It's often a good idea to check for weird characters in a text file. Something like this: cat file.txt | perl -C63 -pe '$_ = join("\n", split(//, $_));' | sort

It's often a good idea to check for weird characters in a text file. Something like this: cat file.txt | perl -C63 -pe '$_ = join("\n", split(//, $_));' | sort | uniq -c I've found lots of errors that way: BOMs in the middle of a file, different kinds of dash, quotation marks or brackets that don't match, ...