Repository navigation
Releases
v1.5.0
Installation
- Download
smart-storage-analyzer.sppkg - Upload to your SharePoint App Catalog
- Deploy tenant-wide or to specific sites
Added
-
One-step setup
Adding the app to a site now creates a ready-to-use Storage Analyzer page in Site Pages with
the web part already placed on it. No manual page creation or web part placement is needed. -
Bulk install script
scripts/Install-StorageAnalyzer.ps1adds the app to many sites in one run (PnP PowerShell), with a
-WhatIfpreview, an explicit-AllSitesswitch (or-SiteUrlsfor specific sites), an exclude list, optional upgrade of older installs, and a CSV results log.
Changed
- The app is now installed per site rather than deployable tenant-wide, so the page can be created
when the app is added.
v1.4.0
Installation
- Download
smart-storage-analyzer.sppkg - Upload to your SharePoint App Catalog
- Deploy tenant-wide or to specific sites
Added
- Localization — 29 languages
The entire web part interface — buttons, tooltips, column headers, warning and
error messages, and the Excel/CSV export column headers — is now localized.
Supported languages: English, Arabic, Chinese (Simplified and Traditional),
Czech, Danish, Dutch, Finnish, French, German, Greek, Hebrew, Hindi,
Hungarian, Indonesian, Italian, Japanese, Korean, Norwegian, Polish,
Portuguese (Brazil and Portugal), Romanian, Russian, Spanish, Swedish, Thai,
Turkish, Ukrainian, and Vietnamese. The language is selected automatically
from each user's own SharePoint UI language, with no configuration required;
an unsupported language falls back to English.
v1.3.1
Added
-
Total Storage Size everywhere sizes are shown
Current File Size plus Version History Size, shown first: as a highlighted summary tile in the
Storage Report, as the first size column in the List View, in the List View's Excel/CSV export,
in the Storage Report's Excel Summary sheet, and as a badge on saved scans. Shown only when
version history was measured, since otherwise it would just repeat Current File Size. -
Dark theme support
The web part now follows a dark (inverted) page section or theme variant instead of always
rendering light.
Changed
-
Renamed the three storage-size labels used throughout the app for clarity
What was labeled "Size" / "Total size" is now "Current File Size" (current files,
excluding version history). What was labeled "Version history" / "Version History" is
now "Version History Size" (version history only). The combined figure is "Total
Storage Size". Updated in Tree View, List View, Storage Report, Excel/CSV export, and the
Treemap tooltip. -
Storage Report summary regrouped
Size tiles come first (Total Storage Size, Current File Size, Version History Size), then Files
scanned, Stale, and Very stale. The Excel Summary sheet uses the same grouping, with blank rows
between groups. -
Tree View / List View always shows Version History Size — no checkbox any more
Version history is fetched at no extra cost during the same sweep that loads a folder either
way, so there was nothing the checkbox actually saved by being off. Total Storage Size, Version
History Size, and Version Count are now always-on columns and always factor into the Treemap's
file square sizing. -
The Storage Report's "Include Version History Size" is now checked by default
It's what makes Total Storage Size the true total-storage number, so it starts on; uncheck it
for a faster scan when you only need current-content size. Your choice is remembered across
sessions. (This checkbox is no longer shared with Tree View / List View, which has none.) -
Incomplete version-history totals are marked "≥"
When some files' version history couldn't be measured, Version History Size and Total Storage
Size show "≥" (a floor) in the tiles, saved-scan badges, and Excel Summary sheet. The List View
and its exports mark partially measured folders "≥" and unmeasured folders "Unknown". -
List View pages large folders 200 rows at a time, and its exports use the same column order
as the on-screen table. -
Deleting a saved scan asks for confirmation.
-
Accessibility: every treemap square can be reached with the keyboard, and file squares show
their archival status as text as well as color; treemap and header text switches between black
and white for contrast on any background; info icons are keyboard-focusable; sortable table
headers announce their sort order to screen readers; saved-scan checkboxes and delete buttons
have labels; disabled Home cards stay readable and their buttons explain why they're disabled.
Fixed
- Version Count could read 0 on a library with minor/draft versioning enabled, even when
Version History Size correctly showed real storage for the same files. The count is measured
exactly only when a library needs per-file measurement; on the common bulk-measured path it was
estimated from the file's version label, and that estimate counted only major versions — so a
label like "1.3" (three retained minor versions) read as zero. It now counts major and minor
versions together. - Tree/List View could get stuck loading when you switched folders before the previous one
finished; a late reply for the old folder left the new one's loading bar spinning. - Leaving the Storage Report mid-scan now cancels the scan instead of letting it run on in the
background. - Viewing a saved scan during a live scan no longer labels the live results as that saved scan
(its thresholds, site, and export site URL). - Version columns no longer appear at the "All libraries" level of the List View.
- The Storage Report results table now follows what the scan measured, rather than the
checkbox changed afterward. - Cancel now stops promptly even while waiting out SharePoint throttling or retrying.
- The item counter on the opening screen no longer inflates (it re-added running totals each
page, so a 200,000-item library showed about 4 million). - Libraries where no file has old versions no longer trigger a pointless full version-history
read, which could also mislabel them as unmeasured. - Throttled requests inside a batch now wait out the throttle before being resent, instead of
being resent immediately one by one. - The web part no longer fails to load when browser storage is blocked (Teams, strict privacy
settings); settings just aren't remembered in that case. - Web part startup: rendering is guarded until initialization finishes, and theme changes use
the built-in lifecycle hook (the previous listener was never removed). - A render error in one screen now shows a message with a "Back to Home" button instead of
replacing the whole web part. - Recycle Bin paging skips duplicate rows, so sizes can't be counted twice.
- Site-root library sizes now use current content only, matching the folder totals inside
each library. - Authors past the first 5,000 site users are no longer blank.
- Saving scan history when storage is full frees older listings one at a time until the new
one fits, and no longer deletes them when the failure wasn't about space. - The "Default view on open" property pane field now uses SPFx's standard dropdown control
instead of a hand-built one that relied on an internal, undocumented field-type number. - The Treemap's hover tooltip and on-square text now show a file's version-history size even
when it wasn't individually measured. SharePoint's own bulk version-size field is populated by
a lagging background job and can simply be missing for a given file's row (most often a recently
added or changed one); this used to render identically to a confirmed 0 B, silently dropping the
version-history breakdown. It's now marked "≥", matching how the List View already flags the same
gap. The Treemap's folded-together "Other" cell (the smallest items beyond the ~40 shown
individually) also now includes version-history size in its combined tooltip total, instead of
omitting it entirely.
Performance
- Opening a library reads it once instead of twice; the Explorer keeps recently opened libraries
in memory up to about 500,000 items in total (instead of every library ever opened); library
reads keep a compact copy of each item instead of the raw response. - Saved scans store their file listing in pieces, so a very large report no longer has to be
written (or read back) as one huge record. - Result tables no longer re-sort every row on unrelated clicks.
- A scan fetches the site user list once rather than once per subsite, and discovers subsites in
parallel.
v1.3.0
Added
-
Version-history measurement now escalates automatically to whatever actually works on a given
list, and always measures every eligible file
Some lists don't project SharePoint's bulk version-size field at all; a few report it while
leaving it unpopulated, which previously produced a confident but wrong 0 B for the whole
library. The scan now proves a bulk mechanism actually carries real data (checking known file
sizes against it) before trusting it, escalating through an un-expanded bulk sweep, then a
CAML/RenderListDataAsStreamsweep, and only then a per-file measurement pass — reserved for
files whose version number doesn't already prove they have nothing retained. There is no
longer a "Quick" cap that left some files unmeasured by design; every file that can have
retained versions is now measured, and the per-file pass runs fast enough for that to be
practical (requests are properly coalesced instead of running a few at a time). -
Version History Count total
Alongside the existing Version History Size total, the Storage Report's summary tile and Excel
Summary sheet now also show the total number of retained versions across the whole scan — the
same free per-file count, just summed. -
Stage-by-stage scan progress, with an ETA and an explicit "paused" state
The Storage Report's progress display now names exactly what's happening at every point in a
scan — reading items, checking how a list can report version history, sweeping it in bulk,
validating the result, or measuring per file — each with its own live count and, once there's
enough data to estimate from, a time-remaining estimate. When SharePoint is throttling requests,
the status line says so explicitly with a countdown, instead of the scan appearing to freeze. A
hover tooltip shows additional detail (requests made, files found vs. added to the report,
failed/unmeasured counts) without crowding the main display. -
Independent cross-check of the version-history total
For each library, the version-history total measured per file is compared against an
independent figure derived from SharePoint's own folder-level Storage Metrics. Diagnostic only
(logged, not shown in the UI) — it exists to catch a systematic measurement error before it's
trusted. -
Home screen
The web part now opens on a landing screen with three cards — Tree
View, List View, and Storage Report — each with a screenshot and
a short description of what it does. Clicking a card opens that tool;
every tool now has a Home button to return. A site administrator can
still configure the web part to skip Home and open directly on any one of
the three (or on Home itself) via the property pane's Default view on
open dropdown, which gained Home, Tree View, and List View as
options alongside the existing Storage Report.
Changed
-
The Explorer is now reached as two separate entry points: Tree View and
List View
Previously the web part opened directly into a single "Explorer" screen
with an internal Treemap/List tab. That screen is unchanged internally —
the same tab still switches between Treemap and List without losing your
place — but it's now reached via the Home screen's Tree View or List
View card (whichever is clicked decides which tab it opens on), and the
header no longer shows "Explorer" / "Storage Report" tab buttons, since
navigation moved to Home. -
More accurate folder/library scans on fast tenants, without sacrificing
the throttling protections that made the fallback walk bounded in the
first place
The fallback measurement walk (used when SharePoint'sStorageMetrics
rollup is stale or unavailable) previously stopped at a folder-count
budget derived from Concurrent API requests (scanConcurrency * 150,
capped at 2400) — on a fast tenant this budget was often exhausted in a
few seconds, well before the 45-second wall-clock deadline that was
already meant to be the real limit, cutting a scan short and reporting an
approximate "≥" result for no real reason. The wall-clock deadline is now
the sole practical bound; the folder count is a much higher safety valve
(20,000) that exists only to catch pathological/runaway cases. -
Folder/file listings that hit SharePoint's own paging limit are no
longer silently under-counted
A folder or file listing beyondgetJsonPaged's internal page-count
safety valve (previously 50 pages / 250,000 items, now 400 pages / 2
million items) was truncated with only a console warning — the resulting
total was short by whatever fell past the last page, with no indication
to the user. This is now surfaced the same way other incomplete
measurements are: assizeApproximatein Tree View / List View, and as a
counted, detailed skipped-folder entry in the Storage Report. -
Saved Storage Report history no longer trims itself, and its "Partial"
badge is gone
A saved report used to keep only its Stale/Very-stale rows once it
exceeded 50,000 files, flagged with a "Partial" badge — a mitigation for
loading every saved report's full listing just to render the history
list. History storage is now split into metadata (always loaded) and file
listings (loaded only when viewed or compared), so every saved report now
keeps every row. The only remaining trim is genuine storage exhaustion —
browser quota running out evicts the oldest report's listing (never its
summary) to make room, shown as a "No file list" badge instead.
Comparing two reports now also degrades per-field instead of all-or-
nothing: size and file-count deltas always work (they only need
summaries); new/resolved archival-candidate counts are shown only when
both reports' listings are actually available. -
$batchenvelope throttling no longer amplifies into a burst of
individual requests
A 429 on a coalesced$batchrequest was treated the same as a
structural failure (a rejected header,$batchunsupported) and fanned
out into one request per member — answering "you're sending too much"
with up to 50× more requests, against a tenant that had just asked to
slow down. A throttled batch now retries the whole envelope through the
same shared backoff every other request uses; per-member fan-out is
reserved for failures where the envelope itself will never work. -
Excel Summary sheet values are left-aligned
Numeric values (file/stale counts) previously sat right-aligned next to
left-aligned text values (sizes, dates) in the same column, producing a
ragged edge. Every value on the sheet now aligns the same way.
Fixed
-
A rejected bulk version-history field could either do nothing or cost
far more than necessary
Two separate issues on lists where SharePoint won't project the bulk
version-size field: sharding used to give up on concurrency entirely
rather than risk losing version data, even though the field's absence is
now handled correctly by escalation; and the per-file fallback ran at a
worker count sized for the pre-$batchera (3 workers), so envelopes
never filled and measured roughly 3.5 files/second regardless of the
$batchcoalescing meant to speed it up. Both are fixed — sharding is
re-enabled once the escalation can safely take over, and the per-file
pass now runs enough workers to keep$batchenvelopes full. -
The Storage Report's progress bar and time estimate could both be
confidently wrong while work continued
A stale item-count estimate that got exceeded mid-scan was clamped rather
than treated as unusable, which pegged the progress bar at 100% and could
show "~0s remaining" for many minutes on a long-running scan. Producers
now report no denominator at all once their own estimate is known to be
stale, which the display renders as an indeterminate bar and no time
estimate, rather than a wrong one. -
ReportHistoryService.add()could hang indefinitely on a storage
failure
A synchronous exception from the underlying IndexedDB write (e.g. a
clone failure, or the transaction aborting on quota) was never caught,
so the returned promise neither resolved nor rejected — the scan
appeared to finish but silently never saved, with no warning shown.
Failures are now always caught and reported, and running out of quota no
longer fails the save outright: the oldest report's file listing is
freed first and the save is retried once before giving up.
Latest Build (v1.3.1.0)
Automated build from commit 872fbe6.
Upload smart-storage-analyzer.sppkg to your SharePoint App Catalog to deploy.
v1.2.0
v1.1.0
[1.1.0] — 2026-07-11
Added
-
Cancel a running Storage Report scan
A Cancel button appears while a scan runs. Cancellation is cooperative
(checked between queued folders/libraries/sites), so it takes effect within
moments rather than instantly interrupting in-flight requests. Partial
results collected before cancellation are shown but not saved to history. -
Live file-count progress during a scan
The "files scanned" counter now updates continuously while a single library
is being walked, instead of sitting at 0 until the whole library finishes. -
Partial-scan warnings
A scan that could not read some folders (permission errors, throttling
exhaustion, etc.) or subsites now completes with a visible warning stating
how many were skipped, instead of silently under-reporting totals. -
Scan history: site awareness
Each saved scan now shows which site it was run against. The history list
and the "compare two scans" feature default to the current site, with an
explicit toggle to show reports from other (sub)sites; comparing two
reports from different sites now shows a caution instead of a silent,
potentially meaningless diff. -
Scan history: bounded storage
A scan whose file count would strainIndexedDBquota now saves only its
stale/very-stale rows (flagged "Partial" in the history list) — summaries
and scan comparisons are unaffected, since both only ever need stale-tier
data. -
Scan history: viewing context
Reopening a saved scan now shows its timestamp, site, and displays the
archival-status legend using the thresholds that scan was classified
with, rather than whatever the current Settings happen to be. -
Keyboard accessibility
Treemap folder cells and sortable table headers are now focusable and
operable from the keyboard (Enter/Space), witharia-sortreflecting the
active sort column.
Fixed
-
Invalid concurrency/threshold values could hang the whole web part
A corruptedlocalStoragevalue, or typing an out-of-range number directly
into a Settings spin button, could set scan concurrency toNaN/0—
which the scan queue treats as "never run anything," hanging every scan and
folder load indefinitely, permanently (the bad value was persisted). All
three settings (concurrency, stale/very-stale day thresholds) are now
clamped at every layer they pass through: on read, in the Settings UI, and
in the scan engine itself. -
A failed scan of the site itself could report as an empty success
If the requested site's own document library listing failed (a transient
error, throttling), the scan previously completed with a "successful" empty
report instead of an error. Subsites still fail silently and are now
counted instead (see "Added" above). -
A scan whose results couldn't be saved to history reported as "Scan
failed" even though the scan itself succeeded and its results were on
screen (e.g. an oversized report hittingIndexedDBquota). These are now
reported as separate, non-fatal warnings. -
A folder that failed to load could be cached as permanently empty
A transient error while resolving a folder's contents (throttling, a
permission error) was cached as "this folder is empty" for the cache's
10-minute lifetime, and the Explorer showed no indication anything had
gone wrong. Failed folders are no longer cached, and the Explorer now
shows the underlying error instead of a false empty state. -
Storage Report scans could silently under-report totals
Folders that failed mid-walk (permission errors, throttling, the list view
threshold) were dropped from the report with no indication — see the new
partial-scan warning above. -
Folder/file listings were never paginated
EveryFiles/FoldersAPI call fetched a single page; a folder with an
extremely large number of items could silently lose entries beyond that
page. These calls now follow SharePoint's paging links. -
Picture Libraries and Site Pages were never scanned
The library query filtered to Document Libraries only, while the rest of
the app already treated Picture Libraries and Site Pages as scannable —
both can hold real storage and are now included everywhere. -
Changing archival thresholds didn't update already-loaded folders
The Explorer cached a folder's file list with its Active/Stale/Very-stale
classification already computed; changing the thresholds in Settings and
revisiting a previously-viewed folder kept showing the old classification.
Classification is now computed at display time from the current
thresholds. -
CSV export was vulnerable to formula injection
A file or author name starting with=,+,-, or@would be
interpreted as a formula by Excel/Sheets when the exported CSV was opened.
Such values are now neutralized with the standard leading-quote mitigation. -
Theme updates could apply a stale color
The in-app theme was only recomputed when the site's primary brand color
changed, missing updates to the other palette shades. -
A very large folder listing could crash the Explorer
Computing the largest size in a folder's file list used
Math.max(1, ...values), which can throw once the argument count gets
large enough; replaced with a plain reduction. -
Scan history trimming could get stuck over its own limit
Adding a new scan when history was already over its 10-report cap deleted
only one old report regardless of how far over the cap it was, so it could
never shrink back down. It now deletes as many as needed to settle at the
cap, and a rejected save no longer leaves the database connection open.
Changed
- Export dates are now real dates, not locale strings
Excel exports write typed date cells (sortable/filterable), and CSV
exports use ISOyyyy-mm-ddinstead of locale-formatted strings.
v1.0.0
Added
-
Explorer — Treemap and List views
A WizTree-style treemap of the current library, with click-to-zoom drill-down
into any folder, plus a sortable List view of the same folder's contents
(folders and files together, largest first by default). Both views share the
same drill-down state — switching between them keeps you in the same folder. -
Library switcher and breadcrumb navigation
A button row switches between every document library on the site without
leaving the Explorer. Breadcrumbs let you jump back to any ancestor folder
in one click. -
Folder size and file count shown on treemap cells
Folder cells display their total rolled-up size and file count (recursive,
not just immediate children) directly on the cell — and in the tooltip —
instead of size alone. -
Archival status legend with real thresholds
A shared status legend shown on the Treemap, List, and Storage Report spells
out the actual configured day thresholds — e.g. "Stale (180–364d)" — instead
of just naming the Active/Stale/Very-stale tiers. -
Excel and CSV export
The Explorer's List view and the Storage Report's results can both be
exported to a styled.xlsxworkbook or a plain.csvfile. -
Storage Report — full site scan
Scans a site (and optionally its subsites) for every file, classifying each
as Active, Stale, or Very stale. Includes configurable scope (subsites,
hidden/system libraries) and scan concurrency, a live progress bar with
file count and elapsed timer, and a sortable results table with a toggle to
show only archival candidates. -
Scan history and report compare
Every completed scan is saved automatically (IndexedDB, capped at the 10
most recent). Past scans can be reopened without re-scanning, and any two
saved scans can be compared to see size change, new archival candidates,
and resolved items over time. -
Settings screen
A dedicated in-app Settings page for archival day thresholds, scan scope,
and scan concurrency, plus instructions for setting the web part's default
view via the SharePoint property pane. -
Progress bar with live folder count
Loading a folder's sizes shows a progress bar and a "N of M folders"
counter once the folder count is known, instead of an unlabeled spinner. -
Session cache for folder sizes
Resolved folder sizes are cached for 10 minutes (sessionStorage), so
revisiting a folder — including after a page refresh — doesn't repeat the
same expensive size computation. -
Minimum clickable cell size in the treemap
The smallest items in the treemap are floored to a minimum on-screen size
so they stay clickable instead of shrinking to unclickable slivers.
Fixed
-
Folders not appearing in the treemap
Any item — folder or file — whose resolved size was zero was filtered out
of the treemap layout entirely, rather than rendering as a small but
visible, clickable cell. A folder with a rollup of 0 bytes (empty, or a
failed size lookup) would silently vanish. -
Folders showing 0 files and 0 bytes
The fallback size estimate used when SharePoint'sStorageMetricsrollup
isn't available only summed files directly inside a folder, never its
subfolders — so a folder whose content lived entirely in nested subfolders
(the common case) reported nothing. The fallback now walks the whole
subtree recursively. -
Folders still showing 0/0 after the above fix
StorageMetricsis computed by a periodic SharePoint background job that
can report a stale "0 bytes" for a folder that demonstrably has content.
The rollup is now trusted only when it reports actual content (bytes > 0);
otherwise a live recursive walk is used instead of accepting the stale
zero. -
A folder taking up to a minute to load
The live-walk fallback was opening its own bounded worker pool nested
inside an already-bounded outer loop, multiplying concurrency instead of
capping it and tripping SharePoint's request throttling. Every fallback
walk triggered by one folder load now shares a single queue, so total
concurrency stays capped regardless of how many folders need the fallback. -
Treemap rendering long thin rectangles instead of near-square cells
The squarified-treemap layout algorithm had the "which side is fixed"
logic backwards — it stretched each row across the box's longer dimension
instead of its shorter one, the part of the algorithm that actually keeps
cells square. Swapped to match the correct algorithm. -
Web part icon not appearing in the "Add a web part" picker
The configuredofficeFabricIconFontNamewas not a valid Fluent/Fabric
icon name, so the picker rendered no icon at all. -
Build/serve failing with
ENOENTafter the project rename
config/config.jsonstill referenced the pre-rename source folder and
class names, so a freshgulp bundle/gulp servefailed to find them
even though the running dev server (started before the rename) hadn't
noticed. -
Excel export silently doing nothing on failure
A failed one-time load of the Excel-generation library was cached
permanently with no retry — every future export attempt (from either
export button) would immediately re-fail. Neither export button caught or
surfaced errors either, so a failure produced no visible feedback at all. -
Every page load computing sizes for every library on the site
The library switcher only ever displays library names, never sizes, but
the app was computing a full recursive size rollup for every library up
front regardless — the single biggest contributor to the initial loading
delay. Size computation is now skipped whenever a named default library
("Documents" / "Shared Documents") is found.
Changed
-
Project and web part renamed from "Smart Data Analyzer" to "Smart
Storage Analyzer" throughout — web part title and icon, class names,
bundle output name,localStoragekey prefixes, and the scan-history
database name. -
Explorer defaults to the Treemap view, with an explicit progress
indicator while folder sizes resolve instead of a blank area.
Removed
- Editable site URL in the header. The web part now always analyzes the
site it's placed on; the "Change URL" control and its associated state
were removed to keep the header simple and match standard SPFx usage.