I assume you are trying to perform QC step for RNA-seq data. Basically you want to know how many reads originated from exons and how many came from introns. Suggesting that all the intronic reads are because of genomic contamination is debatable. Sometimes the transcripts are not fully processed. Also, few tissues such as brain have been shown to express lot of intronic reads. SnoRNA create lot of problems during library enrichment process for small RNAs such as microRNAs.

This tool performs comprehensive QC analyssi for RNAseq data: https://code.google.com/p/rseqc/

Additionally, there are many tools available that will take the BAM file and annotation files in gff,gtf,gff3,bed format and will output the number of reads aligned within genes, intergenic regions. You can download most of the annotation information from UCSC genome browser including snoRNA genomic coordinates.

1. HTSeq - http://www-huber.embl.de/users/anders/HTSeq/doc/count.html
2. http://genome.sph.umich.edu/wiki/Bam_read_count
3. bedtools multicov

You will first map these reads against the reference genome of your specie of interest. All the mapped reads will be assigned to some genomic coordinates after mapping. You will then use annotation file that tells whether given genomic coordinates to which reads are aligned to belong to an intergenic region, located inside the gene (can be exonic or intronic), or located in the region that codes for snoRNA.