Command line¶

analyzeText is the main command-line analyzer. It analyzes one or more text files with a given language and pipeline.

analyzeText -l eng-UD_English-EWT -p deepud file1.txt file2.txt

By default, the result is written on standard output in CoNLL-U. Input files must be UTF-8 encoded.

The language (-l) selects the language configuration and, for neural pipelines, the models: see Language models. The pipeline (-p) selects the sequence of processing steps: see Pipelines.

Options¶

Option	Meaning
`-l`, `--language <lang>`	Language to initialize and use. Can be repeated
`-p`, `--pipeline <name>`	Analysis pipeline to use
`--meta <k1:v1,k2:v2>`	Metadata passed to the analysis, e.g. `udlang:eng-UD_English-EWT`
`-d`, `--dumper <name>`	Output handler to use (`text` by default; also `bow`, `bowh`, `fullxml`, `event`, `xmlbow`). The matching dumper must be in the pipeline
`-o`, `--output <dumper>:<dest>`	Where a dumper writes: `stdout` or a suffix appended to the input file name
`-s`, `--split-mode <mode>`	`none` (default), `lines` or `para`: analyze each line or paragraph independently
`--inactive-units <unit>`	Deactivate a process unit of the pipeline. Can be repeated
`--availableUnits`	List the known process units and exit
`--config-dir <dir>`	Directory containing the configuration files
`--resources-dir <dir>`	Directory containing the linguistic resources
`--common-config-file <file>`	Common configuration file (default `lima-common.xml`)
`--lp-config-file <file>`	Linguistic processing configuration file (default `lima-analysis.xml`)
`-c`, `--client <id>`	Linguistic processing client (default `lima-coreclient`)
`-v`, `--version`	Print the LIMA version
`-h`, `--help`	Print the help

Examples¶

Analyze raw text with the full neural pipeline:

analyzeText -l fra-UD_French-GSD -p deepud article.txt

Analyze an already tokenized CoNLL-U file:

analyzeText -l fra-UD_French-GSD -p deepud-pretok corpus.conllu

Run the rule-based named entity recognizer of the legacy English pipeline:

analyzeText -l eng -p ner-rules article.txt

Analyze a file line by line (one sentence or document per line):

analyzeText -l eng-UD_English-EWT -p deepud -s lines sentences.txt

Other commands¶

Command	Role
`lima`	The graphical interface (C++ installation) or a simple file analyzer (Python package)
`limaserver`	An HTTP server exposing the analyzer (see Docker)
`lima_models.py`	Install language models
`compile-rules`	Compile ModEx rules
`readBowFile`	Print the content of a binary bag-of-words file (see Output formats)
`deeplima`	The standalone neural analyzer of the deeplima library (see Training neural models)