
||||
<DT><H3 ADD_DATE="1409779227" LAST_MODIFIED="1470506008" PERSONAL_TOOLBAR_FOLDER="true">Bookmarks Toolbar</H3>
|
||||
|
|
|
|||
26
fetch.py
26
fetch.py
|
|
@ -32,7 +32,7 @@ def fetch_wget(out_dir, link, overwrite=False, requisites=True, timeout=60):
|
|||
"""download full site using wget"""
|
||||
|
||||
domain = link['base_url'].split('/', 1)[0]
|
||||
if not os.path.exists('{}/{}'.format(out_dir, domain)) or overwrite:
|
||||
if not os.path.exists(os.path.join(out_dir, domain)) or overwrite:
|
||||
print(' - Downloading Full Site')
|
||||
CMD = [
|
||||
*'wget --timestamping --adjust-extension --no-parent'.split(' '), # Docs: https://www.gnu.org/software/wget/manual/wget.html
|
||||
|
|
@ -54,7 +54,9 @@ def fetch_wget(out_dir, link, overwrite=False, requisites=True, timeout=60):
|
|||
def fetch_pdf(out_dir, link, overwrite=False, timeout=60, chrome_binary='chromium-browser'):
|
||||
"""print PDF of site to file using chrome --headless"""
|
||||
|
||||
if (not os.path.exists('{}/output.pdf'.format(out_dir)) or overwrite) and link['type'] not in ('PDF', 'image'):
|
||||
path = os.path.join(out_dir, 'output.pdf')
|
||||
|
||||
if (not os.path.exists(path) or overwrite) and link['type'] not in ('PDF', 'image'):
|
||||
print(' - Printing PDF')
|
||||
CMD = [
|
||||
chrome_binary,
|
||||
|
|
@ -76,7 +78,9 @@ def fetch_pdf(out_dir, link, overwrite=False, timeout=60, chrome_binary='chromiu
|
|||
def fetch_screenshot(out_dir, link, overwrite=False, timeout=60, chrome_binary='chromium-browser', resolution='1440,900'):
|
||||
"""take screenshot of site using chrome --headless"""
|
||||
|
||||
if (not os.path.exists('{}/screenshot.png'.format(out_dir)) or overwrite) and link['type'] not in ('PDF', 'image'):
|
||||
path = os.path.join(out_dir, 'screenshot.png')
|
||||
|
||||
if (not os.path.exists(path) or overwrite) and link['type'] not in ('PDF', 'image'):
|
||||
print(' - Snapping Screenshot')
|
||||
CMD = [
|
||||
chrome_binary,
|
||||
|
|
@ -98,7 +102,10 @@ def fetch_screenshot(out_dir, link, overwrite=False, timeout=60, chrome_binary='
|
|||
|
||||
def archive_dot_org(out_dir, link, overwrite=False, timeout=60):
|
||||
"""submit site to archive.org for archiving via their service, save returned archive url"""
|
||||
if (not os.path.exists('{}/archive.org.txt'.format(out_dir)) or overwrite):
|
||||
|
||||
path = os.path.join(out_dir, 'archive.org.txt')
|
||||
|
||||
if not os.path.exists(path) or overwrite:
|
||||
print(' - Submitting to archive.org')
|
||||
submit_url = 'https://web.archive.org/save/{}'.format(link['url'].split('?', 1)[0])
|
||||
|
||||
|
|
@ -129,7 +136,9 @@ def archive_dot_org(out_dir, link, overwrite=False, timeout=60):
|
|||
def fetch_favicon(out_dir, link, overwrite=False, timeout=60):
|
||||
"""download site favicon from google's favicon api"""
|
||||
|
||||
if not os.path.exists('{}/favicon.ico'.format(out_dir)) or overwrite:
|
||||
path = os.path.join(out_dir, 'favicon.ico')
|
||||
|
||||
if not os.path.exists(path) or overwrite:
|
||||
print(' - Fetching Favicon')
|
||||
CMD = 'curl https://www.google.com/s2/favicons?domain={domain}'.format(**link).split(' ')
|
||||
fout = open('{}/favicon.ico'.format(out_dir), 'w')
|
||||
|
|
@ -149,7 +158,9 @@ def fetch_audio(out_dir, link, overwrite=False, timeout=60):
|
|||
if link['type'] not in ('soundcloud',):
|
||||
return
|
||||
|
||||
if (not os.path.exists('{}/audio'.format(out_dir)) or overwrite):
|
||||
path = os.path.join(out_dir, 'audio')
|
||||
|
||||
if not os.path.exists(path) or overwrite:
|
||||
print(' - Downloading audio')
|
||||
CMD = [
|
||||
"youtube-dl -x --audio-format mp3 --audio-quality 0 -o '%(title)s.%(ext)s'",
|
||||
|
|
@ -173,8 +184,9 @@ def fetch_video(out_dir, link, overwrite=False, timeout=60):
|
|||
if link['type'] not in ('youtube', 'youku', 'vimeo'):
|
||||
return
|
||||
|
||||
path = os.path.join(out_dir, 'video')
|
||||
|
||||
if (not os.path.exists('{}/video'.format(out_dir)) or overwrite):
|
||||
if not os.path.exists(path) or overwrite:
|
||||
print(' - Downloading video')
|
||||
CMD = [
|
||||
"youtube-dl -x --audio-format mp3 --audio-quality 0 -o '%(title)s.%(ext)s'",
|
||||
|
|
|
|||
9
index.py
9
index.py
|
|
@ -19,7 +19,12 @@ def dump_index(links, service):
|
|||
link_html.format(**derived_link_info(link)) for link in links
|
||||
)
|
||||
|
||||
template_vars = (datetime.now().strftime('%Y-%m-%d %H:%M'), article_rows)
|
||||
template_vars = {
|
||||
'num_links': len(links),
|
||||
'date_updated': datetime.now().strftime('%Y-%m-%d'),
|
||||
'time_updated': datetime.now().strftime('%Y-%m-%d %H:%M'),
|
||||
'rows': article_rows,
|
||||
}
|
||||
|
||||
with open(os.path.join(service, 'index.html'), 'w', encoding='utf-8') as f:
|
||||
f.write(index_html.format(*template_vars))
|
||||
f.write(index_html.format(**template_vars))
|
||||
|
|
|
|||
14
parse.py
14
parse.py
|
|
@ -168,8 +168,8 @@ def valid_links(links):
|
|||
return (link for link in links if link['url'].startswith('http') or link['url'].startswith('ftp'))
|
||||
|
||||
|
||||
def calculate_archive_url(link):
|
||||
"""calculate the path to the wgetted html file, since wget may
|
||||
def html_appended_url(link):
|
||||
"""calculate the path to the wgetted .html file, since wget may
|
||||
adjust some paths to be different than the base_url path.
|
||||
|
||||
See docs on wget --adjust-extension."""
|
||||
|
|
@ -190,7 +190,13 @@ def calculate_archive_url(link):
|
|||
def derived_link_info(link):
|
||||
"""extend link info with the archive urls and other derived data"""
|
||||
|
||||
link_info = {**link}
|
||||
link_info = {
|
||||
**link,
|
||||
'files_url': 'archive/{timestamp}/'.format(**link),
|
||||
'archive_org_url': 'https://web.archive.org/web/{base_url}'.format(**link),
|
||||
'favicon_url': 'archive/{timestamp}/favicon.ico'.format(**link),
|
||||
'google_favicon_url': 'https://www.google.com/s2/favicons?domain={domain}'.format(**link),
|
||||
}
|
||||
|
||||
# PDF and images are handled slightly differently
|
||||
# wget, screenshot, & pdf urls all point to the same file
|
||||
|
|
@ -203,7 +209,7 @@ def derived_link_info(link):
|
|||
})
|
||||
else:
|
||||
link_info.update({
|
||||
'archive_url': calculate_archive_url(link),
|
||||
'archive_url': 'archive/{}/{}'.format(link['timestamp'], html_appended_url(link)),
|
||||
'pdf_link': 'archive/{timestamp}/output.pdf'.format(**link),
|
||||
'screenshot_link': 'archive/{timestamp}/screenshot.png'.format(**link)
|
||||
})
|
||||
|
|
|
|||
|
|
@ -64,11 +64,11 @@
|
|||
</head>
|
||||
<body>
|
||||
<header>
|
||||
<h1 title="Last modified {}">
|
||||
<h1 title="Last modified {time_updated}">
|
||||
<img src="https://nicksweeting.com/images/archive.png" height="36px">
|
||||
Archived Sites <img src="https://getpocket.com/favicon.ico" height="36px"> <br/>
|
||||
<small>
|
||||
Exported with: <a href="https://github.com/pirate/bookmark-archiver">Bookmark Archiver</a>
|
||||
Archived with: <a href="https://github.com/pirate/bookmark-archiver">Bookmark Archiver</a> on {date_updated}
|
||||
</small>
|
||||
</h1>
|
||||
</header>
|
||||
|
|
@ -76,7 +76,7 @@
|
|||
<thead>
|
||||
<tr>
|
||||
<th style="width: 120px;"><img src="https://getpocket.com/favicon.ico" height="12px"> Starred</th>
|
||||
<th style="width: 45vw;">Saved Article</th>
|
||||
<th style="width: 45vw;">Saved Articles ({num_links})</th>
|
||||
<th style="width: 50px">Files</th>
|
||||
<th style="width: 50px">PDF</th>
|
||||
<th style="width: 60px;font-size:0.8em;">Screenshot</th>
|
||||
|
|
@ -84,7 +84,7 @@
|
|||
<th style="width: 100px;whitespace:nowrap;overflow-x:scroll;display:block">Original URL</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody>{}</tbody>
|
||||
<tbody>{rows}</tbody>
|
||||
</table>
|
||||
</body>
|
||||
</html>
|
||||
|
|
|
|||
|
|
@ -1,12 +1,12 @@
|
|||
<tr>
|
||||
<td>{time}</td>
|
||||
<td><a href="archive/{timestamp}/{archive_url}" style="font-size:1.4em;text-decoration:none;color:black;" title="{title}">
|
||||
<img src="archive/{timestamp}/favicon.ico">
|
||||
<td><a href="{archive_url}" style="font-size:1.4em;text-decoration:none;color:black;" title="{title}">
|
||||
<img src="{favicon_url}">
|
||||
{title} <small style="background-color: #eee;border-radius:4px; float:right">{tags}</small>
|
||||
</td>
|
||||
<td style="text-align:center"><a href="archive/{timestamp}/" title="Files">📂</a></td>
|
||||
<td style="text-align:center"><a href="{files_url}" title="Files">📂</a></td>
|
||||
<td style="text-align:center"><a href="{pdf_link}" title="PDF">📄</a></td>
|
||||
<td style="text-align:center"><a href="{screenshot_link}" title="Screenshot">🖼</a></td>
|
||||
<td style="text-align:center"><a href="https://web.archive.org/web/{base_url}" title="Archive.org">🏛</a></td>
|
||||
<td>🔗 <img src="https://www.google.com/s2/favicons?domain={domain}" height="16px"> <a href="{url}">{url}</a></td>
|
||||
<td style="text-align:center"><a href="{archive_org_url}" title="Archive.org">🏛</a></td>
|
||||
<td>🔗 <img src="{google_favicon_url}" height="16px"> <a href="{url}">{url}</a></td>
|
||||
</tr
|
||||
|
|
|
|||
Loading…
Reference in a new issue