yunpy 0.1.0

A library for phonetic annotation in Middle Chinese

YunPy

This library automatically adds Middle Chinese phonetic glosses to plain text files, and converts the .txt format into annotated Webanno TSV 3.3 or XML.

The converter is based on the PHONO-ML database, which maps Chinese characters to Middle Chinese readings (26 224 entries), rendering each reading in the Baxter-Sagart reconstruction. The tool supports single-file and recursive directory processing via both a command-line interface and programmatic usage in Python.

Features

Lexicon exploration

• Display all known Middle Chinese transcriptions for a single character or a string
• Display detailed information (transcriptions, fanqie, sixtuples) for a single character or a string

Text annotation

• Annotate and convert TXT to WebAnno TSV 3.3 (compatible with annotation tools such as INCEpTION)
• Annotate and convert TXT to XML
• Process a single file or batch process multiple files from a directory

All of this can be done using either CLI or programmatic usage in Python

Installation

pip install yunpy

CLI usage

Lexicon exploration

Display a character's known reading(s):

howtoread 左

Output:

Character 左 has the following readings: tsaX, tsaH

This command can also be called on a string.

howtoread 左右流之

Character 左 has the following readings: tsaX, tsaH
Character 右 has the following readings: hjuwX, hjuwH
Character 流 has the following reading: ljuw
Character 之 has the following reading: tsyi

Display detailed information about a character:

charinfo 左

Output:

Character = 左
Fanqie = ['臧可', '則箇']
Sixtuples = ['果開一上歌精', '果開一去歌精']
Transcriptions = ['tsaX', 'tsaH']

This command can also be called on a string.

charinfo 左右

Character = 左
Fanqie = ['臧可', '則箇']
Sixtuples = ['果開一上歌精', '果開一去歌精']
Transcriptions = ['tsaX', 'tsaH']


Character = 右
Fanqie = ['云久', '于救']
Sixtuples = ['流開三上尤云', '流開三去尤云']
Transcriptions = ['hjuwX', 'hjuwH']

Characters not included in the database will not crash the program.

howtoread a

No reading available for this character: a

charinfo a

The character 'a' was not found in the database.

Text annotation

Annotate and convert to Webanno TSV 3.3

An example .txt file is included. To run Webanno TSV 3.3 single file conversion on it, use the following command:

webanno --file example

webanno -f example

The resulting file will be saved at ./webanno_output/example.tsv.

To annotate any .txt file into Webanno TSV 3.3:

webanno --file ./path_to/my_file.txt

webanno -f ./path_to/my_file.txt

The resulting file will be saved in a ./webanno_output/ directory, under the same name as the original file (./webanno_output/my_file.tsv).

Batch processing all .txt files from a directory:

webanno --directory ./path_to/my_directory

webanno -d ./path_to/my_directory

All resulting files will be saved in a ./webanno_output subdirectory, under the same name as the original directory (./webanno_output/my_directory/).

Annotate and convert to XML

An example .txt file is included. To run XML single file conversion on it, use the following command:

yunpyxml --file example

yunpyxml -f example

The resulting file will be saved at ./xml_output/example.xml.

To annotate any .txt file into XML:

yunpyxml --file ./path_to/my_file.txt

yunpyxml -f ./path_to/my_file.txt

The resulting file will be saved in a ./xml_output/ directory, under the same name as the original file (./xml_output/my_file.tsv).

Batch processing all .txt files from a directory:

yunpyxml --directory ./path_to/my_directory

yunpyxml -d ./path_to/my_directory

All resulting files will be saved in a ./xml_output subdirectory, under the same name as the original directory (./xml_output/my_directory/).

Programmatic Usage

YunPy was designed as a CLI tool first but can also be fully used in any Python script.

Lexicon exploration from yunpy.utils and yunpy.Character:

from yunpy.utils import *

# Get all known readings from a character
readings = get_readings_from_character("左")

# Get all known sixtuples from a character
sixtuples = get_sixtuples_from_character("左")

# Get all known fanqie from a character
fanqie = get_fanqie_from_character("左")

# Get a list of all the characters registered in the database
all_characters = db_coverage()

# Get an informational message about PHONO-ML database
about = about_database()

from yunpy.character import Character

# Check if a character is registered in the database
in_db = Character("左").is_in_db()

# Get a message showing all known readings for a character
howtoread = Character("左").set_character_readings_message()

# Get a message displaying detailed information for a character
charinfo = Character("左").set_message()

Text annotation from yunpy.core:

from yunpy.core import export_webanno_constituents, process_dir

"""Annotate a single file from TXT into Webanno TSV 3.3"""

# Default output path ("./webanno_output")
export_webanno_constituents("./my_text.txt")
# Custom output path
export_webanno_constituents("./my_text.txt", "./my_output_dir")

""" Batch process all TXT files from a directory """

# Default output path ("./webanno_output")
process_dir("./my_input_dir", "webanno")

from yunpy.core import export_xml, process_dir

""" Annotate a single file from TXT into YunPy XML """

# Default output path ("./xml_output")
export_xml("./my_text.txt")
# Custom output path
export_xml("./my_text.txt", "./my_output_dir")

""" Batch process all TXT files from a directory """

# Default output path ("./xml_output")
process_dir("./my_input_dir", "xml")

Output formats

WebAnno TSV 3.3

Input:

關關雎鳩，在河之洲；

Output:

#FORMAT=WebAnno TSV 3.3
#T_SP=webanno.custom.Transcription|value
#T_SP=webanno.custom.Transcriptions|values
#T_SP=webanno.custom.Fanqie|value
#T_SP=webanno.custom.Sixtuples|value


#Text=關關雎鳩，在河之洲；
1-1	0-1	關	kwaen	_	古還	山合二平刪見
1-2	1-2	關	kwaen	_	古還	山合二平刪見
1-3	2-3	雎	tshjo	_	七余	遇開三平魚清
1-4	3-4	鳩	kjuw	_	居求	流開三平尤見
1-5	4-5	\，	p	_	_	_
1-6	5-6	在	*	dzojX|dzojH	昨宰|昨代	蟹開一上咍從|蟹開一去咍從
1-7	6-7	河	ha	_	胡歌	果開一平歌匣
1-8	7-8	之	tsyi	_	止而	止開三平之章
1-9	8-9	洲	tsyuw	_	職流	流開三平尤章
1-10	9-10	\；	p	_	_	_

YunPy XML

Input:

關關雎鳩，在河之洲；

Output:

<?xml version='1.0' encoding='utf-8'?>
<ANNOTATED_TEXT>
 <line>
  <token>
   <character>關</character>
   <transcription>kwaen</transcription>
  </token>
  <token>
   <character>關</character>
   <transcription>kwaen</transcription>
  </token>
  <token>
   <character>雎</character>
   <transcription>tshjo</transcription>
  </token>
  <token>
   <character>鳩</character>
   <transcription>kjuw</transcription>
  </token>
  <token>
   <character>，</character>
  </token>
  <token>
   <character>在</character>
   <transcriptions>
    <transcription>dzojX</transcription>
    <transcription>dzojH</transcription>
   </transcriptions>
  </token>
  <token>
   <character>河</character>
   <transcription>ha</transcription>
  </token>
  <token>
   <character>之</character>
   <transcription>tsyi</transcription>
  </token>
  <token>
   <character>洲</character>
   <transcription>tsyuw</transcription>
  </token>
  <token>
   <character>；</character>
  </token>
 </line>
 </ANNOTATED_TEXT>

Data sources

PHONO-ML Database

PHONO-ML is a database mapping Chinese characters to Middle Chinese phonetic transcriptions, covering 26 224 character/transcription pairs (or 20 276 characters), developed by principal investigator Guillaume Jacques and software engineer Alexander Delaporte.

YunPy includes a copy of PHONO-ML's phonoML_characters.csv table, which provides transcriptions and data about characters.

You can display an informational message about PHONO-ML by using the database command.

Shijing-net corpus

Shijing-net is a digital corpus composed of the poetic Shijing 詩經 Anthology and multiple commentaries, developed within the CHI-KNOW-PO project led by principal investigator Marie Bizais-Lillig, with research engineer Ilaine Wang.

YunPy's example and test files were created from Shijing-net's TXT version.

Authors

• Alexander DELAPORTE (@alxdrdelaporte)
• Qingqing XIE (@Qingqing-ha)

See the AUTHORS file for details.

Licence

Code

YunPy's source code is licensed under the MIT License. See the LICENSE file for details.

Data sources

PHONO-ML Database is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License, and Shijing-net is licensed under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) License.

See the respective data sources for more information.