#! /usr/bin/python3
# Last edited on 2026-06-30 01:56:55 by stolfi
# -*- coding: utf-8 -*-

# Reads from stdin a transcriptionon of the SBJ in simplified hanzi,
# with one recipe per line. Removes form each recipe the fields that
# seem to have been systematically omitted from the Starred Parags
# section of the VMS.
# 
# Assumes that each field to be omitted has tags surrounded in
# ideographic brackets '［］' and is terminated by ideographic
# period '。'. Allows and preerves a locus ID in "<...>" at the
# beginning of each line.
#
# For now, these fields are the [Flavor], [Other name], and [Provenance].
#
# Also replaces 主治 by just 主.
#
# Also removes all #-comments and blank lines, but adds a line "# -*-
# coding: utf-8 -*-" and a warning to not edit this file.

import sys, re
from sys import stdout as out, stderr as err, stdin as inp
from error_funcs import file_line_error

def main():
  inp.reconfigure(encoding='utf-8')
  out.reconfigure(encoding='utf-8')

  nread = 0

  out.write("# -*- coding: utf-8 -*-\n")
  out.write("# Generated by {remove_vms_omitted_fields_from_sbj.py}\n")
  out.write("# Edit the source, not this file.\n")
  for line in inp:
    nread += 1
    process_line(nread, line)
  
  out.flush()
  return 
  # ----------------------------------------------------------------------

def process_line(nread, line):

  def data_error(msg):
    nonlocal nread;
    file_line_error("stdin", nread, msg, line)
    # ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
    
  def excise(name, pat, max_length, tx):
    # Removes from {tx} any substring that matches {pat}. 
    # Fails if the substring is too long.
    # The removed stibstrings are written out as #-comments.
    # Returns the cleaned text.
    
    while ( m:= re.search(pat, tx) ) is not None:
      field = m.group(0)
      if len(field) > max_length: 
        data_error(f"[{name}] field {field!r} is too long")
      if field[-1] != '。':
        data_error(f"[{name}] field {field!r} is missing the period")
      i0 = m.start(0); i1 = m.end(0)
      tx = tx[0:i0] + tx[i1:]
      out.write(f"# {name} = {field!r}\n")
      
    return tx
    # ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

  # Remove end-of-line, leading and trailing blanks:
  line = line.strip()

  # Discard comments and blank lines:
  if re.match(r"[ ]*([#]|$)", line): return

  # Separate recipe ID, if any:
  if ( m := re.fullmatch(r" *<([^<>]*)> *(.*)", line) ) is not None:
    loc = m.group(1); text = m.group(2)
  else:
    loc = None; text = line

  # # Remove Chinese punctuation:
  # text = re.sub(r"[：［］（）、，。；　]", "", text)
  
  # Add a final Chinese period to simplify matching:
  text += "。"

  # Normalize the "mainly for" keyword:
  text = re.sub(r"主治", "主", text)

  # Locate and remove the "Flavor" fields:
  text = excise("Flavor", r"[［]味[^主。]*[。]?", 10, text)

  # Locate and remove the "Provenance" fields:
  text = excise("Provenance", r"[［]生[^主。]*[。]?", 12, text)

  # Locate and remove the "Other name" fields:
  # Note: there is one alias that has 主 that is not a keyword.
  text = excise("Alias", r"[［]一名[^生。]*[。]?", 10, text)

  if loc is not None: out.write(f"<{loc}> ")
  out.write(text)
  out.write("\n")
  return
  # ----------------------------------------------------------------------

main()