Fecal samples are of great value as a non-invasive means to gather information on the genetics, distribution, demography, diet, and parasite infestation of endangered species. Direct shotgun sequencing of fecal DNA could give information on these simultaneously, but this approach is largely untested. Here we use two fecal samples to characterize the diet of two Red-Shanked Doucs Langurs (Pygathrix nemaeus) that were fed a known combination of foliage, fruits, vegetables and cereals. Illumina HiSeq sequencing produced ~70 million paired reads per sample, of which ~10000 (0.014%) and ~44000 (0.066%) respectively corresponded to chloroplast genomes. Sequences were matched against a database of available chloroplast ‘barcodes’ for angiosperms. The results were compared with ‘metabarcoding’ using PCR amplification of the P6 loop of trnL. Shotgun sequencing identified 7 and 9 of the likely 16 diet plants, against 6 and 5 plant species identified by metabarcoding. Metabarcoding produced thousands of reads that were consistent with the known diet, but the barcodes were too short to identify several diet plants to genus. Metagenomics could utilize multiple, longer barcodes that combined had greater power of identification, but rare diet items were not recovered. Read numbers for diet species in metagenomic and metabarcoding data were correlated, indicating that both approaches are useful for determining relative sequence abundance. Metagenomic reads were uniformly distributed across the chloroplast genomes; thus if chloroplast genomes were to be used as reference, the precision of identifications and species recovery would improve further. Metagenomics also recovered the host mitochondrial genome and numerous intestinal parasite sequences in addition to generating data useful for characterizing the microbiome.