texlive.tlpdb.nix: extract binary files, formats, script extensions

pkgs/tools/typesetting/tex/texlive/combine.nix

@@ -153,7 +153,7 @@ in (buildEnv {
   (let
     hyphens = lib.filter (p: p.hasHyphens or false && p.tlType == "run") pkgList.splitBin.wrong;
     hyphenPNames = map (p: p.pname) hyphens;
-    formats = lib.filter (p: p.hasFormats or false && p.tlType == "run") pkgList.splitBin.wrong;
+    formats = lib.filter (p: p ? formats && p.tlType == "run") pkgList.splitBin.wrong;
     formatPNames = map (p: p.pname) formats;
     # sed expression that prints the lines in /start/,/end/ except for /end/
     section = start: end: "/${start}/,/${end}/{ /${start}/p; /${end}/!p; };\n";

pkgs/tools/typesetting/tex/texlive/default.nix

@@ -96,13 +96,12 @@ let
         # tarball of a collection/scheme itself only contains a tlobj file
         [( if (attrs.hasRunfiles or false) then mkPkgV "run"
             # the fake derivations are used for filtering of hyphenation patterns and formats
-          else {
+          else ({
             inherit pname version;
             tlType = "run";
-            hasFormats = attrs.hasFormats or false;
             hasHyphens = attrs.hasHyphens or false;
             tlDeps = map (n: tl.${n}) (attrs.deps or []);
-          }
+          } // lib.optionalAttrs (attrs ? formats) { inherit (attrs) formats; })
         )]
         ++ lib.optional (attrs.sha512 ? doc) (mkPkgV "doc")
         ++ lib.optional (attrs.sha512 ? source) (mkPkgV "source")
@@ -188,11 +187,10 @@ let
           } // lib.optionalAttrs (tlType == "run" && args ? deps) {
             tlDeps = map (n: tl.${n}) args.deps;
           } // lib.optionalAttrs (tlType == "run") {
-            hasFormats = args.hasFormats or false;
             hasHyphens = args.hasHyphens or false;
           } // lib.optionalAttrs (tlType == "tlpkg" && args ? postactionScript) {
             postactionScript = args.postactionScript;
-          };
+          } // lib.optionalAttrs (tlType == "run" && args ? formats) { inherit (args) formats; };
         } // lib.optionalAttrs (fixedHash != null) {
           outputHash = fixedHash;
           outputHashAlgo = "sha256";

pkgs/tools/typesetting/tex/texlive/tl2nix.sed

@@ -89,9 +89,25 @@ $a}
       N
       s/\n / /    # remove newline
       t next-file # loop if previous line matched
+    s/\n/ \n/     # add space before last newline for accurate matching below
 
     / (RELOC|texmf-dist)\//i\  hasRunfiles = true;
     / tlpkg\//i\  hasTlpkg = true;
+
+    # extract script extensions
+    / texmf-dist\/scripts\/.*\.(jar|lua|py|rb|sno|tcl|texlua|tlu) /{
+      i\  scriptExts = [
+        / texmf-dist\/scripts\/.*\.jar /i\    "jar"
+        / texmf-dist\/scripts\/.*\.lua /i\    "lua"
+        / texmf-dist\/scripts\/.*\.py /i\    "py"
+        / texmf-dist\/scripts\/.*\.rb /i\    "rb"
+        / texmf-dist\/scripts\/.*\.sno /i\    "sno"
+        / texmf-dist\/scripts\/.*\.tcl /i\    "tcl"
+        / texmf-dist\/scripts\/.*\.texlua /i\    "texlua"
+        / texmf-dist\/scripts\/.*\.tlu /i\    "tlu"
+      i\  ];
+    }
+
     D # restart cycle from the current line
   }
 
@@ -101,8 +117,50 @@ $a}
   # extract hyphenation patterns and formats
   # (this may create duplicate lines, use uniq to remove them)
   /^execute\sAddHyphen/i\  hasHyphens = true;
-  /^execute\sAddFormat/i\  hasFormats = true;
+
+  # extract format details
+  /^execute\sAddFormat\s/{
+    # open a list
+    i\  formats = [
+
+    # create one attribute set per format
+    # note that format names are not unique
+
+    # plain keys: name, engine, patterns
+    # optionally double quoted key: options
+    # boolean key: mode (enabled/disabled)
+    # comma-separated lists: fmttriggers, patterns
+    :next-fmt
+      s/(^|\n)execute\sAddFormat/    {/
+      s/\s+options="([^"]+)"/\n      options = "\1";/
+      s/\s+(name|engine|options)=([^ \t\n]+)/\n      \1 = "\2";/g
+      s/\s+mode=enabled//
+      s/\s+mode=disabled/\n      enabled = false;/
+      s/\s+(fmttriggers|patterns)=([^ \t\n]+)/\n      \1 = [ "\2" ];/g
+      s/$/\n    }/
+
+      :split-triggers
+        s/"([^,]+),([^"]+)" ]/"\1" "\2" ]/;
+        t split-triggers   # repeat until there are no commas
+
+      p
+      s/^.*$// # clear pattern space
+      N
+      /^\nexecute\sAddFormat\s/b next-fmt
+
+    # close the list
+    i\  ];
+    D # restart cycle from the current line
+  }
 
   # close attrmap
   /^$/i};
 }
+
+# add list of binaries from one of the architecture-specific packages
+/^name ([^.]+|texlive\.infra)\.x86_64-linux$/,/^$/{
+  s/^name ([0-9].*|texlive\.infra)\.x86_64-linux$/"\1".binfiles = [/p
+  s/^name (.*)\.x86_64-linux$/\1.binfiles = [/p
+  s!^ bin/x86_64-linux/(.+)$!  "\1"!p
+  /^$/i];
+}