making it more robust

scripts/sql-to-dot.sed

@@ -1,39 +1,79 @@
-#! /bin/sed -f
+#! /bin/sed -nf
 1i\
 /** @page database Database Schema\
 \
 @dot\
 digraph schema {
 
-# remove newlines
+# get everithing on one single line
-:a;N;$!ba;s/\n/ /g
+H;$!d;$x
+
+# remove all single-line comment lines
+s/\n--[^\n]*//g
+
 # reduce spaces
-s, \+, ,g
+s,\t\| \+, ,g
 
-s| *CREATE *TABLE *\(if *not *exists\)\? *`\?\([a-z]\+\)`\? *|\n    \2\n        [shape=none, margin=0, label=<\n            <table bgcolor="#dddddd">\n                <tr><td bgcolor="#ddddff" colspan="4"><b>\2</b></td></tr>\n|ig
+# remove multiline comments
+:f;s,\(.*\)/\*.*\*/[ \n]*;*,\1,g;tf
 
-s/varchar *( *\([0-9]\+\) *)/varchar[\1]/ig
+# remove empty lines
+s,\n\+,\n,g
 
-s/ *PRIMARY KEY *( *`\?\([a-z]\+\)`\? *) *[,)]//ig
+# remove unknown commands
+s,\(;\|\n\) *\(DELIMITER\|USE\|DROP\|CREATE[ \n]\+DATABASE\)[ \n]\+[^;]*;\+,,ig
 
-:b;s/\([a-z]\+\)\([^;]*\)FOREIGN KEY *( *`\?\([a-z]\+\)`\? *) *REFERENCES *`\?\([a-z]\+\)`\? *( *`\?\([a-z]\+\)`\? *) *\([^,)]*\)[,)]\(.*\)/\1\2\7\n \1:\3 -> \4:\5/ig;tb
+# convert special characters within quotes
+:a;s/^\(\([^"]*"[^",]*"\)*[^"]*"[^"]*\),\([^"]*".*\)/\1\##COMMA##\3/g;ta
+:c;s/^\(\([^']*'[^',]*'\)*[^']*'[^']*\),\([^']*'.*\)/\1\##COMMA##\3/g;tc
 
-:c;s|\n[(,] *`\?\([a-z]\+\)`\? *\([^ ]\+\)*\([^,)]*\) *COMMENT *[`"']\([^`"']*\)[`"'] *[,)]|\n                <tr><td align="left" port="\1"><b>\1</b></td><td align="left">\2</td><td align="left">\3</td><td align="left">\4</td></tr>\n,|ig;tc
+# backup everything to the buffer
+# then analyze only on one create table
+:i
+h
+s,.*\(create[ \n]\+table[^;]*;\).*,\1,ig
 
-s| \+</td>|</td>|g
+# start html table node
-s|<td> \+|<td>|g
+s|CREATE[ \n]\+TABLE[ \n]\+\(if[ \n]\+not[ \n]\+exists[ \n]\+\)\?`\?\(\w\+\)`\?|    \2\n        [shape=none, margin=0, label=<\n            <table bgcolor="#dddddd">\n                <tr><td bgcolor="#ddddff" colspan="4"><b>\2</b></td></tr>|ig
-s|\(<td[^>]*>[^<]\{30,40\}\) \+\([^<]\{20,\}</td>\)|\1<br/>\2|g
-:d;s|\(<br/>[^<]\{30,40\}\) \+\([^<]\{20,\}</td>\)|\1<br/>\2|g;td
 
-s| *COMMENT *= *["']\?\([^"']*\)["']\?|\n                <tr><td bgcolor="#ddddff" colspan="4">\1</td></tr>|ig
+# remove key definitions
+s/[),][\n ]*\(PRIMARY[ \n]\+\)\?KEY[ \n]\+[^(]*([^)]*)//gi
 
-s| \+</td>|</td>|g
+# move foreign keys as relation to the end
-s|<td> \+|<td>|g
+:b;s/\(\w\+\)\([^;]*\)FOREIGN[\n ]\+KEY[ \n]*([ \n]*`\?\([a-z]\+\)`\?[ \n]*)[ \n]*REFERENCES[ \n]*`\?\([a-z]\+\)`\?[ \n]*([ \n]*`\?\([a-z]\+\)`\?[ \n]*)[ \n]*\([^,)]*\)\([,)].*\)/\1\2\7\n \1:\3 -> \4:\5 [label="\6"]##SEMICOLON##/ig;tb
-s|\(<td[^>]*>[^<]\{60,80\}\) \+\([^<]\{30,\}</td>\)|\1<br/>\2|g
-:e;s|\(<br/>[^<]\{60,80\}\) \+\([^<]\{30,\}</td>\)|\1<br/>\2|g;te
 
-s/\n,[^\n]*/\n/g
+# create table rows
-s|;|\n            </table>\n        >];\n|ig
+s|[(,][ \n]*`\?\(\w\+\)`\?[ \n]\+\(\w\+\(([^)]\+)\)\?\)[ \n]*\([^,)]*\)[ \n]\+COMMENT[ \n]*["']\([^"']*\)["'][ \n]*|\n                <tr><td align="left" port="\1"><b>\1</b></td><td align="left">\2</td><td align="left">\4</td><td align="left">\5</td></tr>|gi
+s|[(,][ \n]*`\?\(\w\+\)`\?[ \n]\+\(\w\+\(([^)]\+)\)\?\)[ \n]*\([^,)]*\)|\n                <tr><td align="left" port="\1"><b>\1</b></td><td align="left">\2</td><td align="left">\4</td></tr>|g
+
+# add line breaks for long lines
+s|\(<td[^>]*>[^<]\{30,40\}\)[ \n]\+\([^<]\{20,\}</td>\)|\1<br/>\2|g
+#:d;s|\(<br/>[^<]\{30,40\}\)[ \n]\+\([^<]\{20,\}</td>\)|\1<br/>\2|g;td
+
+# add table comment below
+:k;tk
+s|[ \n]*)[^)]*COMMENT[ \n]*=[ \n]*["']\?\([^"']*\)["']\?[^;]*|\n                <tr><td bgcolor="#ddddff" colspan="4">\1</td></tr>|ig;th
+s|)[^);]*;|\n;|ig
+:h
+
+# cleanup comment below, add line breaksfor long lines
+s|\(<td[^>]*>[^<]\{60,80\}\)[ \n]\+\([^<]\{30,\}</td>\)|\1<br/>\2|g
+#:e;s|\(<br/>[^<]\{60,80\}\)[ \n]\+\([^<]\{30,\}</td>\)|\1<br/>\2|g;te
+
+# close table
+s|;|\n            </table>\n        >];|ig
+
+# convert ##COMMA## to ,
+s|##COMMA##|,|g
+# convert ##SEMICOLON## to ;
+s,##SEMICOLON##,;,g
+
+# print one table
+p
+# get buffer back and remove the table that has just been analyzed
+x
+s,\(.*\)create[ \n]\+table[^;]*;\(.*\),\1\2,ig
+ti
 
 $a\
 }\