Mike Gerwitz

Activist for User Freedom

aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat (limited to 'src/current/tools/csvm2csv')
-rwxr-xr-xsrc/current/tools/csvm2csv112
1 files changed, 112 insertions, 0 deletions
diff --git a/src/current/tools/csvm2csv b/src/current/tools/csvm2csv
new file mode 100755
index 0000000..410d9fa
--- /dev/null
+++ b/src/current/tools/csvm2csv
@@ -0,0 +1,112 @@
+#!/usr/bin/awk -f
+#
+# Compiles a "magic" CSV file into a normal CSV
+#
+# "Magic" CSVs simply exist to make life easier: they permit comments, blank
+# lines, variables, sub-delimiter expansion, and any number of ranges per line.
+# Ranges will be expanded in every combination, making rate tables highly
+# maintainable.
+#
+# Variables are also supported when defined using :var=val. Variables may expand
+# into ranges, 'cause they're awesome. Multiple variables may be delimited by
+# semi-colons, as may multiple values.
+#
+# For example:
+# :foo=1--3
+# $foo;7;9--10:$foo, 5--10
+#
+# Would generate:
+# 1, 5
+# 1, 6
+# ...
+# 5, 10
+# 2, 5
+# ...
+# 9, 5
+# ...
+# 1, 5
+# 1, 6
+# ...
+
+
+function rangeout( i, m, j, me, orig )
+{
+ if ( i > NF )
+ {
+ print
+ return
+ }
+
+ orig = $i
+
+ # check first for delimiters
+ if ( match( $i, /^([^;]+);(.*)$/, m ) )
+ {
+ # give it a shot with the first value
+ $i = m[1]
+ rangeout( i )
+
+ # strip off the first value and process with following value(s)
+ $i = m[2]
+ rangeout( i )
+
+ # we've delegated; we're done
+ $i = orig
+ return
+ }
+
+ # attempt to parse variable (may expand into a range)
+ if ( match( $i, /^\$([a-zA-Z_-]+)$/, m ) )
+ {
+ $i = vars[ m[1] ];
+ }
+
+ # parse range
+ if ( match( $i, /^([0-9]+)--([0-9]+)$/, m ) )
+ {
+ j = m[1]
+ me = m[2]
+ do
+ {
+ $i = j
+ rangeout( i + 1 )
+ } while ( j++ < me )
+ }
+ else
+ {
+ rangeout( i + 1 );
+ }
+
+ # restore to original value
+ $i = orig
+}
+
+
+BEGIN {
+ # we're parsing CSVs
+ FS = " *, *"
+ OFS = ","
+}
+
+
+# skip all lines that begin with `#', which denotes a comment, or are empty
+/^#|^$/ { next; }
+
+# lines that begin with a colon are variable definitions
+/^:/ {
+ match( $0, /^:([a-zA-Z_-]+)=(.*?)$/, m )
+ vars[ m[1] ] = m[2]
+ next
+}
+
+# lines containing ranges (denoted by `--', the en dash, which is a typesetting
+# convetion for ranges), sub-delimiters, or variables must be expanded
+/--|;|\$[a-zA-Z_-]/ { rangeout( 1 ); next; }
+
+# all other lines are normal; simply output them verbatim
+{
+ # this assignment will ensure that awk processes the output, ensuring that
+ # extra spaces between commas are stripped
+ $1=$1
+ print
+}