Since #849 the parts list has a row for every contact block (slave) and
terminal-type element. Users who draw the dots and bends where wires
branch as terminal-type symbols (114_connections) get one empty row per
junction: 109 of 151 rows on one reported cabinet.
Two options, both off by default so existing exports are unchanged:
- leave out the contact blocks: --no-slaves, "no_slaves" in qet_export,
or uncheck the new "Contacts esclaves" element type in the dialog;
- leave out the junctions: terminal-type elements with no label,
designation, manufacturer or manufacturer reference. --no-junctions,
"no_junctions", or "Laisser de côté les jonctions" in the dialog.
A terminal block with a label or a part number stays.
The element type filter had no box for slaves, so every query it built
left them out: the export dialog never listed contact blocks, although
the command line has since #849. The dialog now checks the new box by
default and gives the same rows as --export-bom. The box is unchecked by
default elsewhere, so nomenclature tables keep their rows.
docs/smart-device-bom.md still said slaves and terminals were excluded;
corrected.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
QetSettings::houseStyle()/setHouseStyle() (assistant/house_style),
qet.houseStyle()/qet.setHouseStyle() in the script API, a house_style
field in qet-assistant.json, and qet_about in the qet MCP server
showing it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
QElectroTech (LiveServer):
- print: folios to the default (or a named) printer with no print dialog,
always after asking the user (paper cannot be undone); or to a new PDF
file. Rendered as the PDF export draws a folio (no grid, no terminals).
- save_project (own file, or save as a new file, never over one) and
close_project (only without unsaved changes: never discards work).
- changes: the undo history, each step marked assistant or user, with
"since" to ask what changed after a point.
- snapshot: a copy of the open project as on screen, for a check to read,
without saving it or changing its file.
- qet.elementTextGeometry(): where a symbol's text is drawn on the folio.
qet MCP server:
- qet_live_print, qet_live_save_project, qet_live_close_project,
qet_live_changes, qet_live_layout_check (snapshot + qet_layout_check, no
"run this script?" window); write paths never overwrite.
- Layout rules from the house style, for file and live checks alike:
label_on_wire, four_way_junction, misaligned_branch.
- The element index is saved under ~/.cache/qet-mcp: a new server process
loads 8839 symbols in 0.4 s instead of parsing them for 6.9 s; the
change check is a third faster.
- qet_about lists the installation's standard symbols
(standard-symbols.json in QElectroTech's data folder) when there are any.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
open_project opens a .qet and makes it current, as File > Open does but
with no file dialog and no error windows: a missing, unreadable or
foreign file is refused in the answer. A project already open is only
made current. switch_project makes another open project current, by its
index or its file. status now lists every open project (index, title,
file, folios, unsaved changes, current), so the assistant sees what it
can switch to. MCP: qet_live_open_project, qet_live_switch_project; both
paths go through the workspace policy.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An assistant in live mode could only work on the project already open.
new_project makes one as File > New does -- this installation's
new-folio defaults, optional title and number of folios -- and makes it
the current project, so the next request works on it. An optional
absolute path saves it at once, never over an existing file: QElectroTech
refuses, and the MCP server refuses before asking (a new never-overwrite
class in its path policy, so an "overwrite" flag cannot get round it).
Pages and links across pages already worked through scripts
(qet.addFolio, setFolioTitle, linkElements across folios); the tool's
description now says so.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A request with "timing": true gets queue, confirm and exec times in its
answer; with QET_LIVE_PERF_LOG set, QElectroTech also appends one JSON
line per request, with the time until the folio is next painted. The
qet MCP server logs its side of each live call to the same file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
qet_element_search matched every word as a substring of the names and
the path, so "NC contact" offered a remanence coil ("nc" in "remanence",
"contact" in its folder) and an outlet before any auxiliary contact,
"normally closed contact" found nothing, and "emergency stop" ranked a
maker's part and two assembly-plan drawings above the push button.
- A word of one or two letters must now be a whole word. Longer words,
and short ones with a digit, still match inside a word, so German
compounds ("schutz", "klemme") and "3p" find what they found before;
on 37 sample queries no result is lost.
- NO, N/O, "normally open", "normalement ouvert" are one word, and NC,
N/C, NF, "normally closed", "normalement fermé" another.
- Symbols under *_graphics, *_manufacturers_articles and
*_miscellaneous_unsorted come after every other match.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Schemas now get saved with a UUID and do not have to be looked up according to the form. They can be renamed and it gets reflected in existing elements. The GUI for
editing them is now much easier, e.g. forms can be reshuffled. A formula can be applied. Many options now also for conductors and sheets.
Discussion #1158 asks for a limit on the wires per terminal. Before any
limit, two read-only checks let anyone count their own projects:
- crowded_terminals: terminals with more than four wires (two double
ferrules, one either side of the screw), folio reports left out;
- reports_with_several_wires: folio report arrows with more than one
wire, the rule plc-user proposes for reports.
Both are info, not errors: cable, busbar and single-line symbols carry
more wires on purpose, and a third of the shipped examples' reports have
several. Over the 24 examples: 23 crowded terminals in 10 projects
(the worst, 19, is a cable symbol in affuteuse_250h.qet) and 130 reports
in 4 projects. Plain SQL over the conductor table, so they work on any
current QElectroTech.
test_qet_mcp: the checks are read-only, the exact-answer test lists them,
and a coil wired to five others is reported once with its count (checked
to fail with the threshold raised). 398 tests pass.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015FPuYPS4T7QuEwjNu22rXD
A wire is straight only when its two terminals are exactly in line, and
qet_edit placed a symbol only by its origin, so an assistant had to guess
where the terminals would land -- and usually missed by a few pixels.
- place_element adds a symbol with one of its terminals in line with
another symbol's, gap pixels away on the side that terminal faces, and
notes when the new symbol's terminal faces the wrong way to meet it.
- align_terminal moves a placed symbol across so two terminals line up.
- align_elements lines up symbols' boxes (left/center/right/top/middle/
bottom), as Edit > Align does.
- distribute_elements spaces symbols' origins evenly, or a pitch apart,
keeping symbols on the grid on it.
place_element and align_terminal need qet.terminalPosition(); on a build
without it the edit stops before running, naming the missing call, and
every other op still works there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015FPuYPS4T7QuEwjNu22rXD
A wire is straight only when its two terminals are exactly in line, and a
symbol is placed by its origin with its terminals at an offset from it, so
symbols an assistant places "under each other" by eye land a few pixels
apart and the wire jogs. Nothing told it so.
qet_layout_check reads every symbol and every wire's drawn path through
the scripting API (the default path is not saved in the file) and reports
a 0-100 score with: wires that jog where moving one symbol would make them
straight, extra bends, wires through symbols, overlaps, symbols off the
grid and crossings. "fixes" is one move_element per symbol, planned
together so they can be applied in one qet_edit call: a straight wire pins
its symbols, a move never lands a symbol on another or across a wire, and
symbols lined up with each other go onto the grid together. Styles: iec
(columns), nfpa (rungs) or auto. Read-only.
Over the 23 shipped examples, applying the fixes never lowers a score and
a second check proposes no further moves on 21 of them; a sloppy
assistant-style drawing goes from 40 to 100 in one round, in both styles.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015FPuYPS4T7QuEwjNu22rXD
Reported on #1178 after a first try of live mode on macOS:
- The English interface showed the live-mode setting in French. The
strings were in lang/qet_en.ts but untranslated. This fills in English
for live mode, the script buttons, the macro recorder and the Connect
an AI assistant dialog (101 entries); other features' untranslated
English entries are left for their own change.
- The setting is on Configurer QElectroTech > Général > Projets, but the
start-up warning, the MCP server's error and its README all said
"Configurer > Général". All three now name the Projects tab, and the
README and the server name the English menus and buttons.
- qet_live_screenshot grabbed the whole viewport, so a tall window around
a landscape folio came back mostly empty. It now crops to the folio's
frame and title block where they are on screen (the whole viewport if
the folio is scrolled out of sight), and says so in cropped_to_folio.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Two exports of one project differed in the creation and modification
dates (the time of the export, in local time) and in the document id
(random). With SOURCE_DATE_EPOCH set (reproducible-builds.org),
--export-pdf now uses the time it names, in UTC, and a document id
derived from the project: version 5 of the project uuid with the
SHA-256 of the project file, so it is the same for the same file and
changes when the file does. Qt has no setter for the dates, so
PdfLinks::setDocumentDate() rewrites them in the document information
and the XMP metadata after the file is written, fixing the XMP /Length
and shifting the xref table. Without the variable nothing changes.
Only --export-pdf reads it; the print window's PDF export does not, as
a person exporting by hand wants the real date.
qet_export (misc/qet-mcp) gains "reproducible" and "source_date_epoch".
It reports whether the build honoured the variable, since an older one
ignores it.
tst_pdfreproducible exports examples/741.qet four times and requires
identical bytes and the epoch's date; it fails with either this change
or the hash seed fix reverted. Projects with cross-reference links also
need their links in a fixed order, which is a separate fix.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
setFolioBorder(folio, "preset", "tabloid-landscape") picks the column and
row counts and whole-number sizes that fill the sheet best without going
over it, as one ChangeBorderCommand. The title block and headers are
measured from the folio, so it holds for any template on either edge.
Sizes stay whole because the folio properties panel edits them in whole
pixels. Among fills within 1.5 pt of the sheet, the size nearest the
current one wins: from a new folio tabloid landscape is 23 x 70 by 12 x 82,
not 9 x 179 by 5 x 197. A0-A5, letter, legal, tabloid/ledger, each
-portrait or -landscape; folioPresets() lists them, and folioBorder() reads
"width" and "height" of the frame and title block.
Every preset exports on its sheet; landscape ones rely on the previous
commit, which turns a wide page before matching it to a sheet.
The MCP server checks the name up front, probes for folioPresets(), and
the op's note says what was chosen and the frame's size in points.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
A new conductor gets the default two or three straight segments, which
run through whatever symbol or wire lies between its terminals; scripts
then fix it segment by segment with moveConductorSegment().
ConductorRouter finds an orthogonal path on the folio grid that leaves
and enters each terminal in its own direction, keeps clear of every
element's rectangle, stays inside the border, and charges for bends and
for running along or crossing other wires. qet.routeConductor() and
qet.routeConductorBetween() apply it through Conductor::setPathPoints(),
which pushes the same ChangeConductorCommand a handle drag does, so the
path is saved, survives a reload, and one undo restores the default.
Where no route exists the wire keeps its path and the call says so.
qet-mcp: add_conductor takes "route": "avoid", and a route_conductor op
reroutes an existing conductor (by terminal or by uuid). The router
methods are required only by an edit that routes.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
qet_recording_list and qet_recording_read fetch what QElectroTech's macro
recorder saved: each step from the undo history as structured changes
(consecutive folio states diffed with qet_diff), and the overall change
from before.qet to after.qet. qet_recording_check runs a script on a copy
of before.qet from where the person started -- the folio on screen (a
proxy answers qet.currentFolio(), which a Qt object will not let a script
replace) and the selection -- and says whether the result matches
after.qet, or what differs. qet_recording_remove deletes one.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
QElectroTech now advertises its live channel in the "live" part of
qet-assistant.json instead of a live-session.json of its own. When nothing
is listening, the error now tells apart QElectroTech not running, live
mode switched off, and the start-up warning not accepted.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
QElectroTech now writes qet-assistant.json with its folders, features,
script calls and stored scripts. The server reads it to find the scripts
folder -- right even when QElectroTech runs with --data-dir, where the
per-platform guess was wrong -- and qet_about shows it, never the live
token. The initialize reply now carries instructions: what QElectroTech
is, headless and live, the usual order of tools, start with qet_about.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
qet_live_screenshot returns an MCP image the assistant can look at;
--call prints an image part as a data: URI instead of failing on it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
qet_live_status, qet_live_run_script and qet_live_run_stored talk to a
running QElectroTech through the channel its LiveServer opens, only when
live mode is switched on in its settings and accepted at this start. They
find it through live-session.json in QElectroTech's data folder and send
its token on every request; with no session, the error says which of the
three switches is missing. Running needs QET_ENABLE_SCRIPTING=1 as editing
does; asking for the status does not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Checked against a real Windows package under Wine: QElectroTech keeps its
data in %APPDATA%\QElectroTech\QElectroTech, next to which the scripts
folder goes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
qet_script_api lists every qet.* call, asked of the QElectroTech that will
run it (qet.apiSignatures() where the build has it, names otherwise), with
the header format that makes a script a button.
qet_script_test runs a script's text on a copy of a project and returns the
qet_diff of what it changed, its log and its errors with their line.
qet_script_install stores a script, and optionally an SVG icon, in the
scripts folder QElectroTech reads; with test_project it tests first and
stores nothing if the test fails. qet_script_list, qet_script_read and
qet_script_remove complete the set.
The folder is the server's choice, never the call's: ids are checked as
file names, and storing or removing needs QET_ENABLE_SCRIPTING=1 like an
edit, because a stored script runs when the user clicks it. A header is
checked with the same rules as QElectroTech's own, so nothing is stored
that would get no button.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The "one text per potential" switch (onetextperfolio) lives in a folio's
conductor defaults, which no scripting call or qet_edit op could reach, so
callers patched the saved .qet afterwards (#1178).
qet.setConductorDefault(folio, property, value) sets onetextperfolio or any
setConductorProperty name on a folio's defaults, or with folio -1 on the
project's defaults that new folios copy. A change to onetextperfolio
re-shows or hides the conductor texts at once, as the Folio properties
dialog does. Not on the undo stack, like both dialogs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The MCP server gives each QElectroTech run a private HOME and writes the
element collection path for it (elements_dir) into
~/.config/QElectroTech/QElectroTech.conf there. Only Linux reads that
file: Qt keeps settings in the registry on Windows and in the system
preferences on macOS. So on those systems elements_dir did nothing, and
every run read the user's own settings (#1178).
When QET_SETTINGS_DIR names a folder, QElectroTech now keeps its
settings in <folder>/QElectroTech/QElectroTech.ini on every system. It is
set in main() before the first setting is read. Without the variable
nothing changes.
The server sets it for each run and writes the collection path to the
.ini as well as the .conf, so an older QElectroTech keeps working on
Linux. The path is written with forward slashes: Qt reads a backslash in
these files as an escape. The README notes that elements_dir needs this
on Windows and macOS, and that long arguments go on stdin with "-"
(Windows refuses a command line over 32,767 characters).
tst_settingsdir: a script places a symbol that only the folder's settings
can resolve, while the usual settings file points at an empty collection.
It fails on master and passes here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On Windows every tool that starts QElectroTech (qet_export, qet_edit,
qet_query, qet_check, qet_continuity, qet_project_new) failed, for two
reasons:
- The server ran a copy of the executable from a temporary folder, to get
its own SingleApplication key. A Windows program loads its DLLs from its
own folder, so the copy died before main() with 0xC0000135 (DLL not
found). Every flag the server passes is a CLI export flag or --run, and
main.cpp handles both before it constructs SingleApplication, so on
Windows the original is now run. The copy stays elsewhere.
- It set QT_QPA_PLATFORM=offscreen. The Windows packages ship only the
qwindows platform plugin, so Qt found none and stopped at a message box
nobody could close: every call hung until its timeout. Windows now keeps
its default platform; the export flags and --run open no window.
Checked under Wine (qet-wine-smoke) on the fork's CI Windows build, run
through python.org's embeddable Python: before, qet_export ended with
exit 3221225781; after, a PDF export, a qet_query (98 elements, as on
Linux) and a qet_edit placing a common:// element all succeed. The hang
was isolated by launching the same export from bash (works) and from
Python with one change at a time: only dropping QT_QPA_PLATFORM made it
work. Four unit tests pin both choices per platform; each fails with its
fix removed. Suite 262/262 none skipped on Linux.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The MCP server (misc/qet-mcp) was only reachable from a source checkout.
It now ships with every package, next to the program, and finds that
QElectroTech and its element collection from where it sits:
- make install (Linux distributions, snap, flatpak):
<prefix>/share/qelectrotech/mcp/qet_mcp.py, executable, with its README.
- Windows installer: a new "AI assistant (MCP)" component (on by default,
~200 KB), installed to <folder>\mcp. The workflow stages files/mcp, so
the portable folder and the MSI, which packs all of files/, carry it too.
The server learns the Windows layout (<root>/mcp beside <root>/bin, whose
program is QElectroTech.exe, and <root>/elements), in addition to
<prefix>/share/qelectrotech/mcp. A copy saved anywhere else, even beside
some bin/ folder, is not taken for an installation.
The installer strings are given in all 29 installer languages: French
translated, the others in English until translated, which is what NSIS
would fall back to anyway but without its warning 6040 per language.
Checked: make install into a scratch prefix, then from the installed
script with nothing configured and no qelectrotech on PATH, an SVG export
and a qet_edit placing a common:// element both succeeded. makensis 3.10
compiles the installer with 0 warnings before and after (removing the
French description gives warning 6040), and the installer contains
mcp/qet_mcp.py. Suite 274/274 none skipped; each layout check was
removed in turn and a test failed. Snap/flatpak: installed by the same
CMake rule; launching their QElectroTech from outside is untested.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
qet_export, qet_edit, qet_query, qet_continuity, qet_check and
qet_project_new took the QElectroTech executable as a per-call argument
and ran whatever executable file it named, with the call's own paths as
arguments. The workspace policy exempted it as configuration, but it is
chosen by the model on every call, so text inside a project could steer
an assistant into starting another program.
The server now finds QElectroTech itself: QET_BINARY, then the install it
ships in (<prefix>/share/qelectrotech/mcp/), then PATH. "binary" becomes
optional; when given it must be that same file (after resolving
symlinks) or one listed in QET_MCP_BINARIES. QET_MCP_ALLOW_ANY_BINARY=1
restores the old behaviour, as QET_MCP_ALLOW_ANY_PATH does for paths.
"elements_dir" defaults to the installed collection and, when given, must
be in the workspace, that collection, or QET_MCP_ELEMENTS.
Both checks live in enforce_path_policy(), the one place tool arguments
enter. test_configuration_paths_are_exempt asserted the old exemption and
is replaced by BinaryPolicy (13 tests) and a stdio test of the original
reproduction. Each new check was removed in turn and the tests failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The "$id" of an add_folio or insert_folio held the index the folio had
when it was made, so a later insert_folio or remove_folio in the same
qet_edit run shifted it, and ops naming "$id" edited the wrong folio --
the case #1115 fixed for folios given by uuid, left open for these.
The script now also keeps such a folio's uuid (qet.folioUuid()) and
resolves "$id", where an op takes a folio, through qet.folioIndex() at
the moment it is used. On a build without folioUuid() it falls back to
the stored index, as before, and nothing new is required of the binary.
The op's reported result is still the index.
Test: add_folio "$f" (index 1), insert_folio at 0, set_folio_title
"$f": the title lands on the third folio -- on the second without this
change. Two script-generation tests updated for the new expression.
255/255.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A web chat in a browser cannot start a local program, so it cannot run
this server. Two ways round that, documented in the README:
- the Claude desktop app runs stdio servers; a step-by-step setup;
- a chat that can execute Python can upload qet_mcp.py and run
`--call <tool> '<json>'` (or `-` to read the arguments from stdin).
--call goes through the same dispatcher as the stdio server, so the
workspace policy applies unchanged. Exit status 0 success, 1 the tool
reported an error, 2 the call was malformed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review of the previous commit: an add_conductor with both terminal uuids
wrong logged two notes under one op index and the second replaced the
first, and neither said which end it was about. Notes of one op are now
joined, and a terminal note starts with its argument ("from_terminal:",
"to_terminal:", "terminal:"). The integration test's both-ends case fails
with the old replacing behaviour.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
"terminal", "from_terminal" and "to_terminal" take the terminal's uuid
(as qet_element_info lists it) in place of its index. It names a
terminal of the op's own element -- for add_conductor, of that end's
element, "$name" references included -- and is turned at run time into
the index the call takes by qet.terminalIndex(); if the element has no
such terminal the op fails with a note. Unlike the index, a sort by
position, it is defined between two terminals at the same point. The
lookup is required only when a uuid is used, so index-only edits still
run on older builds.
README: the terminal uuids; wires of older projects now get a lasting
uuid (#1107) instead of staying unnamed.
Tests: script generation for every terminal-taking op, "$name" and folio
uuid resolution, the index still passed through; through the binary,
bobine_ka_a_remanence (file order A2, A1; index order A1, A2) wired A2 to
A1 by uuid lands on exactly those ends, and an unknown uuid stops the run
with its note. 255/255 against a build with qet.terminalIndex(); a build
without it is reported as missing it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review of the previous commit:
- The load fallback compared a saved uuid with occurrence 0 only, so a
wire on the second of two terminals at one point of a symbol was lost
once the definition was replaced, and one on the first could go to
either of the pair (Element::m_terminals is sorted, not in definition
order). Element::parseTerminal() now records each terminal's rank among
the terminals of the definition at the same point, derivedUuid() uses
it, and fillMissing() starts from the same rank.
derivedUuidFoundAfterReplacement runs on perceuse.qet and industrial.qet
too: 154/156 and 670/671 wires without the rank, all with it.
qet-mcp: the first save of an older project now rewrites its wires from
the numbered form to the uuid form, and qet_diff keyed the two forms
differently, so an untouched resave showed every wire removed and added
(4 failures in test_qet_mcp.py). A uuid end is now resolved to the same
key as a numbered one: the terminal's definition position, moved to where
the wire docks, is the placed symbol's <terminal> record.
test_conductor_key_same_in_both_forms fails without it; 253/253 pass on
this build and on the previous stage's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Laurent Trinques