new: drive the license menu, parsing and sniffing from data/licenses.yml

2026-09-18 13:52:24 +02:00
parent 12407c8eac
commit dd1c70c91a
6 changed files with 496 additions and 100 deletions
@@ -0,0 +1,139 @@
+## License knowledge of `pkh new`, in one place: the wizard menu labels
+## (src/new/questions.rs), the spellings accepted by License::parse and the
+## SPDX URL template (src/new/options.rs), and the license-file sniffing
+## inputs (src/new/detect.rs) all read this table, so the three lists —
+## previously kept in sync by comments only — cannot drift apart anymore.
+## Adding or changing a curated license is one entry below.
+##
+## Keys:
+##   id:             SPDX identifier: written to debian/copyright, returned
+##                   by the license sniff and substituted into
+##                   license_url_template (minus a trailing '+' of the
+##                   "or later" spellings)
+##   menu:           label offered by the wizard license question (the
+##                   free-text "Other (enter a SPDX identifier)" entry
+##                   stays in Rust — it is UX, not data)
+##   spellings:      inputs accepted by License::parse, matched
+##                   case-insensitively; each entry must accept its own id
+##                   (a consistency test in options.rs locks ids, spellings
+##                   and the License enum together)
+##   detect_markers: marker sets driving the LICENSE/COPYING text sniff of
+##                   detect.rs. A set matches when every marker of `all`
+##                   occurs in the lowercased license text and none of
+##                   `unless` does; an entry matches when any of its sets
+##                   does. The sets are written to be mutually exclusive:
+##                   the `unless` markers keep multi-license texts on the
+##                   entry carrying the stronger reference (a MIT-named
+##                   file also quoting the GPL or the Apache license is a
+##                   GPL/Apache file) and keep GPL sets off LGPL texts,
+##                   whose name contains theirs. The entry order below
+##                   (menu order) therefore only breaks ties.
+##
+## detect_files (top level): the candidate license file names the sniff
+## reads, in preference order, shared by every license (the case-variant
+## directory scan around them stays in Rust).
+##
+## license_url_template: the SPDX license page URL, with {id} substituted
+## for the debian/copyright reference paragraph.
+##
+## The behavioral lock for the markers is the LICENSE_TEXTS test table in
+## src/new/detect.rs: a bad marker edit fails those tests, not packages.
+
+license_url_template: https://spdx.org/licenses/{id}.html
+
+detect_files:
+  - LICENSE
+  - LICENSE.md
+  - LICENSE.txt
+  - COPYING
+  - COPYING.txt
+
+# Entries in wizard-menu order.
+licenses:
+  - id: MIT
+    menu: MIT
+    spellings: [MIT]
+    detect_markers:
+      - all:
+          - mit license
+        unless:
+          - apache license
+          - general public license
+      - all:
+          - permission is hereby granted, free of charge
+        unless:
+          - apache license
+          - general public license
+  - id: Apache-2.0
+    menu: Apache-2.0
+    spellings: [Apache-2.0]
+    detect_markers:
+      - all:
+          - apache license
+          - version 2
+  - id: GPL-2.0+
+    menu: GPL-2.0+
+    spellings: [GPL-2.0+]
+    detect_markers:
+      - all:
+          - general public license
+        unless:
+          - version 3
+          - lesser general public license
+  - id: GPL-3.0+
+    menu: GPL-3.0+
+    spellings: [GPL-3.0+]
+    detect_markers:
+      - all:
+          - general public license
+          - version 3
+        unless:
+          - lesser general public license
+  - id: LGPL-2.1+
+    menu: LGPL-2.1+
+    spellings: [LGPL-2.1+]
+    detect_markers:
+      - all:
+          - lesser general public license
+        unless:
+          - version 3
+      - all:
+          - lesser general public license
+          - version 2.1
+  - id: LGPL-3.0+
+    menu: LGPL-3.0+
+    spellings: [LGPL-3.0+]
+    detect_markers:
+      - all:
+          - lesser general public license
+          - version 3
+        unless:
+          - version 2.1
+  - id: BSD-2-Clause
+    menu: BSD-2-Clause
+    spellings: [BSD-2-Clause]
+    detect_markers:
+      - all:
+          - redistribution and use in source and binary forms
+        unless:
+          - endorse or promote
+          - isc license
+          - permission to use, copy, modify, and/or distribute this software
+  - id: BSD-3-Clause
+    menu: BSD-3-Clause
+    spellings: [BSD-3-Clause]
+    detect_markers:
+      - all:
+          - redistribution and use in source and binary forms
+          - endorse or promote
+        unless:
+          - isc license
+          - permission to use, copy, modify, and/or distribute this software
+  - id: ISC
+    menu: ISC
+    spellings: [ISC]
+    detect_markers:
+      - all:
+          - isc license
+      - all:
+          - permission to use, copy, modify, and/or distribute this software