paperlint 2.0.0

package/fixtures/paper-typography/clean-paper/paper.tex

@@ -0,0 +1,29 @@
+\documentclass{acmart}
+\begin{filecontents*}{refs.bib}
+@inproceedings{reachable,
+  title  = {A paper a reader can actually open},
+  author = {Doe, Jane},
+  year   = {2026},
+  doi    = {10.1145/3843282.3844423}
+}
+@article{alsoreachable,
+  title = {Reachable through a url instead},
+  author = {Roe, Rick},
+  year  = {2025},
+  url   = {https://example.org/paper}
+}
+@misc{reachableviaarxiv,
+  title = {Reachable because the note carries the arXiv id},
+  author = {Poe, Pat},
+  note  = {arXiv:2310.05736}
+}
+\end{filecontents*}
+\begin{document}
+\title{A clean paper}
+\section{Introduction}
+\label{sec:intro}
+Consistent cross-references: Figure~\ref{fig:one} and Figure~\ref{fig:two}.
+Section~\ref{sec:intro} is written out, never abbreviated with a sign.
+Effect sizes are written with a leading zero: $p < 0.05$ and $d = 0.21$.
+The arXiv identifier 2310.05736 must not be read as a bare decimal.
+\end{document}

package/fixtures/paper-typography/messy-paper/paper.tex

@@ -0,0 +1,27 @@
+\documentclass{acmart}
+\begin{filecontents*}{refs.bib}
+@inproceedings{unreachableone,
+  title  = {No doi, no url, no arXiv id},
+  author = {Doe, Jane},
+  year   = {2026}
+}
+@article{unreachabletwo,
+  title  = {Also unreachable},
+  author = {Roe, Rick},
+  year   = {2025}
+}
+@misc{reachable,
+  title = {This one is fine},
+  note  = {arXiv:2310.05736}
+}
+\end{filecontents*}
+\begin{document}
+\title{A messy paper}
+\section{Introduction}
+\label{sec:intro}
+The literal sign appears here: §~\ref{sec:intro}.
+The macro form renders as the same sign: \S\ref{sec:intro} and \S~\ref{sec:intro}.
+Mixed cross-references: Fig.~\ref{fig:one} next to Figure~\ref{fig:two}.
+Bare decimals with no leading zero: $p < .05$ and $d = .21$.
+The arXiv identifier 2310.05736 must still not be counted.
+\end{document}

package/fixtures/pdf-facts/README.md

@@ -0,0 +1,22 @@
+# PDF fixtures for the pdf.js reader
+
+Real PDFs that `src/pdf-facts.harness.mjs` and `skills/render-paper/extract-pdf-facts.harness.mjs`
+read. They were measured with poppler and with pdf.js on 2026-09-24 (issue #61); the harnesses pin
+poppler's numbers, so a pass means agreement with the tool pdf.js replaced.
+
+| file               | built from                                                          | what it carries                                                                                                                   |
+| ------------------ | ------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------- |
+| `t3-all.pdf`       | `t3-all.tex`, pdflatex with the cm-super map removed                | four Type 3 fonts and nothing else: every T1 font falls back to bitmaps                                                           |
+| `t3-mixed.pdf`     | `t3-mixed.tex`, pdflatex                                            | two embedded Type 1, one Type 3, and `Times-Roman` NOT embedded — `pdffonts` prints it `Type 1 Custom no no yes`                   |
+| `ttf.pdf`          | `ttf.tex`, LuaLaTeX with the OFL font Arsenal SC                    | a composite TrueType font (`CID TrueType`) and a non-embedded Helvetica; two pages, the last a stub                               |
+| `corrupt-font.pdf` | `t3-mixed.pdf` with the embedded CMR10 program's bytes damaged      | poppler still calls CMR10 embedded (with a syntax warning); pdf.js cannot parse the program and reports it not embedded           |
+| `encrypted.pdf`    | `t3-all.pdf` saved with an AES-256 user password (PyMuPDF)          | pdf.js refuses it with `PasswordException`; the reader reports `encrypted`                                                        |
+| `hidden-text.pdf`  | `hidden-text.tex`, pdflatex                                         | two-column body text beside near-white, rotated and render-mode-3 text; page 2 holds only hidden text, so banal calls it blank    |
+
+`test/e2e/banal.mjs` also reads these files, with the real banal: it pins what banal 1.2 printed for
+each one on poppler's `pdftohtml` 24.02.0 (2026-09-25), so a pass there means rpp's pdf.js-written
+XML measures the same as poppler's. `hidden-text.pdf` exists for that test: each kind of hidden text
+changes banal's answer if it is counted.
+
+The `.tex` sources are kept for provenance and are not built by any test. `ttf.tex` expects
+`ArsenalSC-Regular.ttf` beside it (SIL Open Font License).

package/fixtures/pdf-facts/hidden-text.tex

@@ -0,0 +1,28 @@
+% hidden-text.tex — text a page-layout measurement must NOT count, beside text it must.
+% Built with pdflatex (TeX Live 2023). Three kinds of hidden text, each in its own size, so that
+% counting any of them changes what banal measures (its page types and columns):
+%   10 pt black body text             — counted
+%   near-white (gray!8) text, 12 pt    — banal drops light text itself, from the fontspec colour
+%   rotated 90 degrees, 17.28 pt       — pdftohtml writes it with width 0, so banal drops it
+%   invisible (render mode 3), 20.74 pt — pdftohtml does not write it at all
+% Page 2 holds only hidden text, so banal must call it blank.
+\documentclass[twocolumn,10pt]{article}
+\usepackage[T1]{fontenc}
+\usepackage{lmodern}
+\usepackage{xcolor}
+\usepackage{graphicx}
+\usepackage{lipsum}
+\pagestyle{empty}
+\newcommand{\invisible}[1]{\pdfliteral direct{3 Tr}#1\pdfliteral direct{0 Tr}}
+\begin{document}
+\lipsum[1-4]
+
+{\color{gray!8}\large Lightgrey watermark text that a reviewer never sees. \lipsum[5-9]\par}
+
+\noindent\rotatebox{90}{\parbox{9cm}{\LARGE Rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text}}
+\rotatebox{90}{\parbox{9cm}{\LARGE Rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text rotated margin text}}
+
+{\huge\invisible{Invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer invisible text layer}\par}
+
+\lipsum[10-11]
+\end{document}

package/fixtures/pdf-facts/t3-all.tex

@@ -0,0 +1,8 @@
+\pdfmapfile{-cm-super-t1.map}
+\documentclass{article}
+\usepackage[T1]{fontenc}
+\newcommand\para{The quick brown fox jumps over the lazy dog, measuring fonts in portable documents. }\newcommand\lipsum[1]{\para\para\para\para\para\para\para\para\par}
+\begin{document}
+\section{Type 3 test}
+\lipsum[1-3] \textbf{Bold words} and \textit{italic words}.
+\end{document}

package/fixtures/pdf-facts/t3-mixed.tex

@@ -0,0 +1,9 @@
+\pdfmapfile{-cm-super-t1.map}
+\pdfmapline{=ptmr8r Times-Roman "TeXBase1Encoding ReEncodeFont" <8r.enc}
+\documentclass{article}
+\newcommand\para{The quick brown fox jumps over the lazy dog, measuring fonts in portable documents. }\newcommand\lipsum[1]{\para\para\para\para\para\para\para\para\par}
+\begin{document}
+\section{Mixed fonts}
+\lipsum[1] {\fontencoding{T1}\selectfont Some T1 bitmap text here that goes Type 3.}
+{\fontfamily{ptm}\selectfont Times text, not embedded on purpose.}
+\end{document}