moritz-gross · Oct 19, 2026 · Sep 27, 2024 · Sep 27, 2024 · Sep 27, 2024 · May 30, 2025
diff --git a/.github/workflows/translator-alert.yml b/.github/workflows/translator-alert.yml
@@ -136,11 +136,16 @@ jobs:
               ? Array.from(shaSet).map(sha => `- https://github.com/${context.repo.owner}/${context.repo.repo}/commit/${sha}`).join("\n")
               : "(No commits specifically touched monitored files)";
 
+            const fileLinks = matchedList.map(file => {
+              const encodedPath = file.split("/").map(encodeURIComponent).join("/");
+              return `- [${file}](https://github.com/${context.repo.owner}/${context.repo.repo}/blob/${context.sha}/${encodedPath})`;
+            }).join("\n");
+
             const body = `
             Changes detected in english files.
 
             Changed files:
-            ${JSON.stringify(matchedList, null, 2)}
+            ${fileLinks}
 
             See changes in the commits affecting monitored files:
             ${commitUrls}

diff --git a/AGENTS.md b/AGENTS.md
@@ -12,7 +12,7 @@ but add common mistakes of AI agents here instead.
 - Common per-language files:
   - `ClearSpeak_Rules.yaml`, `SimpleSpeak_Rules.yaml`
   - `SharedRules/`, `unicode.yaml`, `unicode-full.yaml`, `definitions.yaml`, `navigate.yaml`
-- `build.rs` can bundle rules into `rules.zip` when `include-zip` is enabled (always the minimized tree). Shared zipper: `src/rules_archive.rs`, invoked as `cargo run --bin package-rules -- Rules <output> [--minimize]`.
+- `build.rs` can bundle rules into `rules.zip` when `include-zip` is enabled (always the minimized tree). The `lang-*` / `braille-*` features (all on by default via `rules-all`) choose which `Languages/` and `Braille/` directories it embeds; every rules directory needs a feature listed in `lang-all` / `braille-all`, and `build.rs` fails without one. Shared zipper: `src/rules_archive.rs`, invoked as `cargo run --bin package-rules -- Rules <output> [--minimize]`.
 - CI packages `Rules.zip` (verbatim) and `Rules-minimized.zip` (comments stripped from `Languages/**/unicode.yaml` and `unicode-full.yaml` only). Each language/braille subdir is stored as `<name>/<name>.zip`; top-level and `Intent/` YAML stay loose. Inner language zips use BZIP2; the outer downloadable archive uses DEFLATE so `unzip` works. Both omit `Languages/zz`; the `test-rules-package` job restores `zz` from git after unzip so unit tests still run against release-like archives.
 
 ## Translation Conventions

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -0,0 +1,56 @@
+# Contributing to MathCAT
+
+Thanks for your interest in MathCAT! This is a short intro to get you started.
+For more detail, see the [full documentation](https://daisy.github.io/MathCAT/).
+
+## Ways to help
+
+- **Report bugs / suggest features:** open an [issue](https://github.com/daisy/MathCAT/issues).
+  Include the MathML input, the output you got, and the output you expected
+  (the [MathCAT Demo](https://daisy.github.io/MathCATDemo/) is handy for this).
+- **Code:** fixes and improvements to the Rust library (`src/`) or the rules (`Rules/`).
+- **Translations:** translators have their own guide. Start with
+  [information for translators/rule writers](https://daisy.github.io/MathCAT/helpers.html).
+  If you want to add a new language, please open an issue first.
+
+## Getting set up
+
+1. Install [Rust](https://www.rust-lang.org/tools/install).
+2. Fork and clone the repo.
+3. Build and run the tests:
+   ```bash
+   cargo build
+   cargo test
+   ```
+4. For the Python tooling (e.g. translation audits), use [uv](https://docs.astral.sh/uv/).
+   See [`PythonScripts/README.md`](PythonScripts/README.md).
+
+The [Developer Guide](docs/developers.md) has more on testing and coverage.
+
+## Pull requests
+
+- Keep PRs focused: one fix or feature per PR.
+- **Add relevant unit tests whenever it makes sense.** A bug fix should usually include
+  a test that would have caught it. Trivial changes (e.g. fixing a typo in a comment)
+  don't need one.
+- Make sure `cargo test` and `cargo clippy` pass. CI runs both and treats warnings as errors.
+- Don't change expected outputs (speech/braille/navigation strings) in existing tests
+  unless that change is the point of your PR. If so, explain why in the description.
+- Link the issue your PR addresses, if there is one.
+
+## Using AI tools
+
+Using AI to help write code is fine. You are still responsible for what you submit:
+review it, understand it, and test it.
+
+Issues, PR descriptions, and comments should be **written by a human**. If you want to
+include AI-generated text, put it in a clearly labeled block at the end, for example:
+
+> **AI summary:** .
+
+Nobody should have to read through walls of generated text to find out what you mean.
+
+## Questions?
+
+Open an issue or ask in the relevant issue/PR thread. We're happy to help.
+
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -12,12 +12,49 @@ exclude = ["src/main.rs", "docs", "BrailleDocs", "PythonScripts"]    # should ha
 
 
 [features]
+default = ["rules-all"]
 "rule-coverage" = ["dep:serde_json"]
 "include-zip" = []
 "enable-logs" = ["android_logger"]
 "tts" = [ "natural-tts" ]
 "no-unsafe" = ["sxd-document-no-unsafe/no-unsafe", "sxd-xpath-no-unsafe/no-unsafe"]
 
+# Which Rules/Languages and Rules/Braille directories `include-zip` embeds. Every one is on by default.
+# To embed fewer, turn off default features and name the ones you want, for example
+# `default-features = false, features = ["include-zip", "lang-en", "braille-ueb"]`.
+# `en` is always embedded because the library falls back to it, and `build.rs` fails if a rules
+# directory has no feature here or is missing from `lang-all` / `braille-all`.
+# MathCAT itself never asks for ASCIIMath or LaTeX, but an application that copies math in those
+# notations does so through the braille API, so it needs `braille-asciimath` / `braille-latex`.
+# These only change what is embedded: without `include-zip`, the Rules directory is read from disk.
+rules-all = ["lang-all", "braille-all"]
+lang-all = ["lang-de", "lang-el", "lang-en", "lang-es", "lang-fi", "lang-fr", "lang-hu", "lang-id", "lang-nb", "lang-pl", "lang-pt", "lang-ru", "lang-sv", "lang-vi", "lang-zh"]
+braille-all = ["braille-asciimath", "braille-cmu", "braille-french", "braille-latex", "braille-nemeth", "braille-russian", "braille-swedish", "braille-ueb", "braille-vietnam"]
+lang-de = []
+lang-el = []
+lang-en = []
+lang-es = []
+lang-fi = []
+lang-fr = []
+lang-hu = []
+lang-id = []
+lang-nb = []
+lang-pl = []
+lang-pt = []
+lang-ru = []
+lang-sv = []
+lang-vi = []
+lang-zh = []
+braille-asciimath = []
+braille-cmu = []
+braille-french = []
+braille-latex = []
+braille-nemeth = []
+braille-russian = []
+braille-swedish = []
+braille-ueb = []
+braille-vietnam = []
+
 
 [dependencies]
 sxd-document-no-unsafe = "0.4.2"
@@ -44,16 +81,16 @@ elements_rs = { version = "0.2.8", default-features = false }
 
 # Runtime extracts inner language/braille zips (BZIP2). The package-rules binary also
 # writes DEFLATE outer archives for Windows/unzip-compatible Rules.zip downloads.
-# wasm uses the same features: zip 8.x bzip2 is pure Rust (libbz2-rs-sys), not clang C.
-zip = { version = "8.2", default-features = false, features = ["deflate", "bzip2"] }
+# wasm uses the same features: zip 9.x bzip2 is pure Rust (libbz2-rs-sys), not clang C.
+zip = { version = "9.0", default-features = false, features = ["deflate", "bzip2"] }
 
 android_logger = {version = "0.15.1", optional = true}
 natural-tts = { version = "0.3.1", optional = true }
 
 [build-dependencies]
 bitflags = "2.6"
 yaml-rust = "0.4"
-zip = { version = "8.2", default-features = false, features = ["deflate", "bzip2"] }
+zip = { version = "9.0", default-features = false, features = ["deflate", "bzip2"] }
 
 [dev-dependencies]
 tempfile = "3.23"

diff --git a/Rules/Intent/general.yaml b/Rules/Intent/general.yaml
@@ -347,7 +347,7 @@
 -
   name: limit
   tag: [msub, munder]
-  variables: [NoSpacesBase: "translate(*[1], ' \u00A0\u2004\u2005\u2006\u2007\u2008\u2009\u200A','')"]   # Sometimes these have (weird) spaces in them
+  variables: [NoSpacesBase: "translate(*[1], ' \u00A0\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u2061','')"]   # Sometimes these have spaces or an invisible function-application character in them
   match: "*[1][$NoSpacesBase='lim' or $NoSpacesBase='limsup' or $NoSpacesBase='liminf']"
   replace: 
   - intent:
@@ -809,6 +809,22 @@
       variables: [TableProperty: "'determinant'"]
       replace: [x: "*[2]"]
 
+-
+  # TeX's gathered environment is one derivation split across display lines,
+  # not a system whose rows are independent equations.
+  name: mtable-gathered-property
+  tag: mtable
+  match: "contains(concat(' ', normalize-space(@class), ' '), ' gathered ')"
+  replace:
+  - with:
+      variables:
+      - TableProperty: "'lines'"
+      replace:
+      - intent:
+          name: "lines"
+          children:
+          - x: "*"
+
 -
   name: mtable-equations-property
   tag: mtable

diff --git a/Rules/Intent/linear-algebra.yaml b/Rules/Intent/linear-algebra.yaml
@@ -39,6 +39,49 @@
       - x: "*[1]/*[2]"   # grab the contents of the bracketed expr
       - x: "*[2]"
 
+-
+  # TeX4ht commonly attaches the norm's subscript to the closing bar only:
+  # <mrow><mo>∥</mo>...<msub><mo>∥</mo><mi>p</mi></msub></mrow>.
+  name: norm-with-subscripted-closing-bar
+  tag: mrow
+  match:
+    - "count(*) >= 3 and *[1][.='∥'] and *[last()][self::m:msub and *[1][.='∥']]"
+  replace:
+  - intent:
+      name: "subscripted-norm"
+      attrs: "data-intent-property='concat(data-intent-property, \":function:\")'"
+      children:
+      - x: "*[position() > 1 and position() < last()]"
+      - x: "*[last()]/*[2]"
+
+-
+  # The same TeX4ht shape can use two ordinary vertical bars for each double bar.
+  name: norm-with-doubled-bars-and-subscripted-closing-bar
+  tag: mrow
+  match:
+    - "count(*) >= 5 and *[1][.='|'] and *[2][.='|'] and *[last()-1][.='|'] and *[last()][self::m:msub and *[1][.='|']]"
+  replace:
+  - intent:
+      name: "subscripted-norm"
+      attrs: "data-intent-property='concat(data-intent-property, \":function:\")'"
+      children:
+      - x: "*[position() > 2 and position() < last()-1]"
+      - x: "*[last()]/*[2]"
+
+-
+  # Canonicalization groups ||x||_p as an indexed row containing a nested |x|.
+  name: norm-with-canonicalized-doubled-bars
+  tag: msub
+  match:
+    - "*[1][self::m:mrow and count(*)=3 and *[1][.='|'] and *[2][self::m:mrow and IsBracketed(., '|', '|')] and *[3][.='|']]"
+  replace:
+  - intent:
+      name: "subscripted-norm"
+      attrs: "data-intent-property='concat(data-intent-property, \":function:\")'"
+      children:
+      - x: "*[1]/*[2]/*[2]"
+      - x: "*[2]"
+
 -
   name: vector-magnitude
   tag: mrow

diff --git a/Rules/Languages/hu/definitions.yaml b/Rules/Languages/hu/definitions.yaml
@@ -351,6 +351,10 @@
     ## Other : Not tested, don't appear in https://w3c.github.io/mathml-docs/intent-core-concepts/
     "modified-variable": "silent= "
   }
+  # The connector word to say for functional 'IntentMappings' (in English, "of")
+  # This is likely the word used for U+2061 in unicode.yaml, but for technical reasons needs to be defined here
+  # vector of length=1 because scalars not supported
+- FunctionApplicationWord: ["a(z)"]
 
   # Names of functions that in terse mode don't say "of" (or it's equivalent in other languages)
 - TerseFunctionNames: {
@@ -359,30 +363,30 @@
 
 - NavigationParts: {
     # These are the parts of a formula that can be navigated to
-    "large-op": "base; lower limit; upper limit",
-    "mfrac": "numerator; denominator",
+    "large-op": "alap; alsó határ; felső határ",
+    "mfrac": "számláló; nevező",
     "fraction": "számláló; nevező",
     "msqrt": "gyök",
     "square-root": "gyök",
-    "mroot": "root; root index",
+    "mroot": "gyök; gyök index",
     "root": "gyök; gyök index",
-    "msub": "base; subscript",
-    "sub": "base; subscript",
-    "logarithm-with-base": "base",
-    "indexed-by": "base; subscript",
-    "msup": "base; superscript",
-    "say-super": "base; superscript",
-    "skip-super": "base; superscript",
-    "power": "base; exponent",
-    "msubsup": "base; subscript; superscript",
-    "munder": "base; under limit",
-    "mover": "base; upper limit",
-    "munderover": "base; under limit; upper limit",
+    "msub": "alap; alsó index",
+    "sub": "alap; alsó index",
+    "logarithm-with-base": "alap logaritmus",
+    "indexed-by": "alap; alsó index",
+    "msup": "alap; felső index",
+    "say-super": "alap; felső index",
+    "skip-super": "alap; felső index",
+    "power": "alap; kitevő",
+    "msubsup": "alap; alsó index; felső index",
+    "munder": "alap; határérték alatt",
+    "mover": "alap; határérték felett",
+    "munderover": "alap; határérték alatt; határérték felett",
 
     # words for moving into and out of one of the parts (e.g., "move right 'out of' numerator, 'in' denominator")
     # it's a hack to put them here, but at least they are grouped with the other navigation parts
-    "in": "in",
-    "out": "out of",
+    "in": "ban/ben",
+    "out": "ki",
   }
 
 - KnownWords: {