About

Path2Insight (p2i) is a modular and scalable python module which aims at offering a unified and comprehensive set of processing tools for analyzing file paths. P2i supports static file systems analysis without requiring access to the original physical storage. Basically, a scan of the storage’s content exported as a text file suffices to explore the saved resources. There is also no need to access the content of the files as the p2i module import file paths as strings.

Once loaded, the file paths are stored in-memory as a python object enabling: preprocessing, text processing and descriptive analysis of folders and files.

Preprocessing: Sample, sort and select files based on multiple criteria (e.g. parent folder, depth).

Text processing: Chunk file paths into tokens (full path, stem and name), n-grams or complete paths with the help of several extensible tokenizers. Also, taggers offer the option to aggregate files based on their structure and content (which prepare paths for further analysis such as entity recognition or classification tasks).

Descriptive analysis: P2i implements counters for tokens, stems and extensions. It supports also statistical features such as X2 tests on the distribution of extensions, stems and names. Further, a representation of the complexity of the folder structure is facilitated by folder-depth analysis functionalities.

The table below shows how Path2Insight differ and complement functionalities offered by lower-level python modules (pathlib and os.path).

Functionality	P2i	Pathlib	os.path
Preprocessing	Pathlib + Sampling, sorting, selection	match, joinpath	Normcase, norm path
Descriptive statistics	Counters: stem, extension, name. Taggers. Tokenizers	os.stat	os.stat
Text processing	Pathlib + Tokens, n-grams, taggers, lower, upper,…	Stem, name, parent, extension drive, …	Split
Access or modify information on the system	No, can be linked to additional metadata (datetimes, users) by joining on the full path	Yes, chmod, current folder. …	Yes, user, size, datetimes, descriptors, …

P2i is dependency free (only pathlib2 is required for Python 2.7 users), fast and scalable path processing toolkit. It is compliant with the major data analysis python modules such as pandas, scikit-learn, nltk and matplotlib to extent the analytical possibilities of path2insight.

Example

Import the module and load a demo dataset with static file paths (or use path2insight.walk to collect from you file system).

>>> import path2insight
>>> from path2insight.datasets import load_ensembl

>>> filepaths = load_ensembl()

>>> path2insight.depth_counts(filepaths)
Counter({3: 1, 4: 11, 5: 39424, 6: 5543, 7: 2733, 8: 3388})

>>> path2insight.token_counts(filepaths).most_common(10)
[('txt', 31977),
 ('gene', 13798),
 ('ensembl', 12727),
 ('dm', 12500),
 ('homolog', 7380),
 ('fa', 5890),
 ('chromosome', 5011),
 ('feature', 4878),
 ('dna', 4608),
 ('90', 3404)]

>>> path2insight.extension_counts(filepaths).most_common(10)
[('.gz', 44427),
 ('', 3094),
 ('.bb', 847),
 ('.nsq', 349),
 ('.nin', 349),
 ('.nhr', 349),
 ('.tsv', 336),
 ('.psq', 250),
 ('.pin', 250),
 ('.phr', 250)]

>>> path2insight.select_re(filepaths, level5='micro.*')
[PosixFilePath('/Volumes/release-90/variation/VEP/microtus_ochrogaster_vep_90_MicOch1.0.tar.gz'),
 PosixFilePath('/Volumes/release-90/variation/VEP/microtus_ochrogaster_refseq_vep_90_MicOch1.0.tar.gz'),
 PosixFilePath('/Volumes/release-90/variation/VEP/microtus_ochrogaster_merged_vep_90_MicOch1.0.tar.gz'),
 PosixFilePath('/Volumes/release-90/variation/VEP/microcebus_murinus_vep_90_Mmur_2.0.tar.gz'),
 PosixFilePath('/Volumes/release-90/rdf/microtus_ochrogaster/microtus_ochrogaster_xrefs.ttl.gz.graph'),

>>> path2insight.distance_on_token(filepaths[0:10])
array([[ 0.        ,  2.        ,  1.41421356,  3.        ,  3.        ],
       [ 2.        ,  0.        ,  2.44948974,  3.31662479,  3.31662479],
       [ 1.41421356,  2.44948974,  0.        ,  3.        ,  3.        ],
       [ 3.        ,  3.31662479,  3.        ,  0.        ,  1.41421356],
       [ 3.        ,  3.31662479,  3.        ,  1.41421356,  0.        ]])

Cite

[1] A. Lefebvre and M. Spruit, “Designing laboratory forensics” in 18th IFIP WG 6.11 Conference on e-Business, e-Services, and e-Society, I3E 2019, 2019.

Authors

• Armel Lefebvre
• Jonathan de Bruin