feat: _detect_nvchecker_source maps forge+registry hosts

2 files changed, 106 insertions, 0 deletions

diff --git a/mkhint b/mkhint
index 7d4c505..20b20a8 100755
--- a/mkhint
+++ b/mkhint
@@ -843,6 +843,71 @@ _registry_name_from_url() {
     printf '%s' "$fallback"
 }
 
+# Detect an nvchecker source from a package's DOWNLOAD/HOMEPAGE haystack.
+# Args: <haystack> <prgnam>. Prints the TOML body (no [label] header). Empty
+# output means "unrecognised", the caller emits the commented stub.
+_detect_nvchecker_source() {
+    local haystack="$1" pkg="$2"
+
+    # ── owner/repo forges ─────────────────────────────────────────────────
+    if [[ "$haystack" =~ (github\.com|gitlab\.com|bitbucket\.org|gitea\.com|codeberg\.org|pagure\.io)/([A-Za-z0-9._-]+)(/([A-Za-z0-9._-]+))? ]]; then
+        local host="${BASH_REMATCH[1]}"
+        local owner="${BASH_REMATCH[2]}"
+        local repo="${BASH_REMATCH[4]}"
+        repo="${repo%.git}"; owner="${owner%.git}"
+        case "$host" in
+            github.com)
+                printf 'source = "github"\ngithub = "%s/%s"\nuse_latest_release = true\n# use_max_tag = true  # if the repo publishes no Releases\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner" "$repo"
+                return 0 ;;
+            gitlab.com)
+                printf 'source = "gitlab"\ngitlab = "%s/%s"\nuse_max_tag = true\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner" "$repo"
+                return 0 ;;
+            bitbucket.org)
+                printf 'source = "bitbucket"\nbitbucket = "%s/%s"\nuse_max_tag = true\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner" "$repo"
+                return 0 ;;
+            gitea.com)
+                printf 'source = "gitea"\ngitea = "%s/%s"\nuse_max_tag = true\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner" "$repo"
+                return 0 ;;
+            codeberg.org)
+                printf 'source = "gitea"\ngitea = "%s/%s"\nhost = "codeberg.org"\nuse_max_tag = true\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner" "$repo"
+                return 0 ;;
+            pagure.io)
+                # pagure field is the repo only, on pagure.io the first path
+                # segment IS the repo (no owner), so it lands in $owner.
+                printf 'source = "pagure"\npagure = "%s"\nuse_max_tag = true\n# prefix = "v"  # uncomment if tags are v-prefixed\n' "$owner"
+                return 0 ;;
+        esac
+    fi
+
+    # ── language registries ───────────────────────────────────────────────
+    local url; url=$(printf '%s' "$haystack" | grep -oE 'https?://[^ ]+' | head -1)
+    local name
+    case "$haystack" in
+        *pypi.org*|*files.pythonhosted.org*)
+            printf 'source = "pypi"\npypi = "%s"\n' "$pkg"; return 0 ;;
+        *registry.npmjs.org*|*npmjs.com*)
+            name=$(_registry_name_from_url "$url" "$pkg")
+            printf 'source = "npm"\nnpm = "%s"\n' "$name"; return 0 ;;
+        *rubygems.org*)
+            name=$(_registry_name_from_url "$url" "$pkg")
+            printf 'source = "gems"\ngems = "%s"\n' "$name"; return 0 ;;
+        *crates.io*)
+            name=$(_registry_name_from_url "$url" "$pkg")
+            printf 'source = "cratesio"\ncratesio = "%s"\n' "$name"; return 0 ;;
+        *metacpan.org*|*cpan.org*)
+            printf 'source = "cpan"\ncpan = "%s"\n' "$pkg"; return 0 ;;
+        *hackage.haskell.org*)
+            name=$(_registry_name_from_url "$url" "$pkg")
+            printf 'source = "hackage"\nhackage = "%s"\n' "$name"; return 0 ;;
+        *packagist.org*)
+            printf 'source = "packagist"\npackagist = "%s"\n' "$pkg"; return 0 ;;
+        *cran.r-project.org*)
+            printf 'source = "cran"\ncran = "%s"\n' "$pkg"; return 0 ;;
+    esac
+
+    return 0   # empty output → caller stubs
+}
+
 # Append an nvchecker [pkg] section to NVCHECKER_CONFIG, auto-detecting the
 # source from the package's .info DOWNLOAD/HOMEPAGE. No-op if section exists.
 add_nvchecker_section() {
diff --git a/tests/mkhint_test.sh b/tests/mkhint_test.sh
index e000820..075f542 100755
--- a/tests/mkhint_test.sh
+++ b/tests/mkhint_test.sh
@@ -1974,6 +1974,47 @@ assert_contains "gems name parsed"     "$nv_out" '^gems rails$'
 assert_contains "npm name parsed"      "$nv_out" '^npm left-pad$'
 assert_contains "fallback to PRGNAM"   "$nv_out" '^fallback mypkg$'
 
+# ── T-NV2: _detect_nvchecker_source per host ──────────────────────────────────
+echo ""
+echo "T-NV2: _detect_nvchecker_source emits correct body per host"
+d_out="$MOCK_BASE/nv_detect.txt"
+: > "$d_out"
+echo "### github" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://github.com/o/r/archive/v1.tar.gz" r >> "$d_out"
+echo "### gitlab" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://gitlab.com/o/r/-/archive/1/r-1.tar.gz" r >> "$d_out"
+echo "### bitbucket" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://bitbucket.org/o/r/get/1.tar.gz" r >> "$d_out"
+echo "### gitea" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://gitea.com/o/r/archive/1.tar.gz" r >> "$d_out"
+echo "### codeberg" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://codeberg.org/o/r/archive/1.tar.gz" r >> "$d_out"
+echo "### pagure" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://pagure.io/r/archive/1/r-1.tar.gz" r >> "$d_out"
+echo "### npm" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://registry.npmjs.org/lp/-/lp-1.tgz" lp >> "$d_out"
+echo "### cran" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://cran.r-project.org/src/contrib/foo_1.tar.gz" foo >> "$d_out"
+echo "### unknown" >> "$d_out"
+_detect_nvchecker_source "DOWNLOAD=https://example.com/x.tar.gz" x >> "$d_out"
+echo "### end" >> "$d_out"
+
+assert_contains "github source"        "$d_out" 'source = "github"'
+assert_contains "github latest_release" "$d_out" 'use_latest_release = true'
+assert_contains "github max_tag comment" "$d_out" '# use_max_tag = true'
+assert_contains "github prefix comment"  "$d_out" '# prefix = "v"'
+assert_contains "gitlab source"        "$d_out" 'source = "gitlab"'
+assert_contains "gitlab field"         "$d_out" 'gitlab = "o/r"'
+assert_contains "bitbucket field"      "$d_out" 'bitbucket = "o/r"'
+assert_contains "gitea field"          "$d_out" 'gitea = "o/r"'
+assert_contains "codeberg host"        "$d_out" 'host = "codeberg.org"'
+assert_contains "pagure field"         "$d_out" 'pagure = "r"'
+assert_contains "npm field"            "$d_out" 'npm = "lp"'
+assert_contains "cran field"           "$d_out" 'cran = "foo"'
+# unknown host → empty body between ### unknown and ### end
+unk=$(awk '/^### unknown/{f=1;next} /^### end/{f=0} f' "$d_out")
+assert_exit_code "unknown host empty body" 0 "$( [[ -z "${unk//[[:space:]]/}" ]]; echo $? )"
+
 # ─── SUMMARY ──────────────────────────────────────────────────────────────────
 teardown