added logging; introduced main program options; added new target "build-all"

build.py

@@ -1,105 +1,414 @@
-#!/usr/bin/python
+#!/usr/bin/env python
 
-import os
 from pathlib import Path
 import subprocess
-import sys
+from enum import Enum
+from shutil import which
+from typing import Annotated
+import logging
+
+import typer
+import yaml
+from pypdf import PdfReader, PdfWriter
+
+# start logger
+logger = logging.getLogger(__name__)
 
 # define directories
-base_dir = os.getcwd()
-content_dir = f"{base_dir}/content"
-config_dir = f"{base_dir}/config"
-output_dir = f"{base_dir}/output"
+base_dir = Path.cwd()
+config_dir = base_dir / "config"
+content_dir = base_dir / "content"
+cover_dir = base_dir / "cover"
+output_dir = base_dir / "output"
+minizine_dir = base_dir / "promo" / "minizine"
+
+# define executables and default options
+pandoc = which("pandoc")
+if not pandoc:
+  logger.critical("pandoc is not installed.")
+  exit(1)
+pandoc_opts = []
+typst = which("typst")
+if not typst:
+  logger.critical("typst is not installed.")
+  exit(1)
+typst_opts = ["compile", "--root", "."]
+
+# helper class for available output formats
+class Format(str, Enum):
+  PDF = "pdf"
+  PNG = "png"
+
+# instantiate typer app
+app = typer.Typer(
+  no_args_is_help=True, 
+  help="Build script to make a professionally typeset book from markdown source files that can either be read on screen (PDF or EPUB) or be used in a project with Tredition.",
+  context_settings={"help_option_names": ["-h", "--help"]}
+)
+
+
+# ##################
+# INTERNAL FUNCTIONS
+# ##################
+
+def _get_input_files(target_format: str) -> list:
+  # read filenames from file if such a file exists
+  list_file = config_dir / f"source-files.{target_format}.txt"
+
+  # read file if it exists
+  if list_file.is_file():
+    with open(list_file, "r") as file:
+      return [line.strip() for line in file if line[0] != "#"]
+
+  # use all files otherwise; ignore tredition flag
+  else:
+    paths = content_dir.glob('**/*.md')
+    return list(map(str, sorted(paths)))
+
+
+def _run_command(executable: str, infiles: list, ex_opts: list = [], outfile: str = ""):
+  # put command together
+  outfile = [outfile] if outfile else []
+  command = [executable] + ex_opts + infiles + outfile
+  logger.debug(f"Executing command {command}")
 
-def run_command(command):
   # run shell command
-  stdout = subprocess.check_output(command)
-  print(stdout)
-
-def clean(args):
-  pass
+  try:
+    stdout = subprocess.check_output(command, stderr=subprocess.STDOUT)
+  except subprocess.CalledProcessError as e:
+    logger.critical(f"Command failed with return code {e.returncode}.")
+    logger.critical(f"Command output: {e.output}")
+    raise typer.Exit(code=e.returncode)
+  else:  
+    logger.debug(f"Command finished with return code 0.")
+    logger.debug(f"Command output: {stdout}")
 
 
-def epub(args):
+def _clean_dir(path: Path):
+  for file in path.iterdir():
+    # if file: delete
+    if file.is_file():
+      file.unlink()
+    # if dir: delete recursively
+    else:
+      # clean dir recursively
+      _clean_dir(file)
+      # delete dir unless it's the output_dir
+      if file != output_dir:
+        file.rmdir()
+
+
+def _add_presuffix(path: Path|str, presuffix: str) -> Path:
+  # cast past to Path
+  if isinstance(path, str):
+    path = Path(path)
+
+  # add leading dot if nexessary
+  if presuffix[0] != ".":
+    presuffix = "." + presuffix
+
+  return path.parent / f"{path.stem}{presuffix}{path.suffix}"
+
+
+def _merge_pdf_with_covers(frontcover: str, content: str, backcover: str, keep_coverless: bool=False):
+  pdflist = []
+
+  # check if files exist     
+  for filename in [frontcover, content, backcover]:
+    if Path(filename).is_file():
+      pdflist.append(filename)
+  
+  # merge files wwith pypdf
+  merger = PdfWriter()
+  for pdf in pdflist:
+    merger.append(pdf)
+
+  # write merged file
+  outfile = str(_add_presuffix(content, ".with-covers")) if keep_coverless else content
+  merger.write(outfile)
+  merger.close()
+
+
+def _get_data_from_yaml(inputfile: str, key: str) -> str|list|dict:
+  with open(inputfile, "r") as file:
+    data = yaml.load(file, Loader=yaml.Loader)
+  return data.get(key)
+
+
+
+# #############
+# CLI COMMANDS
+# #############
+
+@app.command()
+def pdf(
+  tredition: Annotated[
+    bool,
+    typer.Option("--tredition", "-t", help="Build PDFs in accordance with Tredition's requirements for print projects. Will produce a content file without covers and the book's envelope including the spine.")
+  ] = False,
+  keep_coverless: Annotated[
+    bool,
+    typer.Option("--keep-coverless", "-k", help="Do not delete the content-only PDF file after merging it with the covers (does not apply if 'tredition' is set).")
+  ] = False
+):
+  """
+  Build the printed book's PDF version (usually a single file containing front cover, content, and back cover).
+  """
+  logger.info("Building PDF.")
+  # define executable
+  executable, executable_opts = pandoc, pandoc_opts
+  
+  # add pandoc defaults file to options 
+  defaults_file = str(config_dir / f"pandoc-defaults.{Format.PDF}.yaml")
+  if Path(defaults_file).is_file():
+    executable_opts.append("--defaults")
+    executable_opts.append(defaults_file)
+  else:
+    logger.critical(f"Pandoc defaults file {defaults_file} not found.")
+    raise typer.Exit(code=1)
+
+  # get input files 
+  input_files = _get_input_files(Format.PDF.value)
+
+  # if tredition flag is set: remove empty first and/or last page 
+  if not tredition:
+    for i in (0, -1):
+      if input_files[i].endswith("leere seite.md"):
+        del input_files[i]
+
+  # get output filename (to be used later)
+  outfile = _get_data_from_yaml(defaults_file, "output-file")
+  if not outfile:
+    logger.error("Could not get output filename from defaults file.")
+    typer.Exit(2)
+
+  # add presuffix ".tredition" to output filename if flag is set
+  if tredition:
+    outfile = str(_add_presuffix(outfile, ".tredition"))
+    executable_opts.append("--output")
+    executable_opts.append(outfile)
+
+  # make PDf
+  _run_command(executable, ex_opts=executable_opts, infiles=input_files)
+  
+  # make PDF covers
+  cover(tredition=tredition, format=Format.PDF, frontcover=not tredition, backcover=not tredition, envelope=tredition)
+  
+  # add front-/backcover to PDF unless tredition flag is set 
+  if not tredition:
+    _merge_pdf_with_covers(str(output_dir / "frontcover.pdf"), outfile, str(output_dir / "backcover.pdf"), keep_coverless)
+
+  # check metadata -> try veraPDF
+
+
+@app.command()
+def epub(
+  tredition: Annotated[
+    bool,
+    typer.Option("--tredition", "-t", help="Build files in accordance with Tredition's requirements for ebook projects. Will produce a main EPUB file without covers and a separate PDF with the cover page.")
+  ] = False,
+  validate: Annotated[
+    bool,
+    typer.Option("--validate", "-V", help="Validate completed EPUB file (requires 'epubcheck' executable in path).")
+  ] = False
+):
+  """
+  Build the book's EPUB version (usually a single EPUB file with front cover).
+  """
+  logger.info("Building EPUB.")
   # define pandoc defaults file
-  pandoc_defaults_file = f"{config_dir}/pandoc-defaults.epub.yaml"
+  defaults_file = str(config_dir / "pandoc-defaults.epub.yaml")
 
   # define executable and options
-  executable = "/usr/bin/pandoc"
-  executable_opts = ["-d", pandoc_defaults_file]
-  executable_opts.append("--verbose")
+  executable, executable_opts = pandoc, pandoc_opts
+  executable_opts.append("--defaults")
+  executable_opts.append(defaults_file)
 
-  # take specific content file as input
-  input_paths = [Path(content_dir) / "01 - frontmatter" / "30 - widmung.md", Path(content_dir) / "01 - frontmatter" / "50 - vorwort.md"]
-  input_paths += sorted(Path(content_dir).glob('02 - mainmatter/*.md'))
-  input_paths += sorted(Path(content_dir).glob('03 - backmatter/*.md'))
-  input_files = [str(p) for p in input_paths]
+  # make frontcover
+  format = Format.PDF if tredition else Format.PNG
+  cover(format, tredition, backcover=False, envelope=False)
+  
+  # add cover to pandoc options (no-tredition only)
+  if not tredition: 
+    executable_opts.append("--epub-cover-image")
+    executable_opts.append(str(output_dir / "frontcover.png"))
+  
+  # get input files
+  input_files = _get_input_files("epub")
+
+  # get output filename
+  outfile = _get_data_from_yaml(defaults_file, "output-file")
+
+  #  modify output filename (tredition only)
+  if tredition:
+    outfile = str(_add_presuffix(outfile, ".tredition"))
+    executable_opts.append("--output")
+    executable_opts.append(outfile)
 
   # put together command and run it
-  command = [executable] + executable_opts + args + input_files
-  print(f"command is {command}")
-  run_command(command)
+  _run_command(executable, ex_opts=executable_opts, infiles=input_files)
+
+  # validate epub
+  if validate and which("epubcheck"):
+    _run_command(which("epubcheck"), infiles=[outfile])
   
 
-def cover(args):
-  # define executable and options
-  executable = "/usr/bin/pdflatex"
-  executable_opts = [f"-output-directory={output_dir}"]
-  
-  # get input files from content dir
-  cover_dir = f"{base_dir}/cover"
-  input_files = [f"{cover_dir}/cover.tex"]
-
-  # put together command and run it
-  command = [executable] + executable_opts + args + input_files
-  print(f"command is {command}")
-  run_command(command)
-
-
-
-def pdf(args):
-  # define pandoc defaults file
-  pandoc_defaults_file = f"{config_dir}/pandoc-defaults.pdf.yaml"
+@app.command()
+def cover(
+  format: Annotated[
+    Format,
+    typer.Option("--format", "-f", help="Set output format.")
+  ] = Format.PDF,
+  tredition: Annotated[
+    bool,
+    typer.Option("--tredition", "-t", help="Build covers in accordance with Tredition's requirements. Produces a PDF with the ebook cover and a PDF with the printed book's envelope.")
+  ] = False,
+  frontcover: Annotated[
+    bool,
+    typer.Option("--frontcover", "-c", help="Include the front cover when building.")
+  ] = True,
+  backcover: Annotated[
+    bool,
+    typer.Option("--backcover", "-b", help="Include the back cover when building.")
+  ] = False,
+  envelope: Annotated[
+    bool,
+    typer.Option("--envelope", "-e", help="Include the printed book's envelope when building.")
+  ] = False,
+  all_covers: Annotated[
+    bool,
+    typer.Option("--all", "-a", help="Build all covers in all (sensible) formats.")
+  ] = False
+):
+  """
+  Build only the cover(s).
+  """
+  logger.info("Building cover(s).") 
+  if all_covers:
+    envelope, frontcover, backcover = True, True, True
+    tredition = True
+    format = Format.PDF
 
   # define executable and options
-  executable = "/usr/bin/pandoc"
-  executable_opts = ["-d", pandoc_defaults_file]
-  #executable_opts.append("--verbose")
+  executable, executable_opts = typst, typst_opts
 
-  # get input files from content dir
-  input_paths = sorted(Path(content_dir).glob('**/*.md'))
-#  input_paths = sorted(Path(content_dir).glob('03 - backmatter/*.md')) #DEBUG
-  input_files = [str(p) for p in input_paths]
+  # check for nonsensical flag combination
+  if tredition and format == Format.PNG:
+    logger.warning("PNG can't be used as output format in combination with --tredition; setting format to PDF.")
+    format = Format.PDF
 
-  # put together command and run it
-  command = [executable] + executable_opts + args + input_files
-  print(f"command is {command}")
-  run_command(command)
+  # select files
+  if tredition:
+    paths = cover_dir.glob("*tredition*.typ")
+  else:
+    paths = [path for path in cover_dir.glob("*.typ") if not "tredition" in path.name]
+
+  # remove input files with unset flags   
+  if not envelope:
+    paths = [path for path in paths if not "envelope" in path.name]
+  if not frontcover: 
+    paths = [path for path in paths if not "frontcover" in path.name]
+  if not backcover: 
+    paths = [path for path in paths if not "backcover" in path.name]
+
+  # iterate over files
+  for path in paths:
+    # get input file name
+    input_files = [str(path)]
+    # set output file name
+    output_file = str(output_dir / f"{path.stem}.{format.value}")
+
+    # run command
+    _run_command(executable, infiles=input_files, outfile=output_file, ex_opts=executable_opts)
+
+  # call function again if flag "all" is set (but don't recurse)
+  if all_covers:
+    cover(Format.PDF, tredition=False, frontcover=True, backcover=True, envelope=False, all_covers=False)
+    cover(Format.PNG, tredition=False, frontcover=True, backcover=True, envelope=False, all_covers=False)
+
+
+@app.command()
+def minizine(
+  format: Annotated[
+    Format,
+    typer.Option("--format", "-f", help="Set output format.")
+  ] = Format.PDF
+):
+  """
+  Build the minizine for the book.
+  """
+  logger.debug("Building Minizine.")
+
+  # define executable and options
+  executable, executable_opts = typst, typst_opts
+
+  # iterate over input files
+  for path in minizine_dir.glob("*.typ"):
+    # get input file name
+    input_files = [str(path)]
+    # set output file name
+    output_file = str(output_dir / f"{path.stem}.{format.value}")
+    # run command
+    _run_command(executable, infiles=input_files, outfile=output_file, ex_opts=executable_opts)
+
+
+@app.command("build-all")
+def build_all():
+  """
+  Build every available target, with and without Tredition specs (where applicable).
+  """
+  logger.debug("Building everything.")
+  clean()
+  minizine(format=Format.PDF)
+  minizine(format=Format.PNG)
+  cover(all_covers=True)
+  epub()
+  epub(tredition=True)
+  pdf()
+  pdf(tredition=True)
+
+
+@app.command()
+def clean():
+  """
+  Remove all files from output directory.
+  """
+  logger.info("Cleaning output directory.")
+  _clean_dir(output_dir)
+
+
+@app.callback()
+def main(
+  log_level: Annotated[
+    str,
+    typer.Option("--log-level", "-l", help="Set log level.")
+  ] = "INFO",
+  outdir: Annotated[
+    str,
+    typer.Option("--output-dir", "-o", help="Specify output directory.")
+  ] = "output/"
+):
+  # set user log level
+  if not getattr(logging, log_level.upper()):
+    log_level = "INFO"
+
+  # configure logger
+  logging.basicConfig(
+    level=getattr(logging, log_level.upper()),
+    format="{asctime} {levelname}: {message}",
+    style="{",
+    datefmt="%Y-%m-%d %H:%M:%S"
+  )
+  logger.debug("Logger configured.")
+
+  # set output directory
+  p = Path(outdir).expanduser()
+  globals()["output_dir"] = p if p.is_absolute() else base_dir / p
+  od = globals()["output_dir"]
+  logger.debug(f"Output directory is {od}")
+  od.mkdir(parents=True, exist_ok=True)  
 
 
 if __name__ == "__main__":
-  # get command line parameters
-  args = sys.argv[1:]
-  if len(args) == 0:
-    print("Please specify at least one parameter")
-    exit(1)
-  
-  # call function corresponding to first argument
-  if args[0] == "clean":
-    clean(args[1:])
-  elif args[0] == "pdf":
-    pdf(args[1:])
-  elif args[0] == "cover":
-    cover(args[1:])
-  elif args[0] == "epub":
-    epub(args[1:])
-  elif args[0] == "all":
-    pdf(args[1:])
-    cover(args[1:])
-    epub(args[1:])
-  else:
-    print("Parameter not recognized")
-    exit(2)
-
-  
+  app()