Extract tableid.

[pspp] / parse-all-xml

diff --git a/parse-all-xml b/parse-all-xml
index 98a4258ddbd50c36bbc47fbca083df5491837844..c1282e03ff3bf6476afb81d3824526303762f18c 100755 (executable)
--- a/parse-all-xml
+++ b/parse-all-xml
@@ -1,5 +1,13 @@
 #! /bin/sh
-for d in `ls -1 unzipped/*/*.xml |grep -vE 'notes|table|warning|chart|model'`
-do
-    ./parse-xml $d
-done | sort -u
+
+# Parse the structure members that do not represent models or graphs.
+# Also skip those with borderProperties, which indicate the non-"light"
+# format.
+lightTables=`ls -1 unzipped/*/*.xml |grep -vE 'notes|table|warning|chart|model' \
+          | xargs grep -EL '<([a-z]*:)?(model|graph|borderProperties)'`
+
+if test -n "$1"; then
+    for d in $lightTables; do
+       ./parse-xml $d $1
+    done | sort | uniq -c | sort -rn
+fi