1531263bd235979267e369fe1b5f6fea7d257158
7 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
|
1531263bd2 |
frontend: let a figure be sized by its height as well as its width
A writer with a tall panel does not want to shrink it from the left edge: what
they know is "this one is too tall". The format had a width and nothing else, so
the only way to make a figure shorter was to guess a percentage and check the
page — and the previous commit's answer, "type 638px", asked them to do the
arithmetic themselves.
So `{height=300px}` joins `{width=…}`, in the same attribute block, and the
editor grows a second box beside the first. Both are typeable as well as
pre-set, both write through the reader's own parse function, and the hint beside
them says what the picture will measure on the page.
Three rules, and the point of all three is that a figure cannot be stretched:
* **a width is exact, a height is a cap.** A width has a reference — the column,
or a length the writer chose — while nothing measures a height against
anything, and enlarging a picture past its own pixels only makes it soft. In
the flow of a sentence a height is always a downscale, which comes to the same
thing as exact.
* **given one, the other follows from the picture's own shape** — `max-height`
for a height, `width` for a width — so the height is never written by hand.
* **given both, they are a box to fit inside**, not a box to fill. Two numbers
that need not have the picture's proportions can only mean "no bigger than",
and the check asserts the drawn rectangle still matches the file's ratio.
Whichever box is left empty keeps the other's setting, so a size can be set in
two steps, and clearing either goes back to the natural size.
`check-image.mjs`: 26 checks to 36. Two of the new ones found real bugs while
being written. The first: with both a width and a height in one `{…}` block,
neither side parsed — the patterns were anchored to the whole block, so a figure
with two numbers silently printed at its natural size. The attribute block is now
split into declarations first. The second was mine: returning "contain" for a
height-only figure left a 1276×2000 box around a small picture, which the caption
then sat under. A height is a cap, and a cap needs no contain.
The MCP tool takes the same two arguments, so a model can write either number,
and a percentage height is refused with the reason it cannot work: a sentence is
flowed text, so there is nothing for it to be a percentage of and the browser
ignores it. v8 is deployed and DSH has verified both numbers round-tripping.
|
||
|
|
48f0c12fd2 |
frontend: let the writer set a figure's size in the editor
Figures come out of a plotting tool at whatever size they were made. On the
paper this was written for, the sixteen of them are 1441, 1857, 2409 and 4169
pixels wide — every one wider than the 1276px column — so with no size they all
printed at the full width, and the near-square ones (1857×1986, 1974×2068) took
1365px of height, a page each. Measured in the browser, all sixteen came out at
exactly 1276px: not one of them was the size its author wanted.
`{width=…}` has been in the syntax all along, so this is the control that was
missing, not the feature. A row that is a figure now carries 宽度 in its header:
25/50/70/100%, or any value typed in — the size a panel wants is what it
measures, not a round number — with 原尺寸 as the empty value. Beside it, what
the picture will measure on the page (`原图 1857×1986 · 显示 893×955`), worked
out against the column actually rendered behind the dialog, which is the number
the writer is deciding with.
Two rules keep it honest. The value goes through `parseImageWidth` — the same
function that parses the attribute when the paper is printed — so what the
control shows is what will be printed, and a value it cannot read is refused
with the forms that work instead of stored as a size that quietly does nothing.
And the attribute is written into the sentence content by `withImageWidth`,
which edits only the first picture: the control is offered for a figure, and a
sentence with words around its picture is not one. Reading the size back out of
the content rather than keeping a copy of it in the draft means there is one
place it can be wrong instead of two.
`check-image.mjs` grows from 15 checks to 24, driving the real control: that
only figure rows offer one, that it shows what each is set to, that typing 30%
converts to 383×239 immediately, that what was typed is what gets stored, that
the paper then prints it at 30% of the column, that clearing it goes back to the
natural 480×300, and that "宽一点" is refused while the sentence keeps its size.
The MCP side of the same rule was wrong and is fixed with it: `image_upload(width=
"60")` wrote `{width=60px}`, and the reader takes a bare number as a share of the
column — `parseImageWidth` only reads pixels when the value says `px`. A model
asking for 60% got a 60-pixel figure. v7 is deployed with the corrected reading,
and smoke_mcp.py checks it.
|
||
|
|
ebaf9e558b |
backend: let a model put a figure in the paper, and set a formula's alignment
Two things the editor gained and MCP could not do, both because the tool layer
had been told about the domain's columns by hand.
**A figure.** Bytes go to MinIO and the sentence keeps a Markdown address, so a
picture is ordinary text — which means the writing tools already carry it, and
the only half a model cannot do is the upload: it has bytes and no way to send
multipart. `image_upload` takes base64 and answers with the URL, the Markdown
and the *sentence object* the editor produces when a picture is dropped in, so a
figure is written by putting that object into `sentences` unchanged. It does not
fetch a URL the caller names: a service inside the cluster that will GET any
address it is handed is an SSRF primitive aimed at everything beside it.
The width argument is the reader's rule or it is refused. `{width=70%}` is
parsed by `frontend/src/utils/rich.ts`, and a value it cannot read is not an
error there — the picture just prints at its natural size. A size that silently
does not apply is worse than a refusal, so the syntax and the range (pixels
1–4000, per cent 1–100) are mirrored in `app/mcp/tools/images.py`, with the
usual comment naming the other side.
`paragraph_text` also had to learn what a figure is: a sentence holding nothing
but an image is a 插图 — the view prints it on its own line with the brackets'
text as its caption — and the text a tool hands back broke lines around a
display formula but not around a picture. Same rule as `figureOf`, same pair of
implementations, same warning about changing both.
**A formula's alignment.** `paper_sentence.formula_align` reached the editor and
the REST API and stopped there: `SentenceSpec` had no such field, so a model that
sent one got a successful call and a centred formula. The column is now declared,
carried through every write path and reported by every read path.
That is the second column lost the same way, so the fix is structural rather
than another two edits. `SENTENCE_FIELDS` is read from
`app.schemas.paper.SentenceInput` instead of written out, the payload is built
by one `sentence_payload()`, and the three append paths spread
`SentenceCreate(**item.model_dump())` rather than copying fields. One edit
remains — declaring the field in `specs.py` — and smoke_mcp.py's first step
compares the two field sets, so forgetting it now fails the suite instead of
losing the field. `sort` is declared too, which it never was: a model that sent
one had it dropped in silence.
smoke_mcp.py: 99 checks to 119. The figure step uploads a real 1x1 GIF (the same
bytes twice, and the key is the same object), writes the returned sentence as it
came back, checks the paragraph prints it alone on its line and that an image
inside a sentence does not break the line, and takes the four refusals — not a
picture, an unparseable size, an out-of-range one, not base64. The alignment step
writes a right-aligned formula, moves the sentence to left on its own, and takes
the paragraph to another position to prove the setting travels with it.
Deployed as v6, and the pod now carries MINIO_ENDPOINT (the in-cluster Service)
plus the two credential keys from the Secret; without them the image tool answers
with the missing variable names and every writing tool still works.
|
||
|
|
9f2bbe50eb |
backend: let a model ask for a line break, and read the 参考文献 list back
The paper model gained two things while the MCP layer was not looking, and both were invisible from a tool call. `paper_sentence.line_break_before` — the flag a paragraph needs so a formula, a list of items or an author line can start a line of its own — was not in `SentenceSpec`, and `sentence_inputs` dropped it on the floor: a model could write the field, watch the call succeed, and get prose back. It is now part of every write path (`paragraph_write`, `paper_write`, `sentence_add`, `sentence_update`) and of every read path (`sentence_row` reports it, so a model can see the lines before moving one). Two bugs fell out of doing it properly: * `paragraph_move` rebuilt its sentences without the flag, so moving a paragraph that carried a formula turned it into prose. It also passed the *source* sorts through, which interleaved the moved sentences with the target's own — the merged paragraph read in an order neither writer wrote. Both are fixed; the tool now appends, as its own description promised. * `document_text` printed a formula run into the sentence before it. The paper view renders `$$…$$` as a block, so the seam rules are now the same in both places, with a comment on each side naming the other. `PaperDocumentRead.references` and `reference_style` — the per-work 参考文献 list the backend now assembles and formats — were reachable over REST but not over MCP: `paper_document(format="json")` returned citations and nothing to point them at, even though docs/MCP.md already described the list. The json form now carries the entries (index, reference_id, text, missing, citations) and the style name, so a model quotes the bibliography instead of assembling a second one out of the same fields. smoke_mcp.py grows from 76 checks to 99 over two new steps — the line-break step pins the display-formula rule, the marker toggle and the moving case; the bibliography step cites one work from two sentences and checks that the text, the entry and both markers agree on one number. |
||
|
|
15955f552d |
backend: give the reference library the MCP tools it was missing, and fix the one that broke
The reference library landed with a full REST API and no MCP surface at all: an agent could cite `reference_id: 42` — the citation is validated now — but could not look up what 42 was, create it, or find out whether the library already had it. That is the citation half of "generate the paper, then write it in" missing, and it is the half an agent is worst at doing by hand. Twelve tools, mirroring the REST surface plus one addition each way: - `reference_*` — search, read, create, update, delete, plus `reference_import`, which takes a whole pasted bibliography in one call and reports what it skipped (an existing DOI or an identical title) and what it could not read, one bad entry at a time rather than losing the other nineteen; and `reference_citations`, which answers "what is this cited for" before a rename or a delete. `reference_list`'s keyword search across title, container, DOI and author names is the de-duplication path that `reference_import` uses. - `reference_tag_*` — the five CRUD tools. Deleting a tag drops the labelling and leaves the literature, which is the opposite of what the field library needs and the same choice the REST route makes. Two things had to be fixed to get here, both found by testing rather than reading: `template_list` was crashing. The rename that turned a heading into a title and its use into a placement changed `TemplateListItem.field_count` to `placement_count` and left the tool reading the old attribute, so the call died with `Error executing tool template_list` — unreadable, and live in the cluster. The suite passed 47 checks while this was broken because it never called `template_list`: the uncovered tools were exactly the read-only ones, so the write paths were all green and the browse path was not. The suite now covers all 41 tools — 76 checks — and step 5 asserts the shape each read tool returns, not merely that it returned. `DOMAIN_REFUSALS` in the tool registry held one class, `UnknownReference`. The tag rules arrived with the tag tools and raise two more, so 「标签不存在:9」 and 「标签「必读」已存在」 would have reached a model as `Error executing tool …`. The tuple is now the list of every refusal a tool is expected to translate, and a new one belongs in it on the day it is written. `reference_update` takes its changes as one nested object rather than 38 keyword parameters: a nested model keeps `model_fields_set`, so an omitted field stays absent while an explicit null clears the column. Flattened, the two are indistinguishable — every omitted field arrives as None — which is the same problem `paper_update` solves by rebuilding its payload from what it was given. Verified against the deployed service, not a local process: 76 checks pass over the NodePort, then the same 76 over stdio and against a local HTTP server, and the REST suite still passes its 55. The image is rebuilt as v4 and the Deployment follows it, because `imagePullPolicy: IfNotPresent` means re-pushing a tag the node already holds changes nothing. |
||
|
|
103b5544ba |
deploy: roll the MCP server onto the renamed schema
The running pod predates the table rename and fails every tool that touches the template tables, so the image is rebuilt as v3 and the Deployment follows it. The manifest is bumped with it — `imagePullPolicy: IfNotPresent` means re-pushing the same tag would keep serving the cached layers, so a rebuild needs a new tag to reach the cluster at all. Verified by handshake against the running server: `field_list` answers with the 45 library titles. |
||
|
|
7e7820f823 |
deploy: run the MCP server in k3s under the paper-doc namespace
The server has to be reachable by the Harness, Claude Code and Codex, and all three of those are clients of an address rather than of a source tree — a stdio child process only serves whichever machine spawned it. So it is packaged and deployed: namespace `paper-doc`, one Deployment, one NodePort Service at `http://192.168.1.88:30931/mcp` behind a bearer token, talking to the in-cluster TiDB Service rather than the node's NodePort. The namespace is `paper-doc` and not `paper_doc`, which is not a preference: a namespace name is an RFC 1123 label, underscores are outside that alphabet, and the API server rejects it outright. The database keeps its underscore. Two routes were added to the server for the deployment, and they are the only unauthenticated ones. `/healthz` answers "is this process alive" without touching anything, `/readyz` answers "can it serve" by running `SELECT 1`. The split is the point: a failing readiness probe takes the pod out of the Service's endpoints and puts it back when TiDB returns, while a failing liveness probe kills a process that was working correctly. There is no container runtime on this host — k3s runs containerd and neither `ctr` nor `crictl` builds from a Dockerfile — so `deploy/build-mcp-image.sh` builds inside a throwaway privileged docker-in-docker pod, mounting the working tree through a hostPath so the build context is the tree you are looking at, uncommitted edits included. It reads the registry credentials out of the cluster's own pull secret, so no password is in the script or in a shell history. One Dockerfile serves both entry points, because the API and the MCP server share the domain layer and differ by one argv. Verified against the deployed Service, not against a local process: 47 MCP checks pass over the NodePort, and the Node MCP SDK — the one the Harness's mcp-client plugin uses — lists all 29 tools and calls them. The same 47 pass in stdio and against a local HTTP server, and the REST suite still passes its 45. |