media-scraper: Fix for new path arrangements, and re-scrape

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

diff --git a/library/wikimedia/chess-b-N.download-log b/library/wikimedia/chess-b-N.download-log
new file mode 100644 (file)
index 0000000..e029bed
--- /dev/null
+++ b/library/wikimedia/chess-b-N.download-log
@@ -0,0 +1,3 @@
+
+2020-09-26 16:29:19 UTC: downloaded into library/wikimedia/chess-b-N.wikitext from https://commons.wikimedia.org/wiki/File:Chess_adt45.svg?action=raw
+2020-09-26 16:29:19 UTC: downloaded into library/wikimedia/chess-b-N.svg from https://upload.wikimedia.org/wikipedia/commons/4/43/Chess_adt45.svg

diff --git a/library/wikimedia/chess_adt45.svg b/library/wikimedia/chess-b-N.svg
similarity index 100%
rename from library/wikimedia/chess_adt45.svg
rename to library/wikimedia/chess-b-N.svg

diff --git a/library/wikimedia/chess_adt45.wikitext b/library/wikimedia/chess-b-N.wikitext
similarity index 100%
rename from library/wikimedia/chess_adt45.wikitext
rename to library/wikimedia/chess-b-N.wikitext

diff --git a/library/wikimedia/chess-w-B.download-log b/library/wikimedia/chess-w-B.download-log
new file mode 100644 (file)
index 0000000..6929c9a
--- /dev/null
+++ b/library/wikimedia/chess-w-B.download-log
@@ -0,0 +1,3 @@
+
+2020-09-26 16:29:18 UTC: downloaded into library/wikimedia/chess-w-B.wikitext from https://commons.wikimedia.org/wiki/File:Chess_blt45.svg?action=raw
+2020-09-26 16:29:19 UTC: downloaded into library/wikimedia/chess-w-B.svg from https://upload.wikimedia.org/wikipedia/commons/b/b1/Chess_blt45.svg

diff --git a/library/wikimedia/chess_blt45.svg b/library/wikimedia/chess-w-B.svg
similarity index 100%
rename from library/wikimedia/chess_blt45.svg
rename to library/wikimedia/chess-w-B.svg

diff --git a/library/wikimedia/chess_blt45.wikitext b/library/wikimedia/chess-w-B.wikitext
similarity index 100%
rename from library/wikimedia/chess_blt45.wikitext
rename to library/wikimedia/chess-w-B.wikitext

diff --git a/library/wikimedia/chess_adt45.download-log b/library/wikimedia/chess_adt45.download-log
deleted file mode 100644 (file)
index 2585641..0000000
--- a/library/wikimedia/chess_adt45.download-log
+++ /dev/null
@@ -1,3 +0,0 @@
-
-2020-09-13 22:57:30 UTC: downloaded into library/wikimedia/chess_adt45.wikitext from https://commons.wikimedia.org/wiki/File:Chess_adt45.svg?action=raw
-2020-09-13 22:57:30 UTC: downloaded into library/wikimedia/chess_adt45.svg from https://upload.wikimedia.org/wikipedia/commons/4/43/Chess_adt45.svg

diff --git a/library/wikimedia/chess_blt45.download-log b/library/wikimedia/chess_blt45.download-log
deleted file mode 100644 (file)
index 2bc072b..0000000
--- a/library/wikimedia/chess_blt45.download-log
+++ /dev/null
@@ -1,3 +0,0 @@
-
-2020-09-13 22:57:29 UTC: downloaded into library/wikimedia/chess_blt45.wikitext from https://commons.wikimedia.org/wiki/File:Chess_blt45.svg?action=raw
-2020-09-13 22:57:30 UTC: downloaded into library/wikimedia/chess_blt45.svg from https://upload.wikimedia.org/wikipedia/commons/b/b1/Chess_blt45.svg

diff --git a/library/wikimedia/chess_w-b.download-log b/library/wikimedia/chess_w-b.download-log
new file mode 100644 (file)
index 0000000..8b13789
--- /dev/null
+++ b/library/wikimedia/chess_w-b.download-log
@@ -0,0 +1 @@
+

diff --git a/library/wikimedia/files.make b/library/wikimedia/files.make
index f9f16091d941027c3259bca14b4d26630d3e70c6..a617cf785b097d3e0ddd1b55b5cd4ac253fd9fbb 100644 (file)
--- a/library/wikimedia/files.make
+++ b/library/wikimedia/files.make
@@ -1,6 +1,6 @@
-LIBRARY_FILES += library/wikimedia/chess_blt45.usvg
-library/wikimedia/chess_blt45.usvg: library/wikimedia/chess_blt45.svg library/wikimedia/LICENCE
+LIBRARY_FILES += library/wikimedia/chess-w-B.usvg
+library/wikimedia/chess-w-B.usvg: library/wikimedia/chess-w-B.svg library/wikimedia/LICENCE
        $(LIBRARY_PROCESS_SVG)
-LIBRARY_FILES += library/wikimedia/chess_adt45.usvg
-library/wikimedia/chess_adt45.usvg: library/wikimedia/chess_adt45.svg library/wikimedia/LICENCE
+LIBRARY_FILES += library/wikimedia/chess-b-N.usvg
+library/wikimedia/chess-b-N.usvg: library/wikimedia/chess-b-N.svg library/wikimedia/LICENCE
        $(LIBRARY_PROCESS_SVG)

diff --git a/media-scraper b/media-scraper
index ce9a6384065447b1997cc0db9e4382b791c7684b..2d8fcfe39d5ffac044b6af8ad566412793b896c0 100755 (executable)
--- a/media-scraper
+++ b/media-scraper
@@ -170,10 +170,11 @@ foreach my $groupname (sort keys %$groups) {
   foreach (split(/\n/, $gcfg->('files'))) {
     s/^\s+//;
     next if m/^\#/ || m/^$/;
-    m/^\S+/;
-    my $ministem = $&;
-    my $rstem = cfg_affixes $gcfg, 'stem', $ministem;
-    my $lstem = lc $rstem;
+    m/^(\S+)\s+(\S+)/ or die;
+    my $lministem = $1;
+    my $rministem = $2;
+    my $lstem = cfg_affixes $gcfg, 'item', $lministem;
+    my $rstem = cfg_affixes $gcfg, 'stem', $rministem;
     my $lbase = "$basename/$lstem";
     my $lupstream = "$lbase.svg";
     my $lprocessed = "$lbase.usvg";