Full Page PDF Snap – Save Webpage as PDF version history - 25 versions
Full Page PDF Snap – Save Webpage as PDF by Silence
Be careful with old versions! These versions are displayed for testing and reference purposes.You should always use the latest version of an add-on.
Latest version
Version 2.33.4
Released Aug 7, 2026 - 226.06 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Older versions
Version 2.33.2
Released Aug 7, 2026 - 225.27 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.33.0
Released Aug 7, 2026 - 224.37 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.32.2
Released Aug 7, 2026 - 223.67 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.32.1
Released Aug 7, 2026 - 222.5 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.32.0
Released Aug 7, 2026 - 221.64 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.31.20
Released Aug 7, 2026 - 220.6 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.31.19
Released Aug 7, 2026 - 218.88 KBWorks with firefox 109.0 and later, android 127.0 and laterCHANGED
• The filename now includes the title. The default was
{site}{date}{time}{n} — website, date, time, running number. Anyone
finding a capture a year later could not tell what it was about. It is now
{title}{site}{date}{time}.
• That title comes from the page's publisher metadata rather than the window
title, which almost always carries suffixes ("… - PubMed") that merely
consume space. The length limit rises from 40 to 60 characters; at 40 the
title broke mid-word.
• Existing installations follow on update, but only where the template and
length were never changed. A pattern you chose yourself is a decision and
is left alone.Source code released under MIT License
Version 2.31.18
Released Aug 7, 2026 - 218.22 KBWorks with firefox 109.0 and later, android 127.0 and laterCHANGED
• The filename now includes the title. The default was
{site}{date}{time}{n} — website, date, time, running number. Anyone
finding a capture a year later could not tell what it was about. It is now
{title}{site}{date}{time}.
• That title comes from the page's publisher metadata rather than the window
title, which almost always carries suffixes ("… - PubMed") that merely
consume space. The length limit rises from 40 to 60 characters; at 40 the
title broke mid-word.
• Existing installations follow on update, but only where the template and
length were never changed. A pattern you chose yourself is a decision and
is left alone.Source code released under MIT License
Version 2.31.17
Released Aug 7, 2026 - 215.38 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• Fixed menu columns on either side are kept in the PDF. On app-style
interfaces the capture is clipped to the area that scrolls the content — the
menu column sits beside it and was falling outside. The clip now extends to
include it. Where a column is taller than the window, it is captured as far
as it fits.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.16
Released Aug 7, 2026 - 212.17 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• Fixed menu columns on either side are kept in the PDF. On app-style
interfaces the capture is clipped to the area that scrolls the content — the
menu column sits beside it and was falling outside. The clip now extends to
include it. Where a column is taller than the window, it is captured as far
as it fits.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.14
Released Aug 7, 2026 - 205.79 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• Fixed menu columns on either side are kept in the PDF. On app-style
interfaces the capture is clipped to the area that scrolls the content — the
menu column sits beside it and was falling outside. The clip now extends to
include it. Where a column is taller than the window, it is captured as far
as it fits.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.13
Released Aug 7, 2026 - 205.05 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.31.12
Released Aug 7, 2026 - 204.53 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• Fixed menu columns on the left are kept in the PDF. Where a column is
taller than the window, it is captured as far as it fits — a visible menu
cut off at the bottom is more useful than none at all.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.10
Released Aug 7, 2026 - 203.85 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• On menu columns that scroll on their own, the bottom entries were missing.
Such columns are often built with the scrollbar hidden, so they were not
recognised as a separate area and were only captured as far as they fitted
the window. They are now recognised by position and shape — tall, narrow, at
the edge. Other elements with hidden overflow, which is most cards and
sections on a page, are deliberately left out.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.9
Released Aug 7, 2026 - 203.07 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• On menu columns that scroll on their own, the bottom entries were missing.
Such columns are often built with the scrollbar hidden, so they were not
recognised as a separate area and were only captured as far as they fitted
the window. They are now recognised by position and shape — tall, narrow, at
the edge. Other elements with hidden overflow, which is most cards and
sections on a page, are deliberately left out.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.8
Released Aug 7, 2026 - 202.76 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• On menu columns that scroll on their own, the bottom entries were missing.
Such columns are often built with the scrollbar hidden, so they were not
recognised as a separate area and were only captured as far as they fitted
the window. They are now recognised by position and shape — tall, narrow, at
the edge. Other elements with hidden overflow, which is most cards and
sections on a page, are deliberately left out.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.Source code released under MIT License
Version 2.31.6
Released Aug 7, 2026 - 202.48 KBWorks with firefox 109.0 and later, android 127.0 and laterFIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the tSource code released under MIT License
Version 2.31.4
Released Aug 7, 2026 - 200.77 KBWorks with firefox 109.0 and later, android 127.0 and laterNEW
• A link map beside the PDF. A JSON file listing every link on the page: where
it sits in the image, where it leads, and which part of the page it belongs
to — navigation, footer, content, pagination. Made for AI tools that have to
find their way to a target: on an encyclopaedia page, 741 of 1,528 links
turned out to be page furniture and could be ruled out in advance. Optional,
costs one extra file per capture.
• Time anchor (needs internet, off unless switched on). Before saving, a public
random value is fetched from the decentralised drand network and written into
the capture. Because nobody could have known that value beforehand, it shows
the capture is not older than that moment — without having to trust your
device clock. The footer then also reads "not before <time>".
• In multi-page mode a page now fills an A4 sheet by default. Multi-page
output used to cut at a fixed pixel height, so printed pages did not fit
the sheet and were scaled. A4 derives the page height from the captured
width. The fixed pixel height remains available — anyone who set it keeps
it.
• The page is scrolled through once before capturing. Pages that load more
content while scrolling used to grow in the middle of a capture, which put
some sections into the PDF twice. Now the page is walked through once without
photographing, until its height stops changing, and the capture then starts
from the top.
FIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• In multi-page mode, captures in black and white or greyscale came out
almost empty: a few lines of noise at the top and nothing below. The chosen
colour depth was not carried over when the PDF was assembled, so the file
header announced colour while the viewer found a fraction of the expected
data. Short pages were affected the same way. On long single-page captures
it was never visible.
• On menu columns that scroll on their own, the bottom entries were missing.
Such columns are often built with the scrollbar hidden, so they were not
recognised as a separate area and were only captured as far as they fitted
the window.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where thereSource code released under MIT License
Version 2.31.1
Released Aug 7, 2026 - 199.17 KBWorks with firefox 109.0 and later, android 127.0 and laterVersion 2.31.1 — 7 August 2026
This release covers three versions (2.30.0, 2.31.0, 2.31.1), because the last
published build was 2.29.0.
NEW
• A link map beside the PDF. A JSON file listing every link on the page: where
it sits in the image, where it leads, and which part of the page it belongs
to — navigation, footer, content, pagination. Made for AI tools that have to
find their way to a target: on an encyclopaedia page, 741 of 1,528 links
turned out to be page furniture and could be ruled out in advance. Optional,
costs one extra file per capture.
• Time anchor (needs internet, off unless switched on). Before saving, a public
random value is fetched from the decentralised drand network and written into
the capture. Because nobody could have known that value beforehand, it shows
the capture is not older than that moment — without having to trust your
device clock. The footer then also reads "not before <time>".
• The page is scrolled through once before capturing. Pages that load more
content while scrolling used to grow in the middle of a capture, which put
some sections into the PDF twice. Now the page is walked through once without
photographing, until its height stops changing, and the capture then starts
from the top.
FIXED
• The context menu appeared in German regardless of language. Translations were
present in all nine languages but never used. This affected every build so
far. The same applied to two notifications and three places in the settings.
• On pages with a fixed menu column on the left, that column was missing from
the PDF and left an empty area behind. The cause was not the detection but
its parent container: when that was classified as an intrusive overlay and
hidden, the column went with it.
• Pages with endless loading (timelines, search results, forums) produced an
error and no PDF at all, even though the capture had already got a long way.
The page footer is now recognised as the end and the capture finishes there.
Where there is no footer, the state at the start of the capture applies. In
both cases a PDF is produced, and a note says that it is shorter than the
page has since become.
ABOUT LANGUAGES
The interface follows the browser language. A fixed language can be chosen in
the settings; "Automatic" is the default. Nine languages: English, German,
Spanish, French, Italian, Portuguese (Brazil), Russian, Chinese (simplified),
Japanese.Source code released under MIT License
Version 2.29.0
Released Aug 4, 2026 - 182.94 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.28.0
Released Aug 4, 2026 - 182.06 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.26.0
Released Aug 3, 2026 - 157.15 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.22.0
Released Aug 3, 2026 - 155.27 KBWorks with firefox 109.0 and later, android 127.0 and laterSource code released under MIT License
Version 2.21.0
Released Aug 3, 2026 - 150.04 KBWorks with firefox 109.0 and later, android 127.0 and laterVersion 2.4.0 — 3. August 2026
NEW
• Select an area with the mouse. Click "Select an area", drag the rectangle —
the size shows while you drag. Escape, right-click or a drag too small to read
cancels without producing a file.
• A switch for banners and pop-ups in the popup itself, with your choice
remembered. It used to live only in the settings and the context menu.
• Consent dialogs are now recognised even when they are not fixed in place, and
the scroll lock they set is released. That was never cosmetic: on one news site
the page reported 900 pixels of height instead of 43,101, so a full-page
capture would have quietly collapsed to a single screen. Nothing is clicked
away on your behalf, and every change is put back after the capture.
• Citation details in the PDF: authors, journal, volume, DOI, licence and the
time of retrieval including the time zone. As a visible line, in the document
properties, and as an attached RIS record for Citavi, Zotero and EndNote. Only
the page already open is read — no service is queried.
• Page breaks in multi-page output fall between lines instead of through them,
with an A4 setting that fits printed paper. The searchable text survives the
break; previously it was dropped entirely in multi-page mode.
FIXED
• Area selection produced a broken image: horizontal streaks, vertically
squashed, and with no searchable text. On cropping, the old image width was
left in place while the canvas shrank, so the tiling code read at the wrong
measure. The text layer is now cropped along instead of discarded.
• The provenance line was invisible whenever a text layer was present.
• Citation details were not collected at all in the browser — a start-up error
that only a test of the whole script could find.
• Error pages and access walls were turned into citations.
• With non-Latin scripts the citation line collapsed into punctuation; it now
points at the attached RIS record instead.
• The button still carried its old label, because the locale files take
precedence and had not been kept in step.Source code released under MIT License