tei2html.py

#!/usr/bin/env python3
# -*- coding: utf-8; mode: python -*-

"A converter from TEI to HTML."

__version__ = "1.0"
__date__ = "20190408"
__author__ = "sgfroerer@mpiwg-berlin.mpg.de"

from utils.load_config import load_config, check_executable, exec_command, ToFile
from utils.run_xslt import run_xslt
import utils.libeoaconvert as libeoaconvert

import logging
import argparse
from pathlib import Path
from os import environ, mkdir
from shutil import rmtree, copytree, ignore_patterns
from lxml import etree

# things to be done
# assign ids top to bottom for the following elements:
# div1 div2 div3 note item table EOAfigure EOAequation formula theorem

BASE_DIR = Path( __file__ ).resolve().parent
SCRIPT_NAME = Path( __file__).stem

DEFAULT_INPUT_DIR = \
    Path(environ['INPUT_DIR'] if 'INPUT_DIR' in environ else './input')

DEFAULT_OUTPUT_DIR = \
    Path(environ['OUTPUT_DIR'] if 'OUTPUT_DIR' in environ else './output')

DEFAULT_DEPENDENCIES_DIR = \
    Path(environ['DEPENDENCIES_DIR'] if 'DEPENDENCIES_DIR' in environ else './dependencies')

EOA_SCRIPTS_DIR = \
    Path(environ['EOA_SCRIPTS_DIR'])

def check_executables():
    check_executable( "saxon" )

def copy_dir(
        src,
        dst,
        **opts
):
    print( "'{}' -> '{}'".format( src, dst ) )
    if Path(dst).exists():
        rmtree(
            dst
        )
    copytree(
            src=src,
            dst=dst,
            **opts
    )

if __name__ == '__main__':
    parser = argparse.ArgumentParser(
            formatter_class=argparse.ArgumentDefaultsHelpFormatter
    )
    parser.add_argument(
            "-c", "--config",
            default = BASE_DIR / "config" / "eoaconvert.cfg",
            help="Name of config file"
    )
    parser.add_argument(
            "-l", "--log-dir",
            default = DEFAULT_OUTPUT_DIR / "logs",
            type = Path,
            help="logfile"
    )
    parser.add_argument(
            "--log-level",
            default = "INFO",
            help="log level: choose between DEBUG, INFO, WARNING, ERROR, CRITICAL"
    )
    parser.add_argument(
            "--root-dir",
            default = DEFAULT_OUTPUT_DIR / "html_from_tei",
            type = Path,
            help=""
    )
    parser.add_argument(
            "-p", "--param",
            action = 'append',
            default = [],
            help="xslt params"
    )
    parser.add_argument(
            "-x", "--xsl",
            default = "tei2html.xsl",
            help="name of the xsl file in '{dir}/'".format(
                dir = BASE_DIR / "tei2html"
            )
    )
    parser.add_argument(
            "-f", "--filename",
            default = Path("*.xml"),
            type = Path,
            help = "xml file inside PUBLICATION_DIR, or absolute path. Patterns like '*.xml' are also acceptable"
    )
    parser.add_argument(
            "-o", "--output-dir",
            type = Path,
            default = DEFAULT_OUTPUT_DIR / "html_from_tei",
            help="the main stylesheet 'tei2html.xsl' will write html files here"
    )
    parser.add_argument(
            "-!", "--overwrite",
            action = "store_true",
            default = False,
            help="overwrite OUTPUT_DIR, if existing"
    )
    parser.add_argument(
            "--output-file",
            type = Path,
            help="in case of calling a classical stylesheet with output, redirect here"
    )
    parser.add_argument(
            "-w", "--webdesign",
            default = '../webdesign_platform',
            type = Path,
            help="location to the webdesign (relative to OUTPUT_DIR, or absolute)"
    )
    parser.add_argument(
            "-s", "--static",
            default = '../publication_static',
            type = Path,
            help="static files of publication (relative to OUTPUT_DIR, or absolute)"
    )
    parser.add_argument(
            "PUBLICATION_DIR",
            default = Path("input/example/tei"),
            help = "directory containing the publication (including resources like pictures, etc.)",
            nargs = '?', # (optional)
            type = Path,
    )

    args = parser.parse_args()

    CONFIG_FILE = args.config

    log_dir = args.log_dir

    CONFIG = load_config(
            CONFIG_FILE,
            args.log_level,
            (log_dir / SCRIPT_NAME) . with_suffix( ".log" ),
    )

    check_executables()

    print( "xslt params: " + str(args.param) )

    publ_dir = args.PUBLICATION_DIR
    publ_file = args.filename

    tei_filename = \
        publ_file if publ_file . is_absolute() else list(publ_dir . glob (str(publ_file)))[0]
    
    logging.info( f"tei_file: {tei_filename}, publ_dir: {publ_dir}" )

    output_dir = args.output_dir

    if not tei_filename.is_file():
        raise( Exception(
            f"not a valid input file: {tei_filename}"
        ) )

    if output_dir.exists():
        if args.overwrite:
            rmtree( output_dir )
        else:
            raise( Exception( f"output directory already existing: '{output_dir}'!" ) )
    if not output_dir.exists():
        mkdir( output_dir )
        ## copy webdesign:
        copy_dir(
                DEFAULT_DEPENDENCIES_DIR / "webdesign_platform/dist",
                output_dir / "webdesign_platform",
        )
        ## copy publication static files:
        static_dest_dir =  output_dir / "publication_static"
        static_dest_dir.mkdir(
                parents = True,
                exist_ok = True
        )
        copy_dir(
                publ_dir,
                static_dest_dir,
                ignore = ignore_patterns( "*.xml" )
        )

    run_xslt(
            input_file = tei_filename,
            xslt_file = BASE_DIR / "tei2html" / args.xsl,
            params =
                [
                    f"output_dir={output_dir}",
                    f"webdesign_url={args.webdesign}",
                    f"publ_static_url={args.static}",
                ] +
                args.param,
            output_file = args.output_file,
            exec_command_args = {
                'output_to' : ToFile( log_dir / SCRIPT_NAME / "saxon.log" )
            },
    )

# finis