libreoffice: generate-libreoffice-srcs.{sh->py}

6 files changed, 278 insertions, 80 deletions

diff --git a/pkgs/applications/office/libreoffice/default-gen-shell.nix b/pkgs/applications/office/libreoffice/default-gen-shell.nix
index c3cb23d8639..36603097123 100644
--- a/pkgs/applications/office/libreoffice/default-gen-shell.nix
+++ b/pkgs/applications/office/libreoffice/default-gen-shell.nix
@@ -19,9 +19,11 @@ stdenv.mkDerivation {
     builder = ./download-list-builder.sh;
   };
 
+  buildInputs = [ python3 ];
+
   shellHook = ''
     function generate {
-      ./generate-libreoffice-srcs.sh | tee libreoffice-srcs.nix
+      python3 generate-libreoffice-srcs.py > libreoffice-srcs.nix
     }
   '';
 }
diff --git a/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.py b/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.py
new file mode 100755
index 00000000000..9af2ef2fd88
--- /dev/null
+++ b/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.py
@@ -0,0 +1,269 @@
+#!/usr/bin/env python3
+
+"""
+Converts the LibreOffice `download.lst` file into a Nix expression.
+
+Requires an environment variable named `downloadList` identifying the path
+of the input file, and writes the result to stdout.
+
+todo - Ideally we would move as much as possible into derivation dependencies.
+"""
+import collections, itertools, json, re, sys, os
+
+def main():
+
+    print('[')
+
+    for x in get_packages():
+
+        print('{')
+
+        print('  name = "{}";'.format(x['tarball']))
+        print('  md5 = "{}";'.format(x['md5']))
+        print('  brief = {};'.format('true' if x['brief'] else 'false'))
+
+        if 'subdir' in x:
+            print('  subDir = "{}";'.format(x['subdir']))
+
+        print('}')
+
+    print(']')
+
+
+def get_packages():
+    """
+    All of the package data: What's parsed from download.lst,
+    plus our additions.
+    """
+    return apply_additions(get_packages_from_download_list(),
+                           get_additions())
+
+
+def get_additions():
+    """
+    A mapping from package name (the all-caps identifiers used in
+    `download.lst`) to a dict of additional attributes to set on the package.
+    """
+    with open('./libreoffice-srcs-additions.json') as f:
+        return json.load(f)
+
+
+def apply_additions(xs, additions):
+    for x in xs:
+        yield dict_merge([x,
+                          additions.get(x['name'], {})])
+
+
+def get_packages_from_download_list():
+    """
+    The result of parsing `download.lst`: A list of dicts containing keys
+    'name', 'tarball', 'md5', 'brief'.
+    """
+
+    def lines():
+        for x in sub_symbols(parse_lines(get_lines())):
+
+            interpretation = interpret(x)
+
+            if interpretation == 'unrecognized':
+                print_skipped_line(x)
+            else:
+                yield dict_merge([x,
+                                  interpretation])
+
+    def cluster(xs):
+        """
+        Groups lines according to their order within the file, to support
+        packages that are listed in `download.lst` more than once.
+        """
+        keys = ['tarball', 'md5', 'brief']
+        a = {k: [x for x in xs if k in x['attrs']] for k in keys}
+        return zip(*[a[k] for k in keys])
+
+    def packages():
+        for (name, group) in groupby(lines(), lambda x: x['name']):
+            for xs in cluster(group):
+                yield {'name': name,
+                       'attrs': dict_merge(x['attrs'] for x in xs),
+                       'index': min(x['index'] for x in xs)}
+
+    for x in sorted(packages(), key=lambda x: x['index']):
+        yield dict_merge([{'name': x['name']},
+                          x['attrs']])
+
+
+def dict_merge(xs):
+    """
+    >>> dict_merge([{1: 2}, {3: 4}, {3: 5}])
+    {1: 2, 3: 4}
+    """
+    return dict(collections.ChainMap(*xs))
+
+
+def groupby(xs, f):
+    """
+    >>> groupby([1, 2, 3, 4], lambda x: x % 2)
+    [(0, [2, 4]), (1, [1, 3])]
+    """
+    for (k, iter) in itertools.groupby(sorted(xs, key=f), f):
+        group = list(iter)
+        yield (f(group[0]), group)
+
+
+def get_lines():
+
+    download_list = os.getenv('downloadList')
+
+    with open(download_list) as f:
+        return f.read().splitlines()
+
+
+def print_skipped_line(x):
+
+    print('Skipped line {}: {}'.format(x['index'],
+                                       x['original']),
+          file=sys.stderr)
+
+
+def parse_lines(lines):
+    """
+    Input: List of strings (the lines from `download.lst`
+    Output: Iterator of dicts with keys 'key', 'value', and 'index'
+    """
+    for (index, line) in enumerate(lines):
+
+        x = { 'index': index, 'original': line }
+
+        result = parse_line(line)
+
+        if result == 'nothing':
+            pass
+        elif result == 'unrecognized':
+            print_skipped_line(x)
+        else:
+            yield dict_merge([x,
+                             result])
+
+
+def parse_line(line):
+    """
+    Input: A string
+    Output: One of 1. A dict with keys 'key', 'value'
+                   2. 'nothing' (if the line contains no information)
+                   2. 'unrecognized' (if parsing failed)
+    """
+
+    if re.match('\s*(#.*)?$', line):
+        return 'nothing'
+
+    match = re.match('\s*export\s+([^:\s]+)\s*:=\s*(.*)$', line)
+
+    if match:
+        return {
+            'key': match.group(1),
+            'value': match.group(2).strip()
+        }
+    else:
+        return 'unrecognized'
+
+
+def sub_symbols(xs):
+    """
+    Do substitution of variables across all lines.
+
+    >>> sub_symbols([{'key': 'a', 'value': 'x'},
+    ...              {'key': 'c': 'value': '$(a)yz'}])
+    [{'key': 'a', 'value': 'x'}, {'key': 'c': 'value': 'xyz'}]
+    """
+
+    xs = list(xs)
+
+    symbols = {x['key']: x for x in xs}
+
+    def get_value(k):
+        x = symbols.get(k)
+        return x['value'] if x is not None else None
+
+    for x in xs:
+        yield dict_merge([{'value': sub_str(x['value'], get_value)},
+                          x])
+
+
+def sub_str(string, func):
+    """
+    Do substitution of variables in a single line.
+
+    >>> sub_str("x = $(x)", lambda k: {'x': 'a'}[k])
+    "x = a"
+    """
+
+    def func2(m):
+        x = m.group(1)
+        result = func(x)
+        return result if result is not None else x
+
+    return re.sub(r'\$\(([^\$\(\)]+)\)', func2, string)
+
+
+def interpret(x):
+    """
+    Input: Dict with keys 'key' and 'value'
+    Output: One of 1. Dict with keys 'name' and 'attrs'
+                   2. 'unrecognized' (if interpretation failed)
+    """
+    for f in [interpret_md5, interpret_tarball_with_md5, interpret_tarball]:
+        result = f(x)
+        if result is not None:
+            return result
+
+    return 'unrecognized'
+
+
+def interpret_md5(x):
+    """
+    >>> interpret_md5("ODFGEN_MD5SUM", "32572ea48d9021bbd6fa317ddb697abc")
+    {'name': 'ODFGEN', 'attrs': {'md5': '32572ea48d9021bbd6fa317ddb697abc'}}
+    """
+
+    match = re.match('^(.*)_MD5SUM$', x['key'])
+
+    if match:
+        return {'name': match.group(1),
+                'attrs': {'md5': x['value']}}
+
+
+def interpret_tarball(x):
+    """
+    >>> interpret_tarball("FREEHAND_TARBALL", "libfreehand-0.1.1.tar.bz2")
+    {'name': 'FREEHAND',
+     'attrs': {'tarball': 'libfreehand-0.1.1.tar.bz2', 'brief': True}}
+    """
+
+    match = re.match('^(.*)_TARBALL$', x['key'])
+
+    if match:
+        return {'name': match.group(1),
+                'attrs': {'tarball': x['value'], 'brief': True}}
+
+
+def interpret_tarball_with_md5(x):
+    """
+    >>> interpret_tarball_with_md5("CLUCENE_TARBALL",\
+        "48d647fbd8ef8889e5a7f422c1bfda94-clucene-core-2.3.3.4.tar.gz")
+    {'name': 'CLUCENE',
+     'attrs': {'tarball': 'clucene-core-2.3.3.4.tar.gz',
+               'md5': '48d647fbd8ef8889e5a7f422c1bfda94', 'brief': False}}
+    """
+
+    match = {'key': re.match('^(.*)_TARBALL$', x['key']),
+             'value': re.match('(?P<md5>[0-9a-fA-F]{32})-(?P<tarball>.+)$',
+                               x['value'])}
+
+    if match['key'] and match['value']:
+        return {'name': match['key'].group(1),
+                'attrs': {'tarball': match['value'].group('tarball'),
+                          'md5': match['value'].group('md5'),
+                          'brief': False}}
+
+
+main()
diff --git a/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.sh b/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.sh
deleted file mode 100755
index 82c5013f160..00000000000
--- a/pkgs/applications/office/libreoffice/generate-libreoffice-srcs.sh
+++ /dev/null
@@ -1,77 +0,0 @@
-#!/run/current-system/sw/bin/bash
-
-# Ideally we would move as much as possible into derivation dependencies
-
-cat <<EOF
-[
-EOF
-
-write_entry(){
-  echo '{'
-  echo "  name = \"${name}\";"
-  echo "  md5 = \"${md5}\";"
-  echo "  brief = ${brief};"
-  eval "echo -n \"\$additions_${name%%[-_.]*}\""
-  eval "test -n \"\$additions_${name%%[-_.]*}\" && echo"
-  echo '}'
-  saved_line=
-}
-
-saved_line=
-cat "$(dirname "$0")/libreoffice-srcs-additions.sh" "$downloadList" |
-while read line; do
-  case "$line" in
-    EVAL\ *)
-      echo "${line#* }" >&2;
-      eval "${line#* }";
-      saved_line=
-      ;;
-    \#*)
-      echo Skipping comment: "$line" >&2;
-      ;;
-    *_MD5SUM\ :=*)
-      if test -n "$saved_line"; then
-        tbline="$saved_line"
-      else
-        read tbline;
-      fi;
-      line=${line##* };
-      line=${line##*:=};
-      if [ "${tbline#*VERSION_MICRO}" != "$tbline" ]; then
-         verline=${tbline##* };
-         read tbline;
-         tbline=${tbline##* };
-         tbline=${tbline##*:=};
-         md5=$line
-         name=$tbline;
-         name="${name/\$([A-Z]*_VERSION_MICRO)/$verline}"
-      else
-         tbline=${tbline##* };
-         tbline=${tbline##*:=};
-         md5=$line
-         name=$tbline;
-      fi
-      brief=true;
-      write_entry;
-      ;;
-    *_TARBALL\ :=*)
-      line=${line##* };
-      line=${line##*:=};
-      line="${line#,}"
-      md5=${line:0:32};
-      name=${line:33};
-      name="${name%)}"
-      brief=false;
-      if test -n "$name"; then
-        write_entry;
-      else
-        saved_line="$line";
-      fi
-      ;;
-    *)
-      echo Skipping: "$line" >&2;
-      ;;
-  esac
-done
-
-echo ']'
diff --git a/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.json b/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.json
new file mode 100644
index 00000000000..75d4cced92a
--- /dev/null
+++ b/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.json
@@ -0,0 +1,3 @@
+{
+  "LIBGLTF": {"subdir": "libgltf/"}
+}
diff --git a/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.sh b/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.sh
deleted file mode 100644
index d505aa7a4cb..00000000000
--- a/pkgs/applications/office/libreoffice/libreoffice-srcs-additions.sh
+++ /dev/null
@@ -1 +0,0 @@
-EVAL additions_libgltf='  subDir = "libgltf/";'
diff --git a/pkgs/applications/office/libreoffice/still-gen-shell.nix b/pkgs/applications/office/libreoffice/still-gen-shell.nix
index 991492e2d98..c5c17c2aaad 100644
--- a/pkgs/applications/office/libreoffice/still-gen-shell.nix
+++ b/pkgs/applications/office/libreoffice/still-gen-shell.nix
@@ -19,9 +19,11 @@ stdenv.mkDerivation {
     builder = ./download-list-builder.sh;
   };
 
+  buildInputs = [ python3 ];
+
   shellHook = ''
     function generate {
-      ./generate-libreoffice-srcs.sh | tee libreoffice-srcs-still.nix
+      python3 generate-libreoffice-srcs.py > libreoffice-srcs-still.nix
     }
   '';
 }